Вы находитесь на странице: 1из 631

HVDC Systems and Their Planning

Publisher:
PTD H 1T
Rev. 4

For In-House
Use Only
© Siemens AG
HVDC Systems and Their Planning

Authors:

Karl-Werner Kanngiesser
Dr. Hartmut Huang
Hans Peter Lips
Norbert Christl
Georg Wild
Dr. Dietmar Retzmann
Dr. Peter Riedel
Dr. Kadry Sadek
Fr. Christa Storner
Fr. Martina Gumbert

Publisher:
PTD H 1T
Rev. 4
For In-House
Use Only
Ó Siemens AG
Index

1. Theoretical Foundations ......................................................................................................1

1.1 Transmission of Electrical Energy......................................................................................1


1.1.1 Historical Considerations........................................................................................................1
1.1.2 Advantages and Limitations of Alternating Current Transmission .......................................3
1.1.3 Characteristics of HVDC ........................................................................................................6
1.1.4 Possibilities for Utilization of HVDC ....................................................................................8
1.1.5 Economic Aspects of Utilization of HVDC........................................................................ 44
1.1.6 HVDC Systems in operation or under construction............................................................ 58

1.2 The Main Components of the HVDC Station................................................................. 61


1.2.1 Layout of the HVDC Station ............................................................................................... 61
1.2.2 Weighting of the Major Components .................................................................................. 62
1.2.3 Degree of Freedom in Project Planning .............................................................................. 63

1.3 The Function of the HVDC Converter............................................................................ 65


1.3.1 The Three-Pulse Commutation Group ................................................................................ 66
1.3.2 Voltage Formation ............................................................................................................... 67
1.3.3 The Commutation Process................................................................................................... 70
1.3.4 The Converter in Operation ................................................................................................. 75
1.3.5 The Twelve-Pulse Converter Group.................................................................................... 79

1.4 Reactive Power Demand of the Converter...................................................................... 83


1.4.1 The Uncontrolled, Ideal Converter ...................................................................................... 83
1.4.2 Control Reactive Power ....................................................................................................... 84
1.4.3 Commutation Reactive Power............................................................................................. 86
1.4.4 Fundamental Frequency Reactive Power ............................................................................ 87
1.4.5 Reactive Power as a Function of Load ................................................................................ 88
1.4.6 Reactive Power Control ....................................................................................................... 90

1.5 AC-Side Harmonics ........................................................................................................... 92


1.5.1 Line Current of the Wye-Connected Bridge........................................................................ 92
1.5.2 Line Current of the Delta-Connected Bridge ...................................................................... 94
1.5.3 Line Current of the 12-pulse Group..................................................................................... 95
1.5.4 The Influence of Control Angle and Overlap...................................................................... 96
1.5.5 Non-Characteristic Harmonics ............................................................................................ 99

i
1.6 DC-side Harmonics.......................................................................................................... 101
1.6.1 Direct Voltage of the 6-pulse Converter Bridge................................................................ 101
1.6.2 Direct Voltage of the 12-pulse Converter Group .............................................................. 102
1.6.3 Noncharacteristic Harmonics............................................................................................. 105
1.6.4 Effects of DC-Side Harmonics .......................................................................................... 105

2. Components of the HVDC Station................................................................................. 107

2.1 AC Switchgear.................................................................................................................. 107


2.1.1 Functions of the AC Switchgear........................................................................................ 107
2.1.2 Current and Voltage Rating ............................................................................................... 108
2.1.3 Design Considerations ....................................................................................................... 109

2.2 AC Filter Circuits and Capacitor Banks....................................................................... 110


2.2.1 Functions of the AC Filter Circuits and Capacitor Banks................................................. 110
2.2.2 Sizing of AC Filter Circuits and Capacitor Banks with Respect to Reactive Power ....... 110
2.2.3 Sizing of AC Filter Circuits with Respect to Harmonics.................................................. 120
2.2.4 Filter Circuit Arrangements ............................................................................................... 131
2.2.5 Steady-State Load of the Filter Circuit Components ........................................................ 139
2.2.6 Transient Load of the Filter Components.......................................................................... 143
2.2.7 Design Considerations ....................................................................................................... 148

2.3 Converter Transformers ................................................................................................. 153


2.3.1 Functions of the HVDC Converter Transformer............................................................... 153
2.3.2 Current and Voltage Rating ............................................................................................... 155
2.3.3 Selection of the Short-Circuit Voltage .............................................................................. 156
2.3.4 Insulation Coordination ..................................................................................................... 159
2.3.5 Transformer Design ........................................................................................................... 160
2.3.6 Experience in Existing Systems......................................................................................... 163

2.4 Thyristor Valves............................................................................................................... 164


2.4.1 Historical Considerations................................................................................................... 164
2.4.2 Thyristors as Switches ....................................................................................................... 168
2.4.3 The Multiple Series Connection - Special Requirements and Solutions.......................... 174
2.4.4 Valve Control and Monitoring........................................................................................... 179
2.4.5 Valve Cooling .................................................................................................................... 182
2.4.6 Valve Mechanical Design.................................................................................................. 190

ii
2.4.7 System/Valve Interfaces..................................................................................................... 194

2.5 Smoothing Reactor........................................................................................................... 197


2.5.1 Functions of the Smoothing Reactor ................................................................................. 197
2.5.2 Sizing of the Smoothing Reactor....................................................................................... 198
2.5.3 Arrangement of the Smoothing Reactors .......................................................................... 200
2.5.4 Reactor Design Alternatives .............................................................................................. 202

2.6 DC Filter Circuits............................................................................................................. 203


2.6.1 Function of the DC-Side Filter Circuits ............................................................................ 203
2.6.2 Design Criteria for DC Filter Circuits ............................................................................... 203
2.6.3 Calculation of Equivalent Disturbing Current................................................................... 206
2.6.4 Induced Low Order Harmonics ......................................................................................... 206
2.6.5 Filter Circuit Configurations.............................................................................................. 210
2.6.6 Steady-state Stresses of DC Filter Circuits........................................................................ 211
2.6.7 Transient Stresses of DC Filter Circuits ............................................................................ 212
2.6.8 Design Considerations ....................................................................................................... 213
2.6.9 Development Trends.......................................................................................................... 213

2.7 DC Switchgear.................................................................................................................. 215


2.7.1 Functions of the DC Switchgear........................................................................................ 215
2.7.2 Rating of DC Switching Devices....................................................................................... 225
2.7.3 Design Considerations ....................................................................................................... 235

3. HVDC Overhead Lines and Cables............................................................................... 238

3.1 The HVDC Overhead Line............................................................................................. 238


3.1.1 Introductory Comments ..................................................................................................... 238
3.1.2 Types of HVDC Overhead Lines....................................................................................... 239
3.1.3 Minimum Clearances......................................................................................................... 247
3.1.4 Conductor Surface Gradient and Corona .......................................................................... 249
3.1.5 Electric and Magnetic Fields ............................................................................................. 263
3.1.6 The Insulation of the HVDC Overhead Line .................................................................... 269

3.2 The HVDC Cable............................................................................................................. 281


3.2.1 Introductory Remarks......................................................................................................... 281
3.2.2 Cable Types and their Application .................................................................................... 281
3.2.3 Field Strength Distribution in the Dielectric ..................................................................... 287

iii
3.2.4 Special Problems with the Operation of an HVDC Cable................................................ 291
3.2.5 Overvoltage Stresses .......................................................................................................... 295
3.2.6 Protection against Overvoltages ........................................................................................ 299

4. HVDC Earth Electrodes.................................................................................................. 302

4.1 The Function of the Earth Electrodes in HVDC Systems........................................... 302


4.1.1 Ground Reference Point for the System Voltage .............................................................. 302
4.1.2 Continuous Current Load................................................................................................... 304
4.1.3 Temporary Current Load ................................................................................................... 304
4.1.4 Temporary Overcurrents .................................................................................................... 305
4.1.5 Transient Overcurrents....................................................................................................... 306

4.2 Rating Criteria for Earth Electrodes............................................................................. 307


4.2.1 Current Carrying Capacity................................................................................................. 307
4.2.2 Current density................................................................................................................... 308
4.2.3 Earth electrode resistance................................................................................................... 308
4.2.4 Electrode Voltage............................................................................................................... 309
4.2.5 Step Voltage ....................................................................................................................... 309
4.2.6 Touch Voltage.................................................................................................................... 310
4.2.7 Risk of Corrosion............................................................................................................... 310
4.2.8 Effects on AC Systems ...................................................................................................... 312
4.2.9 Effects on Telecommunication Systems ........................................................................... 312
4.2.10 Compass Declination ......................................................................................................... 313
4.2.11 Useful Life.......................................................................................................................... 314

4.3 Site Selection for Earth Electrodes ................................................................................ 315


4.3.1 Ground Characteristics....................................................................................................... 315
4.3.2 Distance from Civil Works................................................................................................ 317
4.3.3 Proximity to the Right of Way of the HVDC Overhead Line........................................... 318
4.3.4 Accessibility....................................................................................................................... 319
4.3.5 Environmental Considerations .......................................................................................... 319

4.4 Design of Earth Electrodes ............................................................................................. 320


4.4.1 Horizontal Land Electrode................................................................................................. 320
4.4.2 Vertical Land Electrode ..................................................................................................... 322
4.4.3 Cathodic Submarine Electrodes......................................................................................... 323
4.4.4 Anodic Submarine Electrodes ........................................................................................... 324

iv
4.4.5 Anodic Coastal Electrode .................................................................................................. 325

4.5 Anode Material................................................................................................................. 326


4.5.1 Iron ..................................................................................................................................... 326
4.5.2 Silicon-Iron......................................................................................................................... 326
4.5.3 Platinum-Plated Titanium.................................................................................................. 327
4.5.4 Graphite.............................................................................................................................. 327
4.5.5 Coke ................................................................................................................................... 327

4.6 Project Stages ................................................................................................................... 328


4.6.1 Site Selection...................................................................................................................... 328
4.6.2 Definition of the Design..................................................................................................... 328
4.6.3 Construction and Operation of a Prototype Electrode....................................................... 329
4.6.4 Construction and Startup of the Earth Electrodes ............................................................. 330

4.7 Electrode Line .................................................................................................................. 331


4.7.1 Electrode Line as Separate Overhead Line........................................................................ 331
4.7.2 Electrode Line as Shield Wire ........................................................................................... 332
4.7.3 Insulation of the Electrode Line......................................................................................... 333
4.7.4 Electrode Cable .................................................................................................................. 334

5. Control Technique in HVDC Systems........................................................................... 335

5.1 Converter Regulation Functions .................................................................................... 335


5.1.1 Current Regulation............................................................................................................. 335
5.1.2 Extinction Angle Regulation ............................................................................................. 336
5.1.3 DC Voltage Regulation...................................................................................................... 337
5.1.4 Control Characteristics of an HVDC System.................................................................... 339
5.1.5 Regulation Functions of the Transformer On-Load Tapchanger...................................... 342

5.2 Master Control Functions............................................................................................... 345


5.2.1 Power Regulation............................................................................................................... 345
5.2.2 Frequency Regulation ........................................................................................................ 346
5.2.3 Reactive Power Regulation................................................................................................ 347
5.2.4 AC Voltage Regulation...................................................................................................... 349

5.3 Dynamic Regulation Functions ...................................................................................... 349


5.3.1 Damping of Electromechanical Oscillations..................................................................... 349

v
5.3.2 Damping of Subsynchronous Oscillations ........................................................................ 350
5.3.3 Suppression of Low-Order Harmonics.............................................................................. 351

5.4 Regulation Hierarchy ...................................................................................................... 352


5.4.1 Converter Group Regulation.............................................................................................. 353
5.4.2 Pole Regulation .................................................................................................................. 356
5.4.3 Substation Regulation ........................................................................................................ 358

5.5 Control Functions ............................................................................................................ 360


5.5.1 Start-up and Shut-down of the System .............................................................................. 360
5.5.2 Connection and Disconnection of a Converter Group ...................................................... 361
5.5.3 Change of Energy Direction .............................................................................................. 363
5.5.4 Modification of System Configuration.............................................................................. 365
5.5.5 Tap Changer Control.......................................................................................................... 369
5.5.6 Switching of Reactive Power Units................................................................................... 370

5.6 HVDC Protection Functions........................................................................................... 372


5.6.1 Principles of Protection...................................................................................................... 372
5.6.2 Protection Functions of Current Regulation...................................................................... 374
5.6.3 Protection Functions of g and Ud Regulation .................................................................... 375
5.6.4 Travelling-Wave Protection of Transmission Lines ......................................................... 376
5.6.5 Differential Protection........................................................................................................ 377
5.6.6 Overcurrent Protection....................................................................................................... 382
5.6.7 Other Special HVDC Protection Functions....................................................................... 383
5.6.8 Protection of Equipment .................................................................................................... 389

5.7 Measurement .................................................................................................................... 390


5.7.1 Requirements on the Measuring System ........................................................................... 390
5.7.2 Quantities to be Measured ................................................................................................. 390
5.7.3 Fault Recorder.................................................................................................................... 391

5.8 Telecommunications ........................................................................................................ 392


5.8.1 Functions of the Telecommunication System ................................................................... 392
5.8.2 Operation without Telecommunications ........................................................................... 395
5.8.3 Telecommunication Systems ............................................................................................. 396

5.9 General Requirements for HVDC Control Systems.................................................... 405


5.9.1 Functionality....................................................................................................................... 405

vi
5.9.2 Flexibility ........................................................................................................................... 406
5.9.3 Reliability........................................................................................................................... 406
5.9.4 Redundancy........................................................................................................................ 407
5.9.5 Automatic Switch-over ...................................................................................................... 407
5.9.6 Telecommunications.......................................................................................................... 408
5.9.7 Power Supply and Auxiliary Facilities .............................................................................. 409

5.10 User Interface ................................................................................................................... 410


5.10.1 Operating System, Operating Location ............................................................................. 410
5.10.2 Operating and Monitoring Functions ................................................................................ 410
5.10.3 Signal System..................................................................................................................... 412
5.10.4 Signal Archive, Process Data Archive .............................................................................. 414

5.11 Hardware and Software Design ..................................................................................... 415


5.11.1 HVDC Control Design....................................................................................................... 415
5.11.2 Operating and Monitoring System..................................................................................... 417
5.11.3 Control and Regulation System, Valve Base Electronics ................................................. 417
5.11.4 Signal Exchange with Switchgear, Event Detection, and Recording of Analog Values . 418
5.11.5 Fault Location (DC Transmission Line), Station Clock ................................................... 419
5.11.6 HVDC Protection Equipment............................................................................................ 419
5.11.7 Remote Diagnostics ........................................................................................................... 419
5.11.8 Standard Software Module Blocks, Graphical Planning and On-line Documentation .... 420

5.12 Project Planning and Testing of the Control System................................................... 421

6. The HVDC System Performance ................................................................................... 423

6.1 System Environment........................................................................................................ 424


6.1.1 The AC Networks .............................................................................................................. 424
6.1.2 Environment of the HVDC Station ................................................................................... 428
6.1.3 Location of Earth Electrodes ............................................................................................. 434
6.1.4 The Environment of HVDC Overhead Lines.................................................................... 434

6.2 Steady-state Performance of the HVDC ....................................................................... 438


6.2.1 Nominal Data of the HVDC .............................................................................................. 438
6.2.2 Limiting Data of the HVDC System ................................................................................. 441
6.2.3 Reactive Power Behaviour................................................................................................. 446
6.2.4 Harmonics on the AC Side ................................................................................................ 449

vii
6.2.5 DC-Side Harmonics........................................................................................................... 452
6.2.6 Power Line Carrier Disturbances....................................................................................... 456
6.2.7 Radio Interference.............................................................................................................. 458
6.2.8 Acoustic Interference ......................................................................................................... 461
6.2.9 System Losses .................................................................................................................... 464

6.3 Transient Performance of HVDC Systems ................................................................... 469


6.3.1 Definition ........................................................................................................................... 469
6.3.2 Switching Processes in Normal Operation........................................................................ 469
6.3.3 AC Network Faults ............................................................................................................ 474
6.3.4 Converter Faults................................................................................................................. 481
6.3.5 DC-Side Station Faults ...................................................................................................... 484
6.3.6 HVDC Line Faults ............................................................................................................. 488
6.3.7 Electrode Line Fault........................................................................................................... 494
6.3.8 Line Faults in Metallic Return Operation.......................................................................... 494
6.3.9 Peculiarities of HVDC Multiterminal Systems................................................................. 498

6.4 The Dynamic Performance of the HVDC ..................................................................... 502


6.4.1 Definition of Terms............................................................................................................ 502
6.4.2 Voltage Stability................................................................................................................. 503
6.4.3 Steady-State and Transient Stability.................................................................................. 513
6.4.4 Subsynchronous Stability................................................................................................... 520
6.4.5 Higher Frequency Stability ................................................................................................ 524
6.4.6 Dynamic Interactions with Near-by Power Plants ............................................................ 528
6.4.7 Dynamic Performance of HVDC Multiterminal Systems ................................................ 532

6.5 System Reliability............................................................................................................. 535


6.5.1 Definitions.......................................................................................................................... 535
6.5.2 Severity Code ..................................................................................................................... 538
6.5.3 Reliability Requirements ................................................................................................... 539
6.5.4 Calculation of Anticipated Reliability............................................................................... 540

7. Insulation Coordination and System Studies for HVDC........................................... 541

7.1 Introduction...................................................................................................................... 541

7.2 Arrester Protection Configuration................................................................................. 543


7.2.1 Historical Development ..................................................................................................... 543

viii
7.2.2 Customary Arrester Schemes of Today............................................................................. 544
7.2.3 Arrester Protection Scheme for a Back-to-Back-Link ...................................................... 546
7.2.4 Filter Protection Scheme.................................................................................................... 546
7.2.5 Selection of the Arrester Protection Scheme..................................................................... 547
7.2.6 Functions of the Arresters in an HVDC Station................................................................ 548

7.3 Characteristics of Metal Oxide Arresters ..................................................................... 553


7.3.1 Arrester Material and Mechanism of Conduction............................................................. 553
7.3.2 Current-Voltage Characteristic .......................................................................................... 554
7.3.3 Arrester Data ...................................................................................................................... 556
7.3.4 Arrester Protection Level ................................................................................................... 559

7.4 Continuous Operating Voltage....................................................................................... 561


7.4.1 The Valve Arrester (V) ...................................................................................................... 561
7.4.2 The Bridge Arrester (M) .................................................................................................... 562
7.4.3 The Converter Group Arrester (CB).................................................................................. 563
7.4.4 Other Arresters ................................................................................................................... 564

7.5 Temporary Overvoltages ................................................................................................ 566


7.5.1 Temporary Overvoltages on the AC Side.......................................................................... 567
7.5.2 Temporary Overvoltages on the DC Side.......................................................................... 569

7.6 Transient Overvoltages ................................................................................................... 571


7.6.1 Switching Surge Overvoltages on the AC Side................................................................. 571
7.6.2 Switching Overvoltages on the DC Side........................................................................... 574
7.6.3 Lightning-Surge Voltages .................................................................................................. 578
7.6.4 Steep-fronted Surge Voltages ............................................................................................ 579

7.7 Protection Level and Test Voltages................................................................................ 580


7.7.1 Arrester Protection Level ................................................................................................... 580
7.7.2 Safety Factor....................................................................................................................... 581
7.7.3 Test Voltages...................................................................................................................... 582
7.7.4 Testing of the MO Arresters .............................................................................................. 585

7.8 Insulation Dimensioning ................................................................................................. 588


7.8.1 Introductory Remarks......................................................................................................... 588
7.8.2 Minimum Clearances in Air .............................................................................................. 588
7.8.3 Creepage Distance of Outdoor Insulators.......................................................................... 590

ix
7.8.4 Oil-Paper Insulation of Transformers and Smoothing Reactors....................................... 592

7.9 System Studies .................................................................................................................. 594


7.9.1 Introductory Remarks......................................................................................................... 594
7.9.2 Basis of the Studies ............................................................................................................ 594
7.9.3 Events to be Studied........................................................................................................... 596
7.9.4 Tools for Studies ................................................................................................................ 603

8. HVDC Bibliography ........................................................................................................ 617

8.1 Comments ......................................................................................................................... 617


8.2 HVDC Books .................................................................................................................... 617
8.3 EPRI Publications............................................................................................................ 618
8.4 IEC Publications .............................................................................................................. 618
8.5 CIGRE Publications ........................................................................................................ 618

x
HVDC Transmission Systems and Their Planning

1. Theoretical Foundations

1.1 Transmission of Electrical Energy

1.1.1 Historical Considerations

The history of transmission of electrical energy began with direct current. In 1882, high
voltage electrical energy was transmitted over a long distance for the first time under the
direction of Oskar von Miller, namely from Miesbach to Munich. At that time, high
voltage meant 2 kV! Nevertheless, it was proven that it was possible to transmit electricity
over a relatively long distance when a higher voltage was used than was customary in
distribution systems. Limitations of direct current were also shown at this time.
Conversion to a higher voltage or vice versa was only possible by making a
transformation via mechanical energy in the form of rotating transformers, i.e. motor
generator sets.

In this respect, alternating current offered undeniable advantages due to the simple
possibility of transformation of voltage. The only question was: What to do with
alternating current? It was not until the transition to 3-phase ac current and Dolivo-
Dobrowolski's ingenious invention, the alternating current asynchronous motor, that this
system appeared to make sense. Nevertheless, the argument over the two systems
continued on for decades. Alternating current, however, continued to be considered
completely unsuitable for the transmission of electrical energy over large distances. In
1891 a spectacular alternating current transmission from Lauffen at the Neckar to
Frankfurt proved otherwise, thus providing important points in the question of systems in
favor of alternating current.

A short passage from the speech by Prof. Heidinger on the occasion of the hundred-year
jubilee of the first alternating current long-distance transmission is appropriate at this
point:

"It was a great moment of our civilization, a flash of ingenuity, a reflection of the divine
spark in the creation. Today we simply say it was a leap of innovation.

-1-
But what great things this leap of innovation triggered for the good and, unfortunately,
also the bad. Of course much preceded it, a centuries long development in intellect, in
knowledge of nature, and finally, in the nineteenth century, a great focusing on natural
science in the area of electricity. All of these prepared the way for the great, courageous,
visionary experiment of Oskar von Miller and the participating companies and that
spectacular transmission of energy. Many workmen prepared the way and participated
such as Dolivo-Dobrowolski and Charles Brown along with Oerlikon and those who
helped finance the project such as Rathenau with the AEG; but he - Oskar von Miller -
was the great architect."

Even though the city fathers of Frankfurt decided in favor of direct current following the
Frankfurt Exhibition, alternating current transmission began, as can be seen in Figure 1
below, its relentless march of victory. Larger and larger levels of power were needed and
were transmitted at higher and higher levels of voltage but still the idea of direct current
transmission was considered over and over and new ways of realizing this goal were
sought after again and again.

Figure 1.1: Development of alternating current technology


(Dotted Line: square of voltage)

-2-
What was really so special about direct current? What characteristics make it so attractive
for the task of energy transmission? To determine this, first consider the fundamentals of
transmission of electrical energy by alternating current.

1.1.2 Advantages and Limitations of Alternating Current Transmission

Let us consider an alternating current system which consists of two partial networks with
the generators G1 and G2 which are linked by a transmission line with the reactance X.

Figure 1.2: Alternating Current Transmission

The transmission power P is the difference between the power generated by the two
generators P1 and P2, respectively, reduced by power which is consumed locally resulting
in the following equation:

U1 ´ U 2
P= ´ SinQ (Equation 1)
X

-3-
According to Equation (1), the transmitted active power P equals the product of the two
network voltages U1 and U2 divided by the line reactance X - values which in practical
operation change only to a minor degree - and finally is multiplied by the sine of the angle
between the voltage vectors U1 and U2 of the two network voltages. This angle cannot be
influenced directly; it depends upon the power balance in the two connected systems. Thus
an excess of power results in the acceleration of the system, that is, a leading angle, while
a power deficit results in a decelerating of the system, i.e., a lagging angle. This does have
the undeniable advantage that the two linked systems help each other. But it can also lead
to undesired displacements of power which, among other things, can result in an overload
of the transmission line. The sensitivity of alternating current transmission to disturbances
of the power balance and of the uncontrollable load flow over connecting lines are two of
the fundamental technical properties which make direct current transmission attractive.

This is supported by another typical characteristic of alternating current transmission: The


transition from one condition to another does not occur abruptly, but rather in the form of
electro-mechanical oscillations. The rotating masses of the generators and motors as well
as (in a far lesser degree) the electrical energy storages of the two linked systems play a
role in this.

Figure 1.3 shows such a process by means of an example of a generator connected by a


high voltage line to the network. The dampened oscillation of the rotor armature angle
theta can be recognized. The resynchronizing torque R and the dampening torque D are
defined in the oscillation equation which is given under the graph. The former is
responsible for the return of the system to the static resting position and provides the limit
of stability. The latter causes damping of the transition oscillations.

-4-
q Rotator angle
Tq Damping constant
Ta Start-up constant
D Damping torque
R Resynchronizing torque

Figure 1.3: Behaviour of a synchronous machine in a large network N.

Electromechanical oscillations in systems linked by relatively long high voltage lines can
endanger the stability of the connection and in certain transient conditions, at the very
moment when assistance from the neighbouring network is urgently needed, can cause an
overload of the line and force a disconnection. Even should the above conditions not result
in a disconnection, the swings of active and reactive power are very disturbing and can
endanger the entire network system.

This provides another point in favour of direct current. DC power transmitted is, for
practical purposes, independent of events in the linked ac systems, eliminating the inherent
power fluctuations which may endanger the stability.

-5-
In the case of ac transmission over great distances, the demand for reactive power is, to a
large degree, dependent upon load and can present significant problems. These problems
can require the installation of intermediate substations to compensate for the reactive
power demand in relation to transmission power.

While series compensation or parallel compensation with shunt capacitor banks and shunt
reactors, respectively, can be of help in transmission through overhead lines, a technical
limit is encountered in the case of transmission through submarine cables. Even at
relatively short distances, the reactive power consumes the greater part of the current
carrying capacity of the cable.

In this situation, HVDC of course clearly has the advantage that a periodic charge reversal
in the cable capacitance and therefore any charging current are eliminated. Transmission
distances of several hundred km can be bridged without problem.

With relatively short a.c. transmission lines the transfer of short circuit power could cause
an other problem. The short circuit level in one system part will unavoidably be increased
through the contribution of the generators of the other system part. The switchgear
aparatus have to cope with the aggrevated short circuit requirements. Also in this respect
HVDC offers a perfect and elegant solution: short circuit power, which basically is
reactive power, cannot be transferred through a dc circuit.

1.1.3 Characteristics of HVDC

These disadvantages and limitations of the characteristics of a.c. transmission compared


with HVDC troubled generations of engineers until a technically and economically
feasible solution was found. These characteristics will now be discussed in greater detail.

Figure 1.4 depicts two ac systems which are linked together, in this instance by means of
an HVDC system consisting of a rectifier station in network 1, a direct current line and an
inverter station in network 2. The transmitted power P is dependent upon the difference
between the two direct voltages Ud1 and Ud2 which are formed by the converters and upon
the ohmic resistance of the dc line R. If the nominal direct voltage is defined as the
average (1/2 * (Ud1 + Ud2)), the equation for the determination of P is derived as follows:

-6-
P=
(U 2
d1 - U 2d 2 )
2R (Equation 3)
U d = kU v ( cos a - u k / 2)

Where:
a = delay angle
uk = transformer short circuit voltage
Uv = valve-side transformer voltage

Figure 1.4: Transmission of power through an HVDC System

Asynchronous Connection

In transmission equation (3), there is no term which relates to the frequency or to the phase
angle between the two ac system voltages. This shows that the transmission with dc is
completely independent of frequency and phase position of the two ac systems to be
linked. An HVDC transmission system can indeed be used to link systems with different
frequencies.

Limitation of Short Circuit Power

Since reactive power cannot be transmitted over a dc link, the contribution of one system
to the network short circuit power of another system is eliminated. Thus, a second system
can be connected by HVDC transmission without increasing the short circuit power.

-7-
Control of Load Flow

Since the direct voltages of the converters, as we will see later, can be set very exactly and
for practical purposes without any delay by means of electronic control, the load flow can
be controlled freely through an HVDC transmission system. Thus it is possible to maintain
a constant flow of energy independent of the momentary power balance of the linked ac
systems. Electromechanical oscillations in the one system do not influence the
transmission of energy and have no effect on the second ac system to which it is linked. If
necessary, the energy flow can be stopped very quickly or its direction can be reversed.

Enhancement of Stability

The fast and exact control of the flow of energy over an HVDC system can also be used to
create a positive damping of electromechanical oscillations by modulation of the
transmitted power. Hence, an HVDC system can effectively lend support to the stability
of a network to which it is connected or to a parallel ac connection.

Other Regulating Functions

The exact and undelayed control of the converters makes it possible to utilize an HVDC
system for the implementation of a whole series of regulatory functions. Thus the
regulation of frequency of a "weak" ac network can be superimposed upon the control of
active power. By influencing the converter reactive power demand, the ac voltage at the
connecting point of the HVDC system can be controlled, etc.

1.1.4 Possibilities for Utilization of HVDC

The characteristics of HVDC have led to a series of typical applications described below.
For the present, consideration of the economic aspects of costs and losses will be omitted.
These shall be treated in Section 1.1.5

-8-
Figure 1.5 shows, in schematic form, eight typical applications of HVDC.

Figure 1.5: Applications of HVDC


1.) Long distance transmission through overhead lines.
2.) Transmission through sea cables.
3.) Back-to-back link.
4.) Power plant unit connection.
5.) Power supply in densely populated areas.
6.) Superimposed multiterminal dc system.
7.) Tapping of overhead line.
8.) Voltage converter.

1.1.4.1 Long Distance Transmission through Overhead Lines

With growing consciousness of environmental concerns, the use of renewable energy


sources is becoming more and more urgent even if these sources are located far from
consumer centers. HVDC offers a technically reliable and economically feasible
possibility for transmission of very great amounts of energy over long distances.

-9-
Almost all large HVDC systems with more than 1,000 megawatts of nominal power
connect large hydropower plants to distant centers of power consumers. Figure 1.6
presents in schematic form the large HVDC long-distance transmission systems. Some of
these systems are still using mercury-arc valves and some have been converted to thyristor
valves or have been enhanced by thyristor groups.

Besides the technical aspects such as precise controllability, absence of stability problems
and ease of fault clearing, economic considerations are especially important for this
application of HVDC systems.

DC overhead lines need only two conductors. This system is simple in design, tower
weight is less than in the case of ac, and significantly less right of way space is needed.
One must always keep in mind that, in comparison to a bipolar dc line, a double ac line
must be used in order to achieve the same availability. This is because in HVDC systems
monopolar operation with earth return is possible.

The possibility that electrical and magnetic fields affect are harmful to human health is
much discussed today. The extent of any risk is still largely uncertain and therefore is still
a matter of speculation. All we can be sure of at this point is that direct fields are far less
critical than alternating ones.

- 10 -
1 kA
100 kV 1000 km

2
1 2
3 1 2
1

PI NR CB IT EK

PI: Pacific Intertie (western U.S.) 1360 km


1: Original construction with mercury-arc valves ±400 kV, 2000 A
2: Upgrade with thyristor valves to ±500 kV, 2000 A
3: Extension with parallel groups to ±500 kV, 3200 A
NR: Nelson River (Manitoba / Canada) 890 / 930 km
1: Bipole 1 with mercury-arc valves ±450 kV / 1800 A
2: Bipole 2 with thyristor valves ±500 kV / 2000 A
CB: Cahora Bassa (Mozambique / Africa) 1414 km
with two monopolar dc lines ±533 kV / 1800 A
IT: Itaipu (Brazil) 800 km
Two bipoles each ±600 kV / 2625 A
EK: Ekubastus - Central Russia 2400 km
(has been under construction for 20 years) ±750 kV / 4000 A

Figure 1.6: HVDC transmissions using overhead lines.

Figure 1.7 shows schematically the tower diagrams of a 380 kV alternating current double
circuit line and a ±400 kV bipolar HVDC line. Both systems are sized for 1,200 MW
nominal transmission power. The significantly greater cost of the alternating current
option can be imagined. Figure 1.8 shows one-half of an analogous example for
transmission of 4,000 MW. Two 765 kV alternating current lines are compared with two
600 kV monopolar HVDC lines. The significantly lower cost of the dc lines are even more
apparent in this example.

- 11 -
Figure 1.7: 1,200 MW transmission: HVAC 2 x 380 kV; HVDC ±400 kV

Figure 1.8: 4,000 MW transmission: HVAC 2 x 765 kV; HVDC ±600 kV


(Represented are only one ac line and one dc pole)

- 12 -
These examples illustrate the most important structural configurations of HVDC overhead
bipolar and monopolar transmission lines. The monopolar configuration is more expensive
in terms of material, and necessitates the use of two rights of way. However, it has the
significant advantage that in the event of a tower failure of one line, 50% of the
transmission power is available - assuming that operation with one line using ground
return is permitted.

Depending upon the degree of system reliability required and the sensitivity which is
tolerated for transient or permanent line faults, various types of HVDC overhead lines can
be constructed (see Figure 1.9).

Figure 1.9 is a table showing five line configurations with the same transmission power
(same conductor cross section). Each configuration differs in remaining transmission
capacity after (permanent) line faults.

Remaining Transmission Capacity

Tower Loss of one pole Relative Costs


Variants Configuration Ground return Tower breakage in p.u.

permitted not permitted

single 0.5 0 0 1
bipolar (1)
line

double 1 1 0 1.14
line

1½- 1 1 0 1.07
line

monopolar 0.5 0 0.5 1.26


line (1) (1)

two lines 1 1 1 1.36


(bipolar or
homopolar)

Figure 1.9: HVDC Overhead Line Configurations

- 13 -
An increase in reliability means an increase in cost of transmission lines. It is assumed that
the current carrying capacity of a conductor is twice the nominal current. For a more
detailed discussion of the "economic cross section" and thermal design of an HVDC
overhead line, see chapter 1.1.5.2. The figures in parentheses (1) assume that the two
station poles are switched in parallel (see chapter 2.7). If fast switching is provided, the
"one and a half" line is equivalent to a double circuit line. This table assumes that it is
possible to divide out the total cross section into two lines without exceeding the
permissible surface field strength (see chapter 3.1). Use of the homopolar line
configuration (two-line system with same polarity on one tower) can be a way to
overcome this potential problem.

Cahora-Bassa is an example of an existing plant with two monopolar lines. It is one of the
most outstanding examples of long distance energy transmission of any type. As is shown
in Figure 1.10, energy is generated by a hydropower plant at the lower run of the Zambezi
and is transmitted across 1400 km into the load center of the Republic of South Africa at
Johannesburg. The Cahora-Bassa gorge offers ideal conditions for a hydropower plant.
Figure 1.11 shows the 160 meter high arch shaped dam which is only 300 meters in
length. Since the area is almost free of human habitation, no great outlays were necessary
for resettlement.

Because of the economical transmission of energy by HVDC, energy from North


Mozambique can economically compete with energy generated in the coal-fired power
plants in the Johannesburg area. The Cahora-Bassa output is currently 2,000 MW.
Practically the entire output is transmitted over the HVDC system. In this system, a
transmission voltage of more than a million volts (±533 kV) was used for the first time.
For reasons of reliability, a decision was made for two monopolar lines separated by a
distance of approximately 1 km (Figure 1.12).

- 14 -
Figure 1.10: The Cahora-Bassa HVDC System

Figure 1.11: The arch-shaped dam of the Cahora-Bassa hydropower plant in the
Zambezi gorge

- 15 -
Figure 1.12: One of the two monopolar 533 kV direct current lines of the Cahora-Bassa
HVDC

Of course even the most secure line design and construction is useless against the
destructive rage of terrorists, which is adding to the economic distress of the country.
Cahora-Bassa used the energy of the Zambezi for less than five years before the lines were
blown up! It is hoped that the reconstruction which has been decided upon may be
accompanied by the pacification of Mozambique.

- 16 -
There was a very similar task in the northern part of Canada. Energy generated in a chain
of power plants on the Nelson River is needed in the load center of the province of
Manitoba, in the area of the capital city of Winnipeg (see Figure 1.13). Since the early
1970's, an initial bipolar HVDC using mercury-arc valves has been in operation. In the
course of expansion of the power plants, a second HVDC bipolar transmission system was
installed. In this installation, water-cooled thyristor valves were used for the first time.

Figure 1.13: The Nelson River HVDC in the Canadian province of Manitoba links
hydropower plants in the north with the capital city of the province
Winnipeg over a distance of 900 km

- 17 -
Figure 1.14 demonstrates the progress of technology. The left side shows the converter
building for Bipole One with mercury-arc valves, designed for 1,620 MW. The right side
shows the converter buildings of the thyristor plant which is designed for 2,000 MW.
Since the end of the 1970's, Nelson River Two has been in operation with a high degree of
availability and an extremely low thyristor failure rate.

Figure 1.14: The converter buildings of the two HVDC bipoles in the Dorsey inverter
station

- 18 -
One of the most modern HVDC overhead line transmission systems is the Gezhouba-
Shangai system. This was the first HVDC system in China. It transmits 1,200 MW from
the hydropower plant Gezhouba on the Yangtzi across a distance of more than 1,000 km to
the load center of Shanghai. This plant is remarkable because of the valve design. Due to
possible earthquake stresses, the valves are suspended from the roof structure of the
converter building. (Figure 1.15)

Figure 1.15: Thyristor valves suspended from the roof structure

- 19 -
1.1.4.2 Cable Transmission

Aside from the short-lived experiment with a three-point HVDC system in the inner city
of London, all existing HVDC transmission systems using cable are sea cable connections.
At distances of more than 40 or 50 km, alternating current cannot be used for cable
transmission due to the large charging current. Many HVDC sea cable connections have
been constructed because of the desire to establish an electric connection between the
mainland and an off-shore island, between two islands, or across an inlet of the sea. The
smallest is the first commercially used HVDC, the Gotland connection, which carried 20
MW. The largest is the 2,000 MW Cross Channel system with two bipoles of 1,000 MW
each.

With few exceptions, solid cables have been used in existing systems. For practical
purposes, it has no limit with respect to length and it can be laid in depths of 500 meters or
more. However, nominal current and nominal voltage are limited with respect to the
distribution of field strength in the dielectric (See chapter 3.2). The development of
HVDC cable technology can be clearly seen from the nominal rating of cable transmission
systems which have been constructed.

Year System Nominal Voltage Current per Conductor Cable Capacity


(kV) (A) (MW)
1954 Gotland I 100 200 20
1985 Cross Channel II 270 926 250
1990 Fenno-Skan 400 1250 500

Figure 1.16: Data for existing HVDC cable systems

Only a few completed systems feature pure cable transmission. In spite of the very low
cost of HVDC cable in comparison to ac cable, cable transmission systems still cost
several times as much as overhead line transmission systems. When the HVDC converter
station is not directly on the coast, a transition from cable to overhead lines is usually
provided. The nominal voltage is always dictated by the cable. Therefore optimization of
the overhead lines is possible only to a very limited extent.

- 20 -
1.1.4.3 Back-to-Back Links

The properties of HVDC which make it possible to transmit energy and to control it
precisely, even when the ac networks have different frequencies or are operated at the
same frequency but under differing regulating parameters, have led to the construction of
several back-to-back links. Examples are in the U.S. at the borders of the three major
power systems and in Europe at the former border between UCPTE and the RGW
systems. The Dürnrohr link, which has been in operation since 1983, and the Southeast
Vienna and Etzenricht links, both of which went into operation in 1993, can be cited as
examples.

The expression back-to-back indicates that the rectifier and inverter stations are connected
directly without a dc line in the middle. There is no viable competition in ac technology
for this application. The economic justification of a back-to-back link is derived from the
general advantages to the power systems:

- increased reliability of power supply


- reduced spinning reserve
- full economic use of the power plants
- levelling of daily and seasonal peaks

Figure 1.17 shows the former map of Europe with the major power systems which were
linked at Dürnrohr for the first time.

- 21 -
Figure 1.17: Interconnected Power Systems of Europe (1993)

Since no consideration has to be given to the optimization of an overhead line in a back-


to-back link, the system can be designed completely on the basis of an optimized converter
plant. The 550-MW Dürnrohr back-to-back link fully uses the current carrying capacity of
a 100 mm thyristor. At a nominal current of 3,800 A, the direct voltage is only 145 kV.
This configuration made it possible to use a very compact valve design.

Figure 1.18 is an interior view of the converter building of the Dürnrohr back-to-back
link. Figure 1.19 is an aerial view of the plant. Note the surprisingly small space required
by the actual HVDC in comparison to the two 380 kV open air switchyards and the filter
circuits.

- 22 -
Figure 1.18: Converter valves of the Dürnrohr back-to-back link

Figure 1.19: Aerial view of the Dürnrohr back-to-back link

- 23 -
The Chateauguay back-to-back link in the Canadian province of Quebec has a similar
mission to accomplish. A synchronous coupling of the Canadian and US power systems
would have led to major stability problems. The problem of transmitting energy between
the two systems was overcome by using a back-to-back link consisting of two 500-MW
blocks.

It is interesting to note that the system also contains two static var compensators (SVC)
which assure the maintaining of the ac voltage on the US side. The SVCs comprise
thyristor controlled reactors (TCR) whose reactive power consumption is continuously
controlled by means of antiparallel connected thyristors as well as thyristor switched
capacitors (TSC) whose effective capacitance is changed in steps. Thus the active and the
reactive power flows between the two interconnected a.c. systems can be determined in
accordance with the system requirements by means of coordinated control of both, the
HVDC and the compensators, which has been applied for the first time in this installation.

Figure 1.20 shows a photo of the Chateauguay plant.

Figure 1.20: Chateauguay back-to-back link

- 24 -
1.1.4.4 Tapping of an HVDC Long-Distance Line

A frequent criticism of HVDC systems is that there is no simple and economic possibility
of splitting off a small amount of power for local consumers from a long distance line with
high transmission capacity. Such applications are quite conceivable if one keeps in mind
the major task of HVDC long distance transmission, which is to transmit the output of
renewable energy sources across large distances to consumer focal points. It is quite
conceivable that areas of low population density will be crossed by HVDC lines and that
the supply of these areas with clean energy might be urgently needed to reduce use of
wood as fuel. Economical splitting off of very small quantities of power in relation to the
entire system is required precisely in such places.

In figure 1.21, the two basic possibilities to connect an HVDC tapping station are
presented. In the parallel arrangement it is necessary for the converter valves to be
designed for the full transmission voltage at a nominal current of a few percent of the
transmission current. This results in extremely expensive valves. Similar considerations
hold true for the valve windings of the converter transformers. In addition, the
vulnerability of the main system must be considered. Each commutation failure in the
tapping station converter results in a transient short circuit of the main transmission and in
an extreme overcurrent load on the tapping station.

In the series arrangement, however, the converter valves and the valve-side transformer
windings should be designed for the full nominal current of the system but only for a
voltage of a few percent of the transmission voltage. This results in a design with
favourable cost. Another advantage is the insensitivity of the entire system. A
commutation failure in the converter of the tapping station is practically unnoticed in the
main transmission. Overcurrent load on the tapping station does not occur. Thus series
connection will probably be preferred in the future.

Present proven technology can be used in most cases for series tapping so far as the
feeding into an existing network is concerned. Otherwise forced commutated converters
will have to be used. The current source converter promises the best results. This
technology, however, will not be discussed further at this point.

- 25 -
Figure 1.21: Tapping of an HVDC Long Distance Transmission
a.) Parallel Tap
b.) Series Tap

The converter valves of a series tap will be arranged on an insulated platform and the
converter transformer must be insulated between the network and valve windings for the
full direct voltage.

The only protective breaker needed is a fast by-pass switch. This switch does not have to
handle any breaking of direct current. In addition, several disconnecting switches will be
needed, especially if provision is to be made for selectively looping into one of the two
line poles.

Although there are no existing series tapping systems, the technology and know-how are
available.

- 26 -
1.1.4.5 Multiterminal HVDC Systems

Since the beginning of HVDC development the idea of an HVDC multiterminal system
has been considered. An example would be a supernet spanning a continent. Nevertheless,
all HVDC systems have been constructed as point-to-point systems until very recently.
Systems currently being constructed hardly deserve to be called multiterminal systems. It
may therefore be asked why there should be multiterminal systems.

Transmitting electrical energy across great distances between nodes of an interconnected


network or between separate networks can be accomplished by several separate point-to-
point HVDC's, as well as by an HVDC multiterminal network.

Consider as an example the connecting of four ac networks A through D:

Figure 1.22: HVDC Connection of Several Networks


a.) By separate point-to-point HVDC's
b.) By a multiterminal HVDC

- 27 -
In comparison with separate HVDC point-to-point systems, the HVDC multiterminal
network has the following basic advantages:

- The number of HVDC converter stations and their total power is lower.
- Transmission losses are lower since energy transport over more than two HVDC
stations in series is avoided. In case a), the energy flow would take place from A to D
through six stations.
- In a mesh network (additional line from A to D), a current distribution is automatically
realized in the HVDC network which corresponds to minimal line losses.

The HVDC multiterminal network has the following significant disadvantages:

- A dc-side fault or a long-lasting commutation failure in one of the converter stations


will affect the entire system (transient loss of 50% or 100% of transmission capacity).
- The loss of one converter station requires the immediate re-balancing of the current
reference values in order to avoid a breakdown of the entire transmission system.
- The greater costliness of HVDC lines is only an apparent disadvantage. In case a) one
would need to take into consideration the corresponding share of the more expensive ac
lines and their losses as part of the HVDC system.

Except for special situations, an HVDC multiterminal network will always be the more
economical approach. Special cases where this might not hold true could be free transport
capacity in the ac network which is not otherwise usable, or inability to obtain rights of
way for the HVDC lines.

To achieve the same or even a better degree of reliability that one can obtain today from
point-to-point HVDC systems, however, several prerequisites must be fulfilled.

- Minimization of inverter commutation failures


- Very fast and redundant telecommunication connections
- Selection of a suitable control system
- Use of HVDC circuit breakers

- 28 -
There are three fundamental types of network configuration in which an HVDC
multiterminal system can be designed:

- The line network a)


- The star shaped network b)
- The meshed network c)

Figure 1.23 shows three network configurations of HVDC multiterminal systems.

The cost for lines increases from a) to c).

Vulnerability of the system to a permanent line fault such as tower breakage decreases
from a) to c). In case a), a line interruption between B and C would divide or split the
system into two point-to-point HVDC systems. The value of these two partial systems
would be presumably very small for the entire system. In case b), a line interruption would
always result in the splitting off of one HVDC station. The others would be able to
continue the exchange of energy. In case c), a line interruption would have no influence at
all on the exchange of energy between the HVDC stations. Line losses would simply be
higher.

- 29 -
Redundancy of transmission lines is a basic feature in the mesh network. But line
redundancy can be also built into line and star-shaped networks by the use of double lines
(see also section 1.1.4.1 and figure 1.9).

Of course, the advantages of a redundant system are only meaningful if it is possible to


quickly and reliably isolate a section in which there is a fault. This involves fault clearing
by HVDC breakers. The faulty line section is separated out without support by the
converter control. Current continues to flow in the system pole in which the disturbance is
present. The first breaker, which opens, commutates the current into the parallel
(redundant) current path, made available by the system. The second breaker, which opens,
interrupts residual current, which is passing through the fault point to ground.

As could be confirmed in simulator studies, the exchange of energy can be re-established


within approximately 100 ms. This is true whether the line fault is a transient or a
permanent one.

The demands made on an HVDC breaker are not at all exorbitant. An ac circuit breaker
pole enhanced by a commutating circuit and an energy absorber can perform this function
(see chapter 2.7). At the moment of contact opening, a breaker of this type generates a
sharply increasing counter voltage which spreads in the form of a travelling wave to both
sides in the dc system. In this way, a considerable portion of the energy stored in the line
inductances is transferred into the line capacitances. The energy which is picked up by the
breaker does not exceed the capacity of modern metal oxide surge arrestors.

It should be pointed out at this point that no HVDC breakers have been placed into service
in HVDC plants which are in operation today or are under construction. The main reason
for this is that none of the systems owns a true line redundancy. (The two Nelson river
HVDC's are normally operated as two separate bipoles.) But undoubtedly the HVDC
breaker will be available after a 500 kV prototype breaker has been successfully tested in
the Pacific Intertie HVDC system (fig. 1.24).

- 30 -
Figure 1.24:.500 kV HVDC Breaker

An HVDC breaker is not only important for clearing dc line faults, it can also assume
essential tasks in the event of station faults. This is particularly true when a "small"
inverter (or an inverter group in groups connected in parallel in a station) suffers a
commutation failure. The fault current increases to high values, because all rectifier
stations of the multiterminal network feed into this "sink." The chance that such an
inverter will recover after the first commutation failure is very remote. A multiple
commutation failure, however, means in most cases that the energy exchange in the entire
multiterminal system - not only in one pole - is temporarily interrupted.

Only when the inverter regains its commutation capability, following a current reduction,
and has built up its counter voltage again, does the energy exchange resume. Therefore it
appears to be very important that the disturbance is cleared quickly. For this reason, use of
a fast HVDC breaker is called for in this situation.

- 31 -
In summary, the following can be stated:

- The use of HVDC multiterminal systems appears to be advantageous technically and


economically.

- The system of lines must be designed with redundancy in order to achieve a high
degree of reliability, either as a mesh network or by the use of double circuit lines.

- Suitable control methods and a fast telecommunication system are prerequisites.

- Fault clearing can be significantly improved by the use of HVDC breakers in line and
station branches.

- The necessary elements and the know-how are available.

1.1.4.6 Feeding Energy into High-Population Areas

Feeding electrical energy into high population areas presents increasing difficulties.
Cables with increasingly high voltage have to be utilized for the provision of power for
cities and industrial complexes. It seems probable that the much more economical dc cable
should be used for this purpose. It not only offers advantages from the standpoint of cost
but also with respect to right of way width and litigation of land. It is easy to imagine an
HVDC cable ring around an inner-city distribution network. Several HVDC inverter
stations would feed energy from peripheral power plants into the distribution network.
Figure 1.25 shows an example of an HVDC multiterminal system in series configuration.
This configuration might be superior in this special application both technically and
economically to the customary parallel configuration.

- 32 -
P = Power Station
N = Inner-city distribution network

Figure 1.25: Energy supplied to a densly populated area by means of a ring-shaped


HVDC system
Left: Energy flow
Right: Configuration of a multiterminal system with stations
connected in series

An HVDC application of this type could make it necessary to construct an HVDC inverter
station in a thickly populated area. This will be a compelling reason for a compact design.
The utilization of metal encapsulated, SF6 - insulated technology, which has been
successfully used in several hundred ac switchgear applications, appears to be a reasonable
approach.

As has been demonstrated in many studies, insulation of high direct voltage places far
higher demands on the design and manufacture of insulators and cleanness of the gas
volume. In addition to these fundamental and intense investigations, design considerations
have led to the development of a compact design. From this technology, a compact HVDC
station as illustrated in fig. 1.26 was developed. In this example, various components of
the converter station are brought together in a multi-story type of construction. AC
switchgear, connecting lines between transformers and the valves and the dc switchgear up
to the smoothing reactor are configured with metal-encapsulated SF6-insulated technology.
The ac filter circuits are mounted on the roof of the building. Using a design of this type it
is possible to achieve an area requirement of approximately one square meter per MW.

- 33 -
Figure 1.26: Section through the side view of a compact HVDC station with air
insulated converter valves.
SW = Visual and wind screen
F = AC filter circuits
P = Thyristor valves
T = Transformers

As revolutionary as the concept might appear, one must keep in mind that the compact
technology, particularly the SF6-insulated dc connections and the dc switching devices are
not yet available in tested configuration. Ground laying development work is still
necessary before commercial application.

1.1.4.7 HVDC Voltage Converter

Combined cable/overhead line systems were discussed in section 1.1.4.2. It was pointed
out that in these systems, cable limits determine the transmission voltage. If lines are very
long (some thousand kilometers) and the power to be transmitted is very high (some
thousand MW), the optimal transmission voltage would certainly be ±600 kV, perhaps
even ±800 kV. If it was necessary to limit the entire transmission to ±400 kV or ±500 kV.

- 34 -
because cable voltage is limited to 400 kV or 500 kV, the result will be considerably
higher investment costs and power losses. Therefore a conversion of voltage at the
transition from overhead lines to cable appears to be economically attractive.

It seems apparent that the voltage conversion is very expensive if an "inverted" back-to-
back link is used for this purpose. This is demonstrated in figure 1.27A, using the example
of a 750/500 kV voltage conversion. The entire transmission power has to be converted in
two conversion processes which are connected in series.
The converter rating is PBB = 2 * Pd.

Figure 1.27A: Voltage conversion by Inverted Back-to-Back Link

One advantageous option in ac technology is the linking of two voltage levels using an
auto-transformer. The smaller the ratio of voltages of the two levels, the smaller the rated
power of the auto-transformer. A corresponding system is also possible for transforming
direct current.

In Figure 1.27B, a dc voltage conversion in the most economical auto-connection is


depicted. It can be seen from the converter symbols presented, that the converter rating
corresponds to only one third of that of the inverted back-to-back link. In general the
following equation holds true:

PA = PBB(1 - U2/U1)

- 35 -
A system of this type can also be used to split out power by connecting an additional
transformer to the three-phase busbar. On the other hand sufficient short circuit power
must be available at the busbar for the commutation of the converters or it must be made
available by a synchronous condenser. It appears reasonable that the use of forced
commutated converters in these cases should be considered.

Plants of this type have not yet been constructed. The elements for such plants, however,
are available and have been proven successful.

Figure 1.27B: Voltage converter in auto-connection

- 36 -
1.1.4.8 Generator Converter Unit Connection

The most important renewable energy source for the foreseeable future is hydropower. In
North and South America, Asia and Africa, there are large numbers of suitable sites where
many thousands of MW in some cases several tens of thousands of MW can be harnessed.
In most cases, these sites are situated far from the focal points of energy consumption. The
costs for energy transportation assume decisive significance and require the optimization
of HVDC technology for this task. There appears to be a technological limit for the
increase of transmission voltage for the purpose of lowering transmission costs at around
800 kV. This makes it all the more important to look for additional possibilities for
lowering transport costs. There are two starting points for this effort:

- Reduction of construction costs (investment)


- Improvement of the overall efficiency (degree of source utilization)

Unit connection of generator and converter can make a significant contribution to both
starting points.

In figure 1.28, power from a hydro plant is transported using conventional HVDC
technology. It is characteristic for this type that the three-phase busbar serves as the
interface between the power plant and the HVDC system. Filter circuits are connected to
this busbar for absorbing the harmonic currents which are generated by the converter.

Figure 1.28: Generator of a hydropower plant and rectifier station of a conventional


HVDC long-distance transmission system.

- 37 -
If the generator is able to absorb these harmonic currents, the filter circuits will become
unnecessary. In that case, a three-phase busbar is also not necessary - and when the ratings
of generator and converter group are equal - a separate block transformer becomes
unnecessary.

In figure 1.29, a combination of this type of generator and rectifier in unit connection is
shown. The rectifier-transformer becomes the block-transformer. Not only investment
costs, but also losses of the rectifier station can be reduced in this way by 30% to 40%.

Figure 1.29: Unit connection of generator and twelve pulse converter group

In addition, unit connection allows to choose the generator frequency at will and thus to
find the economically optimal design of generator and transformer. In addition, it is
possible to operate the machinery with variable speed. By this means, it is always possible
to operate at optimal turbine efficiency depending on water head and power demand. This
means that the maximum electric energy can be obtained from a given quantity of
hydraulic energy. Unit connection also promises considerable advantages in the case of
pump storage power stations. The optimal speed differs by about 20% between pump and
turbine operation. The converter can furthermore be used for starting-up the machinery for
pump operation.

A very specific problem emanates from the operation of a unit connected system with
partial load. It is common practice in large hydro power stations to connect or disconnect
single machine sets in order to adapt the generated power to the load demand or to the
available hydro energy, respectively. In this way it is ensured that all the running machine
sets are operated close to their rated power and thus with good turbine efficiency. An

- 38 -
HVDC transmission system with a rectifier station in conventional connection, i.e. with an
ac collector bus, does not present any problem with this kind of partial load operation; all
converter groups participate with equal percentage in the total power, the transmission
voltage is maintained at the nominal value. The situation is different, however, with a unit-
connected HVDC system, where the disconnection of one or several machine sets has
significant consequences on the operation of the HVDC system.

Figure 1.30 shows a unit-connected hydro power station with 4 machine sets. The two
converter groups of each station pole are connected in series. When one machine set is
shut down the dc voltage in the affected pole is reduced to half. The consequence is a
significantly reduced transmission efficiency. The advantage of the series connection, on
the other hand, is the maintained current balance between both poles, which means that no
current through earth has to be tolerated. The rating of the converter groups is for the full
transmission current and a part (in our example half) of the transmission voltage; this leads
to the lowert possible valve costs.

Figure 1.30: Series connection

- 39 -
Figure 1.31 shows the same hydro power station, however, the two converter groups of
each station pole are connected in parallel. When one machine set is shut down, the direct
current in the affected pole is reduced to half, while the transmission voltage of both poles
is maintained. Consequently the transmission efficiency is not reduced (even slightly
increased). The difference between the two pole currents is flowing through earth. If earth
return is not permitted, always an equal number of machine sets has to be operated in both
poles, which means that the power steps between two economic modes of operation is
doubled. A further disadvantage of the parallel connection is the rating of the converter
groups for the full transmission voltage but only for a portion (in our example half) of the
transmission current. High specific valve costs will be the consequence, particularly if
more than two converter groups have been connected in parallel. Also the necessary
splitting of the smoothing reactor contributes to increased station costs.

Figure 1.31: Parallel connection

- 40 -
Figure 1.32 shows an other variant, the so-called group connection. Always two (or more)
generators are connected in parallel via generator breakers, forming a group of generators,
which is connected with one 12 pulse converter group. The disconnection of a single
generator reduces the direct current of the pertaining converter group accordingly but
leaves the terminal voltage unchanged. The converter groups are rated for the full
transmission current which means favourable valve costs. Insofar the group connection
combines the advantages of series and parallel connection. As a disadvantage may be
considered the direct parallel connection of generators; the speed autarchy of the
individual machine sets is lost; power oscillations between the generators of one group
cannot be excluded.

Figure 1.32: Group connection

- 41 -
A further variant is shown in figure 1.33 which avoids this disadvantage, the so-called
variable connection. In the upper part of the power range the station operates in the mode
of series connection with the full speed autarchy of the individual machine sets. When one
generator shall be disconnected, the pertaining converter group is connected to the
remaining generator, whose output is now shared by the two paralleled converter groups.
The full dc voltage is thus maintained, the transmission current in the affected pole is
reduced to half. This operation mode equals that of the parallel connection. The converter
groups are rated for full transmission current and half transmission voltage, which means
favourable valve costs. A parallel operation of generators is avoided, the full speed
autarchy of the machine sets is maintained.

Figure 1.33: Variable connection

- 42 -
Which one of the presented arrangements is the optimal solution, cannot be stated in
advance, it has to be established in comperative studies for each concrete project.

Independent from the chosen arrangement there are a number of control functions, which
can differently be allocated to the components of a unit-connected HVDC system
comprising turbine, generator, rectifier and inverter. One of the possibilities is presented in
figure 1.34. The hydroturbine assumes the function of regulating the rotary speed.
Generator voltage and valve voltage are set by means of the generator excitation in such a
manner that the firing angle alpha of the rectifier always corresponds to the nominal value.

The direct current is regulated by the rectifier, and the direct voltage is regulated by the
inverter.

Figure 1.34: Principle diagram of generator-converter-unit connection


R = Regulator for given control functions

It seems quite certain that operation with variable speed is possible for large hydropower
generators without significant additional costs for machinery. The advantages indicated
can thus be fully utilized. It is still unclear, whether variable speed operation is technically
permissible in conventional steam power plants, and if so, whether it will be
advantageous. In contrast, considerable cost savings would have to be obtained in gas
turbine power plants if the machinery (without gear) is operated at nominal rotating speeds
of 5,000 or 6,000 revolutions per minute.

- 43 -
1.1.5 Economic Aspects of Utilization of HVDC

In this section, the question of whether in a given situation HVDC transmission might be
the most economically attractive alternative will be addressed. The boundaries are being
extended with respect to transmission power and distance and also with respect to several
environmental parameters.

1.1.5.1 Range of Power

Since long distance transmission by HVDC using overhead lines is only feasible at a
distance of several hundred km, a transmission voltage of greater than 250 kV, usually
more than 400 kV, must be assumed. In addition, monopolar transmission probably has to
be ruled out for the future, considering the pretended risks of current flow through earth.
The upper voltage limit is currently around 600 kV, although experts believe it possible to
handle 800 kV with the application of a few new developments, primarily in the area of
transformers.

In the case of transmission current, economical design of thyristor valves is of primary


importance. Since thyristors make up less than half of the cost of valves, the smaller the
thyristor power the more expensive a particular valve becomes. The lower limit can be set
at approximately 1,000 A nominal direct current. The upper limit is determined by the
largest available HVDC thyristor which is currently the 100 mm diameter thyristor, having
3,300 to 4,000 A nominal current.

With these preparatory remarks, the lower and the upper limits of the power range for an
HVDC long distance transmission, using overhead lines, can be established as follows:

±250 kV x 1000 A = 500 MW


to
±600 kV x 3300 A = 4000 MW
(±800 kV x 3800 A = 6000 MW)

These data apply to a bipolar configuration. Of course the transmission power can also be
increased at will by arranging parallel configurations.

In the case of cable transmission, the power which can be transmitted through each cable
is determined by the current state of development of cable technology. Today one can
assume around 450 kV and 600 MW, but technology is changing. Of course several cables

- 44 -
can be connected in parallel so that the upper limit of transmission current is determined
by the largest available thyristor. The range of current can somewhat be expanded at the
lower end, since the valve cost in cable transmission (at least at transmission distances
beyond about 100 km) is clearly of less importance. In contrast to transmission with
overhead lines, monopolar operation can be considered acceptable with sea-cable
transmission. In fact, this is almost the rule.

The lower and upper power limits of HVDC sea-cable transmissions can be established as
follows:

250 kV x 800 A = 200 MW


to
±400 kV x 3300 A = 2600 MW *
(±500 kV x 3800 A = 3800 MW) *
* with 2 or 3 cables in parallel

Larger amounts of power can be achieved with bipoles in parallel.

In back-to-back links, current is determined by the nominal current of the largest available
thyristor, that is, 3,300 A to 4,000 A. The power limit per "block" (a unit with one twelve-
pulse group on each side) is determined by the largest available transformer pole. The
transition from the three-winding transformers, which are customary today, to two-
winding ones will facilitate an increase in block power. The following figures give a
"reasonable" range of power, that is, a range of power which excludes the extreme values:

30 kV x 3300 A = 100 MW
to
250 kV x 4000 A = 1000 MW

These figures apply per block. Of course, it is possible to increase the power by using
several blocks.

A range of power for the other potential applications presented in sections 1.1.4.4 through
1.1.4.8 cannot yet be established in a meaningful way, since these applications will be, if
at all, in the future. Transmission tasks must be analyzed on a case by case basis as they
occur to determine the feasibility for application of HVDC technology.

- 45 -
1.1.5.2 Optimal Transmission Voltage

Transmission voltage in the case of cable transmission is predetermined by cable


specifications. In the case of back-to-back links, voltage is determined by the specified
power and the maximum thyristor nominal current. In long-distance transmission using
overhead lines, however, voltage can be freely chosen. Therefore it can be optimized, with
the goal of minimum total transmission costs.
Considered alone, the total costs of overhead lines are made up of construction costs
(usually expressed as annual costs by using an amortization factor) and the loss costs per
year.

The "load profile" of the transmission must be known in order to determine the loss costs.
(The oftenly specified annual usage hours are not sufficient for the calculation of losses
because most of the losses are dependent on I2.) In addition, we need the value factor for
losses, specifically the loss value for load losses.

Construction costs of a bipolar HVDC overhead line (including erection, but without land
costs) can be estimated using the following cost formula:

KB = A + B * Ud + C * q [in DM/km] (4)


where Ud = Line-Ground Voltage in kV
q = Line cross section (Al) in mm²

The cost factors A, B and C were determined by the HVDC Working Group
[Arbeitsgemeinschaft HGÜ] in the 1960's. They have been adjusted by approximately 5 %
per year.

1960 1990
A 6,000 26,000 DM / km
B 115 500 DM / kV * km
C 30 130 DM / mm² * km

It is interesting to note that the number of conductors per bundle has no significant
influence on construction costs.

Cost formula (4) applies to bipolar lines. The following formula applies for monopolar
lines:
KM = 0.63 KB (5)

- 46 -
This means that configuration of an HVDC bipole with two monopolar lines results in an
increase in line costs of about 25%.

For a double circuit line which consists of two bipolar systems on one tower, the following
formula applies:
KD = 1.65 * KB (6)

This means that a double circuit line is about 17.5 % less expensive than two bipolar lines.
On the other hand, dividing a ±500 kV / 2,000 A bipole into two systems (each conductor
with 1/2 cross section) means approximately 15% additional costs.

In cost formula (4), the transmission voltage is used. The transmission current does not
appear; instead, the conductor cross section q is used. To determine this we need the
current density j.

I dN
q= (7)
j

Since overhead lines for economical reasons are operated far below their thermal limit, j is
an optimization parameter. To make the relationships clear, figure 1.35 shows construction
costs and loss costs of an HVDC overhead line. The construction costs increase linearly
with the cross section. The loss costs are proportional to the line resistance. This means
that losses are inversely proportional to the cross section of the conductor. The sum curve
shows a cost minimum at qopt, the optimal conductor cross section.

- 47 -
Figure 1.35: Cost components of an HVDC overhead line
1) Construction costs
2) Loss costs
3) Total costs

The optimum cross section can be determined by calculating the total costs using the
specified current, voltage, load profile and loss value as constants and q as the variable
parameter. This will in turn also yield the current density j for the conductor. When this is
done for different HVDC lines, quite similar current densities are always found as long as
the loss evaluation remains the same.

In the early days of HVDC, values of about j = 1 A / mm² (aluminum cross section) were
obtained. Today an economic current density between 0.7 and 0.9 A / mm² can be
expected, since the loss value has increased more than the construction costs. It is
therefore recommended to use

j = 0.8 A / mm²

for rough calculations. With this value, the voltage drop along the line is a fixed quantity
per km of line length. The following equation applies:

Du = r * j (8)

- 48 -
A value of 29 W mm² / km should be used for the specific resistance r of aluminum
conductor. Using the recommended value of j = 0.8 A / mm², a voltage drop per km of line
length results as follows:

Du = 23 V / km

This figure permits a quick estimation of the transmission efficiency of an HVDC


overhead line as follows

Du * L
hL = 1- (9)
U dN

This equation depends upon the preceding determinations and also from the line length L
in km and the nominal voltage of the HVDC (UdN is measured at the beginning of the line
on the rectifier side.) The cost formula (4) and the power resistance per pole:

r*L
RL = (10)
q

gives us all the elements needed to determine construction costs and loss costs for an
HVDC overhead line of given transmission power for various combinations of Id, Ud. The
value of Ud, with which the minimum total costs are derived, could be designated as the
optimal transmission voltage.

However, we need to consider two additional parameters:

- The permissible maximum conductor surface field strength (surface gradient).


- The degree to which station costs depend on voltage.

- 49 -
Radio interference caused by the line, corona losses and ion current depend upon the
surface gradient (the electrical field strength on the surface of the conductor), or in the case
of bundle conductors, at the surface of the equivalent substitute conductor. The surface
gradient can be reduced by dividing the conductor into a multiple conductor bundle. There
is a voltage for each conductor cross section, at which the field strength limit is attained.
This limit is well within the range of present transmission voltages. A further increase of
voltage is only possible by enlarging the cross section beyond qopt. The increase in
construction costs will be partly compensated by the simultaneously reduced loss costs.
(The phenomenon of surface gradient and its relationship to conductor shape is discussed
in greater detail in Chapter 3.1).

Station costs, as will be shown below, are also dependent upon voltage. This must be
considered in determining the optimal transmission voltage. The relationship can be
expressed in words as follows:

- Overhead lines require the highest possible transmission voltage


- The maximum allowable surface gradient places an upper limit to the voltage.
- Station costs pull the optimum voltage downward especially in the case of relatively
short lines. In the case of very long lines, station costs are of less significance.

Figure 1.36 reflects the relationships described above. Curve 1 shows the voltage
dependency of station costs. In curve 2, showing the total line costs, the influence of the
maximum surface gradient has been taken into consideration.

- 50 -
This can be seen from the steep upward slope of the line costs 2 beyond the optimal line
voltage (compare curve 3 in figure 1.35). Curve 3 presents the total costs of the HVDC
system which is at its lowest well below the optimum line voltage.

The relationship of curves 2 & 1 to each other is, to a large degree, dependent upon line
length. In the case of line lengths of many thousands of km, the cost influence of the
stations becomes trivial. On the other hand, with lengths below 1,000 km, the influence is
rather strong as is represented in figure 1.36.

Figure 1.36: Costs of an HVDC system with overhead lines in relation to voltage
1) Station costs
2) Line costs
3) Total costs

1.1.5.3 Station Costs

As we have seen, station costs and their relationship to transmission voltage must be
known to determine the optimal HVDC voltage. These costs are also used to carry out the
comparison discussed in section 1.1.5.4 between HVDC and HVAC transmission.

In the 1960's, the HVDC Working Group [Arbeitsgemeinschaft HGÜ] conducted basic
studies for the determination of a cost formula for station costs and their relationship to the
most important parameters. For the "typical" HVDC station with 1,600 MW nominal
power at ±400 kV, 2,000 A, specific costs of 160 DM/kW were determined. The cost
included two converter stations for an HVDC point-to-point connection, including ac
switchgear, buildings, and erection costs. Land costs were not included.

- 51 -
At that time it was not necessary to apply an estimated annual rate of change to the station
costs as it was in the case of overhead line costs. Station costs remained constant over two
decades! The general increase in costs for the "conventional" components of the HVDC
station have been offset by drops in costs for the innovative thyristor valves.

Since 1980, however, the HVDC station costs have also been subject to steady increase.
But this increase is well below that of the overhead lines. For rough calculations we can
use 2.5% per annum. This results in a specific price of the "typical" HVDC station based
on 1990 cost as follows:

kst = 205 DM / kW

Of the many factors which influence the cost of an HVDC station and cause a change in
the specific costs, the three most important are listed here:

- Station Power P
- Transmission Voltage Ud
- Transmission Current Id

The station power P has the strongest influence on the specific costs, but it is very
complex. While for several major components, such as transformers and smoothing
reactors, the usual growth law as given below applies:

3
1
æ P ö4 æ P ö4
K = Ko ç ÷ and k = koç o ÷
è Po ø èPø

There are completely different growth laws for valves, filter circuits, ac switchgears, etc.

Efforts by international committees to find a growth law for HVDC stations have come to
the conclusion that the "natural" growth law applies, with adequate accuracy, also for
complete HVDC stations.

- 52 -
Power dependency of the specific station costs can therefore be computed with the
following equation:

Po
k st = k sto 4 [DM / kW] (11)
P

where ksto = 205 DM / kW


Po = 1600 MW

The dependency of station costs on voltage was discussed in the last section with reference
to its significance for the computation of optimal transmission voltage. Its determination is
just as complex as that of dependency upon power. As a rule, there is only a slight
dependency of costs on voltage in the range up to about 500 kV, at which point the costs
increase steeply. The costs of the valve building contribute significantly to this. The
following formula for the dependency of specific station costs on voltage was found
empirically as follows:

4
æ U dN ö
Fu = 0.985 + 0.015ç ÷ (12)
è 400 ø

where UdN = nominal transmission voltage in kV

In figure 1.37, the dependency of station costs on voltage is presented graphically. The
steep increase in costs in the range above 500 kV is clearly shown (the application of the
curve at Ud greater than 600 kV is speculative.)

Figure 1.37: Dependency of station costs on voltage

- 53 -
The dependency of station costs on current was discussed in section 1.1.5.1. It is based on
the relatively minor influence of thyristor size on valve costs. The smaller the transmission
current, the higher the specific costs of the valve. The influence of current on the specific
costs of the entire station cannot be neglected. An empirical formula was also determined
for the dependency of specific station costs on current as follows:

2
æ 2000 ö
Fi = 0.96 + 0.04ç ÷ (13)
è I dN ø

where IdN = nominal transmission current in A

In figure 1.38, the dependency of station costs on current is graphically depicted. The steep
increase of costs below 1,500 A is clearly evident. Above 2,500 A, the dependency of
station costs on current is slight. Note, however, that transmission currents > 2000 A will
only occur in long-distant transmissions of high nominal power.

Figure 1.38: Dependency of station costs on current

When the influences of the three most important parameters are combined, the following
formula for station costs is obtained:

æ U dN ö ù é æ 2000 ö ù
2
1600 é
4

K St = P * 205*4 *ê 0.985 + 0.015ç ÷ ú*ê 0.96 + 0.04 ç ÷ ú (14)


P ë è 400 ø û ë è I dN ø û

- 54 -
Loss costs are also a factor of total costs of the stations. It has been shown that the losses
per station are in the range of 0.6% to 0.8% of nominal power, regardless of station rating.
This may well be attributable to the fact that the somewhat lower efficiency of
transformers of smaller size is compensated for by the higher efficiency of valves with
over-sized thyristors. It is therefore recommended that a station efficiency of

hst = 0.986 (15)

will be used, with this value encompassing both stations.

We have now presented all elements necessary to determine the optimum transmission
voltage for an HVDC system of a given power and transmission distance (namely, that
voltage at which overall costs are minimized), and to allow the comparison with an ac
transmission system. The results allow to estimate the chances of utilization an HVDC
system.

For more precise tasks such as the determination of a budget price, the cost formulas
derived above are not sufficient since very important cost factors, such as the ac supply
voltage, required degree of compensation for reactive power, etc., are not taken into
consideration.

1.1.5.4 Comparison of HVDC/HVAC transmission

If, for a given transmission task, the overall costs of an HVDC system and the cost of an
HVAC system are determined, a clear statement may be made of which is the more
economical solution.

If comparisons are made using various transmission distances, a diagram such as is


presented in figure 1.39 is obtained. The costs at distance zero are the station costs, which
are naturally much higher for the HVDC system. Due to the considerably smaller line
costs per kilometer, however, the dc curve climbs at a much flatter slope than does the ac
curve. In addition, ac lines of relatively long distance require intermediate substations for
the compensation of reactive power, which must also be taken into consideration in overall
costs.

- 55 -
Figure 1.39: Cost comparison HVAC/HVDC.

The so-called break-even distance at which direct current becomes more economical than
alternating current depends upon transmission power, the cost elements in a particular
country, the evaluation of losses, and interest rates. One can expect to find the break-even
distance between 500 km and 1,000 km. Therefore whenever energy is to be transmitted
over a distance of more than 500 km, a precise evaluation of economic feasibility is
required.

Of course the cost comparison is not the only deciding criterion for one of the two
technologies. The technological characteristics of the HVDC system described in section
1.1.3 can present important arguments, although there are no objective measures for their
evaluation.

Another criterion may be the space requirements. The HVDC stations compare very
poorly. This is primarily attributable to the space requirements of filter circuits and their
switch yards. On the other hand, the HVDC system has decisive advantages in land
requirements of the overhead lines and in the lesser affect on the landscape of the line
rights of way.

The electromagnetic influence upon the surrounding area in the form of radio interference
and harmonics is also important. While only a few disturbances are emitted from ac sub
stations, costly measures must often be used in the case of an HVDC station to comply
with legal or specified EMC (electro-magnetic compatibility) guidelines. This situation is
different with the overhead lines. The radio interference of an HVDC line at a given
decibel value are experienced subjectively as less disturbing than the rhythm of the double
network frequency of pulsing interference emitted by an ac line. In addition, there is the
remarkable phenomenon that in bad weather, radio interference emitted by an ac line
increases, while that of an HVDC line decreases.

- 56 -
Telephone interference emitted by an HVDC line due to converter harmonics must be
considered a disadvantage of HVDC transmission, to the extent they are not suppressed by
dc filter circuits.

Transformers are significant sources of acoustic disturbances. In an HVDC station,


smoothing reactors and filter circuits as well as ventilators of valve cooling plants
represent additional sources of noise. In some circumstances this makes expensive sound
control measures necessary.

In recent years, the influence on and possible damage to human health through electric and
magnetic fields is a high-profile complex of problems. Whatever the actual danger that
may be identified in the future, direct fields of an HVDC line are undoubtedly far less
critical than the alternating fields of an HVAC line.

The ions which can be detected below HVDC lines as ion currents caused by corona
appear to be a specific HVDC problem. They can result in the charging of electrically
insulated metal parts. An actual danger, however, is hard to imagine, and to date no limits
have been established.

Finally it should be pointed out that a cost comparison between different types of HVDC
application should also be considered. Figure 1.40 shows comparisons using the example
of a power plant injection via HVDC.

Figure 1.40: Cost comparison of various configurations of a power plant injection via
HVDC.

- 57 -
1.1.6 HVDC Systems in operation or under construction

Figure 1.41 shows a list of HVDC plants which are in operation (or have been, as in the
case of Cahora Bassa) and those which are under construction. It reflects the status as of
1989.

Year of Power DC Line &


Project Name Supplier Com- Rating Voltage Cable Location
mission [MW] [kV] [km]
Gotland 1 # ASEA 1954 20 ± 100 96 Sweden
English Channel ASEA 1961 160 ± 100 64 England-
France
Volgograd-Donbass * 1965 720 ± 400 470 Russia
Inter-Island ASEA 1965 600 ± 250 609 New Zealand
Konti-Skan 1 ASEA 1965 250 250 180 Denmark-
Sweden
Sakuma ASEA 1965 300 BtB Japan
Sardinia GEC 1) 1967 200 200 413 Italy
Vancouver 1 ASEA 1968 312 260 69 Canada
Pacific Intertie Joint Venture 1970 1440 ± 400 1362 U.S.A.
(GE, ASEA) 1982 1600
Nelson River 1 ** GEC 1) 1972 1620 ± 450 892 Canada
Kingsnorth GEC 1) 1975 640 ± 266 82 England
Gotland Extension ASEA 1970 30 ± 150 96 Sweden
EEL River General Electric 1972 320 2x80 BtB Canada
Skagerrak 1 ASEA 1976 250 250 240 Norway-
Denmark
Skagerrak 2 ASEA 1977 500 ± 250 Norway-
Denmark
Skagerrak 3 ASEA 1993 est 440 350 240 Norway-
Denmark
Vancouver 2 General Electric 1977 370 -280 77 Canada
Shin-Shinano Toshiba 1977 300 2x125 BtB Japan
1992 est 600 3x125
Square Butte General Electric 1977 500 ± 250 749 U.S.A.
David A. Hamil General Electric 1977 100 50 BtB U.S.A.
Cahora Bassa HVDC W.G. 2) 1978 1920 ± 533 1360 Moca.-S.
Africa
Nelson River 2 HVDC W.G. 2) 1978 900 ± 250 930 Canada
1985 1800 ± 500
CU ASEA 1979 1000 ± 400 710 U.S.A.
Hokkaido-Honshu Hitachi 1979 150 125 168 Japan
Hitachi 1980 300 250
1993 est 600 ± 250
Acaray Siemens 1981 50 25.6 BtB Paraguay
Vyborg Russian 1981 355 ± 85 BtB Russia
Russian 1982 710 2x170
1065 3x170
Dürnrohr HVDC W.G. 2) 1983 550 145 BtB Austria
Gotland 2 ASEA 1983 130 150 100 Sweden
Gotland 3 ASEA 1987 260 ± 150 103 Sweden
Eddy County General Electric 1983 200 82 BtB U.S.A.

- 58 -
Year of Power DC Line &
Project Name Supplier Com- Rating Voltage Cable Location
mission [MW] [kV] [km]
Chateauguay HVDC W.G. 2) 1984 1000 2x140 BtB Canada
Oklaunion General Electric 1984 200 82 BtB U.S.A.
Itaipu 1 ASEA 1984 1575 ± 300 785 Brazil
ASEA 1985 2383
ASEA 1986 3150 ± 600
Inga-Shaba ASEA 1982 560 ± 500 1700 Zaire
Pacific Intertie Upgrade ASEA 1984 2000 ± 500 1362 U.S.A.
Blackwater Brown Boveri 1985 200 57 BtB U.S.A.
Highgate ASEA 1985 200 ± 56 BtB U.S.A.
Madawaska General Electric 1985 350 140 BtB Canada
Miles City General Electric 1985 200 ± 82 BtB U.S.A.
Broken Hill ASEA 1986 40 2x±8.33 BtB Australia
Intermountain ASEA 1986 1920 ± 500 784 U.S.A.
Cross - Channel 1986 1000 ± 270 72 France-
(Les Mandarins) CGEE Alsthom England
(Sellindge) GEC 1)
1986 2000 2x± 279
Des Catons-Comerford General Electric 1986 690 ± 450 172 Canada-
U.S.A.
Sacoi ## CGEE Alsthom 1986 200 200 415 Corsica-Italy
### 1992 est 300
Itaipu 2 ASEA 1987 3150 ± 600 805 Brazil
Uruguaiana Freq. Conv. Toshiba 1987 est 53.7 17.9 BtB Brazil-
Uruguay
Sidney Siemens 1988 200 55.5 BtB U.S.A.
Gezhouba-Shanghai Brown Boveri 1989 600 500 1000 China
& Siemens 1990 1200 ± 500
Konti-Skan 2 ASEA 1988 300 285 150 Sweden-
Denmark
Vindhyachal ASEA 1989 500 2x69.7 BtB India
Pacific Intertie Expansion Brown Boveri 1989 1100 ± 500 1362 U.S.A.
McNeill GEC 1) 1989 150 42 BtB Canada
Fenno-Skan ASEA 1989 500 400 200 Finland-
Sweden
Sileru-Barsoor [Independent] 1989 100 ± 100 196 India
200 ± 200
400 ± 200
Rihand-Delhi ASEA 1991 750 ± 500 910 India
1991 1500 ± 500
Radisson-Nicolet *** ASEA 1991 2000 ± 450 1500 Canada-
U.S.A.
Welch-Monticello Siemens 1995 est 300 BtB U.S.A.
1998 est 600 162
Etzenricht # Siemens 1993 est 600 160 BtB Germany-
Czech.
Vienna South-East Siemens 1992 est 550 145 BtB Austria-
Hungary
DC Hybrid Link ABB 1992 est 560 -350 617 New Zealand
Ontario East-West Tie 1000 ± 450 600 Canada
Chandrapur-Ramagundum 1995 est 1000 2x BtB India
Chandrapur-Padghe ABB 1996 est 1500 ± 500 900 India

- 59 -
Year of Power DC Line &
Project Name Supplier Com- Rating Voltage Cable Location
mission [MW] [kV] [km]
Gazuwaka-Jeypore GEC 1) 1997 est 500 BtB India
Santo Tome Freq. Conv. 50 17.3 BtB Argentina-
Brazil
Leyte-Luzun ABB 1997 est 1000 350 440 Philippines
Haenam-Cheju GEC 1) 1993 est 300 ± 180 100 South Korea
Baltic Cable Project ABB 1994 est 600 450 Sweden
Victoria-Tasmania 1995 est 300 300 Australia
Kontek ABB 1995 est 600 400 Denmark
Scotland-North Ireland 1995 est 2x125 150 62 United
Kingdom
Greece-Italy 1996 est 600 400 Italy
Sweden-Poland 1999 est 800 500 245 Sweden
Tian-Guang Siemens 1999 est 1800 500 960 China
Thailand-Malaysia Siemens 1999 300 300 110 Thailand-
Malaysia
Three Gorges I 2002 3000 500 China
Three Gorges II 2005 3000 China
Talcher-Bangalore 2001 2000 500 India
Ib-Valley-Jaipur 2002 3000 500 India
Sassaram 2001 500 BtB India

* 2 valve groups replaced with thyristors in 1977


** 2 valve groups in Pole 1 to be replaced with thyristors by GEC by 1991
*** multi-terminal system larges: terminal is rated 2250 MW

# Retired from service


## 50 MW thyristor tap
### Uprate with thyristor valves

1) Formerly English Electric


2) AEG, BBC, Siemens

Figure 1.41: HVDC systems in operation or under construction or in planning.

- 60 -
1.2 The Main Components of the HVDC Station

The significance of the converter station for the economic feasibility of an HVDC system
was discussed in section 1.1.5. The HVDC station includes components with greatly
differing functions. Only in combination do these various components fulfill the task of
the rectification of three-phase current of the supply network into direct current and at the
load end of the line back into three-phase current of the receiving network. They must
perform these functions with precision and reliability and in compliance with the legal or
specified conditions at the interfaces to the ac systems and to the environment.

1.2.1 Layout of the HVDC Station

Figure 1.42 shows the idealized single line diagram of an HVDC station for bipole long-
distance transmission using overhead lines. For each station pole, there is one twelve-pulse
converter group. The transformers are configured as three-winding units. To compensate
for reactive power, a capacitor bank, a synchronous condenser and two high-pass filters
are used. The dc switchgear also includes pole bridging disconnectors, so that operation
with "metallic" return is possible in the event of loss of one station pole. A power line
carrier (PLC) connection is provided for telecommunication purposes.

This example represents a relatively simple station configuration. The design can be
significantly more complex according to the requirements placed on it.

- 61 -
Figure 1.42: Components of an HVDC station

1 AC switchgear 6 Smoothing reactors


2 AC filter circuits and reactive power sources 7 DC switchgear
3 Converter transformers 8 DC filter circuits
4 Converter valves 9 PLC system
5 Control and protection 10 Earth Electrode

In Chapter 2, the major components of an HVDC station as presented here will be


discussed in detail. At this point, only the following brief statement shall be made.

1.2.2 Weighting of the Major Components

For many considerations, it is important to know the breakdown of overall costs - both
construction costs and losses - of an HVDC station with respect to its major components.
Figure 1.43 shows the breakdown in bar graph form of both the costs and the losses. One
must keep in mind the statement made in section 1.1.5.4 that the major components are
subject to quite different growth laws. For example in a long distance transmission system
with very small nominal current the cost share of valves may be significantly higher than
is shown in figure 1.43a), while its share of loss may be lower than is shown in figure
1.43b). The breakdown depicted might apply for our standard HVDC system with ±400
kV, 2,000 A, but it must always be critically examined.

- 62 -
Figure 1.43: Breakdown of costs (a) and of losses (b) in HVDC stations.
1 Other system components
2 AC switchgear
3 Reactive power-compensation and Filter circuits
4 Converter transformer
5 Converter valves

1.2.3 Degree of Freedom in Project Planning

The determination of optimal transmission voltage discussed in section 1.1.5 is in general


limited to the pre-project phase. In the case of a specific invitation for bids, in most
instances the main data of transmission including overhead line or cable sizing are
specified. Nevertheless, there remains a sufficient degree of freedom for the project
planning engineer to achieve an optimal design of an HVDC station. Several points are
listed here:

- In the case of converter transformers, the question of three-phase or one-phase units


and two-winding or three-winding configuration must be considered. If the transport
weight is not exceeded, the three-phase, three-winding configuration will be the most
favourable. However, if a reserve transformer is to be supplied either due to
specifications or because availability analysis has resulted in the necessity of a spare
transformer - a single phase configuration is usually the more economical. Sometimes
connection of the ac filter circuits to an intermediate voltage tertiary winding must be
considered. This is particularly true for a high network voltage (for example, 525 kV).
The necessity of a large number of filter circuits and capacitor banks would make the
"classic" approach of the busbar connection not only too costly but also very expensive
in terms of space requirement.

- 63 -
- The design of the smoothing reactor is usually left to the bidder. The functions of a
smoothing reactor and the resulting guidelines for its rating will be discussed in
Chapter 2.5. A decision must be made for the design as air insulated or oil insulated
units independent of the size. In making this decision, contamination or noise problems
can be factors.

- In the area of valves, a broad range of structures is theoretically available but usually
the standard configuration of the supplier will be applied. For long-distance
transmission with low nominal current, the selection of the most suitable type of
thyristors might be considered. If a high loss valuation is specified, the use of the
largest available thyristors will make sense because the associated marginal cost
increase is more than offset by the reduced valve losses. In this situation, smaller valve
cooling plants or higher permissible cooling medium entry temperature can be
important considerations.

- In the area of filter circuits, there is a wide range of configuration alternatives. Often,
however, the specified distortion limits and the tolerance parameters specified for the
reactive power balance are so critical that it is difficult to find even one solution which
meets the requirements. (For more detail see Chapter 2.2).

- Reactive power compensation is not a problem in the case of a "strong" network, that is
when there is a high SCR (short circuit power ratio S"k/PdN) at the connection point of
the HVDC station. In the case of a "weak" network, it can present a difficult project
planning task with respect to voltage maintenance and even voltage stability. In this
case, the use of static compensators or synchronous condensers may be considered.

- 64 -
1.3 The Function of the HVDC Converter

The technical characteristics of HVDC presented in Chapter 1.1.3 are always an important
criterion for the decision to use HVDC for a given transmission task. In many cases, they
may even be the decisive criterion. The converter is functioning with practically no delay.
It not only fulfills the basic function of energy conversion reliably and economically, but
also allows the specification of a large number of superimposed control and regulation
functions. These can be simple functions or complex support functions for HVDC-linked
asynchronous networks or synchronously-linked portions of networks.

In order to enable the project planning engineer to make optimal use of the features of
converters and to avoid the implementation of contradictory functions which cannot be
fulfilled simultanously, he must be completely familiar with the functions of the converter.
For this reason, converter theory shall be considered at this point.

Our considerations are restricted on the line-commutated converter which sofar has
exclusively been used for HVDC systems. Althrough forced-commutated converters have
occasionally been proposed for very special applications such as tapping of an HVDC line,
no such scheme is in operation or under construction up to now. This statement holds true
also for the so-called „Capacitive Commutated Converter“ which has been propagated
very recently and which is said to be best suited for inverter stations connected to an
extremely weak ac system node and for those, forming part of an extremely long
submarine cable connection.

- 65 -
1.3.1 The Three-Pulse Commutation Group

Of the numerous converter configurations which have come into use for a wide variety of
applications, HVDC technology uses exclusively the three-phase bridge circuit presented
in Fig. 1.44. In many respects this is the optimal converter connection.

Figure 1.44: Three-phase bridge circuit

The three-phase bridge consists of two three-pulse commutation groups connected in


series with valves 1-3-5 and 2-4-6, respectively. To understand the function of the three-
phase bridge, it is only necessary to consider one commutation group, shown in Fig. 1.45
with the same valve designations as in Fig. 1.44. The star-point loading is of no
consequence, since it disappears when the circuit is expanded into a six-pulse bridge.

- 66 -
The starpoint loading has been taken into consideration by means of delta connection on
the primary side. A symmetrical ac network with no impedance and with sinusoidal
voltage UL is assumed, as is generally customary in converter theory. In addition, a
completely smooth direct current (Id) effected by a smoothing reactor with infinite
inductance (Ld), is also assumed.

Figure 1.45: Converter with three-pulse mid-point connection.

1.3.2 Voltage Formation

Fig. 1.46 shows the ac voltage system of the three-pulse converter. For the present, a
converter transformer without leakage reactances and diode valves are assumed.

Figure 1.46: Voltage formation of the idealized three-pulse converter

The transfer of current from one phase to the following one always occurs precisely at the
intersection point of the voltages. The terminal direct voltage follows the curve indicated
in Fig. 1.46 as a double thick line. The idealized direct voltage Udi which is obtained in
this manner (the shaded area in Fig. 1.46) can be calculated by the following integration:

- 67 -
+ p3
3
U di =
2* p
* u$ ph ò cos wt dwt
- p3

3 p
U di = * u$ ph * 2 * sin
2* p 3

U di = 117
. U ph

If the phase-to-phase voltage Uv is introduced, and if one takes into consideration the fact
that in a three-phase bridge circuit a double portion of dc voltage is formed by means of
the series connection of two commutation groups, the following familiar relationship for
the three-phase bridge circuit will be obtained:

Udi = 1.35Uv (16)

By equipping the converter with controllable valves, i.e. thyristors, the ability to control
voltage is gained. Even if the voltage of the thyristor becomes positive, the current flow
remains blocked. Only when the control impulse is applied does the thyristor become
conductive. Up to this point in time, the direct current is carried by the preceding valve,
the phase voltage of which stands at the dc terminals. It can be seen in Fig. 1.47 that the
shaded area of the sine-shaped half-wave which contributes to the dc voltage lags behind
the valve voltage by the control angle or angle of delay a.

Figure 1.47: Formation of voltage in the controlled 3-pulse converter

- 68 -
The calculation of controlled direct voltage Uda is as follows:

+ p3 + a
3
U da =
2* p
* u$ ph ò coswt dwt
- p3 + a

3 é æp ö æp öù
U da =
2* p
* u$ ph êë sin çè 3 + a ÷ø + sin çè 3 - a ÷øúû

3 p
U da = * u$ ph * 2 * sin * cos a
2* p 3

U da = U di *cos a (17)

It is not difficult to see from equation (17) that at a = 90°, the direct voltage Uda becomes
zero (zero voltage operation) and at further increases of a, it becomes negative (inverter
operation). Fig. 1.48 shows the voltage formation during inverter operation. The latest
point in time at which triggering must occur is just before the intersection of the voltages
of the current-carrying valve and the following valve, the difference being the angle of
advance b.

b = 180 - a

Figure 1.48: Voltage formation in inverter operation

The minimum value for b will be presented later.

- 69 -
A converter equipped with controllable valves can, without changing the current direction,
change its direct voltage continuously from a maximum positive value (a = zero), through
zero to a maximum negative value (b = bmin). This can simply be done through the
selection of the firing angle a.

NOTE: The use of valves which can be switched off (GTO) and the use of forced
commutation is not treated here since neither of these have been applied until
today in HVDC systems.

1.3.3 The Commutation Process

Up to now, the assumption has been made that there is a sudden transition of current from
one valve to the following valve. In actuality, this commutation of direct current requires a
certain amount of time. This is due to leakage inductances of the converter transformer,
which only permit a current change of limited steepness, di/dt. Thus for a short period of
time, the releasing and the receiving phases are carrying current simultaneously. This is
referred to as commutation overlap and its duration is defined as the overlap angle u. Fig.
1.49 again shows a three-pulse converter. But in this case, reactances are present. The
leakage inductances Ls are considered lumped elements on the valve side of the
transformer.

Figure 1.49: Three-pulse converter with transformer having reactances

Assume that valve 1 carries the direct current and that at an arbitrary point in time (after
the voltage intersection) valve 3 receives a control impulse. A current loop will be created
with Uv as the driving voltage. The leakage inductances of phases 1 and 3 will be the
reactances which determine the current. This is simply

- 70 -
a phase to phase short-circuit of the transformer. The short-circuit current flows through
valve 3 in the forward direction and through valve 1 counter to the forward direction. In
valve 1, it is superimposed on the flowing direct current. As soon as the short-circuit
current has achieved the amplitude of the direct current (the composite current is zero),
valve 1 extinguishes. At this point, the commutation process has ended and valve 3 is
carrying the entire direct current. Fig. 1.50 shows this commutation process.

Figure 1.50: The commutation process in rectifier operation


u = Overlap angle

It can be seen that the curve of the direct voltage during commutation is along the average
value of the voltages of valves 1 and 3 as depicted in the figure.

This is understandable since the voltage drop associated with the short-circuit alternating
current occurs due to leakage inductances of both phases. The shaded voltage time area in
Fig. 1.50 is apparently lost with respect to the formation of direct voltage. For this reason,
we speak of the inductive direct voltage change Dx. In relation to Udi, we speak of the
relative inductive direct voltage change dx.

The following equatation is the starting point for the calculation of Dx:

òu dt = L * DI (18)

According to this equation, a specific voltage time area is necessary for a current change
DI in an inductance L. DI in our case is the direct current which is to be commutated.

- 71 -
The larger D I is, the longer the commutation process takes (at a given voltage). The same
effect occurs when the valve voltage Uv, which triggers the commutation, is reduced.

These effects are particularly important in inverter operation as can be seen in Fig. 1.51.
The commutation process has to end before the voltage intersection, so that the thyristor
will regain its blocking ability. The current must have been extinguished, thus we can
define an extinction angle g:

g = b -u

Figure 1.51: Commutation process in inverter operation

At this point, the following dilemma can be understood. The angle of advance b must be
selected or computed such that the minimum extinction angle g is always available after
the commutation process. Even when the direct current, which is to be commutated,
increases and/or commutation voltage Uv decreases, this minimum must be maintained.
The greater the tolerance selected for this purpose, the greater the safety of inverter
operation, but the smaller the direct voltage and the efficiency of the converter. As is so
often the case, the project planning engineer must weigh operation reliability against
economy.

- 72 -
For the calculation of dx, the transformer leakage inductance is also needed. This can
easily be calculated from the transformer data:

U 2vN
w Ls = u k *
SN

where uk = Transformer short-circuit voltage


UvN = valve-side transformer nominal voltage
SN = transformer nominal MVA rating

In addition, S N = 3 * U vN * I vN

where IvN = valve-side transformer nominal current

In the "ideal" three-pulse converter, Iv is a rectangular block, 120° long with Id as the
amplitude. If the "negative half-waves" triggered by the second commutation group are
considered, the following valve-side nominal current is obtained for the three-phase bridge
circuit with the r.m.s value:

2
Iv = * Id
3

Equation (16) continues to hold true for the three-phase bridge circuit.

U di
Uv = We then obtain:
135
.

S N = 1.05 * U di * I dN (19)

and furthermore:

0.524 U di
Ls = * uk * (20)
w I dN

- 73 -
The inductive direct voltage change for the 3-pulse converter is derived from equation (18)
as follows:

Dx = 3 * f * ò u dt

And for the 3-phase bridge circuit:

Dx = 6 * f * ò u dt

Dx = 6 * f * Ls *Id

When Id = IdN:

3
DxN = * 0.524 * uk * Udi
p

Thus the following is obtained for the relative inductive voltage change:

dxN = 0.5 * uk

If we remember the relationship of Id and Uv explained above, the following generalization


can be made:

Id U
dx = 0.5 * uk * * vN (21)
I dN Uv

Note that dx is not dependant on either the control angle a nor the overlap angle u.

The overlap angle u, which, as was seen in Fig. 1.51, is of decisive importance for reliable
operation of the inverter, still has to be determined.

This is solved by a relatively simple process. Consider again the "ideal" converter, which
is operated in the first instance with the control angle a and in the second instance with
control angle (a + u), as is shown in Fig. 1.52.

- 74 -
The direct voltage arising in the two cases is as follows:

Ud1 = Udi * cos a


Ud2 = Udi * cos (a + u)

Figure 1.52: Calculation of overlap angle u

The difference of the two voltages indicated by the shaded area in Fig. 1.52 obviously
corresponds to twice the value of the inductive voltage change Dx depicted in Fig. 1.50:

Ud1 - Ud2 = 2 * Dx

Through transition to the relative inductive voltage change dx:

cos a - cos (a + u) = 2dx


cos (a + u) = cos a - 2dx
u = arc cos (cos a - 2dx) - a (22)

1.3.4 The Converter in Operation

The relationships derived to this point provide the most important elements for the
calculation of converter voltage in operation. It is in fact adequate for a rough calculation
to take Dx/Id as the internal resistance of the converter station, i.e. to assume a loss-free
station. For more precise calculations, however, the change in dc voltage caused by losses
should be taken into consideration.

- 75 -
1.3.4.1 Ohmic DC Voltage Change

There are essentially 2 components which contribute to ohmic dc voltage change:

• Copper losses of the converter transformer


• Conduction losses of the valves

The copper losses of the transformer can be interpreted as resistance which causes a
linearly increasing voltage drop with increasing current. The same holds true for the
conduction losses of the valves to the extent they are caused by the "differential resistance"
of the thyristors and ohmic resistance of the valve reactors and busbars. In addition to this,
however, there is the current-independant threshold voltage of the thyristors (see Chapter
2.4). In modern thyristors, however, this amounts to only about 0.1% of the nominal
voltage. It is therefore recommended that the conduction losses of the valves be interpreted
as ohmic resistance. This results in the following equation:

Vtrcu + Vv
Dr =
Id

or the relative ohmic dc voltage change;

Vtrcu + Vv Id
dr = * (23)
I dN * U di I dN

In this case, the dependency of Uv becomes non-applicable. In the preliminary project


planning stage, the losses are generally unknown. It appears acceptable, however, to use
the following overall value for HVDC stations:

Id
dr = 0.005 *
I dN

- 76 -
1.3.4.2 DC Voltage

From Equations (16), (17), (21,) and (23), the following relationship is derived for the
terminal dc voltage of a converter in three-phase bridge connection, functioning in rectifier
mode:

UdGR = Udi(cos a - dx - dr) (24)

where dx and dr are linearly proportional to dc current. If a and Uv are held constant, a
terminal voltage is obtained which decreases in a linear manner with load current.

For the converter in inverter mode, a terminal voltage which increases with load current
would be expected. The inverter side produces a countervoltage to the dc current
impressed by the rectifier. That will be indeed the case, if the firing angle of the inverter,
the angle of advance b, is held constant. The following relation results:

UdWR = Udi (cosb + dx + dr) (25)

If b and Uv are held constant, a dc voltage which increases linearly with the load current
will be obtained at the terminals of the inverter.

It has become customary, however, to express the inverter dc voltage as a function of the
extinction angle g. This is useful because the extinction angle is usually held constant in
the operation of an HVDC inverter station.

If Fig. 1.52 is considered again, it will become clear that there is a difference of 2*Dx
between Udi* cos a and Udi*cos(a +u), or in relation to Udi, a difference of 2*dx. This
holds true analogously for the inverter with Udi * cos g and with Udi * cos(g + u), where (g
+u) = b, as can be seen from Fig. 1.51.

UdWR = Udi(cos g - dx + dr) (26)

Since dx >>dr (usually by more than a factor of 10), equation (26) indicates that when the
angle of extinction g is held constant, the terminal dc voltage declines with the load (Fig.
1.53c). This means nothing less than that the inverter has a negative internal resistance. In
most cases this will have no significance, particularly if a long line with a relatively high
ohmic resistance connects rectifier and inverter. In relatively short lines and "weak"
networks on the inverter side, this effect can lead to problems of instability in the
operation of the HVDC system, forcing the replacement of the control function g = const

- 77 -
by another one (such as UdWR = const), with which the inverter does not demonstrate a
negative internal resistance.

Figure 1.53: Characteristics of the HVDC converters


a) Rectifier with a = const
b) Inverter with b = const
c) Inverter with g = const

- 78 -
1.3.4.3 The Rectifier in Short Circuit

One may wonder whether the characteristic of the rectifier presented in Fig. 1.53a) will
continue to define a linear curve in the overload range, even in short-circuit condition, that
is, when the direct voltage is zero. Such is, indeed, not the case. When it is kept in mind
that the internal resistance is essentially determined by Dx, it becomes clear that it can be
constant only for normal commutation. The commutation overlap u is very dependent
upon Dx, and thus upon the load current. If the load current has risen to such an extent that
u = 60° el - valid for the three-phase bridge circuit - then the existing short circuit of two
transformer phases takes the voltage necessary for commutation of the next valve and
current transfer is delayed until the preceding commutation is terminated. This is referred
to "spontaneous current delay", area II (Figure 1.54) of the rectifier operation. In this
range, initiation of current is delayed to such an extent, that a is effectively enlarged such
that u is always equal to 60°. The direct voltage drops sharply during the spontaneous
current delay. Only when the current has risen further and the "spontaneous" control angle
has reached 30°, does "multiple commutation" take place. That is, commutation takes
place in both commutation groups at the same time. In range III of the characteristic
(Figure 1.54), the voltage drops again in a linear manner, but with three times the slope as
in area I, and this continues to the short circuit point k. There is a very effective limitation
of short circuit current solely through the commutation processes in the converter. In
normal operation, that is of little significance, since a rectifier works with current control
or at least with current limitation. But this effect can be quite helpful for transient
conditions (see Fig. 1.54.).

1.3.5 The Twelve-Pulse Converter Group

In discussing converter theory, the three-pulse commutation group has been assumed but
the rating equations have been halted at the three-phase bridge. The twelve-pulse converter
group consisting of two three-phase bridges connected in series, fed by two alternating
voltages with phase displacement of 30°, is used exclusively in HVDC technology.

If the two bridges have a common inductance in their individual commutation loop, the
two commutations no longer will run independent of each other when u = 30° el is
attained. A higher voltage drop occurs as is also suggested in Fig. 1.54. A drop of this kind
is imminent when the two three-phase bridges of the twelve-pulse group are fed by a
three-winding transformer, a usual design today. Therefore a good converter is designed in
such a manner that the overlap angle remains u<30° el up to the specified overload. The
means for accomplishing this is the transformer impedance, since it determines Dx and

- 79 -
also determines in large degree the commutation overlap u, as can be seen from equation
(22.)

Figure 1.54: Characteristic voltage curves of the rectifier in overload range.

Aside from this special high current situation, the relationships derived to this point for dc
voltage calculation continue to be fully valid for the twelve pulse group, the resulting
direct voltage only has to be doubled.

- 80 -
Fig. 1.55 shows the three converter arrangements for comparison. The three-pulse
commutation group can be recognized as the fundamental element in all three
arrangements. It would appear appropriate to combine the three valves of one
commutation group to one construction unit. With introduction of thyristor technology in
HVDC systems, however, it has been customary to combine the valves belonging to the
same ac phase to one construction unit.

Figure 1.55: Converter Arrangements


a) Three pulse MP circuit
b) Three phase bridge circuit
c) Twelve pulse group

Fig. 1.55b) indicates a double valve as is in use at Cahora Bassa. Fig. 1.55c) shows the
arrangement which is by far the most common one. The twelve-pulse group is then
composed of three quadruple valve units. The advantage of this arrangement becomes
especially important with the use of single phase three-winding converter transformers.
The three phases can be placed side by side. The valve units can either be set on the floor
of the valve hall or they can be suspended - electrically insulated and damped against
mechanical vibrations - from the roof support structures of the valve hall, as for example
in the GESHA project.

- 81 -
In back-to-back links, other valve arrangements can also be advantageous, such as the
grouping of the six valves of a three-phase bridge, as in the Vienna - Southeast system
(this arrangement promises to be particularly advantageous with three-phase converter
transformers), or the eight element valve, which combines rectifier and inverter valves of
one phase (this arrangement has extremely small space requirements and is being used, for
example, in the Highgate back-to-back link).

- 82 -
1.4 Reactive Power Demand of the Converter

It has been emphasized that no reactive power is needed with energy transmission by
HVDC. This statement refers to the line itself and is of particular importance with very
long overhead lines and with cable transmissions. On the other hand, line commutated
converters, as used in HVDC systems, require for their functioning reactive power which
they draw from the ac network.

The following figures show schematically the development of the fundamental frequency
reactive power of a converter. From top to bottom of each figure, voltage formation and
the conduction of current are depicted for a converter in three-phase bridge connection.

1.4.1 The Uncontrolled, Ideal Converter

The uncontrolled converter which is free of ac side inductances will have the following
values:

a=0 u=0

The commutation of the direct current takes place in the instant of the intersection of the
valve voltages. The fundamental frequency valve current I1 is in phase with the associated
valve voltage. This can be seen in valve pairs 1 and 4 in Fig. 1.56. The phase angle j is 0,
so the fundamental frequency reactive power Q is also 0.

- 83 -
Figure 1.56: Voltage formation and conduction of current in the ideal converter
a=0 u=0
j=0 Q=0

For the sake of completeness, it is noted that a distortion reactive power can also be
defined which is the product of the sine-shaped voltage and the geometric sum of all
harmonic currents. This concept is not customary in HVDC technology, it is not needed to
understand the subject and will not be used further.

1.4.2 Control Reactive Power

Fig. 1.57 shows voltage formation and conduction of current in the controlled converter
which is free of ac side inductances. It will have the following conditions:

a¹0 u=0

- 84 -
It can be seen that the fundamental frequency current I1 is now no longer in phase with the
associated valve voltage. The phase angle j is equal to the control or current delay angle
a. A fundamental frequency reactive power can thus be defined, which is referred to as
control reactive power Qst.

Figure 1.57: Voltage formation and conduction of current in the controlled converter
a¹0 u=0
j¹0 Q = Qst

This relationship shows why it is so important in HVDC systems to operate with the
smallest possible control angles a and b. For this purpose, the inverter may be operated
with extinction angle control, which will insure that g equals gmin. But it is also quite
possible to influence the reactive power demand of an HVDC station by operating with
larger than necessary control angles, either dynamicly or in steady state.

- 85 -
1.4.3 Commutation Reactive Power

Fig. 1.58 shows voltage formation and conduction of current in converters which are
uncontrolled but which have inductance. For this more realistice case, the following
conditions apply:

Figure 1.58: Voltage formation and conduction of current by the diode rectifier
a = 0. u ¹ 0, j = arc cos(1-dx)
Q = Qkomm

In this case, the current commutation does not take place abruptly. It follows a (1 - cos)
curve. As has already been seen, this is a segment of the current from the two-phase short-
circuited transformer. These current blocks, which have an almost trapezoid shape, are
characterized by a phase lag of the fundamental current I1, represented by a thin line. A
cosine j can be specified and the term "commutation reactive power" Qkomm is used. Since
the transformer short-circuit current is determined by the value uk, it is also obvious that
the overlap angle u and also the commutation reactive power are functions of uk and
therefore of dx.

- 86 -
One of the most important parameters for the optimization of the HVDC station is the
transformer short-circuit voltage. The reactive power demand of the converter is of course
only one aspect, but because of the expensive compensation equipment, it is a very
essential one.

1.4.4 Fundamental Frequency Reactive Power

Fig. 1.59 shows voltage formation and current conduction in the phase controlled
converter in three-phase bridge connection with commutation inductance.

Figure 1.59: Voltage formation and conduction of current in the real converter
a ¹ 0, u ¹ 0, j = arc cos(a - dx)
Q = Qst + Qkomm

- 87 -
The fundamental frequency reactive power is composed of control and commutation
reactive power.

The curves show how both, the current delay angle a and the commutation overlap u
contribute to the total fundamental frequency reactive power, which is represented by the
angle j.

In very general terms the following relationship holds true:

Q = P * tan j

This yields the equation for determining the fundamental frequency reactive power of an
HVDC station as follows:

Q = P * tan[arc cos(cos a - dx)] (27)

1.4.5 Reactive Power as a Function of Load

The curve of reactive power demand of an HVDC station with changing active power P
can be calculated from equation (27). The dependency from the direct current is mainly
determined by dx and from the direct voltage by the delay angle a.

In Fig. 1.60, the reactive power demand of a converter is presented under three different
control methods.

Figure 1.60: Reactive power demand of an HVDC converter

- 88 -
If the terminal dc voltage Ud and the firing angle a (or the extinction angle g of an
inverter) are held constant, curve (1) will be obtained. If, however, Uv is held constant (Udi
= const regulation), a linear curve such as (2) will be obtained. The power of a converter
can also be changed even when the (nominal) current is held constant by varying the dc
voltage. Curve (3) shows the reactive power demand for this control method. It is
important to note that the entire area between curves (1) and (3) is available for reactive
power control. Each point within this area can be set by the selection of firing angles a and
b (or g).

The relationships can be seen in the P-Q diagram (Fig. 1.61) which will be obtained if the
active power P and the fundamental frequency reactive power Q are given in relation to
the fundamental apparent power of the converter

S1N = Udi * IdN

Figure 1.61: P-Q-diagram of an HVDC converter

- 89 -
The equations derived in Chapter 1.3 for the formation of the converter voltage can be
seen again in this diagram.

The fundamental problem of the inverter is also clear in Fig. 1.61. Although it feeds active
power into the receiving ac network (-P), it draws reactive power from the ac network
(+Q)!

1.4.6 Reactive Power Control

The possibility of electronic reactive power control as demonstrated in the preceding


section is used only to a very limited degree in HVDC technology. This is due to
economic reasons. Both, control reactive power and commutation reactive power are
increased by the reduction of the dc voltage and the corresponding increase of current.
However, load losses increase with the square of the current. For this reason, application is
limited to the light loads where the necessary filter circuits produce a considerable
overcompensation for the reactive power required by the converter.

Fig. 1.62 depicts the reactive power control of the Dürnrohr HVDC link. In this system, a
compensation to 0 ±60 Mvar was specified. Compliance with the Q limit is achieved by
load-dependent switching of a capacitor bank and of one of the two high-pass filters.
Electronic reactive power control is used only in the light load range. It can be seen that
there is a difference between the connect and disconnect points of the reactive power
elements. This provides a "switching hysteresis" which prevents too many switching
operations or even a "pumping".

- 90 -
Figure 1.62: Reactive power compensation and control of an HVDC back-to-back link

In place of the described reactive power open loop control, it is also possible to utilize
reactive power closed loop control whereby a switching is always effected when the upper
or lower limit value of the measured total reactive power is reached. In this case, the
characteristic of the electronic reactive power regulation would be a horizontal line along
the lower limit value.

- 91 -
1.5 AC-Side Harmonics

Converter valves connect cyclically the dc terminals with the ac terminals of a converter
circuit. The alternating currents on the ac side are therefore composed of sections of the
current on the dc side. The direct current is regulated by the rectifier; therefore the currents
flowing on the ac side are also controlled by the rectifier. The converter acts as a current
source.

1.5.1 Line Current of the Wye-Connected Bridge

As established earlier, a very good smoothing of the direct current is assumed as shown in
Fig. 1.63. If the use of the ideal converter is assumed, the alternating currents of the
converter will be rectangular blocks. The amplitudes are equal to the amplitude of the
direct current. If it is assumed that the converter transformer is in a wye/wye configuration,
these currents, converted by the turn ratio will represent the primary currents IL.

When the current waves are analyzed it will be found that in addition to the fundamental
frequency current IL1 in Fig. 1.63, only odd-numbered harmonics are produced by a 6-
pulse converter. These odd-numbered harmonics have the ordinal number

n=k*p±1

where p is the pulse number of the converter (in this case 6) and k is an integer.

- 92 -
For harmonic number n, the following equation holds true:

1
In0 = * IL1 (28)
n

This means that the amplitudes of the harmonic currents decrease sharply with increasing
ordinal number.

Figure 1.63: Direct current Id and line current IL of the ideal converter in three-phase
bridge connection, transformer in wye/wye connection.

- 93 -
1.5.2 Line Current of the Delta-Connected Bridge

Fig. 1.64 shows the relationships of a 6-pulse converter when the transformer is connected
in wye/delta. The shape of the line current IL is completely different from the shape in Fig.
1.63. In the delta circuit, the valve circuit splits and two-thirds of it flows through the
winding which lies directly between two leading valves and one-third through the other
two series-connected windings. Surprisingly the transformer primary side harmonics are
exactly the same. Even their amplitudes are the same as in the case with a converter
transformer in wye/wye connection. However, the harmonic currents of the fifth, seventh,
seventeenth, nineteenth, etc. harmonics are 180° out of phase with those in the wye/wye
circuit while the other harmonics are in phase.

Figure 1.64: Line current of the ideal 6-pulse converter with transformer in wye/delta
connection

- 94 -
1.5.3 Line Current of the 12-pulse Group

If two 6-pulse converters, one with a wye-wye connected transformer, the other one with a
wye-delta connected transformer, are combined (by placing them in series on the dc side,
for example) the out-of-phase harmonics cancel. The result is a 12-pulse converter system
which has only harmonic currents corresponding to p = 12 on the primary side.

This characteristic is very important for HVDC converters because a series of harmonic
currents is removed from the ac networks. For this reason, 12-pulse converter groups are
used exclusively in newer HVDC systems.

Fig. 1.65 shows one example in which the two converter bridges are fed by a 3-winding
transformer with the two valve windings in different connection.

Figure 1.65: Line current of the ideal 12-pulse converter

- 95 -
1.5.4 The Influence of Control Angle and Overlap

The influence of control and the commutation processes on the harmonic currents must
still be considered. This means that the real converter is being considered. It is found that
the ordinal numbers of the harmonic currents are unchanged. No additional harmonics are
generated by delay in firing and by overlap. But there are significant effects on amplitudes
of the harmonic currents. An increasing overlap (due to a higher short-circuit voltage of
the converter transformer) decreases the amplitude of the harmonic currents. An increasing
degree of voltage control (decreasing the direct voltage toward 0) causes the amplitudes of
the harmonic currents to increase.

In comparison with the theoretical value Ino as calculated by equation (28), which applies
when a = 0 and u = 0, there is a reduction in harmonic currents. A reduction factor Fn can
be specified which will make it possible to determine the harmonic currents of the "real"
converter:

In = In0 * Fn (29)

Figure 1.66 shows the reduction factor for a = 30° as a function of the overlap angle u for
the most important harmonics of a converter in 3-phase bridge connection. Corresponding
reduction factors can be indicated for any firing angle. As long as the assumptions of
converter theory defined at the beginning are retained, Fn can be described with
mathematical precision, the calculation of actual characteristic harmonic currents is not a
problem. However, because the assumptions never fully hold true, there is a limit to the
exactness. It has become customary to determine harmonic currents by curves, with the
harmonic currents In indicated as a percentage of line current and as a function of dx.
Therefore, the quantities uk, Id and Uv are all taken in consideration.

- 96 -
Figure 1.66: Reduction factor Fn for harmonic currents at a = 30° as a function of
overlap angle u.

The curves of fig. 1.67 show the most important characteristic harmonic currents on the ac
side of a six- pulse converter as function of dx with the firing angle a as parameter.

- 97 -
Figure 1.67: Harmonic currents on the ac side of a converter in 3-phase bridge
connection
dx relative inductive voltage change.
a firing angle (in inverter operation the extinction angle g should be
used in place of a)

- 98 -
1.5.5 Non-Characteristic Harmonics

The harmonics in the line current determined under the idealized conditions of converter
theory are called characteristic harmonics. Even if one of the assumptions is abandoned
(the total smoothing of the direct current, for example), no harmonics of a different ordinal
number arise. Only the amplitudes change. The fifth harmonic which occurs in the positive
sequence system becomes smaller with increasing ripple in the direct current. The seventh
harmonic in the negative sequence system, on the other hand, is increased, and so on.

There are, however, other influences in HVDC stations which give rise to harmonics of
other ordinal numbers. Such influences are:

• A negative sequence component contained in the 3-phase ac voltage system


• Deviations among the transformer reactances of the 3 phases
• Deviations of current-inception points of the valves from the equidistant 30°
• Ripple in the direct current if the HVDC system connects asynchronous networks

The first three of these influences cause noncharacteristic harmonics in the harmonic
spectrum of the line current, the latter non-harmonic oscillations.

Algorithms can be developed to calculate the influence of the network negative sequence
component and of reactance nonsymmetry. But this is not the case with deviations of the
control impulses from the equidistant 30° within a 12-pulse group, unless a certain control
impulse pattern is arbitrarily specified. This non-equidistant firing is particularly
undesirable since it can result in even-numbered harmonics and in a direct current
component. DC can produce saturation in the transformer causing undesirable effects on
magnetization current and noise generation.

The last influence listed -- interaction with the opposite station through the direct current
ripple -- may occur if asynchronous networks are linked. This results in "foreign"
frequencies and continuous fluctuations in the harmonic spectrum. The misleading
expression "interharmonics" was coined for this phenomena. It is more accurate to speak
of non-harmonic components.

What is the significance of noncharacteristic and nonharmonic components?

- 99 -
Asymmetries which give rise to non-characteristic harmonics are, for the most part, very
slight. The deviations of the control impulses from the equidistant 30° are usually less than
0.1 degrees. The smoothing of the direct current is generally very good (exception: some
of the existing back-to-back links). It can be expected that the harmonic currents caused by
these effects will be in the range of 1% or less, yet they can cause considerable voltage
distortions. There are no filter circuits provided for their frequencies. It is not unlikely that
there is a parallel resonance at a noncharacteristic frequency formed by the network and
filter circuits.

How can the non-characteristic and non-harmonic components be computed?

The best method and the one which is most often used today is the simulation of the entire
HVDC system and the ac networks on a digital computer. A simulation program for
converters and ac networks such as NETOMAC or EMTP may be used. System data
including the nonsymmetries which are expected are simulated and the network currents
are computed. When the ac network and the filter circuits are adequately simulated, the
exact network voltages are obtained. A Fourier analysis then yields the harmonics in the
desired frequency range, usually up to 5 kHz. The entire range of converter operation
should be considered.

What should be done if harmonic limits are exceeded?

If the project is still in the design stage, one can attempt to impose smaller tolerances in
transformer reactances and in the valve trigger device, the smoothing reactor can be
enlarged, and/or the d.c. filters can be improved. If the system is already in operation,
usually nothing will help except an additional filter for the interfering frequencies.

- 100 -
1.6 DC-side Harmonics

The function of the converter as an electronic switch can also be seen on the dc side of the
converter. The direct voltage originates in the way that the sections of ac voltages are put
together suitably. In the same way that the ac network acts as voltage source for the
converter, the converter, viewed from the dc side, acts as voltage source. This is true not
only for the dc voltage but also for the harmonic voltages which are superimposed on it.
The amount of harmonic currents superimposed on the direct current will depend upon the
impedances of the dc circuit. The alternating currents caused by the two converters of an
HVDC system always are superimposed. Before discussing the phenomena associated
with this, the origination of the dc-side harmonic spectrum shall be considered in more
detail.

1.6.1 Direct Voltage of the 6-Pulse Converter Bridge

The voltage across the dc-side terminals of a converter in 3-phase bridge connection is the
sum of the voltages of two 3-pulse commutation groups as seen in Figures 1.46 through
1.48. If the summation voltage magnitude is plotted versus time, the typical voltage of a 6-
pulse converter is obtained. Fig. 1.68 shows the ac terminal voltages of a converter on the
top and the direct voltage on the bottom. The left traces are for

a=0 u=0

These left side curves apply for an ideal rectifier. The right side shows the corresponding
voltages for

a¹0 u ¹ 0.

These right side curves apply to the controlled real rectifier. The influence of the delay
angle a and the overlap angle u can be clearly recognized even in the direct voltage.

An analysis shows that harmonic voltages of the following ordinal number


m = k * p, k = 1, 2, 3...

occur where p is again the pulse number. In a 3-phase bridge circuit, p is six. Thus the
ordinal number m is a multiple of the frequency of the ac network.

- 101 -
Curves in Fig. 1.68 are equally applicable for wye bridges and for delta bridges. The
difference is in the phase relationship to each other, that is, in the displacement by 30° el.

Figure 1.68: Voltages of (a) ideal and (b) real rectifier in 3-phase bridge connection

1.6.2 Direct Voltage of the 12-pulse Converter Group

If two 3-phase bridges of wye and delta types are connected in series, the ripple will be
clearly reduced as shown by a comparison of Figures 1.69 and 1.68. This is especially
clear for the "real" rectifier in 1.69b. a and u are the same in both figures. The voltage
jumps, which were very pronounced in Fig. 1.68b, are greatly reduced for the 12-pulse
case of Fig. 1.69b. Their height is half relative to the double value of the direct voltage
average.

Figure 1.69: Voltage of the ideal a) and real b) 12-pulse converter group
(Series connection of two 3-phase bridges)

- 102 -
The influence of a and u on the dc-side harmonic voltages can be calculated. The result
for the 12th harmonic is shown in Fig. 1.70. The very definite minimum at u = 15° is
interesting. At this point the direct voltage curve approaches that of a 24-pulse converter.

Figure 1.70: The 12th harmonic as a function of a and u

- 103 -
Fig. 1.71 will help determine the harmonic spectrum on the direct voltage side.

dx Relative inductive change in direct voltage


a Firing angle (in inverter operation, extinction angle g should be used in place of a)

Figure 1.71: Superimposed alternating voltages on the dc-side of a converter in three-


phase bridge connection

- 104 -
1.6.3 Noncharacteristic Harmonics

As is the case with harmonic currents on the ac side, on the dc side there are
noncharacteristic harmonics and nonharmonic voltage components in addition to the
characteristic harmonics, m = k * p. The typical relationship between n and m is retained.
Thus two ac-side harmonic currents of the ordinal numbers ni+1 and ni-1 are associated
with a dc-side harmonic voltage of the ordinal number mi.

Thus it can be seen that the causes for noncharacteristic harmonics and nonharmonic
components listed in Chapter 1.5.5 also apply for the dc side.

1.6.4 Effects of DC-Side Harmonics

The dc circuit of an HVDC system is an internal current circuit. To this degree we need
not be further concerned with harmonic voltages. On the other hand, the alternating
currents superimposed on the direct current have numerous unpleasant effects. Their
amplitude is determined by the impedances of the dc circuit. They experience a frequency-
dependent phase rotation along the dc line. This causes standing waves to occur. The
alternating currents caused by the converters at each end of the line are superimposed on
each other. The effects of these alternating currents superimposed on the direct current are
essentially as follows:

• Ripple
• Current zeros
• Telephone interference

Since the direct current is the item which the two HVDC stations have in common, the
two ac networks are coupled by the waviness of the direct current. If the ac networks are
not synchronous, the harmonic frequency spectrum of one side is transferred into the ac
network of the other side and nonharmonic oscillations with frequency fluctuations arise.
Consideration should be made to these phenomena especially in the case of back-to-back
links, which of course, always connect two asynchronous networks. The direct current
circuit of these back-to-back links often contains a relatively small inductance and in
general do not contain any dc filters.

- 105 -
The greater the ripple, the more the minimum values of the current in the dc circuit will
approach the zero-line. This is especially the case in light load operation. In light load
operation, not only is the average dc correspondingly small, but also the alternating
currents superimposed on the direct current are larger due to the influence of dx (Fig.
1.71). If electronic reactive power control is used in this area converters have to be
operated with large control angles and special attention must be devoted to this
phenomenon. Steady state operating conditions should not be considered alone. Sufficient
margin for the dynamic control processes must be allowed so that zeroes in the current do
not occur. When zeroes occur, the linkage of rectifier and inverter is periodically broken.
At extinction and initiation of current, oscillations of considerable amplitude occur and
control processes are greatly disturbed.

If an overhead line connect the two HVDC stations, there will be always the danger that
alternating currents superimposed on the direct current will cause telephone interference.
Telephones which are not connected by shielded cable, in particular single-pole telephone
lines, are especially at risk when they are running in parallel or at a small angle to HVDC
overhead lines for some distance. The conductivity of the ground also plays a role in the
reception of interference frequencies.

The effect of alternating currents superimposed on the direct current can be reliably
calculated for various operating conditions of the HVDC system including their
distribution along an overhead line by appropriate computer simulation techniques. The
causes of noncharacteristic harmonics are to be found predominantly on the ac side; hence,
the two ac networks must also be simulated.

A true simulation of the dc line up to a frequency limit of several kilohertz is especially


important with respect to possible telephone interference.

What should be done if the specified limits or those considered acceptable for the
alternating currents superimposed on the direct current -- or for the effects of such
alternating currents -- are exceeded? In addition to the measures listed in Chapter 1.5.5 for
the reduction of causes of noncharacteristic harmonics, the installation or improvement of
dc filters and the enlargement of smoothing reactors are important considerations. In the
case of telephone interference confined to a local area, it is sometimes more economical to
cable the section of the telephone line with the interference rather than to resort to
additional filtering measures.

- 106 -
2. Components of the HVDC Station

In Section 1.2, an overview of the major components of an HVDC system was given.
Components of HVDC stations will be presented in greater detail in this chapter. The
following aspects will be considered:

• Functions within the framework of the HVDC system

• Current and voltage rating

• Insulation coordination

• Design alternatives

• Experience in existing systems

Reliability will also be treated, including the part the components play in unavailability of
the HVDC system, the ramifications of faults, redundancy and spare part strategies, etc.

2.1 AC Switchgear

The ac switchgear represents the interface between the HVDC transmission system and
the ac network. The switchgear links important components of the HVDC station together
and performs important control and protection functions. For these reasons, it is important
to consider the ac switchgear as a component of the HVDC station, even though it is often
placed out for bids and awarded separately.

2.1.1 Functions of the AC Switchgear

The basic functions can be defined:

• Normal maneuvering
• Switching as a part of control or regulation functions
• Protective circuit breaking

- 107 -
Normal switchgear operations, such as changing of busbars, separation and grounding of
bays, etc., are not different from those of switchgear in ordinary transformer substations.

The majority of switching operations will not be executed by the operator or by the load
dispatcher if remote control is applied. These are principally done as a part of the control-
functions of the HVDC equipment. In other words, the majority of these switching
operations are done automatically. This is particularly true for switching of filter circuits
and capacitor banks and in some cases for shunt reactors. It can be part of the reactive
power control or during the start-up and shut-down of the HVDC system. High
requirements for precision in timing and for the permissible number of switchings between
two maintenance periods are a result of such functions.

In the event of transformer faults and faults in the filter circuits, capacitor banks, and
switchyard, fault clearing by protective circuit breaking is required. In contrast, true
HVDC faults, short circuits at the dc lines, inverter commutation failures and other dc
problems are cleared by protection functions of the converters. To facilitate quick recovery
of the HVDC system, the power breakers should not be opened at such events. Faults
which disable the converter, such as flashover in the valve area, are an exception. In such
an event the appropriate power breakers should interrupt the short circuit. Additionally,
they serve as back-up protection for the converter protection functions. Since thyristors are
thermally designed and tested for three full cycles of short-circuit current, the power
breaker should disconnect within two cycles.

2.1.2 Current and Voltage Rating

The required breaking capacity of the circuit breakers utilized depends upon the maximum
network short-circuit power. Normal disconnect processes, however, also deserve special
attention.

The breaker in the converter branch must not open while the converter is carrying current.
Otherwise, it would interrupt a dc component. For this reason, the direct current is
normally regulated to zero prior to disconnecting. If a converter group is to be
disconnected during operation from the series circuit of several groups for each pole, the
converter will be switched to bypass operation, in which the system direct current is
bypassed by means of a chain of valves (all valves of one phase).

- 108 -
To protect against a breaker fault due to an unintended opening of the breaker during
operation, a monitoring latch should be provided which transmits the signal for bypass
operation to the converter for every breaker trip signal.

For breakers in the filter and capacitor bank branches, it is important to be aware of the
unusually high capacitive power. Sizing must be done for the case of load rejection. It is
not possible in all cases to maintain reactive power consumption of the converter and thus
to avoid load rejection overvoltage. In particular, if all filter and capacitor banks are
connected and load rejection takes place from full load operation, load rejection
overvoltage can assume very high levels (to be determined by network studies). In such
case, immediate disconnection of filter and capacitor banks is necessary, and the breakers
must be sized for this event. The capacitative power increases with the square of the load
rejection factor and the recovery voltage is the sum of network voltage and the capacitor
voltage which equals the peak value of the load rejection voltage.

Coordination of insulation is not usually based on the limit voltage of the arresters
available, but is according to the standard of the responsible power company.

2.1.3 Design Considerations

There is no preferred switchgear type for HVDC transmission system stations. Single,
double and triple busbars with and without bypass bars, with and without longitudinal
separation, one-and-a-half breaker switching systems, H-switching and other
configurations are found. Encapsulated switchgear is also found, although the open-air
configuration predominates by far.

In each case, it is necessary to devote significant design consideration to the ac switchyard


when planning the open air part of an HVDC station. It is usually possible to save 20% to
30% or more of the space requirement of the initial design by optimizing the arrangement
of the components and line routing.

- 109 -
2.2 AC Filter Circuits and Capacitor Banks

2.2.1 Functions of the AC Filter Circuits and Capacitor Banks

The filter circuits and capacitor banks on the ac side of an HVDC converter station have
essentially two functions:

1. Providing reactive power for compensation of the reactive power required by the
converters. This prevents the reactive power demand of the converter station from
placing too large a load on the connected ac network and causing unallowable
fluctuations in voltage in the ac network.

2. Absorption of the harmonic currents caused by the HVDC converter. This prevents
the harmonic currents from flowing into the ac network causing undesired distortion
of the system voltage. Furthermore telephone interference caused by harmonic
currents in ac overhead lines can be avoided.

2.2.2 Sizing of AC Filter Circuits and Capacitor Banks with Respect to Reactive Power

The demand of the converter for fundamental frequency reactive power expressed as the
sum of control and commutation reactive power was derived and presented in Section 1.4.
It was demonstrated that the reactive power demand is essentially determined by two
parameters, the control angle (a or g) and the transmission current (Id). These relationships
must be considered when sizing the reactive power sources - filter circuits and capacitor
banks - which below will be referred to as Q elements.

2.2.2.1 Required Reactive Power Performance

There are still no uniform criteria for the reactive power performance of an HVDC
station. In many older systems, the reactive power performance, in particular the
relationship to operating conditions, was not specified. Instead, the establishment of
compensation at rated power was felt to be adequate. With the power of HVDC systems
becoming greater and the connection of HVDC stations at relatively weak points in the
network (short circuit ratio SCR < 5), greater importance is being placed on the behavior
of reactive power over the entire range of power transmission.

- 110 -
The reactive power behaviour of an HVDC station can be specified in various ways.

• cos j = const: To be similar to normal ac loads and generators, a rectifier station would
have to operate at a constant power factor such as cos j = 0.8 (lagging), which would
closely approximate the natural behavior of the rectifier. The inverter station would
have to operate as a reactive power source at cos j = 0.8 (leading), which is exactly the
converse of the natural behavior of the inverter. Compensation by means of Q elements
becomes very costly in this case (overcompensation of the reactive power demand of
the converter).

• cos j = f(P): By means of an appropriately specified function, compensation can be


adapted to the response of ac high voltage networks. Networks usually have a surplus
of reactive power during light load. It would be desirable for the HVDC station to have
a high reactive power demand at light load, thus a poor power factor. However, that is
not the case. Due to the required minimum ac filter circuits, more capacitive reactive
power is generated in the light load range than is necessary for compensation. It is
indeed possible to increase the reactive power demand of the converter within certain
limits by special control processes. However, this reactive power behavior can usually
not be realized without the use of shunt reactors.

• Q = 0 ±DQt: In newer systems, full compensation of the reactive power needed by the
converter over the entire power range has become customary , with a tolerance band of
±DQt (for example ±100 MVAr) being allowed. The philosophy behind this is that the
ac network should handle its reactive power problems on its own, but that these
problems should not be aggrevated by the HVDC system.

In the following discussion, the last operating mode where the converter station operates in
a ±D Q tolerance band is assumed. Transferring this principle to other arbitrarily specified
reactive power behaviours should pose no difficulty.

- 111 -
2.2.2.2 Conditions to be Considered

Customarily, an operating range is specified for network data (and other parameters)
within which the design values have to be maintained. For the network voltage at the point
of connection of the HVDC, the operating range is usually ±5%. In addition, the variations
of factors within the converter, such as control angle and valve voltage, must also be
considered. Other conditions may be contained in the specifications such as the change in
ac voltage permitted when a Q element is switched. Other system components can have
limits such as the breaking capacity of the ac circuit breakers.

The following considerations for the sizing of the Q elements and their control are derived
from the above principles:

• The reactive power output of the Q elements varies with the busbar voltage. Thus a 100
MVAr element supplies between 90.25 and 110.25 MVAr if the voltage varies between
95% and 105% of the nominal value.

• At a given transmission current, the control reactive power of the converter varies with
the control angle. Ordinarily, the rectifier regulates the transmission current Id, and the
control angle a is held in a certain range, typically between 13° and 17°, by adjusting
the transformer tap. Both limit values must be considered in determining the reactive
power demand.

• If the inverter operates with extinction angle control (g = const), there will be only one
reactive power demand curve in relation to Id. Recently, however, a control method has
become established in which the inverter regulates the terminal dc voltage by firing
angle control. In this case, the converter transformer tap changer control has to keep g
in a certain range, such as between 18° and 22°. In this case, however, there are two
limiting reactive power curves to be considered, as it is the case in the rectifier.

- 112 -
• Influence of valve voltage Uv on the relative change in dc voltage dx (see Section 1.1.3)
is customarily overestimated. The operating change of Uv will be either slight or even
non-existent if a control method with Udi = const is used, as in the case of many HVDC
back-to-back links. In long-distance transmission, however, the terminal voltage of the
rectifier is held constant, thus the Uv variation with the load must be considered so that
compensation is made for the load dependance of dx. For precise calculations, the
influence of Uv on the reactive power demand must be taken into consideration.

• The permissible change in ac voltage during the switching of a Q element is often


specified. In relatively weak network connection points (SCR < 5), this can mean a
limit in the nominal power of the individual Q elements. In making the computation,
the minimal network short-circuit power has to be combined with the maximum busbar
voltage. In addition, the change in the reactive power generated by the Q-elements
already connected to the busbar, due to the change in voltage, must be considered.

• Another condition, though seldom specified, is the inclusion of minimum switching


hysteresis to avoid too frequent switching of Q elements, referred to as "pumping". For
this purpose, the worst case (combination of maximum busbar voltage and maximum
control angle) must be considered.

• AC circuit breakers have limited breaking capacity for capacitive currents. This
capacity, which is based on switching theory and design, is often not indicated at all in
specifications and must be requested from the manufacturer. It is necessary that the
maximum temporary overvoltage at which a Q element will have to be disconnected
(such as load rejection and the resulting elevation of voltage) will be considered. This
can result in a limitation of the nominal power of a Q element. In this case, more Q
elements will have to be provided, resulting in higher costs and in more ac switchgear
apparatus.

- 113 -
2.2.2.3 Connect and Disconnect Criteria

Connecting and disconnecting of Q elements must take place in such a manner that the
resulting fundamental frequency reactive power of the HVDC station through the entire
power range remains within the specified tolerance range. All system conditions in their
range of variation must be considered.

There are two possible processes:

• Reactive power open loop control, in which Q elements are connected or disconnected
at fixed partial loads. In this simple process, a fixed switching hysteresis can be
specified. Due to consideration of all possible conditions, however, it is necessary to
limit the jump in reactive power during the switching process to a relative small portion
of the tolerance range. This process will be particularly suitable if a large tolerance
range is specified, or if the nominal power of individual Q elements is limited for other
reasons.

• Reactive power close loop control, in which a Q element is always connected or


disconnected when the measured total reactive power of the HVDC station reaches the
upper or the lower limit of the tolerance range. This procedure results in the minimum
switching frequency and allows the use of the largest Q elements. Reactive power close
loop control also results in switching processes occurring at various partial loads,
depending on system conditions. In view of this fact, it will be necessary to check if the
required filter effect is guaranteed at all times (see Section 2.2.3).

2.2.2.4 Light Load Range

Even with minimal transmission power, ac filter circuits have to be connected in the
HVDC stations in order to maintain the guaranteed distortion values. In most cases, this
results in significant overcompensation of reactive power demand at minimal power and
the noncompliance with the tolerance range. The only way to avoid the use of shunt
reactors is to elevate the reactive power demand of the converter by operating with a
higher control angle and/or with reduced dc voltage. Increased losses are associated with
this method, but, because of the low losses in the light load range, this may be acceptable.

- 114 -
Dealing with reactive power in the light load range is a control problem. It is considered in
this chapter only to the degree that the process described in the following section
determines also the light load range.

2.2.2.5 Method for Determining the Reactive Power Behaviour

A graphical procedure is particularly suited to demonstrate these relationships. This can


also be accomplished, of course, by an analytic process, particularly with the aid of a
computer.

The rectifier of a 1800 MW transmission system will be used as an example with


dxN = 0.065; amin = 13°; amax = 17°; and DQt = ±100 MVAr.

In Fig. 2.1, the reactive power demand of the converter is entered as a function of
transmission power for minimum and maximum control angles.

Qstr [MVAr]
1200

1000

800

600
alpha = 17°
400
alpha = 13°

200

0
0 0,2 0,4 0,6 0,8 1 1,2
P/Pdn

Figure 2.1: Reactive power demand of a 1800 MW rectifier station (dxN = 0.065)

- 115 -
This reactive power demand is to be compensated by a number of Q elements of equal
nominal power. The tolerance band of ±100 MVAr is to be maintained. It follows from
this that the power of a Q element at maximum busbar voltage, assumed to be 1.05 p.u.,
will have to be less than 200 MVAr. A value of QCN = 160 MVAr is selected so that

QCmin = 144 MVAr, and QCmax = 176 MVAr.

In addition, it is assumed that the required filtering in the light load range can be
accomplished by a filter circuit with the fundamental frequency reactive power of one Q
element.

Fig. 2.2 shows the total reactive power of the converter station with a = 13° and network
voltage of 1.05 p.u. With rising transmission power, total reactive power follows the lower
curve in Fig. 2.1. Each time the upper limit of the tolerance band is reached, the next Q
element is switched on. It can be seen that the range up to the nominal power is covered
with 5 Q elements; the sixth element would only be needed if the reactive power tolerance
is to be maintained in the overload range > 1.12 p.u.

With reduction of transmission power, a Q element is switched off each time the lower
limit of the tolerance band is reached. It can be seen that switching hysteresis is
maintained throughout the entire power range.

Fig. 2.3 shows the same processes at a = 17° and network voltage of 0.95 p.u.

- 116 -
Q [MVAr]
120

80

40
n=1 n=2 n=3 n=4 n=5

-40

-80

-120
0 0,2 0,4 0,6 0,8 1 1,2
P/Pdn

Figure 2.2: Total reactive power of an HVDC station at a = 13° and 1.05 p.u. network
voltage (QC = 176 MVAr)

Q [MVAr]
120

80

40
n=1 n=2 n=3 n=4 n=5

-40

-80

-120
0 0,2 0,4 0,6 0,8 1 1,2
P/Pdn

Figure 2.3: Total reactive power of an HVDC station at


a = 17° and 0.95 p.u. network voltage (QC = 144 MVAr)
n = number of Q elements connected

- 117 -
It can be seen that the sixth Q element is necessary in this case. Otherwise the tolerance
band would be exceeded above a transmission power of 0.92 p.u. The switching
hysteresis, however, is significantly larger.

By comparing Fig. 2.2 and Fig. 2.3, it can be seen that the switching on and off of Q
elements occurs at different partial loads. This, of course, necessitates the use of reactive
power close loop control.

In Fig. 2.2, the light load range (the reactive power passing the lower limit of the tolerance
band) begins at P = 0.15 p.u. In Fig 2.3, the total reactive power stays in the tolerance band
down to the minimal power (assumed to be 0.1 p.u.). This holds true under the optimistic
assumption that the required filter effect can be achieved for small partial loads with one Q
element. If two Q elements were needed for this, the light load range would begin at 0.42
p.u. in Fig. 2.2 and at 0.28 p.u. in Fig. 2.3.

- 118 -
2.2.2.6 Use of Other Reactive Power Elements

If it becomes clear that the required reactive power behaviour cannot be achieved along
with the other system requirements, the following possible solutions remain:

• Use of a shunt reactor with a reactor rating of


QLN = 0.5 * QCN
always switched simultaneously with the capacitive Q elements. The reactive power
jump can be cut in half. However, switching frequency of the shunt reactor is high.

• Use of a static compensator (TCR) with a reactor rating of


QKN = QCmax
Continuous control of reactive power can be achieved. The reactive power tolerance
band can be reduced theoretically to zero.

• The same effect can be achieved with a synchronous condenser rated at one-half the
nominal power
QRN = ±0.5 * QCmax.
The higher losses and the significant servicing expense are factors which limit
application to cases in which the short-circuit ratio (SCR) must be increased.

All of the measures listed are expensive, require a large amount of space and are
somewhat problematical from the control performance aspect. In addition, they have an
unfavourable influence on system reliability. Therefore they should only be considered if
the requirements cannot be fulfilled by the use of ac filter circuits and capacitor banks
alone.

- 119 -
2.2.3 Sizing of AC Filter Circuits with Respect to Harmonics

Prevention of the penetration of harmonic currents into the ac network can be achieved
theoretically either by series connected or by shunt connected filter circuits (Fig. 2.4).

Series filter circuits have the disadvantage that they have to carry the full current (losses!)
and that all filter components have to be insulated against the full network voltage. In
addition, the series filter circuits have to withstand short-circuit currents and the related
overvoltages in the event of a fault. In the case of a shunt filter, only those harmonic
currents flow through the filter for which the filter was designed. An additional advantage
of the switched shunt filter is the controllability of reactive power. This permits a control
of the reactive power of the HVDC station, as was shown in Section 2.2.2. For this reason,
shunt filters are predominately used in practice on the ac side of the HVDC station.

Figure 2.4: Serial (a) and parallel (b) ac filter circuits

- 120 -
2.2.3.1 Required Filter Effect

Harmonic currents occurring during converter operation, which are superimposed on the
line current may be both, characteristic and non-characteristic. This was discussed in detail
in Chapter 1.5 and the relationships were presented.

Harmonic currents must be calculated with consideration given to the special conditions
listed in Section 2.2.2 for the entire power range of the HVDC system. This applies to the
maximum values for individual harmonics, not only at various partial loads, but also for
various parameters of the system, such as control angle.

AC filter circuits are required to absorb the harmonic currents to such an extent that the
residual currents which flow into the network do not cause any unacceptable voltage
distortions or telephone interference. They must be able to perform this function for any
combination of filters, capacitors, and reactors which might be switched on in order to
achieve the required reactive power balance with the ac network.

In the interest of achieving higher availability, an effort is always made to give all filters
the same design, or at least to limit them to two types.

The determination of filter components (at a specified fundamental frequency reactive


power) is made from voltage distortion and telephone interference considerations.

The frequently used design criteria are:

• Individual distortions Dn. This refers to the relationship of the harmonic voltage with a
frequency of (n * fo) to the fundamental frequency voltage.

Un
Dn =
Ul

Un = Harmonic voltage with frequency n* fo


U1 = Fundamental frequency voltage with frequency fo

- 121 -
In earlier systems, permissible individual distortions between 0.7% and 3% were
specified. In most cases, no distinction was made between individual harmonics. This
applied for all harmonics up to the ordinal number 50 (more rarely 25). It is anticipated
that in the future, specifications will be made on the basis of the CIGRE Study Group 36-
05 recommendations corresponding to the following individual distortion limits:

Multiple of 3rd Harmonic Other odd-numbered Harmonics Even-numbered Harmonics


n % n % n %
3 2 5 2 2 1.5
9 1 7 2 4 1
15 0.3 11 1.5 6 0.5
21 0.2 13 1.5 8 0.2
>21 0.2 17 1 10 0.2
19 1 12 0.2
23 0.7 >12 0.2
25 0.7

25
> 25 : 0.2 + 0.5
n

This represents a tightening of the conditions for higher harmonics and, at the same time, a
welcome relief for the lower harmonics which are often controlled today only with low
order filters.

• Total distortion of voltage. This is the geometric sum of all individual distortions.

¥
Dtot = åD
n =2
n

The upper frequency limit for this parameter is also usually set at n = 50, or more rarely
n = 25.

In earlier systems, Dtot was usually specified between 1% and 3%. CIGRE 36-05
recommends Dtot = 3%.

- 122 -
• Telephone Interference Factor (TIF)

or

• Telephone Harmonic Form Factor (THFF)

In evaluating telephone interference by voltage or current harmonics on ac lines, the


fact is taken into consideration that the human ear, the receiver in the telephone and the
electromagnetic link between telephone and power lines are frequency dependant. For
this reason, weighting factors depending on frequency are used. Two systems of
weighting factors are widely used. One system is the C-Message Weighting System of
BTS (Bell Telephone System) and the other system is the Psophometric Weighting
System of CCITT (Consultative Commission on Telephone and Telegraph).

In Fig. 2.5, the standardized weighting factors of the two systems are presented in
relation to frequency. It can be seen that the two curves are in large degree identical and
that the maximum is at about 1 kHz.

Figure 2.5: Comparison of BTS and CCITT weighting factors

- 123 -
This leads to the conclusion that harmonics with frequencies around 1 kHz are particularly
critical with respect to telephone interference. Even if they have small amplitudes, they
can result in unacceptable telephone interference due to the greater weighting.

Two factors are frequently used in quantitatively describing telephone interference from
harmonics:

BTS Telephone Interference Factor (TIF) and the CCITT Telephone Harmonic Form
Factor (THFF)

¥
1
å (K * Pf * U f )
2
TIF = * f
U f =0

where

Kf = 5*f
Pf = BTS weighting factor
Uf = RMS value of harmonic voltage with frequency f

¥
U= åU
f =0
2
f

The CCITT THFF value is also calculated from the above equation, but the factors Kf and
Pf have different values as follows:

f
Kf = ,
800
CCITT weighting factor
Pf =
1000

Customary values are TIF = 40 or THFF = 1%.


The upper frequency limit which is to be taken into consideration is often specified at 5
kHz, making it necessary to calculate the harmonics up to n = 100!

- 124 -
• IT Product (Interference Current)

Less frequently, other factors such as the IT product, are used for specifying telephone
interference. The IT product is the geometric sum of the product of the harmonic
currents penetrating the network expressed in amperes and the corresponding TIF
values.

å (T * If )
2
IT = f

If = effective value of the harmonic current with a frequency of f


Tf = corresponding TIF value for frequency f

The IT product only makes sense if the distribution of network (residual) harmonic
currents onto the HVAC lines emenating from the busbar of the HVDC station is
known. In such case, the results will be unambiguous if there is only one line going out,
as is the case at the C-side of the Dürnrohr and Etzenricht back-to-back links.

- 125 -
2.2.3.2 Design of the Filter Circuits

The degree of voltage distortion at the ac busbar of an HVDC station depends on the
amplitude of the harmonic currents, which are injected by the converters and by the total
impedance resulting from the parallel connection of the network and filter circuits. Fig.2.6
shows this relationship for the harmonic of the ordinal number n. The individual
distortions Dn determined in this manner must not exceed the specified values for any
harmonic in the established frequency range.

In calculating the harmonic currents, the entire range of dc transmission power and the
range of variation of system parameters, as defined in Section 2.2.2.2, must be considered.
In calculating impedances, the particular filter circuit combination and network
configuration must be considered along with the specified range of variation of network
frequency.

A precise solution to this problem is not possible. Therefore several assumptions have to
be made and simplifications must be introduced, as explained in more detail below.

Z nF * Z nN
Zn Zn =
In F Z nF + Z nN

Un = Zn * In
HVDC Un
Dn =
U lN
Zn
N

Figure 2.6: Equivalent circuit for calculating harmonic voltages and currents on the ac
side of the HVDC station

- 126 -
2.2.3.3 Determination of Network Impedance

Basically there are two methods for determining the network impedance:
• simulation of the real network
• impedance equivalent in the form of a circle or a sector of a circle

Simulation of the real network assumes precise knowledge of the network configuration.
The calculation of network impedance for all harmonics is very complex and time
consuming. In most cases, several scenarios will have to be considered -- at the least, a
heavy load and a light load scenario. Assumptions have to be made for these scenarios
with respect to network configuration, generators and consumers which are connected to
the network. The frequency response should be known for the impedance of each element,
if possible up to 5 kHz. If not, assumptions have to be made. (Measurements made in the
North German High Voltage Network have confirmed that the relationship between
harmonic impedances and fundamental frequency impedance is completely different
between light load and heavy load periods). This method, which is very tedious and which
has many uncertainties, can only be used in simple network configurations. Even then, it
will only be undertaken if compliance with the specified distortion values is critical.

Fig. 2.7 shows an example of the computed impedance curve of a high voltage network.

Fig. 2.7: Calculated impedance of a high voltage network at various frequencies

- 127 -
Impedance equivalent of the network avoids these problems.

An impedance boundary can be determined which includes the network impedance for all
cases. This boundary is designated a resonance circle. Fig. 2.8 shows such a resonance
circle. Rmax corresponds to the first parallel resonant point of the network. Rmin occurs at
series resonance of the network. The angles F1 and F2, which are on the two origin lines,
are a measure for the quality factor of the network at the power system frequency.

X/R = tan F1 = G1 (quality factor at the power system frequency)

The shaded area indicates the range within which network impedance can lie. The values
Rmin and Rmax apply for only one harmonic ordinal number, as was shown in Fig. 2.7.

For this reason, unfavourable results are often obtained when Rmin and Rmax are set the
same for all harmonics. These values are often specified differently for various harmonics
or are assumed differently in the calculation. The primary reason for specification of
different values for different harmonics is to avoid impedances, which never really occur.
This avoids unreasonable filter designs due to unrealistic assumptions.

Figure 2.8: Resonance circle of an ac network

- 128 -
2.2.3.4 Determination of Resulting Harmonic Impedance

The calculation of voltage distortion must be made separately for each harmonic. For this
purpose, the impedance resulting from the parallel connection of the network and the
various combinations of filter circuits must be determined. This impedance must be
multiplied by the calculated harmonic current. If the results of an actual simulation of the
network are available (compare Fig. 2.7), this task is relatively simple. When based on an
impedance equivalent, however, one of the following approaches must be selected:

• Determination of maximum impedance for each individual harmonic (resonance


method)
• Assumption of a very large network impedance, i.e. only the filter impedance prevails
(open network method)
• Combination of the two methods with the assumption of resonance for a few (usually
two) harmonics and ignoring network impedance for all others (selective resonance
method)

In the resonance method, the resonance circle is transformed from the impedance plane (Z
plane) into the admittance plane (Y plane). (See Fig. 2.9.)

- 129 -
Range of Filter Resonant points with
Admittance Yf the ac network
➀ Line A A*
➁ Point A or A*
➂ Line AB or A*B*
➃ Point B or B*
➄ Circle B-B*

Figure 2.9: Circle of network admittances and the resonance conditions

The impedance vector of the filter is transformed into the Y plane at a certain harmonic
ordinal number n.

With both the network and the filter impedances plotted in the admittance plane, the
shortest vector between the filter admittance point and the network admittance boundary is
the smallest admittance possible for the parallel combination of the network and filter.
When this smallest admittance is transformed back (inverted) to the Z plane, it represents
the highest impedance into which the converter harmonic current must flow.
Consequently, it produces the highest possible harmonic voltage.

- 130 -
The open network method represents usually the safe side in the calculation of maximum
network voltage distortion for most harmonics. The filter impedance alone is responsible
for the voltage distortion. If, however, due to the actual impedance from the network and
the filters, there is a parallel resonance for a harmonic frequency, a voltage distortion far
beyond the acceptable range can occur. For this reason, the open network method is no
longer accepted.

The selective resonance method represents a reasonable compromise. It takes into


consideration the fact that the highest voltage distortion (highest harmonic voltage) occurs
with a parallel resonance between filter and ac network. It is unrealistic, however, to
assume that such a parallel resonance takes place at all frequencies. Normally it is
sufficient to consider, in the calculation of total distortion and TIF value, only two
maximum individual distortions from the resonance calculation. The ac network is
assumed to be open for the remaining harmonic currents.

In calculating the distortion values and designing the filter components, any possible
detuning of the filter must be considered. The following can cause a detuning of the filter:
• Deviation of the frequency of the ac network by DfN
• Deviation of component data from the specified value based on temperature
fluctuations
• Deviation of component data from the specified value due to manufacturing tolerance
(including aging)
• Deviation of the filter capacity from the original value due to fuse operation in the
capacitor unit (to the point of triggering of alarm)

2.2.4 Filter Circuit Arrangements

There are a number of suitable filter circuit configurations for an HVDC station.
Determination of the type of filter that should be used depends upon the requirements for
voltage distortion and the data of the ac system. If different types of filter circuits are used
on the ac side of an HVDC station, care must be taken that a qualitatively different
impedance curve is generated by means of parallel connection of the filter circuits.

- 131 -
Figures 2.10 through 2.13 represent circuit diagrams, impedance curves, and phase angle
curves of several commonly used filter circuits which can be divided into three groups:

• Sharply tuned filters which are tuned to one or two frequencies

• High-pass filters which have a rather low impedance over a wide frequency range

• Combination of sharply tuned filters and high-pass filters (multiple-tuned high-pass


filters)

- 132 -
2.2.4.1 Sharply Tuned Filters

Fig. 2.10 shows a single-tuned filter circuit on the left and a double-tuned filter circuit on
the right.

Figure 2.10: Sharply tuned filters

- 133 -
A sharply tuned filter circuit provides a particularly low-resistance current path for the
corresponding harmonic. The residual voltage to be expected is, therefore, particularly
low. However, its impedance is especially sensitive to detuning of the filter circuits.
Detuning of the filter circuits can occur due to the reasons listed in Section 2.2.3.4.

It should also be noted that the precision of the original tuning is limited due to design
restrictions. Therefore a broading of the tuning by installation of a low-ohmic series
resistor is often necessary.

2.2.4.2 High-pass Filters

Fig. 2.11 shows the simplest configuration of a high-pass filter, which is obtained by
connecting a damping resistor parallel to the filter reactor.

The impedance and phase angle curve corresponds qualitatively to the single-tuned
resonance circuit, i.e. a capacitive behaviour below the resonance frequency and an
inductive behaviour above it. Quantitatively, however, there are two important differences:

• The filter impedance is at a relatively low value in a broad frequency range around the
resonance frequency (insensitivity to detuning)

• In the higher frequency range, the filter impedance approaches a limit value determined
by the damping resistance (for absorption of the higher harmonics)

- 134 -
Figure 2.11: High-pass filter

- 135 -
2.2.4.3 Multiple-Tuned High-Pass Filter

Multiple-tuned high-pass filters combine the advantages of sharply tuned filters and high-
pass filters. In comparison to single-tuned filter groups, they have the significant
advantages of being space-saving, having fewer high voltage components and of being
less expensive. Additionally, the requirement for filter redundancy can most easily be met
if only one type of filter - namely the multiple-tuned filter - is used in each station. In the
newer HVDC systems, such multiple-tuned high-pass filters are being increasingly used.

Fig. 2.12 shows a double-tuned high-pass filter in two configurations.

The impedance curves of the two variations demonstrate differences in the higher
frequency range. The version on the right approximates the linear slope of a pure
inductance while the version on the left approximates the value of an ohmic resistance at
high frequencies. Each curve has pros and cons depending on network configuration and
converter operating conditions.

Fig. 2.13 shows a triple-tuned high-pass filter with series resonances at the third, twelfth,
and twenty-fourth harmonics. In addition to the characteristic harmonics of a twelve-pulse
converter and the higher harmonics, such a filter circuit also absorbs low frequency,
noncharacteristic harmonics. This can be important if the resulting impedance on the ac
side has a parallel resonance in this frequency range. This is usually the case with a very
low short-circuit ratio (SCR <3).

- 136 -
Figure 2.12: Double-tuned high-pass filter

- 137 -
Figure 2.13: Triple-tuned high-pass filter

- 138 -
2.2.5 Steady- State Load of The Filter Circuit Components

Steady state loads are those currents and voltages which represent an electrical and thermal
state of equilibrium for the filter circuit components. This implies that the root mean
square of the currents and voltages remains constant over a relatively long period of time.

For ac-side filters of an HVDC station, these loads include fundamental frequency voltage
and harmonic currents, depending upon the operating condition of the converter and the
network impedance. These currents have different effects on the individual filter circuit
components.

2.2.5.1 Fundamental Frequency Voltage

The fundamental frequency voltage of the ac network determines the steady state rating of
filter circuits and capacitor banks. The highest operating voltage, ordinarily 1.05 p.u,
should be used as the steady state condition.

In determining the steady state voltage for the individual capacitor units and coils, three
other effects must be taken into consideration:

• Voltage rise due to the resonance circuit effect

• Overvoltage due to fuse operation

• Harmonic voltages

Overvoltage due to the resonance circuit effect is associated with series connection of C
and L with their voltages in phase opposition. Both voltages are of equal magnitude for the
resonance frequency. For a filter circuit tuned to the nth harmonic, the overvoltage of the
fundamental frequency is as follows:

1
Us =
1 - n12

- 139 -
This effect can be neglected at higher tuning frequencies. In a filter tuned to the 11th
harmonic, the overvoltage is 0.83%. At lower tuning frequencies, however, this
overvoltage is significant.

Overvoltage due to fuse operation should not be neglected. Assuming the customary H-
circuit, in which each phase of a filter capacitor is divided into 4 equal blocks, the
maximum overvoltage is obtained when the permissible number of fuse operations occurs
in one row of the capacitor matrix which forms such a block. If a is the permissible
number of faulted segments corresponding to the warning level of the filter protection
system, and the complete row consists of z segments, then the overvoltage is calculated as
follows:

z
Uw =
z-a

With two permissible faults in a row consisting of thirty segments, the overvoltage is 7%.
The influence of fuse operations on the total capacity and on the fundamental current can
usually be neglected.

Harmonic voltages are caused by the harmonic currents flowing into the filter circuit. They
are superimposed on the fundamental frequency voltage across a capacitor unit. At higher
tuning frequencies (n ³ 11), the harmonic voltages are so small that they can be accounted
for by the standard ratings of the capacitor units and do not have to be considered as
additional voltage loads.

- 140 -
2.2.5.2 Harmonic Currents

Harmonic currents are decisive for the steady state load of the reactors and resistors of the
filter. Three steps are involved in determining them:

• Determine the maximum harmonic currents of the converter at every stage of operation
giving consideration to all conditions as presented in Section 2.2.2.2. The minimum
network voltage (0.95 p.u.) must be used, because at this network voltage highest
harmonic currents occur on the network side. These higher amplitudes correspond to
the transformer tap setting.

• Determine the portion of the harmonic currents flowing into the filter circuit as
presented in Section 2.2.3.2. In making this calculation, an additional amount of 10% is
usually included for harmonic currents emenating from outside sources in the ac
network.

• Divide the so determined harmonic currents among the minimum number of filter
circuits on-line for each operating condition (compare chapter 2.2.2.5).

2.2.5.3 Fundamental Frequency Currents

The fundamental frequency current as determined by the filter capacitor also flows
through the filter reactors and resistors.

Two extreme cases must be used in determining the steady state load of these components:

• Maximum network voltage (1.05 p.u.). At this voltage the fundamental frequency
current is at the maximum value while the harmonic currents are at minimum values.

• Minimum network voltage (0.95 p.u.). At this network voltage the fundamental
frequency current is at the minimum value while the harmonic currents are at their
maximum values.

It is impossible to say in advance which of these two cases will result in the highest steady
state load of individual components.

- 141 -
2.2.5.4 Network Frequency

Network frequency determines the fundamental frequency impedance of the filter


capacitors and hence of the fundamental frequency current. When frequency fluctuations
remain under 1%, as is usually the case, its influence on the steady state load of the
capacitor units can be neglected, since the fundamental voltage does not change
significantly. For the other filter components, the effect of frequency deviations is also
small because a smaller portion of the harmonic currents flows through the filter circuit as
a result of the detuning.

2.2.5.5 Computations

Analytic determination of the steady state load of all filter components in individual steps -
as was presented above for purposes of understanding - is usually not practicable. For this
reason, a digital simulation of the entire system is carried out. Programs like NETOMAC
or EMTP are used to determine current and voltage wave shapes. These are separated by
Fourier analysis into fundamental frequency and harmonics to determine the individual
harmonic voltages and currents.

In making these calculations of the steady state load of individual filter components, the
worst case has to be assumed. In general the following assumptions are made:

• Emergency operation with a reduced number of filters at load points at which all n
filters are used in normal operating conditions, are operated with n-1 filters.

• Predistortion of the network voltages is taken into consideration. In practice, the


harmonic currents of the converter are simply increased by a certain factor, customarily
10%.

• Maximum network voltage at the busbar.

• Variation of network frequency (maximum and minimum frequency).

• Detuning of the filters (tolerances of the filter circuit elements, operation of capacitor
fuses, change of filter elements due to temperature and aging)

- 142 -
• Resonance condition between network impedance and the filter impedance, if
appropriate, is limited to 2 specified harmonic frequencies.

2.2.6 Transient Load of the Filter Components

Transient loads are quickly decaying events which are not part of the steady-state
considerations. With respect to ac filters, these are rapid changes of voltage and current
which last between 100 ms (turn-on voltage) and many seconds (transformer saturation).
Specifically, the following events are transient loads:

• Switching on a filter circuit


• Single-phase busbar fault
• Clearing of a three-phase fault
• Load rejection of the converter station following ac faults
• Separation of an inverter station from the network (islanding)

The object of transient load computations is the determination of the following quantities:

• Maximum value and duration of transient voltages for each filter component

• Maximum value and duration of transient currents for each filter component

• Energy dissipation for each surge arrester and resistor in the filter circuit

• Protection level and insulation level for each filter component

- 143 -
2.2.6.1 Switching on of a Filter Circuit

The maximum values and decay time constants of currents which occur during the
switching on of a filter circuit are determined by the short-circuit power of the network
and the number and nature of other filter circuits and capacitor banks that are already in
operation. It should be noted that several network periods are required before the
resonance circuits become effective. During this time, internal balancing processes occur,
some of which cause significant overloads on individual components.

In making the computation, the assumption is made that the filter is switched on at the
crest value of network voltage. The ac network is considered to be a voltage source with
an internal impedance which is determined by the maximum short-circuit power of the ac
network. Fig. 2.14 shows the equivalent circuit diagram.

At the moment of switching on, the capacitors are unloaded: that means that the turn-on
voltage appears across the filter reactors. Arresters are connected parallel to the filter
reactors. Their ability to dissipate energy and their cooling time constants must be sized in
such a way that they are able to handle the anticipated sequence of switching on and off
for any steady state and variable transmission power.

Figure 2.14: Equivalent circuit diagram for switching on a filter circuit.

- 144 -
2.2.6.2 Single-Phase Ground Fault

AC filter circuits which are connected to the high voltage bus are generally arranged in a
wye circuit which is solidly grounded. For such filter circuits, the single-phase to ground
fault of the high voltage bus at the crest value of the fundamental frequency voltage
combined with a switching surge represent the most severe transient conditions. Lightning
voltage waves are not as critical from a thermal standpoint because the energy content is
relatively minor and the effective duration is too short to charge the filter capacitors to the
full voltage level of the wave. However, a lightning strike can result in a single-pole
ground fault. The maximum current occurs when switching surges flow in and cause an
arcing to ground.

The equivalent circuit diagram for this fault is presented in Fig. 2.15. Both the impedance
of the ac system and the impedance of the short-circuit arc should be considered in making
the calculation. It is assumed that the initial condition is that the high voltage capacitor of
the filter is charged to the voltage which corresponds to the switching surge protective
level of the busbar arrester. At the moment of the arcing, the voltage of the main capacitor
is applied abruptly to the other filter components or to the protective arrester. This arrester
must absorb practically the entire energy of the capacitor.

Figure 2.15: Equivalent circuit diagram for a single phase ground fault

- 145 -
2.2.6.3 Clearing of a Three-Phase AC Fault

The three-phase ac fault, which is rare, does not in itself present a higher duty to the filter
components than the single-phase fault. However, additional duty will be caused if a fault
of this type is cleared after a period of e.g. 100 ms. With the return of voltage, not only
switch-on transients occur for the filter circuit components, but in addition, the
transformers of the HVDC system are driven well into saturation (switch-on rush
intensified by the remaining remanence). This causes very high second harmonics in the
transformer current. In unfavorable circumstances, this second harmonic can coincide with
a parallel resonance from the network reactance and the filter capacitors (Fig. 2.16).

For the filter components, this can mean an overload which lasts several seconds and
consists of highly distorted transient overvoltages and noncharacteristic low-frequency
harmonic currents.

Figure 2.16: Equivalent circuit for stresses caused by a cleared three-phase fault

- 146 -
2.2.6.4 Load Rejection Following AC Faults

In the case of a three-phase ac fault, the converter should not be blocked in order to keep it
ready to resume transmission immediately after clearing the fault. Only in the case of
relatively long lasting disturbances the converter may be blocked. In that case the filter
circuits have to be switched off.

Thus a major stress to the filter circuits can only occur in the event of faulty coordination
which is particularly critical because it is also accompanied by load rejection overvoltage.

2.2.6.5 Separation of the Inverter from the Network

This disturbance which is referred to as "islanding" may accur when an HVDC station is
connected to the network over a single line. In such a case, an inverter can continue to
commutate on its own, if the converter reactive power demand was fully compensated
before the disconnect.

This condition represents an extreme danger to the filter circuits and their surge arresters
since the energy being transmitted through the HVDC system has no other path besides the
filter circuits.

As a protective measure, a bypass in the converter bridges is established immediately


preventing further commutation and thus preventing energy transmission to the ac side.
This case can only be analyzed and the transient duty to the components of the filter circuit
can only be determined in simulator studies with precise simulation of the devices for fault
detection and converter control.

- 147 -
2.2.7 Design Considerations

In the great majority of HVDC stations, the ac filters are connected with appropriate
circuit breakers to the ac busbar. The main capacitor is located at the “hot” end; all other
filter components are “close to ground” (at the center of the wye). All resistors and
reactors are protected against overvoltage by suitable arresters. The capacitors are split into
4 blocks (H-circuit) for the detection of fault events in the capacitor bank. A current
transformer in the cross member will reliably indicate very small asymmetries (loss of an
individual capacitor unit or coil). Additional current transformers at the hot and cold ends
provide differential and phase comparison protection. Figures 2.17 and 2.18 show an
example of the structural design of the ac filter.

- 148 -
Figure 2.17: Side view of a double-tuned filter circuit
3 High voltage reactor
16 High voltage capacitor with current transformer
19 Damping resistor
21 Low voltage capacitor bank
27 Low voltage reactor

- 149 -
Figure 2.18: 120 kV filter circuit

- 150 -
In a very small number of cases, the ac filters are connected to the tertiary winding of the
converter transformer. Since the tertiary winding has a smaller voltage than the primary
winding, the filter circuit needs less space (voltage clearances). This also applies for the
switchgear which is also much less expensive. On this reason, a filter arrangement
comprising a great number of filter circuits with a small fundamental frequency rating
each, can economically be provided. Therefore, such arrangements are found primarily in
HVDC systems of small power. The tertiary winding, however, increases the cost of the
transformer. In addition, this configuration makes tuning more difficult because a portion
of the transformer leakage reactance is effective in tuning and it changes with the position
of the tap changer. Therefore this filter circuit design is significantly more critical.

The structural configuration of the individual components will be discussed in greater


detail below.

2.2.7.1 The Capacitor Bank

All filter circuit capacitor banks constructed to date are composed of standard high voltage
capacitors with rated powers of 50 to 200 kVA per unit. Due to losses, only foil capacitors
with silicone oil are used today. The banks are of a tiered design with 5 to 20 capacitor
units per level in parallel connection. There are between 3 and 10 levels, one above the
other, which are electrically connected in series. In order to avoid corona discharges at the
sharp-cornered capacitor housing, the upper levels are usually surrounded by corona rings.
A cagework frame of this type on tall supporting insulators is extremely vulnerable to
earthquakes. Extensive calculations and vibrating table tests may be necessary if a certain
earthquake resistance is specified. In extreme cases, it may be necessary to provide
overhead suspension for the capacitor framework.

An important problem is the discharging of capacitors following off-switching of the filter


circuit. The high-impedance discharge resistors, which are often built into the capacitor
units, usually are not sufficient for this. In this case, it is necessary to provide inductive
potential transformers parallel to the filter circuit. A calculation must be made to
determine if the specified time sequence of discharge events is permissible for the
potential transformer with respect to temperature.

- 151 -
2.2.7.2 Filter Reactors

Most of the newer HVDC systems have air-insulated filter reactors without iron core.
They are made to be resistant to outdoor conditions through fiberglass reinforced synthetic
resin insulation. Generally these air-core reactors are self-cooled. This can present
problems if the reactors have to be retrofitted with noise-damping hoods in order to be in
conformity with the specified noise level. It is recommended therefore that the noise
problem be studied thoroughly during project planning period.

The insulation level along the winding and to the ground is determined by the surge
arrester of the filter reactor. The steep surge voltage is particularly significant in this
respect.

2.2.7.3 Damping Resistors

All high-pass filters and some sharply tuned filters require damping resistors. Losses arise
in them which must be dissipated. If at all possible, forced-air cooling should not be used
for reliability reasons. Ordinarily, air-cooled resistors are used which are installed in a
housing with ventilation slits.

2.2.7.4 Fencing

Costs will be saved, stability of capacitor structures will be improved, and the visual
impact of the filter circuits will be reduced if open accessibility to the filter plant is
sacrificed and the Filter circuits are fenced in. Of course the customer must be in
agreement with this. In this case, the entrance into the fenced area must be interlocked
with the filter circuit breakers.

- 152 -
2.3 Converter Transformers

Next to the valves, the most significant components of an HVDC station are the converter
transformers. This applies to both, costs and losses. According to the failure statistics of
CIGRE, they are a major cause of outages and non-availability of HVDC systems.
Therefore, the CIGRE study committees responsible for HVDC systems and transformers
have set up a joint working group to establish a reliable basis for the specification, sizing
and testing of HVDC transformers.

2.3.1 Functions of the HVDC Converter Transformer

The converter transformers must transform the voltage of the ac busbar to the required
entry voltage of the converter. They provide two ac systems which are turned toward each
other by 30° el for the two three-phase bridges which are combined into a twelve-pulse
group. At the same time, they insure the voltage insulation necessary in order to make it
possible to connect converter bridges in series on the dc side, as is necessary for HVDC
technology. The transformer main insulation, therefore, is stressed by both, the ac voltage
and the direct voltage potential between valve-side winding and ground.

Fig. 2.19 shows the potential versus time curves for the valve-side transformer windings of
the upper and the lower bridges in an HVDC station with four converter bridges per
station pole. The HVDC station in Cabora Bassa was selected as an example. The positive
station pole with bridges 1, 3, 5, and 7 is depicted. The potential wave shapes are the
same, as well, for an HVDC station with two twelve-pulse groups per pole such as in
Nelson River II or Itaipu. The complicated mixture of alternating voltages of varied
frequencies and direct voltage can be seen in this example.

- 153 -
Figure 2.19: Potential versus time curves for the valve-side transformer windings in an
HVDC station with four three-phase bridges per pole

The converter transformers of an HVDC station are almost always equipped with on-load
tapchangers in order to provide the correct valve voltage required at each load point. They
not only compensate for the internal voltage drops of the HVDC converters, but they also
compensate for deviations of the ac busbar voltage from the design value. An additional
important function of the converter transformer is the limiting of short-circuit current. This
will be considered in more detail in Section 2.3.3.

On the other hand a converter transformer is quite ineffective with respect to damping of
high frequency commutation oscillations.

- 154 -
2.3.2 Current and Voltage Rating

While it is assumed that the converter transformers of an HVDC station encounter a


sinusoidal voltage, the curve of the current is anything but sinusoidal. It is considerably
distorted. If a completely smoothed direct current is assumed and overlap is neglected, the
valve-side lines of the transformer carry an alternating current consisting of blocks of
direct current 120° el wide of alternating polarity. The effective value of this current is as
follows:

2
Iv = * Id (30)
3

The r.m.s. value of Iv in relationship to Id decreases with increasing overlap. However it is


generally customary to ignore this effect.

Equation 16 continues to hold true.

U di
Uv =
1.35

Therefore, the apparent power of the transformer according to equation 19 is as follows:

S = 3 * Iv * Uv = 1.05 * Udi * Id

Until a few years ago, even in international standards such as IEC 146, it was assumed that
the rated current of the transformer could be specified using equation (30). It was thought
that sufficient margin of safety to allow for thermal loads from harmonic currents was
obtained by ignoring the overlap. Today this method is known to be inadequate for large
converter transformers. Transformer manufacturers now require data on the harmonic
currents which are actually present over the entire load range up to 5 kHz. It must be kept
in mind that the harmonics of the six-pulse operation (n = 5, 7, 11, 13, 17, 19 etc.) also
flow in the valve-side transformer windings of twelve-pulse converter groups.
Cancellation of the harmonic currents not contained in the twelve-pulse spectrum takes
place in two-winding transformers at the ac busbar. In three-winding transformers, this
cancellation takes place in the transformer main flux. This means that the network winding
of three-winding transformers only carries the harmonic currents of twelve-pulse
operation.

- 155 -
The reduction of just at 4% of the rated current associated with this will ordinarily not be
taken into consideration. It is hoped that the CIGRE working group mentioned above will
clarify this situation.

To take into account the dc component in the valve-side current is also important for
sizing and loss determination of the converter transformer. It is caused by the deviations of
control pulses from the 30° equidistance in a definite pattern. At this point theory -- and
also practice (since there are no methods for measurement) -- fail. Certainly it is too
pessimistic to assume a worst case pattern leading to the maximum dc component taking
into account the full triggering equipment tolerance. 10 -20 % of this value may be
adequate as the basis for calculating the anticipated transformer saturation.

2.3.3 Selection of the Short-Circuit Voltage

One of the important tasks in project planning for an HVDC station is the selection of the
optimal transformer short-circuit voltage uk. The viewpoint of the transformer
manufacturer plays a roll in this only to the degree that a short circuit voltage above about
22% and below about 12% is associated with a definite increase in manufacturing costs.
The value selected for uk determines the following parameters:

• The leakage inductance Ls as per equation (20) and thus the thyristor short circuit
current. In addition, Ls is part of the dc-side inductance and therefore also keeps down
the required smoothing reactor inductance.

• The relative direct voltage change dx in accordance with equation (21) and hence the
internal voltage drop of the converter station. The larger the internal voltage drop of the
converter station, the higher the rated power of the valves and the converter transformer
will be for a given rated nominal power of the HVDC system.

• The commutation overlap u in accordance with equation (22) and the required angle of
advance in inverter operation.

• Reactive power demand Q of the converter in accordance with equation (27) and the
size of the compensation facilities required, such as filter circuits and capacitor banks.

- 156 -
• The amplitude of the harmonic currents In in accordance with equation (29) and the
required quality factor of the filter circuits to be installed.

Fig. 2.20 shows these relationships in graphic form for an existing HVDC station
(Dürnrohr Back-to-Back Link)

(1) Short-circuit surge current of the thyristors


(2) Required smoothing reactor inductance
(3) Rated power of valves and transformers
(4) Reactive power demand of the converter

Figure 2.20: Relationship of important data of an HVDC station upon the short circuit
voltage of the converter transformer.

The relationships of technical data can only be optimized, however, when the effects on
the overall costs of the station are known.

In Fig. 2.21, the effects on the overall costs of the HVDC station are presented.

- 157 -
Figure 2.21: Change of cost components ➀ through ➃ and of overall costs ➄ of an
HVDC station in relationship to uk

A presentation of this type is valid only for a specific system (it is intended here to indicate
trends). In addition, another important limitation must be made. The overall cost curve ➄
with its tendency to decrease with the increasing short-circuit voltage uk will only be valid
if the cost influence of the thyristor surge current represented in curve ➀ does indeed exist.
This is undoubtedly the case in back-to-back links, since the nominal direct current
approaches the limit set by the thyristor. This limit is generally the short-circuit current.
Following a complete wave, blocking ability is still required. An increase in the short
circuit current through a smaller uk requires reduction of the nominal direct current and
thus a corresponding increase of the nominal direct voltage. This accounts for the
pronounced influence on costs. This influence on costs however is much smaller or it does
not exist at all in the case of long-distance HVDC transmission systems with relatively
small transmission current. The thyristors used in such systems are usually over-
dimensioned with respect to current. In that case, they permit an increase in the short-
circuit current without any problem. But then cost factors ➂ and ➃ are predominant. This
tends to choose a smaller short-circuit voltage uk.

- 158 -
This explains why uk values around 20% are usually found in back-to-back links, while
long distance HVDC systems of relatively small power usually have uk values around
14%.

2.3.4 Insulation Coordination

The insulation coordination of HVDC-systems will be discussed in detail in Chapter 7. At


this point, however, a brief introduction to the principle of sizing of insulation and of
overvoltage protection of the valve side of converter transformers may be useful.

Fig. 2.22 shows the circuit diagram of an HVDC converter with surge arresters for
overvoltage protection. This concept, which was first used in the Cabora Bassa project, is
widely used today. In this scheme, each valve is effectively protected against every kind of
overvoltage entering from outside by a directly parallel-connected surge arrester. This
concept forms the basis of thyristor voltage sizing and insulation coordination.

Figure 2.22: Overvoltage protection of an HVDC station pole with one twelve-pulse
group per pole

- 159 -
For the valve winding in delta connection, this means that in operation there is always one
arrester connected in parallel to the winding. This is because at least one of the valves
connected to the winding is in a conducting condition. When at no-load, however, two
arresters connected in series provide the overvoltage protection of the winding. The sizing
of the insulation should of course take this into consideration.

The same considerations are basically true for the valve winding in wye-connection. The
only difference is that there are two phases connected in series which are protected by one
or two valve arresters, respectively.

The observations made to this point have been limited to protection against overvoltages
along the length of the winding. It can be seen in Fig. 2.19 that the voltage to ground can
be several times greater than the winding voltage. It is apparent in Fig. 2.22 that there is an
unbroken chain of arresters from the terminal with the highest potential to ground. To this
extent the valve windings are protected against overvoltages which occur to ground, i.e.
transverse to the main insulation, by a network of parallel and series-connected arresters.

2.3.5 Transformer Design

Both, single-phase and three-phase converter transformers are used in existing HVDC
systems. Two-winding and three-winding transformers, some of which having an
additional tertiary winding, are encountered. In many systems, the transformers are placed
in the yard; in others, they are directly in front of the valve hall into which their valve-side
bushings penetrate. If the structural configuration is not fixed in the bid specifications, this
represents a great deal of latitude for station optimization.

Several aspects which play a roll in such optimization considerations will be discussed
below.

- 160 -
2.3.5.1 Transportation Weight

In systems of high power, weight can be an important consideration, in particular where


transportation is difficult such as in Cabora Bassa. The relative transportation weights of
the 4 major design types are approximately as follows:

Single-phase - two-winding transformer 1


Single-phase - three-winding transformer 1.6
Three-phase - two-winding transformer 2.2
Three-phase - three-winding transformer 3.6

Theoretically it would also be possible to connect transformer poles in parallel to further


reduce the transportation weight. To date, however, no such configuration is in existence.
Complete bridges or twelve-pulse groups could theoretically also be connected in parallel.

2.3.5.2 Costs

Basically, costs in about the same relationship as was shown for transportation weights
would be expected. Accordingly, six single-phase, two-winding transformers would cost
around 1.7 times as much as one three-phase-three-winding transformer. The cost situation
however becomes very different if spare converter transformers are to be supplied, as is
almost always the case.

It can clearly be stated that in these cases single-phase transformers always result in lower
overall costs than three-phase transformers. The degree of the savings will vary according
to whether the system is a back-to-back link with the same or with different network
voltage on the two sides, or if it is a long-distance transmission with one or two twelve-
pulse groups per station pole. Without taking into consideration possible costs of other
components, for example valve design, there seems to be no reason for application a three-
phase configuration of the converter transformer.

- 161 -
2.3.5.3 Reactance Symmetry

The significance of transformer leakage inductance and its symmetry have already been
discussed. In this regard, a fundamental distinction must be made between design
tolerances and manufacturing tolerances. Design tolerances are always in the range of 1%
even with the modern computational processes of today. This is acceptable because the
deviation always effects all transformers in the same manner. With manufacturing
tolerances, however one must apply stricter standards, because these tolerances lead to
differences between the individual transformers, i.e. to reactance nonsymmetry.

There are three distinct types of reactance nonsymmetry. Nonsymmetry of phases causes
noncharateristic current harmonics, predominantly the third harmonic and its multiples. In
single-phase transformers, the cause is to be sought exclusively in manufacturing
tolerances. This is an additional advantage of this type of construction. In three-phase
transformers, the design tolerance also plays a roll since the leakage relationships of the
middle leg are different from those of the outer legs.

Wye-delta nonsymmetry provokes the occurrence of harmonics due to six-pulse operation


in the network current (fifth, seventh, seventeenth, nineteenth, etc.). In this type of
nonsymmetry, design tolerance also plays a role in all transformer configurations. The
concentric winding arrangement (wye-network-delta) should not be permitted for three-
winding transformers. Instead, the network winding should be divided so that a complete
system is provided on each of the two legs of the single-phase unit. For the three-phase
unit, this requirement means a two tier winding arrangement.

Tap nonsymmetry means a change of the transformer short-circuit voltage while passing
through the tap positions. In Section 2.3.3, the significance of the transformer short-circuit
voltage for the various sizing aspects was discussed. Nevertheless, one can be somewhat
more generous in this point. An upper and a lower limit for short-circuit voltage should be
specified. In contrast, it is very important that within one step the narrow tolerances of the
phase nonsymmetry and of the wye-delta nonsymmetry are maintained.

- 162 -
2.3.5.4 Valve-Side Bushings

A clear recommendation can be given in this area also. If the specified lay-out of an
HVDC station makes it at all possible, the converter transformers should be placed
directly against the wall of the valve hall, with the valve-side bushings projecting into the
hall. In this way, the need for wall bushings is eliminated. These wall bushings are both,
expensive and apparently quite critical with respect to their insulation behaviour over long
periods of time. The insulation contamination can be avoided in the sensitive area between
the converter transformer and the valves. This insulation contamination can be critical
even when in conventional terms there seems to be none pollution at all, as the example of
the Dorsey Station of the Nelson River HVDC shows.

In single-phase three-winding transformers, the horizontal arrangement of the four valve-


side bushings undoubtedly represents a design difficulty. The extra costs associated with
it, however, are offset by the fact that the bushing porcelains can be configured with a
specific leakage path of 1.4 cm per kV, while for open-air it is customarily between 3 and
4.5 cm per kV. For fire prevention purposes, solid bushings should be preferred for the
valve side. These are available today on the market up to 500 kV direct voltage.

2.3.6 Experience in Existing Systems

The CIGRE working group which has already been cited several times has compiled
statistics of outages of HVDC converter transformers. A total of 46 transformer faults
have been recorded. The causes of faults are of great variety. It is not possible to describe a
typical HVDC transformer fault. Nevertheless, this conclusion can be drawn: it is not a
simple task to construct a reliable long-living HVDC converter transformer.

In line with this insight, spare transformers are now almost always included in call for
bids. For the system planner, this presents the additional task of making it possible to
change out a transformer unit quickly. This means not only keeping assembly equipment
available but also the planning of simple transport routes to the site of the spare unit. In
addition, this need may result in constraints for the arrangement of cooling plants, noise
shielding devices and the like. A guideline value might be for a transformer to be changed
out within 24 hours.

- 163 -
2.4 Thyristor Valves

This chapter is intended to give the project planning engineer an idea about all that goes
into an HVDC thyristor valve and why this device is as complex as it is. The primary goal
is to improve the understanding between project engineers and the valve manufacturer.
The intention is not to clarify every detail of valve technology. It is, however, important
that both partners know about the concerns and problems of the other. If this can be
achieved, then it will be possible even in the tense situation of the execution of a contract
to have the objective, goal-oriented communication which is necessary for success.

2.4.1 Historical Considerations

In Chapter 1.3, the function of the HVDC converter was explained and it was shown that
the valve acts as a switch which connects a phase of the ac system for a certain period of
time with the dc side, but which lets the current flow in only one direction. Thus it is a
semi-conductor switch. Since we are dealing with high voltage in the range of several
hundred kV and currents on the order of kA, this has to be a very special switch.

Converter valves using mercury arc technology were developed and tested in Germany
and Switzerland as early as in the 1930's. Today it is widely forgotten that Siemens along
with AEG in 1941 had a contract for the construction of the world's first commercial
HVDC transmission system. This was the 115 km-long land cable connection of a power
plant on the Elbe River to Berlin (60 MW). The transmission was ready to be switched on
in 1945 but the system was dismantled and served as the basis for the development of
Russian HVDC technology. It was not until 10 years later, in 1955, that the Gotland
transmission system constructed by the Swedish company ASEA went into operation (20
MW). In about 1965, Siemens resumed the development of HVDC valves, first with
mercury arc technology. As early as 1967 the first thyristor valve was built, long before the
Swedish competition. Siemens carried on the development at that time with AEG and
BBC in the "Arbeitsgemeinschaft HGÜ" - a very fruitful cooperative effort for all partners.
Essential elements of the first Siemens thyristor valves are still being used today; an
example is the method of stacking thyristors and heat sinks and the liquid cooling.

- 164 -
Initially, transformer oil served as the cooling medium. For this reason, valves for the first
contract (Cabora-Bassa) had a superficial similarity to a transformer. The cooling oil was
also used for ac voltage insulation and the active part was installed in a steel container for
placement out-of-doors. DC voltage insulation was by air. The valves rest on isolating
platforms (Figure 2.23). Recently ABB has been promoting a similar technology, but
using air-insulated valves in the container.

In about 1974, BBC introduced the technology of water cooling into the development of
HVDC thyristor valves. It was commercially used in the Nelson River Bipole II
transmission system for the first time. Siemens manufactured all modules for this system
with thyristor deliveries from AEG and BBC. From this point on, the valves were air-
insulated for indoor installation (Figure 2.24). It was not until 1978 that the other suppliers
built their first trial valve with water cooling. In the meantime, this technology has been
adopted by all manufacturers.

Several concepts still being used today also originate from this period. These concepts
include: snubber circuits, individual thyristor protection, and the control of valves with
indirect light triggering directly from ground potential. These concepts will be discussed in
greater detail in the following sections. Therefore, it is justified to state that even recently
employed valve technology has been proven and successfully used for a long time. There
has just been an evolution of the components to higher and higher capacity.

- 165 -
Figure 2.23: An ac bridge of the Cabora Bassa HVDC system, consisting of three oil-
filled double valves on insulating platforms

- 166 -
Figure 2.24: Air-insulated, water-cooled quadruple valve of the Nelson River II HVDC

- 167 -
2.4.2 Thyristors as Switches

So far, the valve has been treated as an ideal switch in our consideration of the function of
the HVDC converter. This was justified because at the interfaces between the ac network
and the dc line essentially the fundamental frequency characteristic of the converter and
the harmonics up to about 5 kHz were of interest. But nothing is perfect including
thyristors, of which the modern HVDC valve is composed. At this point the essential
relationships which are of interest will be discussed.

Power thyristors are manufactured from highly pure monocrystaline silicon. They are so
called NPNP semiconductors. This means that they consist of four layers which are doped
alternately with P and N (Fig. 2.25a). The outer, highly doped zones are the emitting
zones; the weakly doped, inner layers are the base zones. The control connection G is
located on the P base; J1-J3 designate the junctions between individual zones. The off-
state voltage in the reverse direction is blocked at junction J1 between P-emitter and N-
base. The off-state voltage in the forward direction is blocked at junction J2 between P-
base and N-base. The maximum diameter of the silicon wafer of an HVDC thyristor is
currently 100 millimeters. The wafer is less than 1 millimeter thick and blocks up to 8 kV.

Several of the imperfections of the thyristor in comparison with the ideal switch can be
recognized in the static V/I-characteristic of the thyristor (Figure 2.25b). In the presence of
off-state voltage, an off-state current (several mA) flows both in the forward direction and
in the reverse direction. If a maximum value of voltage is reached in the reverse direction,
the off-state current increases sharply (avalanche effect). This results in local overheating
and destruction. The off-state current also increases sharply in the forward direction upon
reaching a limit value of the off-state voltage (zero-toggle voltage). This causes junction
J2 to be flooded with carriers resulting in an uncontrolled turn-on - an event which also
can lead to the destruction of the thyristor and therefore should be avoided.

- 168 -
Figure 2.25: (a) structure and (b) static V/I-characteristic of a thyristor

- 169 -
Figure 2.25 shows an additional imperfection, the on-state characteristic, i.e. the
characteristic curve of the conducting thyristor. It can be approximated by a constant
portion (designated by the differential resistance Rt). The entire voltage drop of an HVDC
thyristor is of the order of two volts. This means that for typical currents of one to two kA,
considerable power losses must be dissipated. The thyristor is therefore far from being an
ideal switch.

This can also be seen when considering the turn-on behavior more closely. If the thyristor
is switched on by means of a turn-on signal across a control connection G, the PN junction
J2 is first flooded with carriers in the vicinity of the control connection and therefore is
conductive (Figure 2.26a). The increasing current first flows to a small surface, which can
result in local heating and, in the event of high di/dt, it can result in the failure of the
thyristor. The breakdown of the blocking layer expands laterally with the speed of "only"
0.05-0.1 mm/ms. For this reason, it is quite possible for several hundred ms to elapse in
thyristors with large surface area, as are used in HVDC systems, before the entire thyristor
surface is uniformly engaged in conducting the current.
Figure 2.26b illustrates the time sequence of turn-on.

- 170 -
Figure 2.26: Turn on process of the thyristor.
(a) basic structure of the device
(b) time sequence of the most important parameters

- 171 -
In the turn-off process of the converter valve, the thyristor does not function in accordance
with pure theory as presented in Section 1.3 (see Fig. 2.27). When the forward current
becomes zero at time t1, the thyristor is still highly conductive due to charge carriers which
are stored on both sides of the PN junction J1, so that current can continue to flow as
reverse current. It is not until a later point in time that the charge carriers are dissipated
partly through the reverse current and partly through recombination to such an extent that
a space charge zone forms and the junction J1 can withstand blocking voltage.

Figure 2.27: Turn-off process of the thyristor


a) Structure of the component
b) Time sequence of current and voltage

- 172 -
At this point in time, the reverse current has reached its maximum and begins to decay. Its
rate of change causes a voltage peak as determined by the inductances of the commutation
circuit, which is superimposed upon the quasi-stationary blocking voltage UR. The
resulting peaks in the overall voltage can, if they exceed a breakdown voltage level,
present a danger to the thyristor and must be damped by a carrier-storage circuit. The
integral of the reverse current over time is designated as the recovery charge Qrr. This is a
very important parameter for the design of HVDC valves.

Another essential parameter is the recovery time tq of the thyristor. It denotes the period of
time between the zero passage of the commutated current and the zero passage of a
forward recovery voltage of a defined magnitude which the thyristor tolerates without
toggling into the on-state. Since the recovery time is dependant upon temperature,
preceding on-state current, time sequence of current change, magnitude of the forward
recovery voltage and its wave shape, the thyristor must be given a hold-off interval to
allow for achieving the dielectric strength following the carrying of current. This hold-off
interval, tc, is greater than the recovery time, and is customarily expressed as the angle of
extinction. A value of 18° el. has proven useful in the 50 Hz network.

Considering the inadequacies of thyristors, the question must be answered as to why much
more modern components which have become available are not used in HVDC. The GTO
and the IGBT would be candidates. In contrast to thyristors, which can be switched on
through the control connection, the GTO (Gate Turnoff Thyristor) and IGBT (Insulated
Gate Bipolar Transistor) are switched on as well as off through the control connection.
The GTO is a four-layer semiconductor with the zone sequence PNPN like the normal
thyristor. In contrast, the IGBT is a bipolar power transistor with integrated MOSFET
control.

GTOs are available today up to 4.5 kV at 4 kA switch-off current. Their drawbacks are
higher forward and switching losses grading. In addition, multiple series connections with
the voltage, which would then be necessary, and the low inductance connection
requirements are too expensive.

- 173 -
IGBTs would be more favorable with respect to forward losses, but they are available only
up to about 1 kV at a few hundred amperes. With respect to the low inductance connection
requirements, they present even higher limitations. For these reasons, thyristors will
probably continue to be used during the coming years.

2.4.3 The Multiple Series Connection -- Special Requirements and Solutions

Modern thyristors for HVDC applications have a maximum blocking voltage of 5 to 8 kV.
Utilizing these thyristors for an ac bridge of 250 kV requires that the thyristors be
connected in series to achieve the 250 kV blocking voltage of a valve. In practice this
could require up to 100 thyristors.

It has already been shown in Section 2.4.2, that the thyristor is not an ideal switch.
Additional difficulties arise from the multiple series connection, some of which also are
attributable to the inadequacies of the components and of their manufacture. This begins
with the turn-on, which must take place simultaneously for all elements in the series
circuit. A prerequisite for this would be that the gate pulse would be applied
simultaneously to all thyristors. This, however, is not realizable, as will be explained in
greater detail in Section 2.4.4. In addition, the turn-on delay and the gate-controlled turn-
on time of thyristors may not be identical, and it is dependent upon the damping circuit
connected in parallel to the thyristor (Fig. 2.28).

- 174 -
Figure 2.28: Voltage curve of two thyristors (a) and (b) of the same type for differing gate
currents and differing RC snubbers 1 and 2

- 175 -
Unless measures to the contrary are taken, the last of a group of series-connected thyristors
with differing gate-controlled turn-on time would be subjected to the entire valve voltage.
That would result in periodic overvoltage firing, and must be reliably avoided. This is one
function of the RC network connected in parallel to each thyristor in conjunction with the
valve reactor (c.f. Fig. 2.29). The function of the valve reactor also is to limit the increase
of the thyristor current until a significantly large cross section has become conductive.

During the turn-off process, the recovered charge Qrr presents difficulties. It is prone to
variations to a significantly greater extent than the gate-controlled turn-on time. (See Fig
2.30.) ±10% of the mean is typical. As a result, turn-off overshoot at individual thyristors
in the series connection are of varying magnitudes, the voltage zero crossings occur at the
individual thyristors at differing points in time, and a non-uniform sharing of voltage in
the forward direction results. Optimization of the compromise between the technical
requirements and the economic expenditure is an additional criterion for the sizing of RC
snubber circuits and valve reactors.

Each thyristor along with its circuit is a physical entity and has a stray capacitance to
ground. These capacitances are basically different for each installation location. The valve
can be depicted as a capacitive ladder network, which causes a very nonlinear voltage
distribution for steep fronted surges. This effect is mitigated by compensation for the
ground capacities through suitably sized capacitors parallel to individual valve sections.
The term "grading capacitor" has come into use. (Fig. 2.29).

The blocking current of the thyristor, that is, its resistance in off-state condition, also
exhibits a stray band which cannot be neglected. This would result in non-uniform voltage
distribution when direct current is applied. To alleviate this, a grading resistance is
connected to every thyristor in parallel.

- 176 -
Figure 2.29: Components of thyristor valves for HVDC

Key: TE Thyristor electronics


TL Thyristor level
K Heatsink
VS Valve section
n Number of thyristor levels in a valve
k Number of thyristor levels in a valve section
LWL Light guide
KV Coolant distribution
IG Insulating structure

- 177 -
Figure 2.30: Effects of non-uniform recovered charge in series-connected thyristors

- 178 -
2.4.4 Valve Control and Monitoring

Each valve in an HVDC converter contains up to 100 series-connected thyristors, all of


which have differing potential to ground. To turn on the valve, each thyristor must have a
gate pulse of sufficient energy at the same point in time. Today that is an electric signal of
a few amperes with a rise time less than 1 ms. These signals can no longer be generated at
ground potential at an acceptable cost and then transmitted to each thyristor, for example
by magnetic pulse transformers.

Since the mid-1970's, an optoelectronic system has come into use which has proven very
successful (Fig. 2.31). The underlying idea is to bridge the spatial distance and the
potential difference between ground and the individual thyristors with practically no power
and no delay. The necessary ignition energy then has to be made available to each thyristor
individually.

Figure 2.31: Principle of the optoelectronic valve control system

Key: VA Valve control


TÜ Thyristor monitoring
EV Energy decoupling
FPE Valve base electronics
R Converter control
LWL Fiber optic cable
TE Thyristor electronics

- 179 -
The turn-on command from the converter control is converted by the valve base
electronics into light signals which are transmitted through fiber optic cables to each
thyristor. Here, an electronic board performs a variety of functions:

• Decoupling and storing of auxiliary energy from the snubber circuit


• Monitoring of the auxiliary energy storage voltage
• Sensing of thyristor voltage
• Conversion of the light signal into an electrical signal
• Formation of a suitable electric gate pulse
• Formation and transmission of monitoring signals to the valve base electronics

The electronic logic system at each thyristor location determines that all requirements for
turning the thyristor on are met before the gate pulse is released. Tolerances in the timing
sequence cannot be entirely avoided. However, the entire system is dimensioned in such a
manner that the time difference between the arrival of the gate pulse at the first and the last
thyristor of a valve is no greater than 2 ms.

In the event an individual thyristor does not receive a gate pulse due to an electronic
failure, it would be subject to the total valve voltage during a late turn-on and would be
destroyed. To prevent this from occurring, there is a protective turn-on circuit at each
electronic board which is connected directly between anode and gate. The protective turn-
on circuit is independent of the turn-on command of the control system and turns the
thyristor on when its maximum blocking voltage is attained.

The optoelectronic system is also excellently suited for monitoring the thyristor levels and
for special protective functions. A second fiber optic channel per thyristor is needed for
this purpose. The signal transmission from the valve-monitoring circuit to ground potential
takes place in "dialogue" according to the signal diagram shown in Figure 2.32. This is the
method by which reports of thyristor failures during operation and self-monitoring of the
signal path are accomplished. Customarily, component failures are printed showing type
and location of the failure.

- 180 -
Fig. 2.32: Signals between valve base electronics and thyristor electronics (signal
telegram)

- 181 -
2.4.5 Valve Cooling

During operation of HVDC thyristors, heat losses are generated which have to be
dissipated by forced cooling. The cooling medium must have good thermal and good
dielectric properties. While all of the competing companies initially operated using air
cooling, Siemens, along with the Arbeitsgemeinschaft HGÜ (AEG, BBC, Siemens), used
liquid cooling from the start. At Cabora Bassa, oil was used, and in all later systems, de-
ionized water was used. Approximately 5 years ago, the last competing company switched
to water because of its outstanding cooling properties.

Most heat losses in HVDC thyristor valves originate:


• in the thyristors while current is being carried
• in the snubber resistors during the recharging of the snubber capacitors and
• in the valve reactors as copper losses and as iron losses during magnetization and de-
magnetization

The components are cooled by special coolers or directly with water (see Fig. 2.33). Other
heat losses, such as in the electrical connections, are radiated out into the valve hall, from
which they have to be dissipated by suitable measures.

- 182 -
Figure 2.33: Schematic diagram of water circulation in a thyristor valve

- 183 -
Valve losses are not constant, but rather are dependant on the operating condition. The
individual component losses are quite varied. Thyristor losses are essentially determined
by the direct current and the static V/I-characteristic (Fig. 2.34a). The magnitude of
snubber losses is dependent on the number and the amplitude of step changes in the valve
voltage (Fig. 2.34b) and thus on the trigger delay angle. Each voltage change results in a
recharging current in the snubber capacitor and losses in the associated damping
resistance.

Figure 2.34: Influence factors of valve losses


a) Thyristor V/I-characteristic
b) voltage curve under various operating conditions

- 184 -
These dependencies can be represented in one diagram which gives the project planning
engineer an overview of the framework within which he can operate in designing a system
(Fig 2.35). The diagram shows the valve losses as a function of trigger angle, with direct
current as the parameter. Assume a system sized for a nominal trigger angle a = 15°. The
valve losses at this trigger angle and nominal current will be 1 pu. If the same current is to
be conducted with a = 90°, perhaps to compensate for reactive power, in a back-to-back
station the losses increase to 1.6 pu. The cooling circuits in the valve and also the external
cooling system must be rated for this value. As an added margin, the valves would then
have the ability to carry approximately 40% overcurrent continuously when the trigger
angle is limited to 15°. Characteristics such as this are important to many customers.

Figure 2.35: Relationship between trigger angle, direct current and valve losses

- 185 -
On the other hand, if valve cooling is designed for losses at nominal operation, it can be
seen from the diagram that for operation at a = 90°, a current of approximately 0.4 p.u. is
permissible. Of course, this is a qualitative representation intended to illustrate trends.
Precise figures would have to be determined for a specific case based on the technology
used and system data.

Fig. 2.31 shows that in SIEMENS' valves, all thyristor levels are provided with cooling
water in parallel. This distinguishes SIEMENS from the competition, whose heat sinks are
supplied with water to each component in a series manner. Fewer hose connections, but
ones of larger diameter, are required in the arrangement of the competition. The
competition accepts higher cooling water temperature at the last thyristor and an
electrolytic current through all metal parts in the cooling circuit.

In contrast, SIEMENS cools all thyristors with the same water temperature and reduces the
corrosive electrolytic current to practically zero by means of special potential control. This
can be accomplished only with parallel connection. Electrolytic current is permitted only
in major distribution pipes of the valve structure. SIEMENS uses these electrolytic
currents to equalize the voltage distribution in the series connection. To prevent the
possibility of electrolytic corrosion, SIEMENS uses platinum "target electrodes" at
locations where electrolytic currents greater than or equal to 10 mA can flow.

Corrosion in cooling circuits is a very important subject for HVDC because a very high
value is placed on the useful life of the system. The problem lies partly in the fact that de-
ionized, highly pure water with a high oxygen content is quite aggressive. This is a fact
which is not properly appreciated by many manufacturers of conventional cooling systems.
SIEMENS provides the cooling water piping within the valves up to a junction flange.
Everything else - pipe lines in the building, pumps, water preparation system, heat
exchanger, etc. - are generally purchased from external suppliers, often from a
manufacturer in the country of destination who is not familiar with the specific
requirements of HVDC valves.

- 186 -
NOTE: The pure water circuit for the cooling of HVDC thyristor valves is not a
conventional system component. Design of this system must be under the
full control of experienced specialists.

The cooling system of a recently commissioned back-to-back station can be used as a


negative example. In contrast, the pure water circuit of the Nelson River Bipole Station 2,
where SIEMENS (in association at that time with the HVDC joint venture) first utilized
water cooling, has not had any problems to date. Fig. 2.36 shows, schematically, the
proven arrangement of the pure water circuit. A discussion of these components is
provided on the following pages. If any of the directives relevant to corrosion are not
observed, the results can be the same as those of the competition, whose valve cooling
circuits had heavy corrosion damage after three to eight years of operation.

Figure 2.36: Cooling circuit for HVDC valves


Key:
1 Valve p pressure measurement
2 Heat exchanger l conductivity measurement
3 Pump J temperature measurement
4 Ion exchanger q volume flow measurement
5 Fine filter h water level measurement
6 Expansion vessel L leak detection

- 187 -
• Pump
The pump operates at a constant rotation speed, even during standby operation of the
system. Its power is thus included in no-load losses and is a large factor in loss
evaluation. In a pump which is sized too large, it is possible to reduce the power
absorbed by milling down the impeller. An undersized pump cannot be corrected.
Pumps must be constructed of stainless steel with floating ring seals of coal (hard coal).
Grease must be avoided because it will reduce the capacity of the mixed bed filter.

• Heat Exchanger
The heat exchanger should be directly behind the pump in the flow sequence. This
arrangement causes a portion of the dynamic pump pressure to be dissipated so that it
does not impact the filter and the thyristor valves. Because the piping of the latter is
made of synthetic materials, thyristor valves are especially sensitive to pressure at high
operating temperatures. The useful life is influenced by this fact. Titanium-stabilized
stainless steel (CrNiMo) should be specified as the material for the pure water circuit of
the heat exchanger. Heat stressing during manufacture, such as welding, can cause
stainless steel to lose its non-rusting characteristic due to microstructural modifications.

• Ion Exchanger (Mixed Bed Filter)


The ion exchanger should be located in bypass between the cold-water side of the heat
exchanger and the suction side of the pump because the ion exchange resin is sensitive
to temperature. The water demand is approximately 3% of the total amount of flow, the
maximum being 200 l/min. The resin must be chlorine-free. If the resin is reprocessed
after being used-up, great care must be taken to ensure that salts used in the process are
completely removed. Additionally, it must be ensured that resin cannot find its way into
the valves, for example by back-rinsing which is often used in cleaning. Resin would
form deposits on the wires of the snubber resistors, where it would be baked on as a
result of the high temperatures and would make cooling impossible. As a result, the
resistor and the associated thyristor would fail.

- 188 -
• Fine Filter
The purpose of the fine filter, which is situated in front of the valves, is to catch
mechanical particles ³ 50 mm. At high water flow speeds these particles would have a
sandblasting effect and would lead to erosion damage of the metal surfaces. A filter
with a coarser mesh should be placed in front of the fine filter for protection.

• Measurement of Conductivity
Conductivity should be measured in the intake to the valves behind the fine filter. It
should be held as low as possible - £ 0.5 mS/cm at operating temperature - in order to
keep electrolytic currents and losses low. A standard value which has proved useful is
0.2 mS/cm as a mean, with the protective system set to issue a warning at 0.4 and to trip
at 0.5.

• Valves
Usually several units (multiple valves) are provided with cool water by means of a
distribution system. SIEMENS specifies the pressure drop and the minimum flow per
valve. If it is not possible in designing the water distribution system to guarantee a
symmetrical distribution of flow volume to the individual units, an add-on factor
should be used in determining the total flow of water. This factor should be selected in
such a manner that the unit which is located at the most unfavourable position will
receive no less than the minimum quantity of water.

• Expansion Vessel
The low-pressure version of expansion vessel is manufactured of material suitable for
use in pure water circuits. It must be sized for volume changes caused by temperature
and loss of seal integrity. The connection with the pure water circuit is made through
supply and return connections on the warm water side of the valve cooling system. The
expansion vessel performs several functions:

1) Compensation for volume changes from the pure water circuit


2) Removal of air from the pure water circuit
3) Discharge of H2 and O2 dissolved in the pure water
4) Enrichment of the pure water with oxygen at the surface of the water
5) Monitoring of the water supply and/or water loss
6) Assurance of compliance with defined pressure relationships in the water circuit

- 189 -
An explanation of these functions and presentation of all points which must be
observed in the planning and realization of an expansion vessel would require a treatise
of its own and thus is beyond the scope of this presentation. The following are the
important issues:

• The expansion vessel is open to the atmosphere


• The pressure differential with respect to the warm-water side of the thyristor valve
should be ³ 0.5 bar.

• Piping System
The piping system and all elements which come in contact with pure water such as
valves and metering instruments must be of non-rusting stainless steel. In unusual
circumstances, coatings which are compatible with pure water can be permitted if
documentation of sufficient stability is available. For the pipes themselves, titanium-
stabilized CrNiMo steel must be used. This also applies to welding electrodes.
Otherwise, microstructural modifications could occur during welding which would
later result in rust. Over time, rust can become coated on the cooling hoses of the valves
and on the switching resistors and can result in clogging. Pipes made of fiberglass-
reinforced plastic could be considered as an alternative, if compatibility with pure water
has been demonstrated and if the useful life requirements can be met.

2.4.6 Valve Mechanical Design

An HVDC thyristor valve must be designed in such a manner that

• the electrical requirements can be handled


• manufacturing costs are low
• space requirements are low
• fast, simple on-site assembly is possible
• the components are easily accessible during commissioning and service
• effects of component faults remain confined within limits

Additionally, the valve must also be mechanically stable.

- 190 -
It has become customary to not manufacture complete valves in the factory. Instead, the
valve is assembled at the construction site from components which correspond to the
electrically defined valve sections in Fig. 2.29. The components are completely assembled
and tested at the factory. The major activity during production is the valve assembly. The
larger part of an HVDC valve module from the standpoint of cost are the components
supplied by outside sources. The know-how of the design engineer consists of making the
most intelligent arrangement of components possible.

Today, two types of modules are used in HVDC systems, as shown in fig. 2.37 and 2.38.
Both have in common the fact that thyristors and heat sinks are bound together into a
stack, reflecting the electrical wiring diagram directly in the mechanical arrangement. As
far as possible, this idea was also incorporated in the arrangement of the other
components.

- 191 -
Figure 2.37: Thyristor module for back-to-back stations

- 192 -
Figure 2.38: Thyristor module for long-distance transmission

- 193 -
It has become customary to construct HVDC valves not as individual units, but rather to
combine several valves into a common structure (multiple valve). In back-to-back links
with rated voltage up to 160 kV, double valves are used, with the three double valves of a
three-phase bridge on a common insulating table which is insulated against ground for
one-half of the rated voltage.

In long-distance transmission systems, nominal voltages are higher. Therefore the number
of series-connected thyristors is greater and the physical height of the valves is greater.
Specific design is somewhat project-dependant. Double valves, quadruple valves, single
towers, and double towers are a few of the concepts which will be encountered in the
planning of a system. It is important to note that valves for long-distance transmission
systems no longer stand on the floor but rather are suspended from the roof of the valve
hall. This solves the problem of mechanical stability of tall, slim structures and also avoids
the need for special valve foundations.

2.4.7 System/Valve Interfaces

HVDC thyristor valves are an integral part of the entire system. This means that they have
to be designed and tested - electrically and thermally - according to the constraints of the
basic design. This also means, however, that a series of interfaces between the thyristor
valves and other components of the system have to be coordinated. The knowledge and
maintenance of these interfaces is essential to the HVDC department and represents an
important prerequisite for a smoothly running project.

Space will not permit a listing and detailed description of all of the interfaces of the valve.
SIEMENS maintains an index of interfaces which essentially applies to every project. It
has only to be updated with respect to details. Individual categories and considerations for
each will be listed in summary fashion in this paper (see Fig. 2.39).

Figure 2.39: Schematic presentation of interfaces between HVDC valve and other parts
of system

- 194 -
1) Interfaces to the building:
Location and dimensions of points of attachment on the floor of the hall or to the roof;
weight; requirements for the detection of leakage water; dimensions of attachment
points for valve base electronics and thyristor monitoring system, if applicable;
dimensions for cable ducts for fiber optic cables.

2) Interfaces to valve hall buswork:


Location and dimensions of connection terminals; permissible mechanical loads;
responsibility for terminals and corona shielding.

3) Interfaces to the cooling system:


Location and dimensions of the connecting flange for cooling water pipes; flow-rate
and pressure drop of the valves; supply and return flow temperatures; water quality.

4) Interfaces to the auxiliary energy supply:


Terminals in the valve base electronics cabinet; power consumption, connection
voltage.

5) Interfaces to system control and monitoring system:


Terminals in valve base electronics cabinet; definition of various signals; EMI
shielding concept.

6) Interfaces to converter control:


Terminals in valve base electronics cabinet; definition of various signals; EMI
shielding concept.

7) Interfaces to valve hall climate control:


No physical interface; requirements with respect to temperature, humidity, and purity
per specifications.

8) Interfaces to protective system:


Terminals in base electronics cabinet, definition of various signals (leak detection,
loss of redundancy, number of protective firing, etc.), influence upon various
protective settings (overcurrent, water temperature, valve hall temperature, etc.).

- 195 -
The thyristor valves do not have an interface to the operators desk except that they are
represented on the screen as symbols and failure reports from the thyristor monitoring
system for any component failure are printed out. Therefore, the thyristors cannot be
"operated" directly; they are simply there. The best thyristor valves are those which are no
longer noticed after commissioning. SIEMENS uses redundancy in the thyristor valve and
recommends a certain maintenance program so that the system will operate reliably over a
period of many years. This will be discussed briefly in closing.

Even when perfectly designed and operated properly, there will be a certain statistical
failure rate for electrical and electronic components. The greater the number of
components, the greater the probably of failures occurring. To avoid activation of the
protective system and disconnection of the entire system each time individual components
in the HVDC thyristor valves fail, SIEMENS builds in redundancy. This redundancy does
not consist in providing complete valves on a redundant basis, but rather more thyristors
are provided in series than are actually necessary based on the voltage. This makes it
possible for a component to fail occasionally without the valve as a whole becoming
inoperative. The redundancy is sized so that only one annual shutdown for maintenance is
necessary. During this downtime, any defective components are replaced, certain visual
and mechanical checks are performed and the protective settings are checked. Afterwards,
the valves are in like-new condition and continue their function so unobtrusively that their
presence is not noticed.

- 196 -
2.5 Smoothing Reactor

In the design of an HVDC station, smoothing reactor sizing is challenging and presents
several interesting questions: How large should it be? Where should it be arranged? Is a
smoothing reactor needed at all? Before attempting to answer these questions, a few
observations will be presented on the functions of an HVDC smoothing reactor.

2.5.1 Functions of the Smoothing reactor

Limitation of the rate of rise of current in the event of dc-side faults, i.e. line-to-ground
faults or commutation failures of the inverter station, in combination with dead time and
regulating speed of the rectifier current control, results in a limitation of peak short-circuit
current. Since dc current causes an equivalent current on the ac-side, the degree of
disturbance of the ac network is directly dependant upon this limiting function. For the
inverter which has suffered a commutation failure, the limitation of the rate of current rise
is critical to recovery of operation. The lower the rise of current can be held in the dc
circuit, the greater the chance that the next commutation which is due after 30° will take
place successfully.

The limitation of ripple of the direct current has already been discussed in Section 1.6.4 as
being important with respect to frequency transfer between asynchronous networks (non-
harmonic oscillations) and the avoidance of current discontinuities in the light load range.
The smoothing reactor plays a key role in this process, though the leakage inductances of
the converter transformer are also involved.

An increase in operating security can be achieved, because voltage jumps in one of the two
ac systems result in a change of the direct current until current control becomes effective.
The rate of change of direct current is inversely proportional to the effective inductance in
the dc circuit. In order to ensure that commutation failures in inverter operation will not
result due to rapidly rising direct current, the dc-side inductance should not be too low.

- 197 -
Avoidance of telephone interference emanating from the dc overhead line is of course an
essential function of the dc filter circuits, but the smoothing reactor also plays an
important role as a series impedance.

A dc-side resonance at the network frequency must be avoided. This danger exists for sea
cable connections with a length between 30 and 80 km. The resonant circuit can be
detuned through the selection of an appropriate smoothing reactor inductance.

In summary, it can be said that it would be possible, if necessary, to operate an HVDC


scheme without a smoothing reactor at least a back-to-back link, since there is usually a
transformer short-circuit voltage of 20% and the transformer leakage inductances present a
significant inductance in the dc circuit. But that is not the recommendation, in any case. In
long-distance transmissions with overhead line or cable, the smoothing reactor is probably
indispensable.

2.5.2 Sizing of the Smoothing Reactor

While the current and voltage rating of the smoothing reactor can be specified based on
the data of the dc circuit, the inductance is the determining factor in sizing the reactor. In
order to obtain a degree of effectiveness for the reactor which permits a functional
comparison for HVDC stations of arbitrary data, the HVDC Working Group
(Arbeitsgemeinschaft HGÜ) introduced the current slope factor Si. However, this factor
has not found general acceptance in international usage, probably due to difficulties in
understanding the concept. Nevertheless this term will be used in this discussion; there is
no equivalent substitute term.

U dN
Si = [in ms-1] (31)
L d * I dN

where
UdN = nominal dc voltage of the HVDC system per pole in kV
Ld = dc-side inductance in H
IdN = nominal direct current in A

- 198 -
The Si factor describes the rise of the direct current, related to the nominal value, within a
millisecond, when the nominal dc voltage is applied to the dc-side inductance. Therefore,
it characterizes the most important function of the smoothing reactor, i.e. the limiting of
rate of current rise.

There are no standards and also no well defined boundaries for the Si factor. Existing
HVDC systems, ignoring the few extreme cases, show a range of

0.22 < Si < 1

It might be recommended to begin with a value of Si = 0.5 in designing a station. This


value means that in the event of a dc-side short circuit directly behind the smoothing
reactor of the rectifier station, the direct current would rise within 5ms by 2.5 * IdN (thus,
when operating at nominal current, direct current increases to nominal current times 3.5).

Equation (31) gives the value of the dc-side inductance necessary to achieve the selected Si
factor. The leakage inductances of the converter transformer still have to be subtracted to
the extent that they are effective in the dc circuit. To explain this Fig. 2.40 shows the
equivalent circuit of a station pole with one twelve-pulse group.

Figure 2.40: Equivalent circuit for determining Ld

- 199 -
The condition presented is one in which a commutation process is just taking place in the
lower commutation group. This causes two transformer phases, including their leakage
inductances, to be connected in parallel. The dc-side inductance for this case is derived as
follows:

Ld min = 3.5 * Ltr + Ldr (32)

Ignoring operation with very high overcurrents, in which multiple commutations can
occur, Equation (32) indicates the lower value of dc-side inductance. It is a justifiable
mean for back-to-back links with uk = 20%, at least with nominal current, with which an
overlap angle of approximately u = 25° el can be expected. Thus a commutation is actually
taking place almost constantly somewhere in the twelve-pulse group.

For the situation at light load, u » 0°, the following holds true:

Ld max = 4 * Ltr + Ldr (33)

It is a matter of personal choice whether one calculates using equation (32) at uk = 12% or
15%, or if one uses an average value such as 3.7 * Ltr + Ld. For most cases of line-to-
ground faults, Ld is anyhow much larger, since a portion of the line inductance is included.

If a station pole has more than one twelve-pulse group, the predecessor factor to Ltr has to
be multiplied by the number of 12-pulse groups per pole in equation (32) and (33).

2.5.3 Arrangement of the Smoothing Reactors

In an HVDC long-distance transmission system, it seems quite logic that the smoothing
reactor will be connected in series with the dc line of the station pole. This is the normal
arrangement. There are some systems, however, such as Inga-Shaba (Zaire), CU (U.S),
and Skagerrak, in which the smoothing reactor is arranged between the converter and the
station neutral. The advantage, of course, is that the smoothing reactor does not carry any
dc potential. However, the "neutral" of the converter must be fully insulated. In newer
systems, however, only the normal arrangement can be found.

- 200 -
In back-to-back links, three fundamentally different arrangements of the smoothing reactor
are found, as Fig. 2.41 indicates.

Arrangement (a) has the advantage of a fixed connection of the valves to ground potential.
Arrangement (b) avoids a direct potential on the smoothing reactor.
In arrangement (c), only half of the direct voltage to ground occurs in the steady state.
However, to split the smoothing reactor is very expensive, at least in the case of reactors
under oil.

Recommendation: arrangement (a).

Figure 2.41: Possible arrangements of the smoothing reactor in a back-to-back link.

- 201 -
2.5.4 Reactor Design Alternatives

There are basically two types of reactor design:

- air insulated dry-type reactors


- oil insulated reactors in a tank

The air-insulated dry-type reactor is cost-effective for small reactor ratings (Id2 * Ldr). An
advantage of the dry type reactor is that maintaining spare units (to the extent necessary) is
not very expensive because the smoothing reactor usually consist of several partial coils.
On the other hand, dry-type reactors are sensitive to contamination. In earthquake regions,
setting them on post insulators or on an insulating platform is a critical problem.

The oil-insulated reactor is economical for very high rating (Id2 * Ldr). It is resistant to
earthquakes. Only the bushings are endangered by contamination. Maintaining of spares is
expensive. In back-to-back links, spares may reasonably be omitted because

- the probability of a reactor failure is very low


- operation is also possible without smoothing reactor, though with slight trade-offs

The subject of which type of reactor is more favourable with respect to noise generation
continues to be in dispute.

- 202 -
2.6 DC Filter Circuits

2.6.1 Function of the DC-Side Filter Circuits

Harmonic voltages which occur on the dc-side of a converter station cause ac currents
which are superimposed on the direct current in the transmission line. These alternating
currents of higher frequencies can create interferences in neighbouring telephone systems
despite limitation by smoothing reactors. Harmonic currents of lower frequency can cause
risks for human beings and devices through induced voltages. DC filter circuits, which are
connected in parallel to the station poles, are an effective tool for combating these
problems.

DC Filters will always be used if a telephone line is in close proximity to an HVDC


overhead line and adequate shielding is not present on both the telephone line and the
HVDC line. DC filters are not required in HVDC cable connections and HVDC back-to-
back links.

It can often be more economical to shield individually exposed sections of telephone lines
than to install a filter combination to handle the "worst case".

2.6.2 Design Criteria for DC Filter Circuits

The harmonic currents on the HVDC line create a magnetic alternating field. If a
telephone line is in the vicinity of the dc line, harmonic voltages will be induced in the
telephone line by the magnetic field. These harmonics in the telephone line and the
associated acoustic noise can significantly impair the quality of the telephone system.
Therefore it is common practice to use the telephone interference caused by the dc line as
a measure for the design of the dc filter circuits.

- 203 -
The interference voltage induced on the telephone line can be characterized by the
following equation:

å (H )
2
Vin(x) = Z * m * Cm * I m(x) = Z * Ieq
1

where
Vin(x) = Interference voltage on the telephone line at point x (in mV/km)
Z = Mutual coupling impedance between the telephone and HVDC lines
at 1000 Hz
Hm = Weighting factors which reflect the frequency-dependency of the
coupling between telephone and HVDC lines
Cm = "C message" - weighting factors (see also Section 2.2)
Im(x) = Resulting harmonic current of the ordinal number m in the HVDC
line at point x as the vector sum of the currents caused by the two
HVDC stations
Ieq = Psophometric weighted equivalent disturbing current

It is apparent from the above equation that the interference voltage is dependent only on
the equivalent disturbing current and the impedance between the HVDC and the telephone
lines. At the time when the system is being designed, this impedance is often unknown.
Therefore, in practice, often only the equivalent disturbing current is specified as a design
criteria.

The equivalent disturbing current combines all harmonic currents with the aid of
weighting factors to a single interference current. With respect to telephone interference, it
is the equivalent to the sum of all harmonic currents. It also encompasses the factors which
determine the coupling between the HVDC and telephone lines:

• Operating mode of the HVDC system (bipolar or monopolar with metallic or ground
return)
• Specific ground resistance at point x

For this reason, only an approximation of the equivalent disturbing current can be
determined.

- 204 -
There are no uniform standards for the maximum permissible equivalent disturbing
current. It is dependant on conditions in the vicinity of the HVDC lines and the philosophy
of the telephone and power system companies. The intensity of interference currents is
strongly dependant on the operating condition of the HVDC. In monopolar operation,
telephone interference is significantly stronger than in bipolar operation. As an example,
the following values can be used as guideline for the maximum permissible equivalent
interference current:

bipolar operation: Ieq £ 500 mA


monopolar operation: Ieq £ 800 mA (short-time operation)

In light of the above, it is not possible to precisely calculate in advance the equivalent
disturbing current caused by a planned HVDC system in existing or planned telephone
lines. For this reason, the permissible induced interference voltage in an assumed
telephone line, over a length of 1 km parallel to the HVDC system, at a distance of 1 km is
often specified. The ground resistance has to be specified, possibly with different values
for different sections of the prospective line. The equivalent disturbing current can then be
derived from the interference voltage (in mV/km).

Calculation of the equivalent disturbing current assumes knowledge of the distribution of


the individual harmonic currents along the HVDC line. The driving voltages of the
harmonic currents on the dc lines are the harmonic voltages at the HVDC converters of the
two stations.

In chapter 1.6, the origin of harmonic voltages on the dc-side of an HVDC converter was
described and the relationships of data and operating parameters were presented. A
distinction was made between characteristic harmonics of the ordinal numbers m = k * p
(p = 12 in 12-pulse converters) and non-characteristic harmonics, which can have any
other ordinal number.

The causes for non-characteristic harmonics are predominantely on the ac side of the
converter. They were discussed in detail in Section 1.5.5. The relationship between the
ordinal numbers of the harmonics on the ac side (n) and those on the dc side (m) must be
noted:

mi « n(i + 1) and n(i - 1)

- 205 -
2.6.3 Calculation of Equivalent Disturbing Current

The task is now to determine the most unfavourable combination of harmonic voltages
which can occur at one time in view of the operating modes and conditions of the HVDC
system which are to be considered, incorporating the anticipated unsymmetries and
tolerances. "Most unfavourable" means that combination which produces the highest C-
message value of the disturbing current. In making this calculation, the dependency of the
parameters of overhead lines, smoothing reactors, ground electrodes, etc., on frequency
must be taken into consideration.

The harmonic currents which are emanating from a station undergo a frequency-dependent
phase shift. Standing waves are formed along the HVDC line. These standing waves are
the superposition of the waves which are emanating from the two stations. The resulting
harmonic profile is derived from the vector addition of the harmonic currents of the same
frequency, emanating from the two sources, and being present at a particular location. The
phase angle between the two harmonic currents can thus be calculated for characteristic
harmonic currents. For non-characteristic harmonics, an angle of 90° is generally assumed.

However, if the frequencies of the two ac networks are not identical (asynchronous
networks) the method described cannot be used. Fluctuations between the harmonics
occur, forming non-harmonic currents which extend out into the two ac networks. The
only solution in this case is to perform a simulation study to determine the harmonic
profile responsible for telephone interference. In such a study, pragmatic assumptions and
simplifications must be used to limit costs to a reasonable level.

2.6.4 Induced Low Order Harmonics

In addition to telephone interference, harmonic currents on an HVDC overhead line can


also cause risks to human beings and animals, as well as to sensitive equipment. These
risks are caused by induced voltages. Of concern are primarily the low order harmonics
(those with ordinal numbers between 1 and 6). The determining factor is the geometric
sum of these harmonic voltages, as indicated in the following equation:

6
Ul. o. = åU 2
m
m =1

- 206 -
Uniform limit values have not yet been established. CCITT and AT&T have prescribed a
limit for induced voltages to ground for the frequencies of 50 and 60 Hz. The limits are
60 V and 50 V, respectively.

The ordinal numbers 1 to 6 denote non-characteristic harmonics in HVDC systems. One


should expect that their amplitudes would be so small that they could be ignored as has
been the practice in most cases. Measurements on existing HVDC systems have shown,
however, that low order harmonic currents do occur, the amplitudes of which were far
above predicted levels. This was especially true for the third harmonic. The causes for this
discrepancy, according to recent findings, are the leakage capacitances in the converter
station, particularly those of the transformer valve-side windings and the associated
bushings.

Fig. 2.42 shows the classic equivalent circuit for the calculation of dc-side harmonics of an
HVDC converter with the voltage source U12p. This generates the characteristic harmonic
voltages of ordinal numbers k*12. The voltage spectrum will not be altered if the leakage
capacitance Cs is added.

Figure 2.42: DC-side equivalent circuit of the 12-pulse converter

- 207 -
As was seen in Chapter 1.3 ("The Functions of the HVDC Converter"), a 12-pulse HVDC
converter is composed of 4 series-connected 3-pulse commutation groups. It can be seen
from the expanded equivalent circuit (Fig. 2.43) that the leakage capacitances Cs of the
wye and of the delta bridges are decisive factors. There are currents flowing toward
ground, which, in combination with the effective stray inductances Ls, cause voltage
drops. The amplitudes of the four voltage vectors which are to be added are no longer the
same, so that the low order harmonics do not fully cancel each other, leaving particularly
the harmonics of ordinal number k*3 in the total voltage.

In order to avoid unpleasant surprises, at least an estimating calculation should be carried


out. This will allow to estimate the effects of the third harmonic (and its multiples), which
are caused by leakage capacitances. Consideration must be given to the fact that currents
of the second and forth harmonics on the ac side correspond to a current of the third
harmonic on the dc side. If there is an existing network resonance for one of these
frequencies, these currents can cause considerable distortions.

- 208 -
Figure 2.43: DC-side equivalent circuit of a converter composed of four 3-pulse
commutation groups

- 209 -
2.6.5 Filter Circuit Configurations

DC filter circuits are usually arranged as parallel filters between the dc busbars and the
station neutral bus or ground. Series dc filters have been used only in isolated instances
and therefore will not be discussed further in this paper.

The configuration of the dc filters very strongly resembles the filters on the ac side of the
HVDC station. There are several types of filter designs. Single and double tuned filters
with or without the high-pass feature are common. One or several types of dc filters can be
utilized in a converter station. The characteristics of the various filter circuits were
described in detail earlier in Section 2.2, and will not be repeated here again.

Despite the many similarities, there are several important differences between dc and ac
filter circuits.

• AC filter circuits are essentially responsible for providing reactive power of


fundamental frequency. Therefore they are usually designed for higher reactive power
than would be necessary for the required filter effect. This design aspect does not apply
in the case of dc filters.

• The high voltage capacitor of the dc filter must withstand a high direct voltage. The
uniform distribution of voltage among the many series-connected capacitor units,
which can be automatically assumed in the case of ac filters due to capacitive current,
does not apply for dc filters. It is necessary to ensure the distribution of voltage by
means of parallel resistors.

• The ac system, to which the ac filter is connected in parallel, has a wide impedance
range. Thus a resonance can occur under certain network conditions between filter
capacitance and ac network inductance. For this reason a certain damping is necessary
even in the case of sharply tuned ac filter circuits. In contrast, the impedance of the dc
circuit is largely constant and therefore permits the use of sharply tuned filters.

The determination of filter configuration is made on the basis of achieving the smallest
equivalent disturbing current caused by the HVDC line and with minimum filter costs.
Since the characteristic harmonic currents have the largest amplitudes, the dc filters are
usually matched to these harmonics (i.e. ordinal numbers 12, 24, 36, ...).

- 210 -
The portion of cost of the entire HVDC system as represented by the dc filter circuits is
normally not great. The most cost-intensive element in a dc filter is the high voltage
capacitor. Therefore an effort is always to be made in the design of dc filter circuits to
optimize the cost of high voltage capacitors, i.e., to keep them as small as possible. One
means to this is the construction of double or multiple tuned filter circuits which have a
common high voltage capacitor.

In connection with the dc filter circuits the capacitor between the station neutral bus and
station ground must also be mentioned. It represents a low resistance path for transient
overvoltages and also for dc-side harmonics. In particular when the dc filter is connected
to ground, the capacitor prevents the harmonic currents from flowing through earth
electrodes and electrode lines to the station neutral point, causing telephone interference or
inducing unacceptable voltages. The low frequency harmonic currents described in the
previous section, which flow out through leakage capacitances, pass through this capacitor
to the station neutral point, which customarily has a capacitance of several mF.

2.6.6 Steady- state Stresses of DC Filter Circuits

Steady- state stress of the filter circuit capacitors is composed of two components: (1)
direct voltage, which causes a breakdown risk on the dielectric but will not result in any
heat generation, if the resistors built in for voltage grading are ignored, and (2) the
harmonic currents of the tuned frequency. In filter inductors and resistors, these harmonic
currents account for the entire steady state stresses.

In order to determine the maximum steady state stress of individual filter elements,
extensive calculations for various operating conditions are required. In particular, the
following parameters and/or operating conditions should be taken into consideration:

• The specified maximum continuous direct voltage


• The maximum characteristic harmonic voltages which can occur in the steady-state
(function of control angle and system direct current)
• If high-pass filters are provided, the maximum high frequency non-characteristic
harmonic voltages which can occur at the same time

- 211 -
• If low order filters are provided: unsymmetries of the ac side and leakage capacitances
in the station, so that the maximum filter current is achieved
• Harmonic currents flowing from the opposite station
• Fluctuations of ac-network frequency
• Filter detuning as a result of temperature fluctuation, manufacturing tolerance, aging of
the elements, loss of capacitor sections, and failure of capacitor cans
• Operating mode of the HVDC system (bipolar, monopolar, reduced direct voltage, etc.)
• Outage of a filter unit (if several are present)

2.6.7 Transient Stresses of DC Filter Circuits

Filter circuit components are subjected to various transient stresses as a result of faults and
switching procedures. For this reason, the reactors and the resistors in the filter circuit
have to be protected from overvoltages by arresters just as in ac filters. The following two
types of faults ordinarily result in the most severe stresses for dc filter components:

• A flash-over to ground occurring when the voltage at the dc busbar prior to the fault is
equal to the arrester switching surge protective level. The fault causes the capacitor
voltage to be applied as a surge to the filter reactor and the parallel connected surge
arrester, which immediately reaches its voltage limit. Arrester and reactor share the
discharge current of the capacitor, which means a high current stress for all elements.

• A line to ground fault occurring in the HVDC line at a certain distance from the
converter station. Again, the capacitor voltage before the occurrence of the fault rises to
the protective level at switching surge. Capacitance and inductance of the filter circuit
form a low frequency resonance circuit with line inductance to the location of the
ground short circuit. The discharge oscillations which are damped by the line resistance
are very important for the sizing of the filter components with respect to current and
energy.

- 212 -
2.6.8 Design Considerations

Basically the same aspects apply for the components of a dc filter circuit as were presented
for ac filter circuits in Section 2.2.7. The high voltage capacitor represents an exception.
Since only harmonic currents produce heating, the electric power is very low:

50
1 1 2
Qc =
wl * C
å m
* Im
m =1

This means that only a few capacitor units have to be connected in parallel. In contrast, the
direct voltage, which is usually high, requires the series connection of a large number of
capacitors. If the classic tiered arrangement on an insulating table is selected, a super slim
tower will be obtained which is very sensitive to wind and earthquake stresses. This
requires appropriate anchoring.

For this reason it can be more advantageous from a cost standpoint to suspend the high
voltage capacitor from a mast or portal by elastically fixed insulators.

2.6.9 Development Trends

While ac filters simultaneously provide reactive power compensation, dc filters have no


other function than to provide a low impedance path for dc-side harmonic currents and to
filter them out of the HVDC line. In view of the growing demand for more cost effective
designs of HVDC stations, the question of how to reduce telephone interference more
economically has to be addressed.

A more precise and realistic specification with respect to telephone interference could
significantly reduce the costs of dc filter circuits. In the past, limit values were often
specified for telephone interference which were based on the telephone technology of
twenty years ago. The interference effect of harmonic currents occurs primarily in open-
wire analog telephone systems. During recent years telecommunications technology has
developed to the extent that higher harmonic currents could well be permitted on HVDC
lines.

- 213 -
Modern digital transmission technology and mobile phones are both immune to
electromagnetic interference by an HVDC line. The need for dc filter circuits in HVDC
systems appears, at least in countries with well developed telecommunications technology,
to be becoming smaller. It is often more economical to take measures to shield telephone
lines which are still sensitive to interference than to install expensive dc filters. This would
entail converting sections of telephone lines which are subjected to interference from
HVDC overhead lines to cable.

"Active" filters represent a rather new development. In contrast to traditional filters, active
filters have a controlled current source which feeds certain harmonic currents into the line
through a power amplifier. These harmonic currents act to cancel the harmonic currents
from the converter. The converter, however, represents a voltage source with respect to dc-
side harmonics. The theoretically correct action would therefore be to place a controlled
harmonic voltage source in series to the converter and to control it in such a manner that
the harmonic voltages are precisely cancelled. A series source might present great
difficulties since the transmission direct current has to pass it. The active filters placed in
service to date are therefore integrated into the passive filter as an auxiliary device in the
form of a harmonic current source. The high voltage capacitor in the passive filter is used
to decouple the active filter from the high dc voltage.

- 214 -
2.7 DC Switchgear

Switchgear arrangements as known from the ac technique comprising a busbar system


from which a number of transmission lines are going out, each protected by a circuit
breaker, do not exist in HVDC systems to date. Nevertheless, there are - depending on
system type and station design - a number of HVDC specific switching functions which
require a more or less extended dc switchgear. This refers, however, exclusively to long
distance transmission systems. Back-to-back links require no dc switchgear; even if a link
comprises more than one block their dc circuits have no galvanic connection to each other,
switching occurs only on the ac side.

2.7.1 Functions of the DC Switchgear

The following switch types are defined according to their basic functions. The
nomenclature of the CIGRE Working Group 13/14.08 „D.C. Switching Equipment“ will
be used as follows:

· Disconnector D

· Earthing switch E

· Commutation switch S

· Circuit breaker B

With regard to the arrangement within the dc switchgear, the following areas are
distinguished:

· Transmission line L

· Converter C

· DC filter circuits F

· Neutral bus N

- 215 -
Fig. 2.44 shows the extensive dc switchgear in Itaipu, where the switching devices are
labeled according to the above nomenclature. Significant features are the capability of
parallel switching of the two bipoles using the disconnectors DL1 and DL3. It also has the
capability of commutation to the so-called metallic return, a mode of operation in which
the neutral bus is connected via the commutation switch SN2 and the disconnectors DN2,
DN5, and DL2 with one conductor pole. The MRTB, the "metallic return transfer breaker"
commutates the system current from "ground return" to "metallic return".

Figure 2.44: DC switchgear in the HVDC station of a long-distance transmission system


consisting of two bipoles ± 600 kV.

- 216 -
In the following the switching functions of the switch types mentioned are examined in
detail, to the extent they are HVDC-specific.

2.7.1.1 Disconnector

The converter disconnectors DC1, DC2, and DC3 must perform an HVDC-specific
function, when the associated group is to be disconnected during the operation of the
HVDC. After the bypass switch SC1 has been closed and taken over the current (see
2.7.1.3), DC2 is closed and thus an "external" bypass is produced. After the converter
group has been blocked, the bypass switch SC1 and then the disconnectors DC1 and DC2
may be opened. The converter group is then disconnected.

The return to service takes place in the reverse order. Before the bypass disconnector DC2
is opened, DC1 and DC3 are closed. The bypass switch SC1 is closed and then DC2 is
opened. During this process the disconnector must perform a current commutation
function. This can be easily handled by a normal disconnector because of the low loop
inductance.

In the event of a permanent dc line fault, almost full power can continue to be transmitted
if the two substation poles are connected in parallel. However, for this it is necessary to
change the polarity of the converter in the pole affected. This function, implemented at
Cahora Bassa, is shown in Fig. 2.45. The polarity reversal switches are operated with zero
dc voltage and current.

- 217 -
Fig. 2.45: Substation pole with polarity reversal switches for the two converter groups

The parallel switching of the substation poles is accomplished with the disconnectors DL1
of Fig. 2.44. The leakage capacitance of the affected substation pole is connected to the
full dc voltage. At Cahora Bassa, it proved to be necessary to equip the disconnector with
a closing resistor to prevent dangerous overvoltages.

The bipole parallel switches DL1 and DL3 shown in Fig. 2.44 were configured in the
Nelson River Project as rapid disconnectors. The normal ac breaker poles which were used
were able to handle energization of an unloaded line without any problem. Switch closing
resistors, which were envisaged, proved to be unnecessary.

- 218 -
If it is required that dc filter circuits be switched during operation, it must be verified that
normal disconnectors can fulfill these functions, otherwise ac breaker poles must be used.

2.7.1.2 Earthing Switch

Here, there are hardly any HVDC-specific functions. Perhaps it may be necessary to define
a current conduction and switching function for the line earthing switch EL1. Harmonic
currents can be induced by the current-conducting pole in a disconnected bipolar line pole.

2.7.1.3 Commutation Switch

The commutation of currents also occurs in ac switchgears at the time of a busbar change.
However, it has a far greater significance in direct current switchgear in HVDC
substations and demands significant switching capabilities from the switching devices.
Consequently, a type of commutation switch has been defined whose function is the
commutation of a direct current from one existing circuit to another parallel one.

The best-known example is the bypass switch. Fig. 2.46 depicts a 12-pulse converter
group which is part of a substation pole consisting of multiple groups. In operation, the
group disconnectors DC1 and DC2 are closed, while the bypass switch SC1 and the
bypass disconnector DC2 are open. If the group is to be taken out of operation, the dc
voltage is first brought to zero and then a bypass is established by activating a chain of
valves consisting of the four valves of one phase. Then, by closing the bypass switch,
commutation of the direct current is accomplished. The converter is then blocked, and the
additional switching procedures previously described are executed.

- 219 -
Fig. 2.46: Commutation circuit of the bypass switch

At the time of reconnection, the system current is -- as already described -- commutated to


the bypass switch. There are two processes for startup. In process 1, a bypass chain of 4
valves of one phase is formed. The bypass switch commutates the dc current into this path.
Its arc voltage must overcome the countervoltage of the valve chain. When the
commutation is finished and SC1 is completely opened, the transfer to the normal control
at zero voltage operation is performed in the converter. Then the delay angle is ramped
down to the desired value determined by the regulator.

In process 2, the converter is deblocked at zero voltage operation with the bypass switch
still closed and its mean current value is adjusted to the value of the reference direct
current. Thus, the bypass switch is relieved from the direct current, conducting at this
point only the harmonic current (predominantly 12th harmonic). With the opening of its
contacts and interruption of the harmonic current, the commutation process ends.
Operation continues again as in process 1.

The process selected depends on the switching capability of the available bypass switch.
Process 1 is the more elegant; it avoids the high harmonic currents, but requires a high
switching (arc) voltage.

- 220 -
Another commutation switch depicted in Fig. 2.44 is the neutral bus switch SN1 which is
essential when a short-to-ground exists in the associated substation. Part of the current
from the conducting pole flows to the ground through this switch as an auxiliary path to
the earth electrode. In this case, SN1 must commutate the dc current flowing to the
substation ground into the earth electrode line.

A third commutation switch is the GRTS, the "ground return transfer switch", SN2. It is
essential, prior to returning to normal bipolar operation after operation with "metallic
return", that brief operation with "ground return" be utilized as an intermediate step. In
order to commutate the remaining current still flowing over the dc line after the closing of
the MRTB, into the earth electrode line, the GRTS is opened.

2.7.1.4 HVDC Breaker

Apart from the MRTB in the earth electrode line, Fig. 2.44 contains no HVDC circuit
breaker. In fact, circuit breakers have not been used in HVDC systems to date; however, in
HVDC point-to-point transmissions with double line and with meshed HVDC multi
terminal systems HVDC breakers promise significant operational advantages. Their use
may be expected in the future.

It is possible to distinguish two areas of application:

· as line breaker

· as substation breaker

An HVDC circuit breaker as a line breaker is required only if line redundancy is present,
i.e., with a point-to-point HVDC system, in which one double line, or even two single
lines via separate routes link the two HVDC substations , or in an HVDC multi terminal
systems with meshed line systems, e.g., in a ring shaped network.

- 221 -
In the event of a line-to-ground fault caused by lightning or insulator pollution, the HVDC
circuit breakers on both ends of the line section affected by the fault open and clear the
fault without the HVDC converters having to block the dc current. This is significantly
faster than the conventional process consisting of blocking of the converter, a zero current
pause of approx. 200 ms to deionize the fault site, and then restarting the HVDC system.
As the simulator test shown in Fig. 2.47 demonstrates, the power exchange is restored
after approx. 100 ms.

Figure 2.47: Simulator test for clearing a line-to-ground fault by HVDC breakers. The
rectifier substation keeps its current Id1 constant; the current of conductor A is
commutated to conductor B. After approx. 100 ms, the fault incident is
cleared.

It remains an open question whether the reduction of the fault time from approx. 250 ms
(with conventional line protection) to approx. 100 ms using HVDC breakers justifies the
additional costs. It is, of course, necessary to consider that with utilization of HVDC
breakers, also persistent line faults are cleared in 100 ms, which would otherwise require
multiple startup trials or even long-term outages of a pole (e.g., in the case of a fire under
the line).

- 222 -
Fig. 2.48 depicts--likewise in single-pole representation--an HVDC multiterminal system
with ring configuration, the simplest form of the meshed networks. The clearing of a line
fault occurs in the manner described above. A parallel current path into which the current
is commutated with the opening of the breakers is also available here. The current of the
HVDC substations remain unchanged. This has the advantage that restarting, which is
somewhat problematic in a multiterminal system because of the necessary time
coordination, is eliminated.

Fig. 2.48: HVDC multiterminal system with ring shaped network

Fig. 2.48 also shows breakers in the connections to the HVDC substations, i.e., substation
breakers. By analogy with the line breakers it is possible to see that their use makes sense
only where substation redundancy is present, i.e., in a multiterminal system, but also in
point-to-point connections if two or more converters are connected in parallel (to date,
implemented only at the Ekubastus HVDC).

Despite the redundancy, a current commutation function is not possible here. The current
controllers of the remaining converters of the same operational mode do not permit
commutation of the disconnected substation current. Thus a current interruption function
is required, which places significantly greater demands on the breaking capability of the
HVDC circuit breakers. Nevertheless, in future HVDC multiterminal systems, it will be

- 223 -
necessary to demand the use of substation breakers, in particular, when the associated ac
network is a "weak" one, such that frequent commutation failures must be expected. Each
persistent commutation failure implies the interruption of the power exchange in the entire
multiterminal system and a high overload for the substation in question. The quickest
possible disconnection of the faulty substation is needed to save the operation of the
remaining system.

For the sake of completeness it should be added that the use of HVDC breakers in
multiterminal systems as both, line breakers and substation breakers, has been intensively
studied in the aforementioned CIGRE working group.

- 224 -
2.7.2 Rating of DC Switching Devices

In the specification of HVDC switching devices, the following data are essential:

· Current rating
- Steady state current and overload requirements
- Fault current (peak value)
- Turnoff current (dc and ac current, frequency)
- In-rush current (peak value)

· Voltage rating
- Voltage to ground on the top/bottom connector
* dc voltage
* lightning surge voltage
* switching surge voltage

- Voltage across open contacts


* during opening
* continuous dc voltage
* lightning and switching surge voltages

· Switching (with commutation switches and breakers)


- energy to be absorbed per switching operation
- switching cycles
- closing and opening times

In the following, only those rating criteria which are HVDC-specific and not obvious will
be discussed for the various switch types.

2.7.2.1 Disconnector Rating

The different potentials of the "top" and "bottom" connectors of the disconnectors in the
converter area are an interesting aspect. Fig. 2.49 depicts a substation pole of a 600 kV-
HVDC system with two 12-pulse groups. All switching devices shown have a nominal
voltage of 300 kV across their contacts. The potentials of their connectors are either 600,
300 or (0) kV. The substation neutral bus is at zero dc potential, but is insulated for 100
through 200 kV BIL. With single-pole operation, the substation neutral bus may have a dc
potential of 10 to 60 kV. Utilizing pantograph isolators, it is possible to accurately adapt to

- 225 -
these requirements, as the direct current switchgears at Cahora Bassa and Nelson River 2
demonstrate.

Fig. 2.49: Potential on the disconnectors in the converter area

Reference has already been made in Section 2.7.1.1 to the exceptional features of the
parallel disconnector DL2, which applies voltage to the stray capacitances and must
possibly be equipped with a pre-insertion resistor. This also applies to the filter
disconnectors DF1, which must have a significant opening and closing capability if the
filter circuits have to be switched when the dc voltage is present.

- 226 -
2.7.2.2 Earthing Switch Rating

Section 2.7.1.2 discussed the harmonics induced in a switched-off line pole by the other
pole which determine the requirements of the line earthing switch, EL1. The maximum
load must be determined through analysis.

The Nelson River 2 dc switchgear presents an additional exceptional feature. Here, it was
necessary to consider the switching state shown in Fig. 2.50. The two 12-pulse groups of
one pole have gone out of operation and their bypass switches were closed even before
SN1 opened. Then the current of the other pole Id splits into one (larger) subcurrent IdE via
the earth electrode and a "spill current" Idsp via the conductor of the disconnected pole. The
commutation of this current exceeds the capability of SN1 because of the great
inductances of the mesh. By closing the rapid earthing switch EN6, the line becomes
current-free and SN1 is capable of commutating the much greater current flowing via EN6
back into the earth electrode line, since the loop is virtually free of inductance.

Fig. 2.50: Disconnection of the spill current using a rapid earthing switch

- 227 -
A high inrush current will be specified for the earthing switch EF1 to avoid having to wait
for the complete discharge of the filter capacitors before grounding.

2.7.2.3 Commutation Switch Rating

In addition to current and voltage rating, the commutation switch also has to absorb
switching energy. Fig. 2.51 schematically depicts a commutation circuit. The commutation
switch S must apply a switching voltage US greater than that of the countervoltage source
UZ which might be present in the commutation loop. If the effect of the ohmic resistance R
is ignored, as is generally permissible, the switching energy to be absorbed by the switch S
is as follows:

L 2 US
ES = * Id * (34)
2 US - UZ

where Id = current to be commutated


L = inductance effective in the commutation loop
US = switching voltage (arc voltage)
UZ = counter voltage in the commutation loop

Fig. 2.51: Commutation loop

- 228 -
In Eq. (34) the switching voltage is assumed to be a dc voltage. If US is the arc voltage,
this assumption cannot be made since the switching voltage is largely dependent on the
current and would be highly fluctuating.

In an actual situation, lab tests are almost indispensable. Nevertheless, Eq. (34) gives
important indications concerning the switching requirements to be anticipated.

Consider the bypass switch in process 1 (see 2.7.1.3). UZ must be interpreted as the
threshold voltage of the thyristors connected in series in the valve chain. The commutation
process can function only if US is greater than UZ. A solution would be to equip the switch
with a commutation circuit as is used with the HVDC breaker (see next section).

In process 2, UZ can be interpreted as the mean value of the converter voltage. If this is
adjusted slightly in a positive direction (a < 90o), the sign in the denominator of Eq. (34)
will be changed, and the switching energy may be reduced to very small values.

Very generally, the process for determination of the switching energy of a commutation
switch can be represented as follows:

- define the commutation loop


- determine the actual quantities L and R
- determine UZ = UZ0 + IS * R
- verify whether US > UZ

Also the commutation duration is important; it can be estimated as follows:

1
tk = * L * Id (35)
US - UZ

- 229 -
2.7.2.4 Circuit Breaker Rating

Consider a line breaker: Its function is that of a commutation switch; however, much
higher switching energy is required of it than was dealt with in Section 2.7.2.3, because the
commutation circuit is spatially quite extensive and includes the inductance of hundreds,
in certain circumstances more than 1000 km of overhead line. This function can be
managed only if the switch applies a switching voltage US of several 100 kV. This is
impossible with arc voltage alone.

In the 1970's, the HVDC Switch Research Group, which included the companies of the
Arbeitsgemeinschaft HGÜ (HVDC Working Group) plus the ASEA, in cooperation with
the Technical University of Braunschweig developed a dc circuit breaker taking these
conditions into account. In view of the low numbers of units expected to be needed, a
design was conceived which utilizes conventional, commercially available components.
Fig. 2.52 shows the principle of the breaker. The primary contact ➀, provided by an ac
circuit breaker, conducts the line current. In the event of a fault, the breaker contact opens,
the arc voltage excites the commutation circuit ➁, and oscillations build up until the
resonant circuit current reaches the amplitude of the line current and the arc is
extinguished. The dc current now flowing through the resonant circuit causes a linear
increase in the voltage on the capacitor CS, whereby the du/dt has to be limited by the
selection of the capacitance such that a re-ignition of the switch contact ➀ is avoided.
After a few 100 ms, depending on the amplitude of the line current, conduction of the
energy absorber ➂ in parallel to ➀ is attained. The energy absorber ➂ consists of metal
oxide elements with adequate energy dissipation capability. The wye connection of the 3
elements of the energy absorber limits both the voltage across the contact and the voltage
of the two connections to ground to a value of 1.6 x UdN. This provides an adequate
insulation coordination.

- 230 -
➀ primary contact
➁ commutation circuit
➂ energy absorber

Fig. 2.52: The principle of the HVDC breaker

If the derived relationships are applied to the case of the double line and if both, the ohmic
resistance and the capacitance per unit length of the lines are ignored, the simplified circuit
shown in Fig. 2.53 will be obtained.

- 231 -
Fig. 2.53: HVDC point-to-point transmission system with double line

Eq. (34) can be simplified as follows:

L1 + L 2 2
E= * I dl
2

where E is the energy to be absorbed by the breaker S upon commutation of the current Id1
from line 1 into line 2. It is interesting that neither the line voltage Ud nor the inductance of
the smoothing reactor L0 play a role in the energy equation. With regard to the energy to be
absorbed, even the switching voltage US is inconsequential. It is significant only with
regard to the commutation time. For this case, Eq. (35) simplifies to:

1
tk = (L1 + L2) * Idl
US

Since there is naturally great interest in a short commutation time, which is the fault
clearing time, US is selected as high as can be justified from the standpoint of insulation
coordination.

- 232 -
Consideration of transition from the double line point-to-point connection to a meshed
multiterminal system in no way alters the sizing equations; the applicable data must
merely be used for L1 and L2.

The situation is different for the substation breaker, which does not--as already
established--have to perform a current commutation function but rather a current
interruption function. The simplified circuit according to Fig. 2.54 applies here. In this
case, both the smoothing reactor inductance L0 and the system voltage Ud must be
considered.

Fig. 2.54: Current interruption by a substation breaker

The energy to be absorbed by the breaker S during the current interruption is according to
Eq. (34)

1 æ L * L2 ö 2 Us
E= * çL0 + 1 ÷ * Id *
2 è L1 + L 2 ø Us - Ud

This energy can well be a total of 5 to 10 times the amount to be absorbed in the case of
the commutation process. This requirement can nevertheless be fulfilled; however, the
energy absorber becomes rather expensive.

The question remains whether the simplifications made are permissible. It is possible to
answer yes with regard to ignoring the ohmic resistances, but not for the capacitance per
unit length of the lines. The operation of the HVDC breaker is definitely associated with a

- 233 -
sharp rise in the switch voltage, which has the form of a travelling wave that is
propagating along the line.
A line-to-ground short circuit can occur anywhere in the system. Then the system voltage
becomes zero, while the converters continue to operate under the effect of their current
regulation. If the switch is opened and the line surge impedance is equal on both sides,
traveling waves of differing polarity but the same peak value, ½ US, will propagate in both
directions (Fig. 2.55) and will charge the distributed capacitances. The energy required for
this is derived from the energy stored in the conductor inductances. This process is linked
with a current change: C * U2 = L * (I1 - I2)2.

Figure 2.55: Travelling wave generation upon switch opening

- 234 -
This can also be explained in another manner if we consider the lines replaced by their
surge impedance Z. Then a voltage rise corresponds to a corresponding rise in current:

US
ID = (36)
2*Z

The height of the rise in current can be easily estimated. If Z = 400 W and
US = 1.6 * 500 kV = 800 kV, then ID = 1.000 A.

The following is generally true: If the current to be disconnected or to be commutated is


less than ID, the switch voltage will be not reach the turn-on voltage of the energy absorber
and the energy to be absorbed will be zero in the first approximation. If the current to be
disconnected is larger, it will be reduced by ID upon opening of the switch. This reduced
current now remains constant until the return of the travelling wave reflected at the fault
location, whereupon a new current reduction of ID takes place, etc.

This analysis which more nearly approaches reality replaces linear current reduction with
step-wise current reduction. The energy absorption of the switch yields smaller values with
the surge impedance method; however, higher currents to be interrupted diminish the
differences.

2.7.3 Design Considerations

To date, all existing dc switchgears are of the open air type. This statement does not
conflict with the fact that there was once an encapsulated, SF6-insulated back-to-back link
in Manhattan, a trial supported by the EPRI, which was recently discontinued without
success, nor the Gotland II plant, where the connections between the converter
transformers and the wall bushings were encapsulated with SF6-insulation. The nominal
dc voltage of this Gotland plant is 150 kV only; of more interest is the first expansion of
the Pacific Intertie HVDC system where the 400 kV link between the old mercury- arc
converter and the new thyristor bridge was designed as encapsulated, SF6-insulated line.

- 235 -
There are significant problems with SF6 insulation of conducting parts against high dc
voltages. Barrier insulators show problems for which contamination of the gas chamber
by particles is not an adequate explanation. Studies conducted in the late 1970s by the then
BBC-Mannheim in cooperation with the high voltage institutes of the Technical
University of Darmstadt and the Technical University of Munich revealed the following
about SF6: Up to 100 kV no problems were detected; up to 300 kV with careful sizing and
superclean installation a reliable operation was possible; 500 kV functioned only for a
short time, and 600 kV not at all!

2.7.3.1 Disconnector Design

In present dc switchgears, virtually all types of disconnectors are found, often several types
in one plant. Considering the variety of connection potentials depicted in Fig. 2.50, it
would seem that preference should be given to the pantograph disconnector, particularly
since it has only one contamination prone insulator; unfortunately, the pantograph
disconnector is expensive and not always easy to incorporate into a plant design. The fact
that it is possible to equip a pantograph disconnector with a closing resistor was
demonstrated at Cahora Bassa.

2.7.3.2 Earthing Switch Design

For many variants, an HVDC direct current switchgear requires the arrangement of
earthing switches on both sides of the disconnector. Since not all disconnectors permit
this--e.g., the pantograph isolator--earthing switches are occasionally needed as individual
elements.

- 236 -
2.7.3.3 Commutation Switch Design

For commutation switches and for rapid disconnectors, poles of modern ac circuit breakers
are generally used, although for the bypass switch in mercury arc valve converters other
complex designs have been common. Reference has already been made to the problems of
the necessary minimum arc voltage with certain commutation functions.

2.7.3.4 HVDC Breaker Design

Fifty years have passed since the first tentative efforts to develop an HVDC breaker. An
almost incalculable number of switching principles and design solutions have been
proposed. The design elaborated almost 30 years ago by the Arbeitsgemeinschaft HGÜ
has been accepted and is currently generally recognized. Yet there is a certain tragedy in
this solution.

At that time a minimum-oil breaker was selected which builds up a particularly high and
stable arc voltage. The commutation circuit consisted of a simple capacitor. Ten years
later, when a 500 kV prototype was to be built, the minimum-oil breaker no longer
existed. The change to the compressed air breaker was made as a "second best" solution
but the commutation circuit had to expand to a resonant circuit. Today's technology is the
SF6 breaker with an even lower arc voltage. Will we have to develop a new commutation
circuit in 10 years? Or even a new switching principle? Will they require forced
commutation?

- 237 -
3. HVDC Overhead Lines and Cables
3.1 The HVDC Overhead Line

3.1.1 Introductory Comments

The specifications and orders for HVDC overhead lines are generally written and issued
separately from those of HVDC stations. Only in rare cases, such as in the Cabora-Bassa
Project, are the two sections of the project bundled together into a comprehensive package.
Nevertheless, the engineers responsible for the design of HVDC stations must also be
familiar with the most important phenomena which occur in an HVDC overhead line and
their relationship to design, dimensions and other parameters. Familiarity with these
phenomena can also be very useful during preparations for a project, in the optimization of
the system and during meetings with the client.

This section discusses the following topics:


• Different types of HVDC lines
• Required minimum clearances
• Conductor surface gradient and corona
• Electrical and magnetic fields
• Line insulation

This chapter does not discuss questions of cost in detail, please refer to Section 1.1.5.

- 238 -
3.1.2 Types of HVDC Overhead Lines

HVDC systems with overhead lines - in contrast to systems using cables - are almost
always bipolar systems. Monopolar systems, however, are conceivable as the initial stage
of a system which, in the final stage, will be a bipolar HVDC system, or as overhead line
section of a cable transmission system. Nevertheless, a wide variety of types of HVDC
overhead lines have been constructed or at least planned. In the decision to construct a
certain type of HVDC line, both, the reliability of the overall system and construction
costs play important roles. Consideration of the effects of an overhead line on the
environment is also becoming increasingly important.

3.1.2.1 The Bipolar Line

Figure 3.1 shows the


tower for a typical
bipolar overhead
line. The most
important factor in
the design is the
required minimum
clearance. (See
Section 3.1.3)

Figure 3.1: Bipolar HVDC line

In determining the minimum clearance against the tower structure, the swing width of the
insulator strings under maximum wind load must be taken into consideration, as indicated
in the left half of Figure 3.1. This swing width can be reduced, and thus the crossarm can
be kept shorter, if the conductor is suspended on V-strings, as illustrated in the right half
of figure 3.1. The required width of the right-of-way can be reduced by this means, but the
reduced clearance between the conductors has the unfavourable influence of increasing the
conductor surface gradient and the effects which are a function of it. (See 3.1.4).

- 239 -
HVDC overhead lines are conventionally protected against direct lightning strikes by
means of an overhead ground wire over their entire length. On the last few spans near the
HVDC stations, sometimes there are two overhead ground wires to guarantee an absolute
reliable protection against a lightning strike at these sections. Otherwise a steep-fronted
overvoltage surge could enter the station. For this purpose, the top of the tower has a small
crossarm or is split in a Y-shape.

If the line runs through regions with very high specific ground resistance, it may also be
necessary under some conditions to connect the towerfeet to one another by means of an
uninsulated cable - also called a counterpoise - laid in the earth. The purpose of this cable
is to lower the footing resistance and to prevent back flashovers between the tower and
conductor in the event of a lightning strike which hits the tower or the surrounding earth.

3.1.2.2 The Monopolar Line

Figure 3.2 shows the tower of a monopolar


HVDC line, such as the towers constructed for
the Cabora-Bassa System (but with I-strings
rather than V-strings). In this case, the two poles
were constructed on two separate rights of way
running 1 km apart. The intent was to prevent
the loss of the entire system in the case of
extraordinary events such as tornados. (In
practice, this design has proven quite
successful; during the numerous bush fires, the
fire-related flashovers on the one pole were
usually over before the first flashovers occurred
on the second pole).

Figure 3.2: Monopolar HVDC line

- 240 -
One peculiarity of the Cabora-Bassa line is the weakly insulated overhead ground wire,
which forms the return conductor for the PLC (power line carrier) system. Lightning
protection is not adversely affected by this insulation, as confirmed by the operating
results. Although the line runs through an area with the highest level of atmospheric
electrical discharges in the world, there have been practically no lightning strikes to the
main conductor, although there have been innumerable strikes into the overhead ground
wire. The unshielded porcelain insulators of the overhead ground wire have proven to be
resistant to flashovers.

3.1.2.3 The Homopolar Line

Homopolar line is a special form of monopolar line. In terms of its construction and the
configuration of the towers, it is the same as the bipolar line, illustrated in Figure 3.1. But
both conductors have the same polarity, representing one pole of a bipolar system. The
second pole should preferably be run on a separate right of way. By using HVDC switches
on both ends of each line, a redundant system is obtained. In the event of a line-to-ground
fault - whether transient or permanent - the defective conductor is disconnected.
Transmission of full power can be resumed after approximately 100 ms. The remaining
conductor of the affected pole will carry twice the current, but with a line designed for an
economical current density, this is still below the thermal limit.

Figure 3.3: HVDC with 2 homopolar lines

- 241 -
If the system shall also include protection against tower collapse, load transfer switches
must be provided in the converter substations, by means of which the remaining line can
be converted into a bipolar line. In that case, the same reliability can be achieved as with
the installation of two bipolar lines, but this design has disadvantages in terms of the
conductor surface gradient (See 3.1.4).

Figure 3.4: Load transfer switches

- 242 -
3.1.2.4 The Double Line

A double line will also make available a redundant conductor, if there are HVDC switches
as illustrated in Figure 3.3. Naturally, the case of the collapse of a tower is not covered and
would result in the total outage of the HVDC system. The advantage of the double line is
that only one right of way is required, and construction costs are significantly lower than
with two single lines.

Figure 3.5 shows two possible configurations of a double line. At the left, the tower carries
the two bipoles in two levels. At the right, the two bipoles are in one level, to the right and
left of the tower. The latter design is more expensive and in spite of its lower height is
visually more obtrusive; it also requires a wider right of way. On the other hand, with this
design it is easier to build the second system at a later stage while the first bipole is in
operation.

Figure 3.5: Two versions of an HVDC double line with double and single crossarms,
respectively

- 243 -
3.1.2.5 The One-and-a-Half Line

If, for reasons of reliability, the decision is made to give a bipolar line the feature of
redundancy, but not to assume the high expenses for a double line, the one-and-a-half line
may be a suitable solution. In this system, one pole is designed to carry the full current as
in a single line, and the second pole is split into two conductors as with a double line. In
the HVDC substations, load transfer switches are necessary so that bipolar operation can
be continued even in the event of the failure of the common conductor.

No such lines have yet been constructed, and they will probably be considered only for
very long distances in which the cost of the line is the determining factor.

Figure 3.6: One-and-a-half line

- 244 -
3.1.2.6 Line with Neutral Bus Conductor

If even for a short period of time, current through the earth is not permitted in an HVDC
system, availability of the system can be increased by connecting the two substation
neutral buses to one another by a third conductor. In the event of the outage of one pole,
50% of the nominal power can then be transported for an unlimited period of time. The
neutral bus conductor can be designed so that it has approximately one-half the cross
section - corresponding to the thermal limit current - since line transmission losses do not
play any role during emergency operation, and there are no problems involving conductor
surface gradient. The earth electrodes and the electrode lines can also be eliminated.
Nevertheless, the added expense of such a system is significant. The extra cost can be
reduced if the neutral bus conductor is strung on top of the tower, and is assigned the
function of the overhead ground wire.

Lines with neutral bus conductors may be desirable in HVDC multiterminal systems, since
otherwise the failure of one station pole would always require the shutdown of the second
station pole, if return current through the ground is not permitted.

Figure 3.7: HVDC system with neutral bus conductor

- 245 -
3.1.2.7 Comparison of the Different Types of Lines

The following table presents the different types of lines discussed above from the point of
view of redundancy. It shows the transmission capacity which remains in the event of a
line fault.

Remaining Transmission Capacity

Varient Type of Tower Failure of one Pole

With earth return Without earth return Collapse of


tower or fire

3.1.2.1 0.5 0 0 *)
bipolar (1)
line

3.1.2.2 0.5 0 0.5 *)


Two (1) (1)
monopolar
lines

3.1.2.3 1 1 1
Two
homopolar
lines

3.1.2.4 1 1 0
Double line

3.1.2.5 1 1 0
One-and-a
half line

3.1.2.6 --- 0.5 0


Bipolar line
with neutral
bus
conductor

*) When the station poles are switched in parallel, the values in parentheses apply.
Table 3.1: Comparison of the different types of lines

- 246 -
Unfortunately, it is impossible to give a universally valid ranking of the costs of the
various types of lines. Such a ranking is a function of the power being transmitted and of
the line voltage. Each of the possible variants must be calculated for each individual
project, using the line cost formulas indicated in Section 1.1.5, with adjustments as
necessary. Such a process can be used to determine approximately how much the
increased reliability of a redundant line would cost.

3.1.3 Minimum Clearances

As with ac lines carrying nominal voltages > 300 kV, the magnitude of the switching
surge is the determining factor for the minimum clearances (flashover distances) which
must be maintained in HVDC overhead lines. That may seem surprising, since none of the
HVDC systems built and operated so far have been equipped with HVDC breakers. But
given the conventional bipolar HVDC overhead line, in the event of a line-to-ground fault
of one pole, the capacitive coupling of both line poles will also influence the potential of
the intact pole in the same direction. Because of the operating voltage of opposite polarity,
the operating voltage and switching surge add up to an overvoltage of 1.5 to 1.8 p.u. peak
value.

In monopolar lines with a large clearance between the two poles, the capacitive coupling is
either completely absent or very low, and the phenomenon referred to above either does
not exist or is very much attenuated. Nor does this overvoltage occur with homopolar
lines, since the operating voltage of the intact pole and the switching surge are in the
opposite direction. In that case, a reduction of the rated switching surge might be
considered, but caution is advised in this respect, since there is a whole series of internal
and external abnormalities which can result in overvoltages on the dc side. In any case, a
basic study of insulation coordination must be conducted before a reduction of the rated
switching surge and thus of the minimum clearances can be considered.

- 247 -
The relationship between the rated switching surge and the minimum clearance is very
complex. The determining factors include not only the wave form of the switching surge
and the shape and location of the electrodes, but also the atmospheric and climatic
conditions, the influence of which is also a function of polarity. Extensive literature on the
subject describes the studies which have been performed in numerous high voltage
laboratories to define this relationship. The following table is taken from the
"Transmission Line Reference Book HVDC to ± 600 kV" published by EPRI, and
indicates the minimum clearance in air as a function of the overvoltage factor:

Over-voltage Minimum clearance in m at nominal voltage (kV)


factor (p.u.) 250 400 500 750
£ 1.5 0.91 1.37 1.83 3.35
1.6 0.91 1.37 1.98 3.66
1.7 0.91 1.52 2.13 4.11
1.8 1.07 1.62 2.20 4.57

Unfortunately, IEC and VDE standards do not so far include any information on the
minimum clearances which must be maintained in HVDC systems. If the rated dc voltage
of an HVDC system is assumed to be equal to the peak value of the rated voltage to
ground of an ac line, minimum clearances which are much too high are obtained. That is a
result of the much higher overvoltage factors which must be taken into consideration on ac
lines.

The minimum conductor-to-ground clearance must naturally take into consideration the
safety clearances specified in national standards, which are a function of the use of the
land over which the line travels, intersections with roads and transportation routes etc. The
maximum conductor sag at the maximum continuous allowable conductor current and the
least favorable environmental conditions must also be taken into consideration. In
particular when there is a redundant line, and in the event of the failure of one line, the
current in the remaining line can be close to the thermal limit current, resulting in a large
conductor sag.

- 248 -
3.1.4 Conductor Surface Gradient and Corona

Corona is defined as an electrical discharge accompanied by luminous phenomena as the


result of the ionization of the air surrounding a conductor. It occurs when the field strength
exceeds a critical value. A distinction is made between two zones around a conductor:
• the ionization zone
• the space charge zone.

The ionization zone is a thin layer around the conductor; its thickness is on the order of
magnitude of one tenth of a percent of the distance between the electrodes. In this zone,
the high field strength causes charge carriers to collide at high velocity with air molecules
and ionize them. The electrons released are accelerated toward the positive conductor and
away from the negative one, and thus once again collide with air molecules, causing an
avalanche effect to occur.

Some of the charge carriers penetrate into the space between the conductor and ground,
i.e. into the space charge zone, where they are decelerated and ultimately recombine. Since
these charge carriers are constantly replaced from the ionization zone, an ion current and
associated corona losses are produced.

The critical field strength at which ionization begins has been determined to be 29.8
kV/cm. In tests with industrially-manufactured conductor cables, the corona effect was,
however, found to occur at voltages as low as approximately 15 kV/cm. The reason for
this effect at such a low field strength is the presence of small defects and irregularities in
the surface of the conductor, which lead to local areas in which the critical field strength is
exceeded.

The maximum field strength which occurs in the conductor surface can be calculated if the
following parameters are known:
• the conductor-to-ground voltage
• the diameter of the strand
• the number of strands per bundle
• the clearance between the strands in the bundle
• the height of the conductor above ground at midspan

and for bipolar or homopolar lines:

• the clearance between conductors.

- 249 -
Formulas indicated in the literature, some of which are quite complicated, give results
which are only approximately correct. One reason is that with increasing voltage, the
ionization zone expands, so that a conductor bundle ultimately appears as a cable of
correspondingly large diameter.

Reliable computer programs are currently available to calculate the maximum surface field
strength.

3.1.4.1 Corona Losses

The corona losses of an HVDC line are determined essentially by the conductor surface
gradient, i.e. they are defined primarily by the parameters of the line. Results determined
in various long-term field tests and by the approximation formulas which have been
derived to describe the relationship between surface field strength and corona losses are
not too good in agreement.

Even with the same line parameters, the results determined vary widely. For example, the
relationships illustrated in Figures 3.8 and 3.9 taken from the above-mentioned EPRI
Reference Book were determined with a bipolar test line. The line is equipped with a 2 x
46 mm two-conductor bundle, has a distance between poles of 18.3 m and a conductor
height of 15.2 m in the midspan. Figure 3.8 shows the results in fair weather, and Figure
3.9 the results in rain weather. A comparison of the two figures shows that the range of
corona loss values is significantly larger for fair weather than for rain weather.

- 250 -
Figure 3.8: Corona losses Figure 3.9: Corona losses
in fair weather conditions in rain weather

It is apparent that the average value of the corona losses in rain weather is three to five
times higher than in fair weather conditions. For individual measurements, this weather
factor can cause the results to differ by a factor of up to 15. In ac lines, however, the
weather factor can be 50 or higher.

Figures 3.8 and 3.9 also show that the corona losses, in a first approximation, are
proportional to the square of the surface gradient.

- 251 -
Other atmospheric conditions investigated to determine their influence on corona losses
are as follows:
• air temperature
• air pressure
• wind speed
• wind direction
• relative humidity and
• absolute moisture content of the air.

Of these, only the wind speed has a significant effect on the corona losses. Ions are
apparently removed by the wind from the ionization zone and are then replaced by
increased ionization, leading to increased corona losses. To get a feeling for the order of
magnitude, the following rule of thumb has been determined empirically:

Vk = 1.6 + 0.44 × Vw (at ± 600 kV)

where Vk = corona losses in kW/km per pole


Vw = wind speed in m/s

The influence of the other parameters was small or insignificant.

3.1.4.2 Radio Interference

The corona effect causes a wide-band electromagnetic radiation which can lead to
interference, particularly with AM radio transmission. The sources of these radiations are
very different. On the negative conductor, there are Trichel pulses which are distributed
approximately in a uniform manner over the surface of the conductor and contribute very
little to the radio interference measured. On the positive conductor, there are several
mechanisms which can be observed. The major contribution to the high frequency
interference generated by a bipolar line is from "streamers", which are distributed more
randomly.

- 252 -
Radio interference is measured at 1 MHz or, if there is a foreign radiation source at 1
MHz, on a neighboring clear frequency, in which case the results are then recalculated at 1
MHz. The reference value (0 dB) is a field strength of 1 mV/m. Measurements are
conventionally taken at a distance of 30.5 m from the positive conductor. The
measurement instrument to be used is one operating in the "quasi-peak mode".

Figure 3.10 shows a typical lateral profile of radio interference which was recorded on a ±
600 kV test line with 4 * 30.5 mm cables, a distance between poles of 11.2 m, and a
conductor height in the middle of 15.2 m.

An evaluation shows that


the noise level up to a
distance of approximately
50 m is inversely
proportional to the square
of the distance. Beyond 50
m it is inversely
proportional to the
distance itself.

Figure 3.10: Typical RI lateral profile of an HVDC line

For practical purposes,


radio interference is a
linear function of the
voltage or of the maximum
surface gradient, as shown
in Figure 3.11.

Figure 3.11: Radio interference as a function of voltage

- 253 -
In the measured RI curve presented in Figure 3.11, the environmental noise level
predominates at a surface field strength of less than 18 kV/cm. With a linear extension of
the upper portion of the measured curve, we come down to approximately 15 kV/cm,
which corresponds to the field strength of the corona inception under operating conditions.

In an investigation of the environmental factors, RI interference was found to decrease


during rain, namely by 2 to 5 dB. That is remarkable in view of the fact that ac lines have
higher RI noise levels during rain, by up to 25 dB.

In wind, RI interference will increase, if the direction of the wind is from the negative to
the positive pole, the effect can be up to approximately 1 dB per 1 m/s of wind speed.

A significant seasonal effect was observed, with RI interference being lower in the winter.
In this regard, the FGH, in its Dossenwald tests, verified a clear relationship between the
level of RI interference and the absolute moisture content of the air. At low temperatures,
the absolute moisture content is significantly lower, although the relative humidity in
winter is generally higher than in summer.

The frequency spectrum measured does not differ significantly from that of an ac line. The
noise level decreases in monotonic fashion with the frequency. Below 1 MHz, the
decrease is proportional to log f. Above 1 MHz, it decreases approximately at log f2.

It is interesting to observe that, for the RI frequencies generated by an HVDC line, a


signal-to-noise ratio of 10:1 was felt acceptable, while in ac lines, values of 15:1 to 25:1
were required. The lack of modulation with line frequency probably means that the
subjective impression of the noise from HVDC interference is less.

Interference in the TV range, of course, is theoretically conceivable, but measurements


have shown that the noise level in this frequency range is so low that there is no reason to
fear any interference beyond the right of way of an HVDC line.

- 254 -
3.1.4.3 Acoustical Noise

Ionization caused by the corona represents a partial breakdown in air; this high-energy
partial discharge causes a local compression and decompression of the air, which is
propagated as an acoustical wave through the medium air.

In the analysis of the frequency spectrum of the acoustical noise generated by a dc line, it
has been determined that below 500 Hz no typical interference noises are generated by the
line, and that environmental noises predominate in this frequency range. This situation is
illustrated in Figure 3.12, in the form of a frequency spectrum recorded on a ± 600 kV test
line with a variation of the voltage between 350 and 600 kV.

Figure 3.12: Frequency spectrum of the acoustical noise generated by an HVDC line
M.A.F. = Minimum Audible Field = Hearing threshold for test persons aged
20 and 60, respectively.

The noise spectrum of a 525 kV ac line recorded for purposes of comparison, in the
frequency range in question between 1 and 20 kHz, shows values which are lower by 20
dB than the HVDC line at 600 kV in good weather, but values which are higher by 10 to
15 dB in rainy weather.

- 255 -
Figure 3.12 also shows the acoustical noise as a function of the voltage. Measurements
have shown that there is a roughly linear relationship between the acoustical noise level
and the line voltage or surface field strength; this is also true for the subjective acoustical
impression.

Figure 3.13 shows the lateral profile of the acoustical noise of a dc line ± 600 kV under
fair weather conditions determined by the above-mentioned tests. In this case, the
influence of the polarity is also clear; the positive conductor is the dominant source of
interference. This should not be surprising, since the streamer discharges, which are rather
randomly distributed over the surface of the positive conductor, release pulse-shaped
acoustic pressure waves of relatively high energy.

Figure 3.13: Acoustical noise lateral profile of a ± 600 kV HVDC line in fair weather, in
dB over 20 mPa

In an investigation of the effects of weather, it was determined that both radio interference
and acoustical noise decreases slightly during rain. This is exactly the opposite of the
behaviour of an ac line.

- 256 -
The subjective impression of the noise is also an approximately linear function of the
voltage and of the surface gradient. Immediately underneath the positive conductor, the
line was classified as "quite loud" to "intolerable" when the surface gradient exceeded 25
kV/cm. But outside the right of way of the line, there is no reason to expect any
complaints.

3.1.4.4 Ion Currents

While within the ionisation zone, corona current is carried by electrons, current between
the conductors and between the conductors and ground is carried by ions and other
charged particles. It is important to note that in ac lines, this effect does not exist. The
charge carriers which are accelerated in one direction in one half-wave are decelerated in
the next half-wave and then accelerated in the opposite direction, only a few of them
"escaping" into the open space. On the other hand, currents are induced by capacitive
coupling in conductive objects which reach values of 16 mA per 1 kV/m field strength. If a
person stands on the ground under an ac line at 5 kV/m field intensity, the induced current
reaches 80 mA. It is important to keep this value in mind when the effects caused by the
ion current underneath an HVDC line are discussed.
The ion currents generated by an HVDC line cause a space charge. Figure 3.14 shows the
lateral profiles of charge and current density for a ± 600 kV HVDC test line, measured in
the middle of a span with conductors 12.2 m above ground.

Figure 3.14: Lateral profiles of space charge density and current density underneath an
HVDC line

- 257 -
A conductive object, such as a person, in the space charge zone underneath the HVDC line
intercepts a current the strength of which depends on the current density as presented in
Figure 3.14 and an equivalent area of the object. For a person under the test line, a current
of 3 mA has been calculated and verified by measurements.

If the physiological effects of the current flowing through the human body, which range
between "barely perceptible" and "painful shock", are investigated, the respective
thresholds for direct current were higher by a factor of five than for alternating current.
Furthermore, if the relationship of the current flowing through the human body under a
high voltage overhead line - 3 mA compared to 80 mA - is taken in consideration, it can be
concluded that the physiological effects to be expected from a direct current line will be
very much less than those from an ac line.

An additional effect caused by ion current is the charging of objects which are under the
dc line and which are well insulated from ground. For example, tests have been conducted
on the charging of a large vehicle parked under the line, and on the effect of the discharge
caused when a well-grounded person touches the vehicle. The most unfavourable effect
determined was a discharge current peak of 670 mA, which was reduced in 200 ms to < 1
mA. 1300 mA for 30 ms is considered the critical value. Thus the discharge in question is
lower by orders of magnitude. The effect is comparable to the sparks generated by a
person walking over a carpet.

Well-insulated wire fences could be a cause for concern. It has been determined that a
fence running at a distance of 10 m parallel to the HVDC line for several km can cause an
"unpleasant shock". Such a very theoretical case can easily be eliminated by installing a
few longitudinal insulators in the fence and by grounding the fence.

The question of the ignition of fuel caused by electrical discharges is more critical. The
hypothesis is that a vehicle underneath an HVDC line is filled with fuel and struck by
discharge sparks. Under unfavourable conditions, the energy released can be 5 to 10 mJ.
Although under laboratory conditions, an ideal mixture of air and gasoline can be ignited
at 1.2 mJ, more than 100 mJ are necessary under conditions closer to those experienced in
actual practice. Such value can never be reached, even with very large vehicles which are
very well insulated from ground, and even if they are parked under an HVDC line for a
very long time.

- 258 -
3.1.4.5 Comparison of Types of Lines

To get a feeling for the dependence of the phenomena caused by the conductor surface
gradient and the corona effect on the type of line selected, the following variants of a
± 500 kV HVDC line at IdN = 1800 A are compared, in which there are two separate lines
for reasons of reliability:

• two monopolar lines


• two homopolar lines
• two homopolar lines one conductor being grounded
• two bipolar lines

For the monopolar lines, a four-conductor bundle with the strands having a diameter d =
3.18 mm was selected, and for the other lines, a two-conductor bundle with d = 3.18 cm
and alternatively a four-conductor bundle with d = 2.25 cm. In this manner, the same
conductor cross section is available in all the variants. Generally, the conductor height
above ground in the midspan was 15 m, and in the 2-pole lines, the distance between poles
was defined as 12 m.

The following parameters were computed:

• the maximum conductor surface gradient Emax in kV/cm on the surface of the
individual strand
• the radio interference field strength Fo in dB over 1 mV/m, measured at 1 MHz and at a
distance of 30 m
• the corona losses Pc in kW/km line length.

All the values are for fair weather. However, in rain and fog the RI interference decreases,
only the corona losses increase significantly.

Since corona losses are a function of polarity only to a slight degree, while the RI
interferences come primarily from the positive line pole, only the positive pole was
calculated in the monopolar and homopolar variants.

The formulas indicated by E. Uhlmann (Power Transmission by Direct Current; Springer


Verlag 1975) were used for the calculation:

“The conductor surface gradient can be described in general by the equation

- 259 -
Ud
r
E=
m é æ 2H ö ù
2
2H 2H
ln +( m - 1) ln + c ln ê1+ ç ÷ ú
r s' 2 ë è Aø û

r = 0.5 d
s’ = s (for m=2)
= 1.123 * s (for m=4)
c = +1 (for homopolar lines)
= 0 (for monopolar lines)
= -1 (for bipolar lines)
= -0.5 (if one pole is grounded)

Thus, the conductor surface gradient is determined by

· the line-to-ground voltage Ud


· the conductor strand diameter d
· the number of strands per bundle m
· the distance of the strands within the bundle s
· the average height above ground of the conductor H
· the pole-to-pole distance A

The maximum conductor surface gradient of bundle coductors is found to be

é rù
E max = E ê1+( m-1) ú
ë Rû
R = 0.5×s ( for m= 4)
1
= ×s ( for m= 2)
2

The radio interference, emenating from the overhead line can be determined by the
following equation:

Fo = 25 + 10 lg m + 20 lg r + 1.5 (Emax - 22)

Thus, the radio interference is defined as the field strength Fo, measured in a horizontal
distance from the nearest conductor of 30 m, at a frequency of 1 MHz with a band width
of 9 kHz. Fo is obtained in dB over 1 mV/m.

- 260 -
Reference point of this equation which was empirically determined in field tests is a
maximum conductor surface gradient Emax = 22 kV/cm.

The corona losses are determined by the corona current emanating from the conductors of
the overhead line:

I c = c×m×2 0.25( E max - 22 ) ×10 - 3 in A / km


c= 0.15...0.30...0.35 (depending on the quality of the conductor surface)

Reference point of this equation which also was determined empirically in field tests is
again a maximum conductor surface gradient of 22 kV/cm.

The corona losses per km of line length are then determined to

Pc = l × Ic × Ud
l = 1 (monopolar line)
= 2 (homopolar line)
= (1 + k) × 2 (bipolar line)
= 1 (one pole grounded)

2 2H
k = arctan
p A

The results of the calculations according to the above cited formula shows fig. 3.15

- 261 -
Ud =500 kV d= 3.18 cm Ud = 500 kV d=2.25 cm
Emax = 18.7 kV/cm
+ F o = 30.0 dB
Pc = 0.3 kW/km

Emax = 25.1 kV/cm Emax = 20.6 kV/cm


+ + Fo = 36.6 dB Fo = 29.9 dB
Pc = 1.0 kW/km + + Pc = 0.94 kW/km

Emax = 31.9 kV/cm Emax = 27.4 kV/cm


Fo = 40.0 dB
0 + Fo = 47.0 dB
0 + Pc = 1.5 kW/km
Pc = 1.7 kW/km

Emax = 35.2 kV/cm Emax = 30.7 kV/cm


Fo = 45.1 dB
_ + Fo = 51.8 dB _
+ Pc = 9.5 kW/km
Pc = 10.4 kW/km

Figure 3.15: Comparison of types of lines of ± 500 kV HVDC systems with equal power
transmission capacity and line redundancy (two lines in parallel).
Emax = maximum surface gradient
Fo = Radio interference (field strength)
Pc = corona losses

It is apparent that the variant with two bipolar lines is hardly feasible, due to the radio
interference. In the homopolar lines, the variant with the four-conductor bundle should be
selected, since it maintains the interference field strength < 45 dB even when one line
conductor is shut down and grounded.

The conditions are so unfavourable in this case because the nominal transmission power
per line is only 900 MW; at higher power, the higher nominal current requires a lager
cross section, causing a reduction in conductor surface gradient. For purposes of
comparison, the characteristics of a bipole at 1800 A nominal current are presented below.
The line is configured as four-strand bundle with a conductor diameter of 3.18 cm, and
with all the other dimensions the same as above:
Emax = 28.3 kV/cm
Fo = 44.5 dB
Pc = 1.6 kW/km

The findings appear to be very agreeable.

- 262 -
3.1.5 Electric and Magnetic Fields

Both, electric and magnetic fields occur in the vicinity of a high voltage overhead line
carrying current. The potential effects of these fields on human beings have attracted
increasing interest in recent years leading to some heated discussions in the media under
the general category of electrosmog.

An HVDC overhead line also generates electric and magnetic fields, but these are dc
fields, the biological effects of which - if any - are much less than those of power system
frequency ac fields. There are currently no reliable definitions of limits, but there is reason
to expect that limits will be specified or at least calculated data will be requested for future
HVDC projects. Therefore the electric and magnetic fields generated by an HVDC
overhead line are discussed in greater detail below.

3.1.5.1 The Electrostatic Field

The electrostatic field is formed in the space between the live conductor and the ground
when there is no interference from charge carriers. The field strength at the ground level
can easily be calculated with sufficient accuracy. The formulas indicated by N. Knudsen
for the electric field strength under monopolar and bipolar HVDC lines are presented
below:

U 2 H2
Monopolar line: E=
H Inæ 2 H ö H 2 + X 2
ç ÷
è Rø

U 2 é H2 H2 ù
Bipolar line: E= ê 2 - ú
H 1 - K In æ 2 H ö H 2 + ( X - S) úû
2 2
( ) çè ÷ø êë H + X
R

- 263 -
where: U = Conductor-ground voltage in kV
H = Height of conductor above ground
R = Equivalent conductor bundle radius
S = Distance between conductors
X = Lateral distance from the conductor

All distances are in meters.

é ( 2H) 2 + S2 ù
ln ê ú
ê S úû
The coupling factor K is calculated as: K= ë
2H
ln éê ùú
ëR û

D n n* d
The equivalent conductor bundle radius is: R=
2 D

where D = Diameter of the bundle


d = Diameter of the individual conductor
n = Number of conductors per bundle

The result is the electrostatic field strength in kV/m at a lateral distance X from the
conductor.

- 264 -
As a typical example, Figure 3.16 shows the calculated field strength profile under a
monopolar and a bipolar 600 kV HVDC line.

Figure 3.16: Electrostatic field strength profile under an HVDC overhead line
_____ bipolar line ± 600 kV
----- monopolar line 600 kV
A= Distance from the center line of the overhead line

- 265 -
3.1.5.2 The Electric Field Changed by Space Charge

If corona occurs as a result of high conductor surface gradient, the charge carriers
generated by ionization cause an apparent increase of the conductor diameter, i.e. a
decrease of the surface gradient. Simultaneously, however, there is an increase of the field
strength on the ground.

This effect of field strength increase cannot be calculated with any accuracy. It is
significantly influenced by weather conditions, in particular by the direction and speed of
the wind.

Extensive measurements on the EPRI ± 600 kV test line have shown that, due to corona
effects under the negative conductor, the electrostatic field strength was increased by a
factor of three. Under the positive conductor the increase in field strength was less than
50%. The reasons for this asymmetry are suspected to lie in the different ionization
mechanisms of the two polarities.

For example, the following figure shows a comparison of calculated and measured field
strength curves underneath a ± 600 kV test line. The comparison has been taken from the
above-referenced "Transmission Line Reference Book":

- 266 -
Figure 3.17: Calculated and measured field strength underneath a ± 600 kV test line.

The conclusion is that only the electrostatic field can be reliably calculated and therefore
specified. Verification by means of measurements is possible only on lines in which the
conductor surface gradient is so low that no significant corona occurs.

- 267 -
3.1.5.3 The Magnetic Field

The current carrying HVDC overhead line, as all energized lines, is surrounded by a
magnetic field. However, the line currents are not very high (generally £ 2 kA), and the
height of the conductors above ground is generally considerable (³ 10 m). As a result, flux
densities on the surface of the ground are only several 10 mT, which is on the order of
magnitude of the earth's natural geomagnetic field. It is difficult to image any adverse
effects or even health hazards at this low level.

Figure 3.18 - again for a monopolar and a bipolar 600 kV HVDC line - shows the flux
densities which occur on the ground at a line current of 1 kA.

Figure 3.18: Magnetic flux density under an HVDC overhead line


_____ bipolar line ± 600 kV, 1 kA
----- monopolar line 600 kV, 1 kA

- 268 -
3.1.6 The Insulation of the HVDC Overhead Line

The insulation of the overhead line determines the number of flashovers which have to be
expected on a statistical basis, and thus the reliability of an HVDC system. The correct
selection of the insulation requires a compromise between cost and technical performance.
The environmental conditions of the route of the line have a decisive influence on its
reliability. In addition to the local frequency of thunderstorms (the isokeraunic level), air
pollution which is present continuously or discontinuously, in combination with the
moisture in the air, in particular dew and fog, has a major influence on the frequency of
flashovers of a given line insulation. However, if an acceptable number of flashovers per
year is specified, all these effects must be taken into consideration in the selection and
sizing of the line insulation.

- 269 -
3.1.6.1 Flashover Mechanisms

An insulation flashover, i.e. a line-to-ground short circuit on an overhead line, can result
from the following events:
• direct lightning strike
• back flashover
• switching surge
• insulator contamination
• insulation defect
• fire under the line.

A direct lightning strike generally results in an immediate insulation flashover, although


the lightning withstand voltage of the insulators of a 500 kV HVDC overhead line is on
the order of magnitude of lightning voltages. If an HVDC line is to be constructed through
an area where there is significant thunderstorm activity, an overhead earth wire must be
provided. Generally, the first spans of the outgoing line are provided with two overhead
earth wires to ensure protection against lightning strikes at these points and thereby
prevent the entering of a steep fronted impulse wave into the converter substation.

A back flashover of the line insulation may occur if the line runs through an area of high
specific ground resistance. A lightning strike into the tower or the ground near the line can
then increase the potential of the tower to the point where an insulation flashover occurs.
Protective measures include laying of an uninsulated cable in the ground, called a
counterpoise, which connects the footings of the towers.

- 270 -
Switching surges occur in a bipolar HVDC overhead line when one pole experiences a
conductor-to-ground flashover. As a result of capacitive coupling, the potential of the
unaffected pole is elevated, resulting in a overvoltage resembling a switching surge, with a
peak value from 1.4 to 1.9 p.u. It is imperative that such an overvoltage not results in an
insulation flashover under any circumstances because the philosophy of the reliability of
the bipolar HVDC system is based on the idea that any dc-side fault will affect only one
pole and 50 % of the transmission capacity will remain. Accordingly, the switching surge
withstand voltage of the line insulation must be > 1.9 p.u. under the most unfavourable
conditions.

Contamination flashovers are a problem specificly inherent to the HVDC line. Air
contamination can have all sorts of causes, whereas the salt content of the deposits is
particularly critical, since salt in combination with moisture forms a conductive coating on
the insulator surface. The contamination in question includes:

• salt fog in coastal areas


• dust containing salt (deserts or dry regions)
• agricultural deposits (fertilizers, combustion residues)
• industrial pollution (coal mining, salt mining, cement plants, chemical industry)
• pollution from vehicles (highways, non-electrified railways).

Contamination of the insulator surface in itself is relatively noncritical. Only when


moisture is present - in the form of dew, fog or mist - does the coating become conductive.
A leakage current occurs along the surface of the insulator, which can easily be as high as
several 100 mA. As a result of the heat which is generated, some areas dry out. The
voltage, which was previously distributed uniformly, appears across these dry areas, and
there occur partial flashovers. This effect, which also occurs with alternating current, is
particularly critical with direct voltage because there are no zero crossings to allow an
extinction of arc currents. There is thus a significant hazard that a partial flashover will
grow into a total flashover.

- 271 -
An insulation defect can occur with the cap-and-pin insulators which are used almost
exclusively in HVDC systems. It generally begins with the failure of a first one cap-and-
pin insulator, and then several insulators in a string. Such defects are very difficult to
detect. There has long been disagreement in the industry over the relative merits of glass
and porcelain caps, which are otherwise equivalent, in this regard. It is believed that the
sheds shatter immediately when there is a defect in a glass insulator, making it easy to spot
defects from a helicopter, while the penetration of a porcelain cap-and-pin insulator
remains invisible. The only way such defects can be detected is by measuring each string
during inspections. On the other hand, glass cap-and-pin insulators have a certain natural
shattering rate, requiring more frequent inspections and repairs.

Fire under the line is a phenomenon which is very relevant in Africa. This has led to
hundreds of flashovers per year on the Cabora-Bassa line. Flames rising high into the air
ionize the air, causing direct flashovers from the voltage conducting line to earth. A
necessary precondition is the growth of high grass or bushes underneath the line and
frequent bush fires. A similar phenomenon is the widespread practice of burning off sugar-
cane fields. For that reason, the right of way of the line must be carefully considered, and
the regular maintenance and clearance of the right of way under the line must be specified
in the contract.

3.1.6.2 Parameters of Contamination Flashovers

The flashover caused by pollution is critical, in particular because it occurs during normal
operation at the nominal voltage of the system. Therefore, appropriate attention should be
devoted to the individual parameters which affect the probability of a flashover:

• degree of insulator contamination


• type of contamination
• non-uniformity of the contamination along the insulator string
• different degrees of contamination of the top and bottom of the insulator caps
• kind of wetting
• voltage polarity.

- 272 -
The "Equivalent Salt Deposit Density" (ESDD) has been developed as a unit of measure
for the contamination of the insulator surface. The ESDD is that salt coating measured in
mg/cm2 which produces the same conductivity, when completely wetted through, as the
actual layer of contaminants. Since there is generally a very low contamination to be dealt
with on HVDC lines, the unit mg/cm2 is more appropriate as a unit of measure for the
ESDD.

Tests performed in the U.S. by Zafanella and Schneider for EPRI have shown that the
"Critical Flashover Voltage" CFO (50% value) for a cap-and-pin insulator widely used in
HVDC systems decreases from 140 to 55 kV/m when the ESDD (of the bottom-side of the
insulator) increases from 10 to 100 mg/cm2. The tests were performed in a climate-
controlled chamber (clean fog method) and thus include full wetting.

The type of contamination, according to the ESDD definition, is assumed not to play a
role. But it has been shown that different salts with the same conductivity can lead to
somewhat different flashover behaviour. For example, CaCl2 gives a CFO value which is
approximately 10% higher, and KNO3 a value which is approximately 10% lower than
NaCl. The ratio of soluble to insoluble components in the layer of pollution also seems to
play a role. Hence, it is important to know both the degree and type of insulator
contamination.

The non-uniformity of the contamination along the insulator string is determined by the
curve of the field strength which leads to an increased accumulation of charged particles
on the upper and lower ends of an insulator string. The effects on the flashover voltage,
however, are not at all negative. On the contrary, if an insulator set is divided into three
segments with different ESDD, and if the three CFO voltages are measured, the sum of the
CFO voltages is approximately 20% higher than the CFO voltage of the total set
contaminated with the average of the three ESDD values. Although it is difficult to draw
any useful conclusions from this phenomenon, the result is still comforting.

- 273 -
The non-uniform contamination of the upper and lower sides of the insulator caps can be
explained by the natural washing caused by rainfall, which predominately affects the upper
side. Zafanella and Schneider represent the effect on CFO voltage as follows:

O O
CFO = 1 - K * log
in p.u. relative to =1
U U
where O and U = ESDD of the upper and lower sides, respectively
K = 0.38

Thus, for the ratio O/U = 0.1, which becomes established as the final value after a
relatively long period of time, the CFO voltage is higher by 38% than if the upper side was
contaminated to an equal degree.

The nature of wetting significantly influences the effects of contamination. Not every fog
and not even every dew fall results in thorough wetting of the layer of contamination and
the associated lowering of the CFO voltage. A complete soaking is only achieved at 100%
humidity with an effective time > 1 hour with no wind. In comparison with fog chamber
tests, outside tests usually resulted in an increase of CFO voltage of approximately 20%.

Polarity has some influence on contamination effects. The CFO voltage is lower by several
percent for negative test voltage than for positive test voltage. The difference is so slight,
however, that no benefit can be derived from the use of differing rating of insulation on
the two poles of the line.

However, the CFO voltage for negative polarity should always be used as the reference
value.

- 274 -
3.1.6.3 Insulator Design

To date, HVDC overhead lines have been almost exclusively equipped with cap-and-pin
insulators of glass or porcelain. Figure 3.19 shows a typical HVDC insulator with a large
ratio of creepage path to length. A large number of other geometric parameters such as cap
diameter, number and height of petticoats, skew angle, etc., determine the flashover
behaviour, even though their influence cannot be described in detail. In the 1960's, it was
found in the Dossenwald out-door test station of the FGH that of two very similar cap-and-
pin insulators, one was among the best and the other among the worst of many insulators
investigated with respect to dc contamination flashover. For this reason, it is highly
advisable to recommend only insulators which have proven successful in HVDC
application over a period of many years.

Figure 3.19: Glass cap-and-pin insulator for HVDC lines

The geometric arrangement of insulator strings on the tower, namely:

• Single suspended string (I)


• Double string (II)
• V-string (V)

does not have any significant influence on flashover behavior. In the tests by Zafanella and
Schneider, only horizontal tension strings showed a CFO voltage lower (by 6%) in
comparison with suspended strings having the same ESDD. It should be taken into
consideration, however, that horizontal strings favor the wash effect and that the ESDD
will therefore be lower than in the case of suspended strings.

- 275 -
Composite insulators with silicon rubber sheaths may play a special role in the insulation
of HVDC lines in the future. Figure 3.20 shows a cross section of a typical composite
long-rod insulator.

Fiberglass-reinforced resin

Silicon-rubber sheath

Silicon-rubber shed
chemical bond

Vulcanization

Polymer sealing

End fitting

Cleaving wedge

Figure 3.20 Cross section through a composite long-rod insulator (Hoechst-Ceramtec)

Such insulators have been used in the critical 50-km long southern end of the Pacific-
Intertie HVDC in connection with the increase of the voltage from ± 400 kV to ± 500 kV.
Previously it had been necessary to wash the line insulators at regular intervals to prevent
flashovers caused by pollution which after the replacement could be abondened.

The most remarkable characteristic of such composite insulators is the hydrophobia of the
surface. Moisture accumulates into droplets, and no water film is formed. The layer of
pollution is therefore not wetted through.

Surprisingly, the hydrophobia remains effective even when there is significant


contamination of the surface of the insulator, and it returns completely when
contamination is washed off by heavy rains.

- 276 -
The useful life of composite insulators is currently considered problematic. The effects of
UV solar radiation, damage to the surface caused by dust particles in a strong wind (in
effect, sandblasting), and insufficient stability of the internal compounds are indicated as
reasons for premature aging. Such insulators have nevertheless been used in HVDC
systems for more than 10 years.

The following recommendation can be made: the use of composite insulators for HVDC
overhead lines is appropriate where contamination problems can be expected with
conventional insulators (ESDD > 20 mg/cm2), even though there is a risk that the
insulators will have to be replaced after 10 or 15 years.

3.1.6.4 Rating of the Insulation

In spite of the results of extensive testing, the rating of the insulation of an HVDC
overhead line remains a difficult matter. It is clear that insulator contamination and wetting
have a much greater effect on the insulation rating for direct current than for alternating
current. Therefore knowledge of the local pollution and climatic conditions is much more
important than with ac lines. Indeed, such knowledge is even essential for the choice of a
suitable route of an HVDC line.

The following information must be known for each segment of the line:

• expected contamination of the bottom-side of the insulatiors, given as ESDD in mg/cm2


• expected deviations from the average value of contamination along the insulator string
• expected differences between the contamination on the top and bottom of the insulator
• type and amount of soluble and insoluble components in the contamination
• statistically projected frequency of wetting periods, their duration and intensity.

For this purpose - in regions which appear critical - it is essential to have long-term
measurements taken on site, ideally by means of original insulators installed and exposed
to an equivalent direct voltage.

- 277 -
The starting point for the rating of the insulation must be the accepted number - to be
specified - of statistically projected flashovers caused by pollution in a year. If the line runs
through regions of different environmental conditions, it will be appropriate to subdivide
the line into segments.

The number of flashovers expected is equal to the probability of flashovers of an insulator


string in a wetting period multiplied by the number of wetting periods expected per year
and by the number of insulator strings in the line or line segment in question.

Zafanella and Schneider assume a standard deviation of the flashover probability of 6%,
and then calculate the flashover probability per wetting period for the type of insulator
tested, on a critical line segment with 800 insulator strings. They then assume eight
different combinations of the contamination parameters. The results are illustrated in
Figure 3.21. Curve 1 shows the "baseline case", in which the combination of parameters
was that considered to be most likely.

Figure 3.21: Flashover probability W (%) per wetting period as a function of the voltage
stress of the string in kV/m

- 278 -
Figure 3.22 illustrates the application of these results to the Cabora-Bassa line. In this case,
100 wetting periods a year were assumed, and 10 flashovers a year caused by pollution
were considered acceptable. The necessary insulation - expressed as the specific leakage
distance, SLD, in cm/kV - was in turn calculated for the eight hypothetical combinations
of parameters and plotted in the diagram. Cases 5 and 6, which differ from the baseline
case 1 only in terms of the ESDD, are particularly interesting. The curve interconnecting
cases 5.1 and 6 can be considered the sizing function

SLD = f (ESDD)

for all cases with this combination of parameters. The question of which ESDD should be
used must remain unanswered for the Cabora Bassa line, due to the present impossibility
of measurement. For purposes of comparison, several other measurements were used
which are plotted in the diagram as vertical lines. All the measurements - with the
exception of the ESDD of the Sylmar Station, the southern end of Pacific Intertie - are less
than 5 mg/cm2. At 20 mg/cm2, the ESDD of the Sylmar Station which in general terms can
still be classified as "very light pollution", there have been problems with flashovers
caused by pollution.

- 279 -
m

Figure 3.22: Insulation rating (required specific leakage distance SLD) as a function of
the insulator contamination ESDD

The diagram also shows the insulation values of the highest-power HVDC connections,
whereby the values for Pacific Intertie are for the period after the voltage was increased;
the original design was the same as for Cabora Bassa.

3.1.6.5 Operation at Reduced Voltage

At the beginning of this chapter, it was noted that the rating of the insulation of an HVDC
overhead line requires a compromise between cost and the reliability. The question arises
whether the insulation must actually be rated for the most unfavourable case, such as a
dense layer of fog which occurs extremely rarely. But in contrast to ac transmission
systems, HVDC has the ability to operate at reduced voltage, e.g. at 80% or 70% of the
nominal voltage. It may be more advantageous to transmit a reduced power without faults
than to transmit the full power with frequent short-term interruptions. Operation at
reduced voltage must be planned and specified from the outset, since it must also be taken
into consideration in the design of the valves (thyristor RC circuits) and the ac filter
circuits.

- 280 -
3.2 The HVDC Cable

3.2.1 Introductory Remarks

As with HVDC overhead lines, HVDC cables are usually placed for bids and contracts
separately from bids and contracts for HVDC stations. Nevertheless, the engineer
responsible for the design of HVDC stations must also be familiar with the most important
physical phenomena which occur in an HVDC cable, so that these phenomena can be
taken into consideration. This is particularly important for the definition of control and
protection functions. A familiarity with these phenomena is also important in the context
of the selection of systems and the optimization of the principal characteristics of the
system.

The following sections discuss:

• Cable types and their application


• Field strength distribution in the dielectric
• Special problems in the operation of an HVDC cable
• Overvoltage stresses
• Protection of the cable against overloads

3.2.2 Cable Types and their Application

The following section presents a general discussion of several technical aspects related to
the principal characteristics, operating limits, and general application considerations for
HVDC cables.

- 281 -
3.2.2.1 Technical Aspects of the High Voltage Cable

Currently, the theoretical construction of HVDC cables does not differ at all from the
construction of an HVAC, single-core cable. This is surprising in itself, because there are
phenomena which are of decisive importance in HVAC cables but which are almost
negligible in HVDC cables, and vice-versa.

• The cable capacitance in HVAC cables results in a charging current which reaches the
level of the nominal current at approximately 50 km (400 kV) and 80 km (220 kV),
respectively. An increase in the conductor cross section does little to improve the
situation because cable capacitance also increases with the increase in diameter. This
phenomenon does not occur in the HVDC cable; only during start-up to voltage and
shut-down, and when changing the direction of transmission is there any reason to be
concerned about charging and discharging current. On the other hand, cable capacitance
in combination with dc-side inductances of the HVDC system form a series resonance
circuit. The resonance frequency of this resonance circuit must not coincide with the ac
system frequency or its second harmonic.

• The phenomenon of skin effect is a significant consideration only in HVAC cables. In


HVDC cables, this effect plays a role only during current transients and with respect to
dc-side harmonics. Its effect, if any, is in an increased damping.

• Dielectric losses and the aging of the insulation material as a result of the constant
change in polarization occur only in the HVAC cable. Therefore significantly higher
field strengths are permitted in the insulation material of an HVDC cable.

- 282 -
• Eddy current losses in the cable jacket are also a problem which is experienced only
with HVAC cables.

• The strong temperature dependence of the specific resistance of the insulation material,
however, is important only with the HVDC cable where it determines to a decisive
extent the operational behaviour and the design limits of the cable. This phenomenon is
therefore discussed in detail in the following section.

In spite of the disadvantage of this last point, the technical characteristics taken as a whole
present so many significant advantages for HVDC cables that approximately three times
the power per cable conductor can be transmitted with the same expense for material as
compared with HVAC cables.

3.2.2.2 Paper-Insulated Mass-Impregnated Cable

This type of cable is used most frequently in HVDC systems. The insulation consists of
special paper wound in multiple layers, which is impregnated with a viscous insulation
compound. It is assumed that the dielectric, free of cavities, follows the changes in volume
which occur as a function of the temperature. There is practically no migration of the
impregnation agent. Therefore the length of such a cable is theoretically unlimited.
Submarine cables can be layed at a depth of 500 meters with no problem.

These characteristics and the simple construction of the cable led to its use in the first
HVDC systems. The HVDC land cable connection erected in the 1940's from the Elbe
Power Plant to Berlin used a paper-insulated mass-impregnated cable with 200 kV
nominal voltage and 150 A nominal current.

- 283 -
In the 1960's, three different designs of 400 kV paper-insulated mass-impregnated cable
for 1000 A nominal current were subjected to a long-term test in the Dossenwald HVDC
Test Field of the FGH Mannheim. The results were not very satisfactory. Two of the
cables failed after only a few months. The objectives were probably too ambitious, and the
design was too far ahead of the state of the art.

The following table shows the progress in the development of paper-insulated mass-
impregnated HVDC cable:

No. Year Power per Nominal Insulation Conductor Nominal Current Cable
started core voltage thickness cross section current density length
up [MW] [kV] [mm] [mm2] [A] [A/mm2] [km]

1 1954 20 100 7.0 90 200 2.2 100


2 1961 80 100 9.0 390 800 2.1 51
3 1965 100 200 11.8 420 500 1.2 119
4 1969 156 260 18.5 400 600 1.5 27
5 1976 250 250 16.0 800 1000 1.25 125
6 1986 250 270 12.3 900 925 1.03 50
7 1989 500 400 17.5 1200 1250 1.04 200
8 1994 600 450 19.0 1600 1333 0.83 250

The numbers in the left column refer to the following HVDC systems:

1: Sweden - Gotland I (20 MW)


2: England - France I (160 MW)
3: Sardinia - Italy I (200 MW)
4: Vancouver I (312 MW)
5: Skagerrak I (500 MW)
6: England - France II (2000 MW)
7: Fenno-Skan (Finland-Sweden) (500 MW)
8: Baltic Cable (Sweden-Germany) (600 MW)

- 284 -
It should be noted that current density decreases with increasing voltage, i.e. with
increasing thickness of the cable insulation. This decrease is related to the dependence of
the field strength distribution on the temperature gradient in the dielectric, which will be
discussed in the following section. Nevertheless, the development of the paper-insulated,
mass-impregnated HVDC cable has not reached the limits of technology.

Operating experience with paper-insulated mass-impregnated HVDC cables is very


positive, if only problems resulting from defective insulation are considered. However, the
results have been extremely poor if disruptions caused by external factors such as by ships'
anchors or drag nets are included. Therefore in some more recent systems, the cable has
been laid in excavations in the ocean floor, or has been washed into the bottom after
laying.

3.2.2.3 The Gas-Insulated Internal-Pressure Cable

In a paper-insulated mass-impregnated cable, there is a risk that cavities can form in the
dielectric as a result of rapid changes in temperature. These cavities can lead to partial
discharges and thus to premature aging. This risk can be largely prevented by the
introduction of an insulating gas, generally nitrogen, under a specified pressure. The gas is
introduced through a hollow passage in the conductor. The New Zealand HVDC cables
laid in the Cook Strait are gas-insulated, internal-pressure cables. However, cables of this
type have not been used in any other recent projects.

3.2.2.4 The Oil-Filled Cable

The quality of the paper insulation of a cable can be significantly improved by a


pressurized low-viscosity insulating oil. At a specified voltage, a lower insulation
thickness is sufficient. This also results in improved heat dissipation conditions. With a
given insulation thickness, the operating voltages can be increased. Arated voltage of
± 600 kV is considered feasible.

- 285 -
Temperature dependent volume equalization is a problem, it can be guaranteed only for
relatively short cable lengths. For this reason, the 18 km overland cable segment on the
English side of the Channel Transmission System is divided by 3 stop joints.

A highly interesting solution represents the „two core flat oil cable“ which already had
been used in the first stage of the Kontiskan HVDC cable link in 1965. It was applied for
the 23 km long western section of the link. Very recently, this cable type was chosen for
the KONTEC HVDC cable link, comprising a submarine cable section, 55 km in length,
connecting Germany and Denmark, and a land cable section, 120 km in length, on the
Danish side. With this project the two core flat oil cable can surely be regarded as a proven
alternative for HVDC cable links.

With a rating of 600 MW, the KONTEK cable provides the same transmission capacity as
the oil impregnated mass cable of the Baltic Cable link. The rated voltage of the KONTEC
cable is 400 kV, the rated current 1500 A as compared with the data of the Baltic Cable,
450 kV and 1330 A.

The flat cable comprises two cores, each insulated for the full voltage. Basically it could as
well be used as bipolar cable with half the rated current. Both cores are surrounded by a
common lead sheath and the outer armature and thus form a cable with flat-elliptical cross
section. By this design a compensation effect is achieved. In case of a thermal expansion
of the active material, a deformation of the cross section in direction of a circular form
takes place, thus providing more volume for the insulating oil. Vice versa, in case of a
cooling down, the cross section returns to the original flat-elliptical form. The oil pressure
is maintained at about 6 bar and the formation of cavitations in the dielectric is reliably
prevented. A migration of the insulation oil is not effected.

According to the manufacturer, there is no limitation in length of submarine cables; laying


depths up to 150 m are possible. The land cable has to be layed in sections of about 1000
m each which represents the transportation limit over land.

The current density of 0.94 A/mm² of the KONTEK cable is significantly higher than that
of the Baltic Cable. According to the manufacturer, a rated voltage of 600 kV is
considered feasible with this type of cable.

- 286 -
3.2.2.5 Plastic-Insulated Cable

In the 1970's, it was thought that it might be possible to use an extruded plastic cable made
of polyethylene or crosslinked polyethylene for HVDC applications.

Although plastic-insulated cables have long been used in ac high voltage technology, they
have not been used in HVDC technology - apart from the failed attempt in the Kingsnorth-
London system. In HVDC use, space-charge problems in the dielectric led to problems of
the field strength distribution, which cause local increases in field strength and partial
breakdowns.

The industry now believes that it understands this phenomenon and can solve the
problems related to it, but there is no proof so far by the application of a plastic-insulated
cable in an HVDC system.

An entirely different type of cable uses plastic films wound in laminar fashion as a
dielectric. But whether the excellent results achieved in power capacitors can be
transferred to the dc cable remains to be seen.

3.2.2.6 Superconducting Cable

It is uncertain whether superconducting cables will ever be used in power engineering. The
cost of the thermal insulation and cooling systems is too high. The same is true for the so-
called high temperature superconductors which operate at the temperature level of liquid
nitrogen. These conductors are extremely sensitive to magnetic fields, and are difficult to
work into flexible conductors because of their ceramic structure.

Nevertheless, it can be stated that superconducting cables, if they are used, will be high
voltage dc cables. Present technology will not allow currents of 10 or 100 kA to be
handled in the cable terminations and converter stations; such a high current also adds to
the problem of high magnetic fields.

3.2.3 Field Strength Distribution in the Dielectric

The direct voltage drop across the insulation of the HVDC cable is determined by the
conductivity of the insulation material. For paper-insulated mass-impregnated cables, the
voltage across the insulation is determined essentially by the conductivity of the insulating
paper. In cold conditions, the conductivity of the paper is uniform in the entire dielectric.

- 287 -
Thus the following relationship applies for the field strength at a given point of the
dielectric with the radius r:

E r r0
= where r0 = radius of the cable conductor
E0 r

This field strength distribution is the same as in an ac cable in which the capacitance
determines the voltage distribution. This fact is important to the extent that transient
voltages in an HVDC cable always follow the above-mentioned field strength distribution.

In a cable which is carrying current, there is a thermal gradient from the conductor toward
the outer jacket. Therefore the conductivity of the paper changes in the layers of the
dielectric. This condition is described by the following equation:

E r r r r0
= * where r0 = specific resistance of the paper at the inside radius (conductor)
E 0 r0 r

The following relationship applies for the temperature dependence of the specific
resistance r of the paper insulation of the cable:

r2 = r1 * e-a(J1-J2) (where a » 0.1 K-1)

This equation says that when there is a temperature increase by DJ = 23 K, the specific
resistance r2 decreases to approximately 10% of the original value r1 , and to a mere 1%
at DJ = 46 K.

During the operation of the cable, the ohmic heat loss must be discharged outward
through the paper insulation. The inside radius of the dielectric is then at the conductor
temperature. The outside radius is at a temperature which is determined by the temperature
of the surrounding medium and the thermal resistances of the outer cable layers, such as
plastic and lead jackets, steel reinforcements, etc.

- 288 -
Figure 3.23 shows the field strength distribution in the dielectric of a paper-insulated cable
with temperature differences of 0 and 10 and 20 K in steady-state operation. If a transient
(over-)voltage occurs, the "capacitative" voltage distribution that it causes is superimposed
on the steady-state "ohmic" voltage characteristic. This effect is most apparent at the
reversal of voltage polarity on the energized HVDC cable, a process which is very
important for the operation of HVDC cables.

Figure 3.23: Field strength distribution in the dielectric of an HVDC cable

- 289 -
Figure 3.24 shows the voltage distribution
in the cold cable (k), in the hot cable (w)
and the voltage distribution which occurs
immediately after reversal of voltage
polarity (u).

Figure 3.24: Voltage distribution in the dielectric

The field strength E which can be read as the slope of the curves is greatest in the cold
cable on the inside radius r0 (conductor), but in the hot cable it is greatest on the outer
jacket (ra). In the event of the reversal of polarity of the hot cable, the highest field
strength, on the other hand, occurs on the inside radius r0 and is (approximately 50%)
greater than on the cold cable.

Corresponding to the time constants determined by the capacitance per unit length and the
conductivity, the voltage distribution (u) makes the transition into the curve (w), but here it
must be noted that the time constants are a function of the temperature and differ from one
location to another (a few minutes in the interior, up to an hour on the outside).

- 290 -
3.2.4 Special Problems with the Operation of an HVDC Cable

The operation of an HVDC cable produces a number of problems which must be taken
into consideration in the design of an HVDC system, especially in the definition of the
control and protection functions.

3.2.4.1 Energizing of the Cable

The slow ramping of the system voltage, necessary on HVDC systems with overhead lines
to prevent high surges on an open ended line, is required, to an even greater extent, for an
HVDC system using cable. Normal startup with operational inverters must be done
carefully to prevent oscillations of the series resonant circuit formed by station inductances
and cable capacitances.

The charging of the cable capacitance, which can be assumed to be 0.3 to 0.4 mF/km, is
preferably to be done by means of linear voltage ramping. For example, if the voltage on
the 250 km long 450 kV cable of the Baltic Cable HVDC is to be gradually increased to
the nominal voltage in 100 ms, i.e. with a du/dt = 4.5 * 106 V/s, then a charging current of
370 A is required.

3.2.4.2 Change in Direction of Power Flow

All HVDC cable systems currently in operation change the direction of power flow by
reversing the voltage polarity. The solution employed in the first Channel HVDC was the
reversal of the direction of current flow by means of polarity reversal switches. This
method has not been employed again in any subsequent systems. The reason may be that it
was still not possible to avoid a transient reversal of polarity. The cable must anyhow be
designed to withstand this increased stress on the dielectric.

- 291 -
For scheduled changes in the direction of power flow, predetermined current and voltage
ramps must be executed so that the stress on the cable is kept as low as possible.
Adventageously the first phase is a simultaneous reduction of current and voltage to a
current of 40%, which is then to be kept constant. This is followed by a second phase in
which there is a linear reversal in the voltage. The third phase is the simultaneous increase
of current and voltage to the desired level. Such a procedure avoids the current reduction
effect as described below and prevents partial discharges in the cable and also results in an
approximately constant reactive power demand from ac systems.

3.2.4.3 Current Reduction Effect

The insulation of a paper-insulated mass-impregnated HVDC cable is considered to have a


very long useful life, as demonstrated by many systems which have been in operation for
decades. One requirement for a long useful life is naturally that no partial discharges occur
in the cable dielectric which can result in damage and gradual destruction of the insulation.

In the conventional design of HVDC cables with a maximum field strength of 250 kV/cm
in continuous operation and < 400 kV/cm during reversals of polarity, there is no reason to
fear a partial discharge as long as no cavities occur in the cable dielectric. One effect,
however, has recently been discovered which deserves closer attention. With a rapid
reduction of the transmission current, there can be a rapid decrease in the pressure in the
cable causing the formation of cavities. In that case, even at the nominal voltage, there is
the danger of partial discharges which can damage the insulation.

- 292 -
It would be an unacceptable restriction in the operation of HVDC systems to permit their
operation only at a constant power or only with very slow variations in the power in order
to avoid the current reduction effect and thus the formation of cavities in the insulation.
One remedy is the so-called "cable-dependent control", which brings about an initial
power reduction by means of a voltage reduction of up to 20%, followed by a very gradual
reduction of the current accompanied by an equivalent increase of the voltage, until the
voltage has once again reached the nominal value.

This method of power reduction must be viewed in a very positive light, from the point of
view of the ac system, when the voltage reduction is achieved by increasing the firing
angle. The excess capacitive reactive power which normally occurs in the event of a
current reduction is in this case offset by the increased reactive power demand of the
converters.

In itself, the ability to control the power by means of electronic voltage regulation does not
require any modification in the converter design. It does not add to the cost if a slight
temporary increase in the voltage distortion can be accepted. (While the fundamental
component of the current decreases, the absolute values of the harmonic current increase
slightly.) One requirement is that the valves allow operation with an increased firing angle,
i.e. the increased stresses on the RC snubber circuits of the thyristors as a result of the
greater voltage jumps during firing and extinction are allowable.

If such operation is not permitted, configuration of the converter transformer with an


expanded control range of taps will remain as an expensive alternate measure.

- 293 -
3.2.4.4 Resonance Oscillations

The dc circuit of a cable transmission system represents a circuit prone to resonance


oscillation and is only slightly damped. For the characteristic harmonics of the converters,
the cable represents almost a short circuit. Therefore there is no need for dc filter circuits
in pure cable systems.

As a result of fault events, alternating voltages can occur on the dc side. These can be
harmful if they are applied to a series resonance circuit tuned to the same frequency. These
circuits may be formed by smoothing reactor inductance and cable capacitance.

One possible cause of low frequencies is the periodic firing failure of a rectifier valve.
This causes a severely distorted alternating voltage being superimposed on the direct
voltage. The fundamental component of the ac voltage corresponds to the system
frequency. It has an amplitude of approximately 20% to 30% of the nominal dc voltage. In
the case of resonance, a high voltage can occur on the cable. A periodic firing failure must
be detected rapidly by means of a special protection function, and the danger must be
eliminated by blocking the rectifier.

Excitation of a resonance circuit tuned to the first harmonic on the dc side can be initiated
also from the ac side. A resonance for the second harmonic must be expected with a very
low short circuit ratio SCR on the station bus bars. Experience has shown that the
suppression of the second harmonic in the ac current cannot always be accomplished. A
second current harmonic on the ac side always produces a voltage of the first harmonic on
the dc side. If such conditions are present - low short circuit ratio on the station busbar and
a cable of intermediate length - a very detailed simulator analysis must be conducted and
additional filters or a detuning of the resonance circuit by changing the inductance of the
smoothing reactor may be required.

- 294 -
A dc-side resonance for the second harmonic - which can be expected in relatively short
cable connections - can also have undesirable consequences, since there is a third
harmonic voltage existing in many ac networks. Here, too, an off-resonance adjustment of
the resonance circuit by changing the inductance of the smoothing reactor may be
necessary. But if such a measure is not possible or is prohibitively expensive, blocking
filters can be installed in the dc circuit, preferably between the converter and the station
neutral bus.

Finally, it should be indicated that with relatively long cable connections, the dc-side
resonance lies in the range of subsynchronous resonance frequencies. If the rectifier station
is located close to a thermal power station, which has only a relatively weak
interconnection to the ac network, this problem must be studied in greater detail.

3.2.5 Overvoltage Stresses

HVDC cables - in particular paper-insulated mass-impregnated cables - are designed to


withstand high overvoltages. This capability is always tested, generally on a prefabricated
length of cable, by performing long-term tests in addition to the specified type tests. It can
nevertheless not be excluded that overvoltages occur which exceed the dielectric strength
of the cable. Additive or subtractive transient voltage surges can be superimposed on the
cable dc voltage. Due to the comparatively very large cable capacitance, the dc voltage can
be considered a constant voltage source.

Because of the characteristics of the voltage distribution across the cable insulation
illustrated in Figure 3.24, the highest stress is produced by an additive surge voltage in a
cold cable, and by a subtractive surge voltage in a hot cable. This surge voltage is always
on the inside diameter, i.e. at the surface of the conductor.

- 295 -
The following section discusses only the stress caused by surge voltages (lightning surge
voltages and switching voltage surges). With the control and protection equipment used in
modern HVDC systems, it is possible to completely prevent a dc overvoltage.

3.2.5.1 Voltage Surges Caused by Lightning

A lightning strike directly on the cable termination will always be fatal. Therefore, an
effective lightning protection system (overhead ground wires or lightning rods) must be
provided at this point. The protection equipment must be sized to include a wide margin of
safety. The same is true for a dc switchgear at the cable input (transition from overhead
line to cable, possibly with the capability of switching to spare cores), and for the first few
km of the subsequent dc overhead line (e.g. by equipping it with double overhead ground
wires). In this manner, it is possible to prevent direct lightning strikes, at least for strikes
which produce a surge greater than a few kA. Back flashover must also be prevented in the
event of a lightning strike in an overhead ground wire or a tower. Therefore, a low-
resistance grounding of the tower footings on the overhead line segment near the cable is
an absolute necessity (< 10 W).

It is impossible to prevent a direct lightning strike (or back flashover) on overhead line
segments far away from the converter station at a reasonable cost. A voltage surge
travelling in the direction of the cable - in particular with flashovers to the towers - is
significantly being damped. By interacting with the cable capacitance, the impulse
assumes the form of a switching voltage surge (comparable to the effect of the lightning
surge capacitor in earlier HVDC systems with overhead lines).

It is an easy matter to limit a voltage surge having a relatively slow rising front to values
far below the cable withstand voltage by means of a metal oxide surge arrester located at
the cable terminal.

- 296 -
3.2.5.2 Switching Voltage Surges

Switching processes which could cause a switching voltage surge do not occur in HVDC
systems (These remarks do not include HVDC multi-terminal systems or double line
transmission systems with HVDC switches).

Switching voltage surges originating on the ac sice can be transferred onto the dc side,
they are added up by the converter bridges connected in series on the dc side. But they are
always limited to safe levels by the metal oxide surge arresters which are always present in
modern HVDC systems connected parallel to the individual valves and to the converter
groups. (At this point, it should be noted that the insulation coordination of an HVDC
system always starts from the valve, i.e. where there is the lowest withstand voltage.) This
definition applies initially for the cable input.

If an amplitude-limited voltage surge travels through the cable, it will encounter, at the end
of the cable, either the smoothing reactor or the high surge resistance of an overhead line.
Therefore there is a reflection of the travelling wave. But there is no reason to fear a
significant increase in voltage, since the surge travelling through the cable experiences a
significant damping, i.e. a further flattening, and in addition, there is a metal oxide surge
arrester at the end of the cable which limits the increase caused by the reflexion.

3.2.5.3 Stress Caused by Fast Changes in Voltage

This category includes those transient voltage changes which are significant, i.e. on the
order of magnitude of the nominal direct voltage, but do not exceed the rated dc voltage.
These stresses do not include the harmless voltage jumps caused by firing and extinction
processes of the converter valves. Therefore the following rapid changes in voltage will be
considered:

• direct voltage reversal of polarity, and


• collapse of the direct voltage.

- 297 -
Whenever there is an inverter failure resulting in a collapse of the dc voltage because of an
ac system fault near the station, leading to commutation failures, or the activation of a
bypass path for protection purposes, the rectifier responds with a reversal of voltage
polarity. The rectifier temporarily switches to inverter operation, to reduce the dc current,
which has increased as a result of the fault, to the set point value, or to zero current. The
word "fast" must be considered a relative term, since the process of reversing voltage
polarity - even with the "forced retard" - takes more than 10 ms. But this time is still very
short compared to the cable time constant. Reversal of polarity is part of the design
function of the HVDC system and the cable must be capable of withstanding the
associated stress practically an unlimited number of times without degradation.

There are also very fast voltage changes in the form of a direct voltage collaps. This fault
can occur, for example, as a result of a flashover caused by polluted and humidified station
porcellains. In marine cable transmission systems, parts of the system are always near the
coast. Under certain atmospheric conditions, the possibility of flashover cannot be ruled
out even without the presence of an overvoltage. The same is true (and even more likely)
for the cable termination the porcelain of which is particularly sensitive to flashovers
caused by contamination due to its large diameter. On the basis of the information
currently available, flashovers cannot even be prevented by using an extremely long
creepage distance. (It remains to be seen whether the silicon rubber coatings currently
being tested will change this situation.) The result of an insulation flashover is a very
steep subtractive travelling wave in the cable.

Even less favourable are the conditions which occur if there is a back flashover on an
overhead line tower near the cable input, since at this point the incoming travelling wave
results in a reversal of polarity. If the cable section in question is relatively short, the
reflected wave can reach twice the nominal value (opposite polarity!).

- 298 -
3.2.6 Protection against Overvoltages

Depending on the different causes of the overvoltages which may occur on an HVDC
cable, different measures can be taken to protect the cable.

3.2.6.1 System Design

Section 3.2.2.4 above referred to the possibility of hazardous resonant voltages. In any
cable project, consideration must be given to whether the dc circuit of the HVDC has a
resonance for the first or second harmonic. If so, measures must be taken to detune the
resonant circuit by modifications, preferably by increasing the inductance of the smoothing
reactor. A reduction of the resonance frequency by 10-20% should suffice to eliminate the
problem.

If the risk of subsynchronous resonances should occur in an ac network, it must be


considered whether the dc circuit of the HVDC has a resonance for the corresponding
frequency, as may be the case for relatively long cables (several 100 km).

It should be noted that the effective inductance of the dc circuit can also be adjusted by
modifying the short circuit voltage of the converter transformers, which may be more
economical under certain conditions.

- 299 -
3.2.6.2 Lay-out Measures

Lay-out measures are most appropriate to reduce the risk of damage to the cable due to
voltage surges caused by lightning strikes. In pure cable transmission systems, the HVDC
stations, particularly their dc-side outdoor equipment including the cable terminations,
must be reliably protected against lightning strikes by generously-sized lightning
protection equipment.

Combined transmission systems require reliable lightning protection for the transfer
station between the overhead line and the cable, and for the first few kilometers of the
overhead line. Towers supporting overhead lines connected to dc cables must have low
footing resistances to avoid back flashovers.

3.2.6.3 Control Measures

Operational overvoltages and other high stresses must be reliably prevented by suitable
control, regulation and protection functions, as described in Section 3.2.4. Naturally, the
control and protection equipment in the converter stations must be absolutely reliable,
redundant and must be fault-tolerant.

- 300 -
3.2.6.4 Surge Arresters

The measures indicated above cannot and must not be used to replace the overvoltage
protection by means of surge arresters, although they may reduce surge arrester duties and
make them more economical.

Careful insulation coordination and the location of metal oxide surge arresters on both
sides of the cable can reliably prevent danger to the cable insulation caused by voltage
surges.

When sizing the surge arresters with regard to their energy absorption capacity, it may be
beneficial to size them on the generous side. This allows repeated operation of the surge
arresters without causing thermal instability. It also allows the reduction of residual
voltage of the surge arresters in the event of the maximum (coordination) current. This can
mean that the rated (nominal) direct voltage of the system can be increased by reducing the
switching voltage withstand level in relation to the rated dc voltage.

Another way to reduce the protective level is to use metal oxide surge arresters with arcing
gaps. With these surge arresters, after sparkover, a portion of the surge arrester is bridged,
reducing residual voltage, even in the event of high discharge currents.

- 301 -
4. HVDC Earth Electrodes

4.1 The Function of the Earth Electrodes in HVDC Systems

In considering the functions of earth electrodes, monopolar HVDC transmission systems


come to mind first. There are many monopolar HVDC transmission systems. They are
predominately cable transmission systems. Earth electrodes are indeed an essential
component of the monopolar HVDC transmission system, since they carry the
transmission current on a continuous basis. They contribute decisively to the profitability
of low-power HVDC systems, since the costs for a second conductor (with half the
nominal power) are significantly higher, even for transmission over short distances, than
the costs for earth electrodes. This is true even when the costs of their connecting lines
designed for continuous operation are included.

Earth electrodes are also found in all bipolar HVDC systems, and they will certainly be
used in future HVDC multiterminal systems. Therefore the various functions of earth
electrodes in HVDC systems will be discussed below.

4.1.1 Ground Reference Point for the System Voltage

As in any high voltage system, the power circuit of the HVDC system requires a reference
point for the definition of the system voltage to ground as the basis for the insulation
coordination and overvoltage protection. In a bipolar HVDC system, it would conceivably
be possible to connect the station neutral point to the ground mat of the HVDC station to
which the line-side star points of the converter transformers are also connected. But since
the direct currents in the two poles of the HVDC are never absolutely equal, in spite of
current balancing control, a differential current flows continuously from the station neutral
point to ground. There is danger of corrosion not only of the ground mat itself, but also of
all the other metal structures embedded in the earth, such as cables, pipelines, etc.

- 302 -
It is common practice to locate the grounding of the station neutral point at some distance
(10 to 50 kilometers) from the HVDC station by means of special earth electrodes. Figure
4.1 illustrates a bipolar HVDC system with two earth electrodes.

Figure 4.1: Bipolar HVDC system with earth electrodes

- 303 -
4.1.2 Continuous Current Load

In monopolar HVDC systems, the earth electrode must carry the transmission current
continuously, i.e. it must reliably conduct the current from the electrode line into the earth
and vice-versa. During this operation, the soil in the area surrounding the electrode must
not become overheated and moisture must not be extracted excessively from it by osmosis.
Both conditions may cause the soil to dry out irreversibly.

Moreover, the operation of the earth electrodes must not pose any hazards to animals or
human beings or cause any interference with neighbouring installations such as
telecommunications, signalling and safety systems.

The current rating corresponds to the operating current of the HVDC, i.e. it includes all the
specified overloads. The useful life of the earth electrodes should be the same as the useful
life of the system, i.e. approximately 30 years.

In monopolar systems, the direction of the current and thus the type of operation of the
earth electrode (anodic or cathodic) do not change during the life of the system. Therefore
the two earth electrodes of a monopolar HVDC system can be designed very differently.
This is of major importance, in particular for submarine cable transmission systems.

4.1.3 Temporary Current Load

The earth electrodes must be designed for the full system current even in bipolar HVDC
systems. In the event of the failure of one pole - whether of the line or of a station - the
operation of the system can continue with half power while the current is flowing through
earth. This contributes significantly to the availability of a bipolar HVDC system.

Generally, monopolar operation while the current is flowing through earth is limited by
environmental laws and regulations to a certain period of time such as ten minutes, two
hours or 15 days, respectively. These limitations are a decisive factor in the design of the
earth electrodes with regard to their useful life and for the allowable levels of stray
currents and their corrosive effects on nearby metal structures in the ground.

- 304 -
The type of operation of the ground electrodes depends on which of the two poles has
failed; therefore both electrodes must be designed for both types of operation - anodic and
cathodic.

The fact that the current is carried only for a limited period of time does not reduce the
safety requirements with regard to factors such as step voltage or touch voltage, or effects
on telecommunications, signalling and safety systems.

4.1.4 Temporary Overcurrents

In the more modern HVDC power transmission systems, specifications frequently call for
the system to carry a temporary overload. This is intended to reduce the sudden power
drop in the event of the failure of one pole, so that operation of the ac network may be
adjusted to provide reserve power.

The temporary overload is frequently specified in the form of a time/power curve which
asymptotically approaches the specified continuous overload. In some cases, this
continuous overload must be withstood for several days. It is clear that such a temporary
overload must be taken into consideration in the sizing of the earth electrodes.

An even more severe design requirement for earth electrodes is encountered in systems
which provide for parallel switching of the station poles in the event of a long lasting fault
of one pole of the transmission line. The remaining transmission line pole is loaded to
twice the nominal current and the decrease of the transmission power can be limited to a
few percent. One example of such a system is illustrated in Figure 4.2.

Figure 4.2: HVDC with station poles switched in parallel

- 305 -
4.1.5 Transient Overcurrents

Overcurrents caused by lightning strikes or other events trigger transient overcurrents to


ground. This condition is the same as in ac technology except that in ac systems, the short-
circuit currents are much higher. Such transient overcurrents do not represent any
significant load for the earth electrodes as such. The problems lie in the station neutral
point.

Due to the electrode lines which may be several tens of kilometers long, the station neutral
is only grounded with high impedance. The high frequency (transient) impedance of the
earth electrodes and the surge impedance of the electrode line are switched in series for
high-frequency transient events. Consequently, the potential of the station neutral would
be greatly increased in the event of transient overcurrents. This fact must be taken into
consideration for the design of the insulation.

In earlier HVDC systems, the station


neutral point was equipped with a
large capacitor (several mF), a
voltage-dependent resistor and a
surge arrester, as shown in Figure
4.3 using the example of a
monopolar HVDC system. In more
recent systems, the last two elements
are generally replaced by a metal
oxide arrester.

Figure 4.3: Overvoltage protection of the station neutral

- 306 -
4.2 Rating Criteria for Earth Electrodes

The rating criteria for earth electrodes in HVDC systems can easily be derived from the
requirements indicated in the preceding section.

4.2.1 Current Carrying Capacity

In monopolar systems, the earth electrodes must be rated to carry the nominal system
current, including any long-term overload. Short-term overloads and transient overcurrents
can be ignored.

In bipolar systems, the decisive factor is how long the earth electrodes are required to carry
the full system current, including any long-term overload. If parallel switching of the
station poles is provided, twice the nominal current must be accommodated. The
determination is made on the basis of anodic operation. It is also essential to determine
whether a second overload cycle can occur immediately after the preceding one or whether
a minimum currentless pause of several weeks or so after one overload cycle can be
assumed.

The definition of ground current magnitudes and times is especially important in land
electrodes because the thermal time constant of the soil is assumed to be several months.
The maximum temperature of the soil reached during operation must not exceed 85° C in
order to prevent thermal drying- out in the areas adjacent to the earth electrode.

The allowable power loss (in continuous operation) or energy loss (in cyclical operation)
which can be dissipated by the soil before the temperature limit is reached, is determined
on the basis of the following factors relating to the ground at the electrode site:

• Thermal conductivity,
• Thermal capacitance,
• Thermal time constant

- 307 -
4.2.2 Current Density

Current density is defined as the current flow at the interface between the earth electrode
itself and the surrounding medium in relation to the surface of the external contour of the
electrode, i.e. in most cases, the external surface of a coke bed.

To prevent excess stress on the surrounding soil, in particular drying-out caused by


electroosmosis, a value of 1A/m2 should not be exceeded. This empirical limit - which has
not been verified by systematic tests - is generally the determining factor in designing a
land electrode. In the event of short ground current periods (hours), it is possible to exceed
the value indicated above. But if the current density exceeds this value for longer periods,
there will be a danger of irreversible drying-out of the surrounding soil, causing soil with
initially good conductivity to be turned finally into an insulating ceramic.

4.2.3 Earth Electrode Resistance

Earth electrode resistance is defined as the sum of the resistances of the electrode itself,
the earth transition resistance to the surrounding soil, and the resistance of the layers of
earth through which the current flows to the hypothetical potential reference point
("infinite earth").

The further current flow through the ground is assumed to be loss-less; this assumption is
justified because a direct current seeks the path of least resistance and penetrates into very
deep strata of the earth. This current, therefore,sees a conductor with a practically infinite
cross section.

Earth electrode resistance is the governing factor in determining the losses which occur in
the ground. Because current density decreases in homogeneous earth with the square of the
distance, almost all of the losses occur in the close proximity of the earth electrode.

Depending on the current carrying capacity desired, recommended earth electrode


resistances are between 50 and 200 mW.

- 308 -
4.2.4 Electrode Voltage

Since the required earth electrode resistance is dependent upon the rated current, electrode
voltage can be used more appropriately as a design criterion.

The earth electrode voltage is defined as the product of the rated current and the earth
electrode resistance. Recommended values are between 100 V and 200 V.

4.2.5 Step Voltage

The transfer of the current to the ground and the current which runs through the earth
surrounding the electrode results in the formation of a voltage crater, the equipotential
surfaces of which surround the earth electrode like the shells of a hemisphere. As a result
of the decreasing current density, the distance between the equipotential surfaces increases
with the distance from the earth electrode, i.e. the maximum field strength occurs in the
immediate vicinity of the electrode.

To prevent hazards to animals and human beings, the field strength which occurs on the
surface should not be greater than 5 V/m. This value has a large safety margin compared to
the step voltage of 20 V/m which is otherwise permitted.

When calculating step voltage, the maximum short-term earth electrode current which
occurs must be considered. Transient overcurrents may be ignored.

Submarine electrodes require special attention. If they are accessible to swimmers or


divers, field strength must be limited to 3 V/m.

- 309 -
4.2.6 Touch Voltage

Hazardous touch voltages can occur if rural fences run in the vicinity of the earth
electrode. Because wooden posts may be used to anchor such fences, there is sufficient
insulation to ground, so that the potential resulting from a chance contact with ground can
be transported over long distances and result in a hazardous touch voltage. A similar effect
can be caused by transportable irrigation pipes used in agriculture. Connection or
disconnection of the irrigation pipes can be particularly hazardous.

A value of approximately 20 V should be considered as the limit for the touch voltage,
although the public rules generally allow 60 or 75 V for direct current.

Nevertheless, the touch voltage is not considered a sizing criterion for the earth electrode.
This phenomenon must considered carefully in each individual case. If fences or irrigation
pipes bridge a potential difference which could result in touch voltage > 20 V, longitudinal
insulators or insulating flanges should be inserted as a precaution.

4.2.7 Risk of Corrosion

Basically, a risk of corrosion exists for metal structures embedded in the earth over a very
wide circular area around an earth electrode. This is one of the principal arguments against
monopolar HVDC systems - apart from their use in submarine cable transmission systems
- and for a very narrow restriction of the time limit for temporary operation with current
conducted through the earth in bipolar HVDC systems.

The concerns raised have generally been overstated. In the vicinity of large cities, the
direct currents in the ground caused by subways, municipal railways and tram systems are
generally dissipated quite well, although such systems involve vast tangles of cables and
pipelines. Nevertheless, careful consideration must be given to such concerns laid down in
specifications and safety regulations.

- 310 -
The following figures can be used to better evaluate the phenomenon of electrochemical
corrosion as a result of a current flow carried by ions into a surrounding electrolyte
medium.

For 80% of the material of a 1 mm thick metal body buried in the earth (cable jackets,
pipelines etc.) to remain after 35 years of continuous operation of earth electrodes, the
following values could be allowed:

0.2 mA/cm2 surface area for lead or


0.5 mA/cm2 for iron.

These values apply only for metal bodies laid in the earth without any insulation. If the
cable jacket or the pipeline are insulated, and there are locally restricted insulation failures,
a much higher current density will occur at these points and may result in severe corrosion
damage in a very short time. In such cases, therefore, cathodic protection must be
provided. This reduces the potential of the metal body and prevents an anodic current
flow, or reverses it into a cathodic current flow carried by electrons, which is acceptable in
terms of corrosion. But the potential must not drop below -0.85 V (for iron).

In recent years, values between 0.1 and 20 mA/cm2 have been indicated for uninsulated
metal structures laid in the earth, depending on the type of operation and the duration of
the current carried by the earth electrodes.

Measures to reduce corrosion include:

• Insulation of cable jackets and pipelines combined with use of cathodic protection
• Insertion of insulating flanges in pipelines
• Insulation of the shield wires in overhead lines
• Installation of isolating transformers in low voltage distribution systems.

The most effective protection is achieved by maintaining a distance of several km from


equipment which is in danger of corrosion (guideline value ³ 10 km).

- 311 -
4.2.8 Effects on AC Systems

When an ac system is grounded in two places which - due to HVDC operation with
current carried by ground return - are connected to different potentials, there is a direct
current through the transformers and transmission lines of the ac network. The
consequence can be a dc flux in transformers. This results in a higher and asymmetrical
magnetization current and increased noise.

Inductive voltage transformers can also carry such a direct current, which can result in
false tripping.

Generally the effects of the phenomena, which can also affect the HVDC stations
themselves, have been overstated. Very few such problems have actually been
experienced. It is important to select the site of the earth electrodes so that they are far
away, i.e. ³ 10 km, from ac substations. If problems do occur, the grounding of
transformer neutral points can be accomplished by means of a resistance of a few Ohms
which is bridged by a surge arrester with a low clipping voltage.

4.2.9 Effects on Telecommunication Systems

Telephone, telegraph or railway signalling systems operated with direct current are
sensitive to direct currents of other systems in the ground. But since such
telecommunication systems are themselves a thing of the past, they will probably not be a
criterion in making design decisions. If necessary, an old system must be replaced by a
new one.

- 312 -
4.2.10 Compass Declination

A problem which is always relevant to monopolar submarine cable transmission systems


is compass declination. This problem is related only indirectly to the conduction of current
through the water and the sizing of the earth electrodes. Declination is the result of the
uncompensated magnetic field of the cable core. In particular when the cable is laid in the
north-south direction, a ship navigating with a magnetic compass can be "caught" when it
crosses the cable at a small angle.

If a ship navigating with a magnetic compass crosses the cable at a larger angle, there will
be a lateral drift as illustrated schematically in Figure 4.4. Such a compass declination can

have disastrous effects in the vicinity


of the coast and at harbor entrances.
Therefore it may be necessary to lay
pole cables and submarine electrode
cables in a common right of way in
these areas to guarantee magnetic
field compensation.

b This requirement will affect the


selection of the location for
submarine electrodes.
Figure 4.4: Compass Declination

Problems caused to small craft navigation (only small craft still navigate using a magnetic
compass) can be avoided by laying the pole cable in a zig-zag pattern. Cable laid in an
east-west direction does not present any problem to navigation at all.

- 313 -
4.2.11 Useful Life

Basically, the earth electrodes must have the same useful life as the HVDC system of
which they are a part. Under normal conditions, this means that the percentage of material
corrosion in anodic operation must be sufficiently low that there is no adverse effect on the
operational characteristics of the earth electrodes.

The ampere-hours expected during the useful life of the electrodes must be specified, and
the percentage loss of active electrode material considered acceptable must be defined.
The selection of material, which is discussed in greater detail in Section 4.5, also plays an
important role in this process.

If, in practical terms, it is impossible to design the earth electrodes for the full useful life of
the HVDC system, which can very easily be the case for anodic electrodes of monopolar
HVDC systems, the electrodes must be designed so that regeneration is possible, i.e. the
refilling of used anode material. For this purpose, the earth electrodes must be divided into
sections, so that earth electrodes will remain fully operational under the (n-1) principle
even when one section is separated.

- 314 -
4.3 Site Selection for Earth Electrodes

The number and variety of design criteria which must be taken into consideration mean
that it is very difficult to find a suitable site for the earth electrodes of an HVDC system
which meets all the applicable requirements.

4.3.1 Ground Characteristics

The most important requirement for the construction of an earth electrode relates to
suitable soil characteristics. The most important of these characteristics is the electrical
conductivity of the soil, generally expressed by the specific resistance in Ohm meters
(Wm). The ground in question is not only the soil in the immediate vicinity of the earth
electrode, but also the strata farther away, up to several hundred meters deep and several
kilometers away, primarily in the direction of the opposite electrode. Nevertheless, the
specific resistance in the immediate vicinity of the earth electrode is the decisive factor for
electrode resistance and thus losses, because current density is highest at this point.

Figure 4.5 shows a typical ground


strata structure with a surface
stratum which has a very good
conductivity , a conductive sedi-
mentary stratum 200 m thick which
is semi-conductive, and solid rock
underneath that.
Figure 4.5: Ground strata suitable for the use of flat electrodes

Such a stratification structure is particularly suitable for the installation of a flat electrode
(horizontal electrode).

- 315 -
In contrast, Figure 4.6 shows a
stratification structure with a surface
layer 100 m thick which has very
low conductivity. Below, there is a
stratum with very good conductivity.
The only electrode which can be
used with such a stratification
structure is a deep electrode (vertical
electrode).

Figure 4.6: Ground strata suitable for deep electrodes

The specific resistance of the ground can be determined by means of the Schlumberger
method, which uses current and voltage measurements, as illustrated schematically in
Figure 4.7

The further apart the current electrodes are, the deeper the strata which are included in the
measurement. In homogeneous ground and partial horizontal stratification, it is possible to

get a reliable image of the


stratification of the substrate and of
its electrical conductivity.

Figure 4.7: Measurement of the specific ground resistance using the Schlumberger
method

- 316 -
In addition to the current and voltage measurements, test borings must also be made at
various points and the drill cores be examined in the laboratory.

Additional important parameters of the ground characteristics are:

• Thermal conductivity;
• Thermal capacitance;
• Porosity (water and gas permeability);
• Penetration of moisture and influx of water.

To determine these parameters, ground samples must be taken and measurements must be
performed both, in the laboratory and on site. When determining the groundwater level,
consideration must be given to whether the present or future extraction of water, e.g. for
irrigation, may cause a reduction of the groundwater level.

4.3.2 Distance from Civil Works

The term "civil works" as used here includes all objects which are in danger of corrosion
from the direct current flowing in the ground. The important factors are not only the
distance, but also the position of the objects in relation to the principal direction of the
current, i.e. toward the opposite electrode. For long objects such as pipelines, cables and
rails, for example, the angle in relation to the principal angle of the current is important.
The conductivity of the strata near the surface in which these objects are embedded is the
most important factor and determines the degree of the risk of corrosion, i.e. the minimum
distance which must be maintained to eliminate the risk.

- 317 -
Because of these many interdependencies, only very approximate guideline values can be
indicated for the distance which must be maintained from civil works.

• Fences, irrigation systems > 0.3 km


• Small towns > 3 km
• Pipelines, cables, rails > 6 km
• Electrical substations > 10 km

The maintenance of these guideline values naturally does not take the place of a more
detailed investigation of the problems. Local conditions also frequently mean that it is
impossible to observe these distance guidelines. Nevertheless, these guideline values can
be quite useful for a preliminary selection of possible sites for an earth electrode.

4.3.3 Proximity to the Right of Way of the HVDC Overhead Line

The connection of the earth electrodes to the HVDC station over a distance of 10 km, 30
km or more is an important cost factor. Therefore in HVDC systems with an overhead
line, consideration has frequently been given to using the HVDC overhead line towers for
the electrode line, at least over a large portion of the distance. In this case, as indicated in
Section 4.7, the function of the shield wire can be allocated to the electrode line.

Common routing of the HVDC overhead line and the electrode line is also advantageous
because the earth electrode is then being shifted toward the opposite electrode, so that the
direct current flow through the ground hardly affects the HVDC station itself. The
proximity to the right of way of the HVDC overhead line is therefore an important factor
which must be taken into consideration when selecting the site of an earth electrode.

- 318 -
4.3.4 Accessibility

Heavy equipment must be used for the construction of an earth electrode, and large
quantities of material have to be transported. Therefore it is important that the electrode
site is not too far from solidly-built roads and routes which can be used to transport the
expected weight, so that no significant additional road building costs are incurred.

For the subsequent operation of the system, the electrodes must be accessible at all times
for inspections. If the design of the earth electrodes calls for the periodic regeneration of
individual sections, transportation required for this purpose must be possible without
additional construction requirements or special permits.

4.3.5 Environmental Considerations

The search for a suitable site for an earth electrode may lead to a national park or nature
preserve, where, although all the requirements indicated above may be present, it is
impossible to obtain a construction permit. Therefore it is important, before searching for
an earth electrode site, to investigate the possibilities of getting a construction permit and
meeting the requirements of any environmental laws and regulations for the areas being
considered, in particular if recreational or protected areas may be affected.

- 319 -
4.4 Design of Earth Electrodes

Earth electrodes for HVDC systems may be land, coastal or submarine electrodes. In
monopolar HVDC systems, which exist almost exclusively in the form of submarine cable
transmission systems, there are fundamental differences between the design of anode and
cathode electrodes. There are a large number of variations within each category, almost as
many as there are HVDC systems. In the following sections, discussion is limited to the
most important designs.

4.4.1 Horizontal Land Electrode

If a sufficiently large area of flat land with relatively homogeneous ground characteristics
is available, and if the land meets the selection criteria indicated in Section 4.3, the
horizontal ground electrode will be the most economical form of a land electrode.

As shown in Figure 4.8, the


electrode conductor itself, which is
generally made of iron, is laid
horizontally at a depth of
approximately 2 m. It is embedded
in coke which fills a trench having a
cross section of approximately 0.5 x
0.5 m2.

Figure 4.8: Cross section through a horizontal land electrode

- 320 -
The advantage of this design becomes apparent in anodic operation. The passage of the
current from the electrode conductor into the coke bed is carried primarily by electrons,
and is thus not associated with loss of material. The passage of the current from the coke
bed into the earth occurs by ion conduction, i.e. a significant amount of material is lost
because of corrosion. But as shown in Section 4.5 below, the corrosion rate for coke is
much lower than for iron, and there is also a much greater amount of coke available, at
least, if the above-mentioned guideline value of 1 A/m2 of the current density at the
boundary layer will be observed.

The purpose of the cover layer of crushed stone is to refill the excavation by "settling" into
the cavities which are formed by material corrosion on the surface of the coke bed.

The design of a horizontal land electrode illustrated in Figure 4.8 has been used, with
minor variations, in all the HVDC earth electrodes of this type constructed. But there is no
similarity at all in the external shape of the earth electrodes. In fact, there is an almost
infinite number of possibilities of distributing the required electrode length over a given
area. Several typical patterns of horizontal land electrodes are illustrated in Figure 4.9:

Figure 4.9: Typical designs of horizontal land electrodes

- 321 -
The line electrode (a) oriented at right angles to the main direction of current flow offers
the most efficient utilization of material and a particularly uniform current distribution, if a
sufficient number of current feeds are provided along the electrode. Only the two terminal
points are subjected to a slightly higher current density. The disadvantage is the large
amount of land required. This disadvantage can be eliminated by means of a multi-line
electrode( b), but the utilization of the material is not as efficient.

The ring shape (c) is the most economical form. By adding a concentric second ring (d),
the resistance of the earth electrode can be reduced by approximately 20%, although the
cost of material then becomes higher.

Many different examples of the star-shaped electrode (e) have already been constructed,
and they can be adapted quite flexibly to the contour of the terrain and the characteristics
of the ground by means of different lengths or angles of the arms. Because of the shielding
effect near the origin of the star, however, the load is quite non-uniform. Thus the forked
star (f) can be used to better advantage, although it is more complex and expensive to
install.

4.4.2 Vertical Land Electrode

If the ground strata near the surface have a high specific resistance, but underneath, there is
a conductive and sufficiently thick stratum at a depth of several tens of meters, the vertical
deep electrode will be one possible solution.

Figure 4.10 shows, as an example, one of the four deep electrodes at Apollo, the southern
station of the Cabora Bassa HVDC system. The use of coarse-grain graphite as the
material in which the graphite conductor is embedded, not only achieves conductive
contact with the surrounding earth, but also guarantees permeability for gases and
groundwater.
The specified requirements were met, but the electrodes proved not to be suitable for
continuous operation lasting months. Due to the unusually high current density of > 4
A/m2 at the boundary surface to the conductive layer, the soil dried out, and ultimately
melted to form a highly-insulating glass.

- 322 -
Figure 4.10:
Vertical Electrode at Apollo,
the Southern Cahora Bassa
HVDC Station

The two ground electrodes of


the IPP HVDC in the Western
United States can be
considered examples of a
conventionally-sized and thus
much more expensive
deep electrode. No less than 60 deep electrodes with a borehole diameter of 0.35 m were
installed for each ground electrode. The holes are located on the circumference of a circle
900 m in diameter. The deep electrodes were grouped into sections of six. Each section
can be isolated individually, with the remaining 9 sections allowing an unrestricted
operation.

4.4.3 Cathodic Submarine Electrodes

The design and construction of the cathodic submarine electrode of a monopolar HVDC
system with submarine transmission cable do not present any particular problems. Since
there is no material corrosion, a copper cable laid on the bottom should theoretically
suffice. The length of the cable must be designed so that the current density on its surface
does not cause an electrical field of >3 V/m in the surrounding water, which is also safe
for swimmers and divers. For example in the Konti Skan system, a 1000 A electrode was
laid in the Baltic Sea. It was a copper cable 250 m long around a rock island. A field
strength of 2.7 V/m was attained.

Since fish are repelled by a negative electrical field, there is no reason to install any
protective devices.

- 323 -
4.4.4 Anodic Submarine Electrodes

The oxygen generated during the electrolysis of water on the anode is extremely
aggressive, causing corrosion of the anode material. Because of its particularly low
material corrosion rate, graphite proves to be a particularly suitable material. Care must be
taken, however, that the chlorine gas which is generated during the electrolysis of sea
water is able to escape or is flushed away. Unfortunately, graphite is not very resistant to
waves and ice, and therefore designs must be used which are very similar to horizontal
electrodes, as illustrated in Figure 4.11 using the example of a linear submarine electrode
for anodic operation. The prefabricated electrode modules are lowered to the ocean floor
and then connected to the feed cable. When the submarine electrodes are divided into
sections which are connected to the HVDC station by means of separate feed cables, the
electrode can be monitored from the land.

Figure 4.11: Linear submarine electrode (anodic operation)

- 324 -
Such an electrode design is very expensive. Therefore there have been many attempts to
construct submarine electrodes for anodic operation using alternative design principles. In
Sardinia, for example, an electrode was constructed in which a number of platinum-plated
titanium rods extend into the water from a wooden bridge spanning a small inlet. The rods
are adjustable and can be replaced easily. The inlet is protected by a net, so that access by
both, fish and swimmers is prevented. This submarine electrode has been in operation for
almost 30 years. In the Fenno-Skan Project, the anodic submarine electrode consists of a
titanium mesh coated with a noble metal which is simply laid on the ocean bottom.

4.4.5 Anodic Coastal Electrode

The feed cable is a significant cost factor in submarine electrodes. This is particularly true
in the case of an electrode which must be installed far from the coast for safety reasons.
An additional disadvantage is the very difficult access for inspections and for any repairs
which may be necessary. In any case, regular regeneration to extend the useful life of the
electrode is impossible. Therefore the anodically operated earth electrodes in several
submarine cable transmission systems are designed as coastal electrodes.

The conventional design of a coastal electrode is similar to that of a vertical land electrode.
Graphite rods surrounded by a coke bed are installed in boreholes which are sunk along
the coastline. Since sand, which is quite unsuitable to carry current, is generally
encountered at these points, such coastal electrodes are frequently artificially irrigated to
guarantee conductivity and to prevent drying-out.

The advantage of the coastal electrodes is easy accessibility for inspection, maintenance
and regeneration, if necessary. The problem is generally in obtaining the required permits
for construction and operation.

A coastal electrode can also be configured in the form of a horizontal land electrode if the
ground has the necessary conductivity or if the necessary conductivity can be achieved by
irrigating the trench with salt water. In either case, it is assumed that even with a coastal
electrode, the current flow to the opposite electrode takes place almost exclusively through
the water.

- 325 -
4.5 Anode Material

As noted above, the selection of material for an earth electrode operated exclusively
cathodically is no problem. On the other hand, the selection of a suitable electrode material
for earth electrodes, which are operated exclusively anodically or in alternation, is
extremely important. Material loss due to corrosion of electrode material is measured in
kilograms per ampere year, kg/A×a, where a = 1 stands for one year of operation.

4.5.1 Iron

Iron is very economical, mechanically strong, and is easy to work, lay and weld. Its
electrical conductivity is sufficiently high for use as a conductor in an earth electrode. But
these advantages are offset to some extent by the disadvantage of a high level of corrosion
in anodic operation, 9.13 kg/A×a.

With an anodic electrode at 1000 A continuous current, that represents a corrosion loss of
9.13 t/year.

Nevertheless, iron has very frequently been used as a conductor in horizontal land
electrodes, but always embedded in coke. Since the current transmission from iron to coke
is carried primarily by electrons, the corrosion loss in the iron is reduced to 0.09 - 0.45
kg/A×a.

4.5.2 Silicon-Iron

Iron with a high silicon content and the addition of chromium has a significantly lower
corrosion loss in anodic operation than pure iron. It is resistant to chlorine gas and is
therefore suited for submarine and coastal electrodes. In contact with sea water, the
corrosion loss is 0.25 - 1.0 kg/A×a.

The brittleness of the material, however, means that it cannot be used as a long conductor.
Cast bars or plates connected to one another are used. The iron elements are generally
embedded in coke.

- 326 -
4.5.3 Platinum-Plated Titanium

This material, which is very well-suited for submarine electrodes in anodic operation but
is also very expensive, has by far the most favourable corrosion rate of 6-9×10-6 kg/A×a,
i.e. it can be used in a design for a long useful life.

4.5.4 Graphite

Because of its electrical characteristics, graphite is excellently suited for earth electrodes
operated anodically; the corrosion is only 0.05-0.2 kg/A×a.

Graphite is resistant to seawater and is impervious to chlorine gas. Therefore it can also be
used for submarine electrodes, as illustrated in the example in Figure 4.11.

The brittleness and low mechanical strength of graphite are disadvantages. Nevertheless,
the material has proven suitable for the vertical electrodes in Apollo, which are installed at
a depth of 130 m. The conductor was screwed together from individual rods 30 cm in
diameter and several meters long.

4.5.5 Coke

For most land electrodes, and also for some coastal electrodes, the material which is
actually in contact with the surrounding medium is the coke bed. Naturally, in anodic
operation, the coke is also subjected to corrosion, which amounts to 0.5-2.0 kg/A×a.

Coke is cheap, available everywhere, easy to transport and can easily be introduced into
the prepared excavations. The coke bed can be generously sized to achieve a long useful
life.

- 327 -
4.6 Project Stages

Although the specifications and orders for earth electrodes are usually issued separately
from those for the HVDC stations, it is important to define the individual project stages
and to track the progress of the project so that the necessary attention can be devoted to
the interfaces between the two subprojects of an HVDC system.

4.6.1 Site Selection

The various criteria which are important for selecting the site of an earth electrode were
discussed in detail in Section 4.3.

In each of these processes, the practical method consists of a series of steps. The first step
is to get an overview of the conditions in a certain area around the HVDC station. The
next step is to exclude those areas for which there is a "knock-out" criterion in one point or
another, and finally to perform the necessary investigations of the remaining areas to
obtain a basis for evaluating their suitability for the installation of an earth electrode.

An evaluation matrix which includes all the selection criteria can then be used to establish
a ranking of the various sites to be considered. The final decision, of course, is to be made
by the customer.

4.6.2 Definition of the Design

The various designs of earth electrodes were illustrated in Section 4.4. In an optimization
study, the design options must be considered and calculated in the light of the conditions
of the selected or proposed sites. The result is a preliminary estimate of the manufacturing
costs of the various designs, which can then be used as an important decision-making
criterion.

- 328 -
Electrical losses which occur during electrode operation do not have any measurable
influence on the return on investment, but they can be used as an indication of the safety
margin which must be maintained with regard to the drying of the soil.

Maintenance and regeneration costs, if any, are also of major importance, since they are
reflected in the current operating costs.

One aspect of the selection of a suitable design which should not be underestimated is the
extent of existing experience from the operation of similar earth electrodes in other
systems.

4.6.3 Construction and Operation of a Prototype Electrode

In spite of all the basic knowledge gained in recent years and the resources currently
available to perform very complex calculations on increasingly refined models, and in
spite of the extensive experience which has been gained, it is strongly recommended that a
prototype electrode be constructed at the selected site and operated for at least 6 months. In
order for the deeper earth strata to be taken into consideration, the opposite electrode must
be installed a sufficient distance away. The current density at the transition to the earth
must equal the original value. To avoid distortion of the measurements caused by stray
currents in the ground, the total current should not be less than 100 A.

If the selected design allows for the earth electrode to be divided into sections, it is
recommended that one section be installed as a prototype electrode. This prototype
electrode can also be used to test and optimize the manufacturing process.

The long-term operation of the prototype electrode may result in verification of the design
and sizing of the earth electrode, or it may reveal corrections and improvements which
must be made. Construction specifications can then be written.

- 329 -
4.6.4 Construction and Startup of the Earth Electrodes

During the first phase of construction, the excavation must be monitored constantly to
identify any irregularities of the earth such as rocks, faults, slips etc. This monitoring is
particularly important with deep electrodes.

After completion of the earth electrodes, current transmission measurements should be


made between the individual sections of the electrode to verify the correct installation, e.g.
proper contact between the electrode conductor, coke bed and ground over the entire
surface area. The power supply unit of the prototype unit can be used for these tests.

As soon as the HVDC stations and the line are available, the acceptance tests can begin. If
possible, they must include the specified maximum load, i.e. the maximum continuous
current over the specified maximum period of time.

These measurements must also include:

• the earth electrode voltage, plotted over the duration of the test
• the step voltage in the area surrounding the earth electrode, with a plotting of the
equipotential lines
• heating of the ground at various points, plotted over time
• touch voltages on fences, water pipes and similar objects within a radius of several km
• measurements on all objects at risk of corrosion - pipelines, cables, rails - in the
surrounding area. These measurements must include a test of the effectiveness of
cathodic protection.
• observation of switchyards and telecommunication systems in the surrounding area to
identify any adverse effects of the direct current

- 330 -
4.7 Electrode Line

Since the electrodes must be installed several kilometers away from the corresponding
HVDC station for the various reasons indicated above, they must be connected to the
station neutral bus by means of the electrode line. For land electrodes, this line is generally
an overhead line. For coastal or submarine electrodes it is generally an overland cable or a
submarine cable or a combination of the two.

The following section discusses only those phenomena which occur in particular in the
electrode line, i.e. those which were not discussed in Chapter 3, "HVDC Lines".

4.7.1 Electrode Line as Separate Overhead Line

Since an electrode line is always short compared to the HVDC overhead line, and - at least
in bipolar systems - the length of time it carries a current is also short, line losses do not
play a role in economic terms. Therefore, the selection of the conductor cross section will
be guided by thermal considerations.

Figure 4.12 Homopolar Electrode Line

- 331 -
In order to provide a means of locating line-to-ground faults, it is helpful to configure the
line as a homopolar line, with two separately insulated conductors. A simple, though not
very precise method of locating a fault consists of taking comparative measurements of
currents. For this purpose, a temporary current can be made to flow in the electrode lines
by using the dc control to intentionally unbalance the current in the station poles. If there is
an insulation failure on one of the two conductors of the electrode line, a current difference
will be measured. The relative magnitude of this difference can be used as a measure to
determine the distance to the location of the fault.

If the earth electrode is located not too far from the


right of way of the HVDC overhead line, the cost
of the electrode line can be reduced significantly by
using the HVDC overhead line towers for the
electrode line, as shown in Figure 4.12.

Figure 4.13: Bipolar HVDC line with electrode


line

4.7.2 Electrode Line as Shield Wire

The combination of HVDC line and electrode line


can be made even more attractive by using the
electrode line for lightning protection, as illustrated
in Figure 4.14. The high degree of protection
against direct lightning strikes on the HVDC line
which is necessary in the vicinity of the station is
provided by the two conductor cables of a
homopolar electrode line. The low insulation of the
electrode line does not have any negative influence
on the shielding effect.

Figure 4.14: Electrode line used as shield wire

- 332 -
4.7.3 Insulation of the Electrode Line

The voltage drop along a direct current overhead line operated at the thermal limit current
is calculated as approximately 50 V/km. The potential of the station neutral bus therefore
will not exceed a few kV, even with a very long electrode line. One unit of a glass or
porcelain cap-and-pin insulator, which are conventional in HVDC technology, is therefore
completely sufficient to insulate the electrode line.

Insulator flashovers in the event of a direct lightning strike of the conductor, or back
flashovers if lightning strikes a tower or other objects in the vicinity, remain a problem.
Because of the low level of insulation of the line, it is practically impossible to prevent
these flashovers. Unlike an insulator flashover of the HVDC line, however, no blocking is
initiated of the converters, which would be very undesirable because of the frequency of
occurrence which could be expected.

The problem is complicated by the danger of a steady-state arc caused by the continued
flow of direct current. This risk also exists with bipolar operation. A current difference
between the poles of even a few Amperes is sufficient to maintain the flashover arc. Such
an arc would quickly destroy the glass or porcelain cap of the insulator. Therefore
arrangements must be made to eliminate the arc by means of automatic self-extinction.

Figure 4.15 shows the insulator system


which has been developed for the
insulated shield wire of the Cabora-
Bassa HVDC, which consists of one
glass cap-and-pin insulator with arcing
horns.

Figure 4.15 Insulator for an insulated shield wire

- 333 -
It has been shown in laboratory tests that the magnetic forces on the arc are negligible for
currents of a few amperes and that thermal motive forces are also low. Hence, self-
extinction can only be assured by a series of measures:

• The flashover path must be horizontal, to allow the bases of the arc to migrate to the
tips of the arcing horns.
• The arcing horns must be made of stainless steel. If galvanized standard steel is used,
the footings of the arc are held by vaporizing material.
• The shape of the arcing horns must guarantee that no raindrops are kept in the flashover
path, which can also result in a holding of the footings of the arc.
• The electrical connection of the arcing horns must be done with low inductivity so that
even with extremely steep voltage surge wave fronts of up to 5000 kV/ms, there is no
insulator puncture, but only an external flashover.

4.7.4 Electrode Cable

The connection between a coastal electrode and the corresponding HVDC station could
theoretically be made by means of an overhead line. But because of the risk of insulator
flashover as a result of contamination by salt spray, which is always present in coastal
areas, preference is generally given to a connection using a land cable.

The connection of a submarine electrode requires a submarine cable. In that case, one
special problem is the release of chlorine gas during anodic operation of the electrodes.
This factor must be taken into consideration in the design of the cable insulation, fittings
and connections.

- 334 -
5. Control Technique in HVDC Systems

The foundations of HVDC technology and their applications have been presented in
Chapter 1, and the major components of HVDC systems in Chapters 2 through 4. This
Chapter covers HVDC control technique, which unites the functions of the components
with the desired or specified behaviour of the system. In addition to traditional control,
regulating and protection functions, measuring technology and telecommunications, which
is indispensable for transmission over long distances, are also considered.

Note: In this chapter the term „regulation“ is used to describe a close loop control system
or function while the term „control“ is used to describe an open loop control
system or function.

The emphasis in this chapter will be on functions. However, their device-based


implementation in hardware and software will also be presented.

5.1 Converter Regulation Functions

The converter valves, as precise and virtually delay-free control elements, are the most
important actuators of the HVDC control system. Moreover, in most cases the converter
groups have an additional actuator in the form of the transformer tap changer. Although it
does not operate on a continuous basis and there are relatively long periods of idle time, it
nevertheless is responsible for important regulation functions.

5.1.1 Current Regulation

All HVDC systems implemented to date are constant voltage systems. Direct current thus
assumes critical significance as a system control variable. In particular, current regulation
determines

• steady-state transmission power

• changes in transmission power according to size and rate of change

• the dynamic behaviour of the system including temporary overload

• limitation of transient overcurrents determined by amplitude and duration

- 335 -
The loading of all essential components of an HVDC system, with the exception of filter
circuits, is determined by the direct current or an alternating current proportional to the
direct current. Therefore current regulation is also a very essential protective function.

In HVDC two-point systems, the rectifier generally assumes the task of current regulation.
It has occasionally been proposed in the literature to assign the current regulation function
to the inverter. However, there are good reasons to leave this function with the rectifier in
normal operation. Only the rectifier can assume the above mentioned protective function
of current limitation. An inverter cannot exercise any effect on a rise in direct current
caused by a short circuit. However, since the current regulation of the rectifier is needed as
a protective function, it is advantageous to also use it for this purpose during normal
operation. Then it is always active and monitors itself.

The situation is different with HVDC multiterminal systems in which only one converter
group can regulate the dc voltage as a common variable of all converters. All other
converter groups irrespective of operating as rectifier or inverter regulate their individual
dc current.

5.1.2 Extinction Angle Regulation

The idea of keeping the extinction angle of an inverter to the minimum value, gmin, during
normal operation is obvious and has a number of advantages:

• a reliable commutation of the direct current is assured

• the reactive power requirement of the inverter is minimized

• the best possible use of the installed equipment ratings of converter valves and
transformers is obtained

- 336 -
Extinction angle regulation can best accomplish the task of holding the minimum
extinction angle constant for all load and network conditions. Consequently, the
technology of the Arbeitsgemeinschaft HGÜ was to equip the inverter with extinction
angle regulation, but supplemented by current regulation, which comes into operation in
response to a lowering of system dc voltage. The inverter current reference is less than the
normal rectifier current reference by the current margin, DI.

However, extinction angle regulation also has serious disadvantages:

• An inverter operating with a constant extinction angle has negative internal resistance
which can result in stability problems.

• The regulation has relatively high dead time. This is especially true when equidistant
firing is applied and the extinction angle of only one valve of a 12-pulse group is
therefore used as the actual value.

• This dead time requires the setting of slow control dynamics. Extinction angle
regulation reacts inadequately to all fast transient events and must be modified by
techniques such as using other control signals and nonlinear regulator characteristics.

Consequently, the dynamically far superior dc voltage regulation by the inverter has
gained favour with extinction angle limitation being provided as a superimposed control
loop or as an override to increase the extinction angle when needed.

5.1.3 DC Voltage Regulation

Electronic dc voltage regulation of a converter is extremely fast when the actual value is
obtained directly at the converter terminals, i.e., between the converter and the smoothing
reactor. The use of dc voltage regulation also has the advantage that the inverter no longer
has negative internal resistance. This improves the stability of the HVDC operation. An
inverter controlled in this manner contributes to the voltage stability in the ac network.
When a voltage increase occurs on the ac busbar of the inverter substation, the converter
reacts with an increase in the angle of advance b and thus of the extinction angle g. This
results in an increase in the reactive power demand which opposes the ac voltage increase.
The reverse is also true.

- 337 -
The electronic dc voltage regulation of an inverter also has disadvantages:

• In order to have an adequate control margin, the nominal extinction angle must be
selected a few degrees greater than gmin

• Because of the larger extinction angle, the reactive power demand of the inverter
substation is greater

• For the same reason the rated power of converter valves and transformers are greater
than with g = gmin

Improved operational behaviour offsets these increased "costs". In addition to the much
faster reaction to transient changes of set points and system disturbances, on the average,
larger extinction angle results in a statistically lower number of commutation failures.

Of course, an HVDC two-point system can also be operated with electronic dc voltage
regulation by the rectifier and current regulation by the inverter. However, this has the
disadvantages referred to above with regard to the protective function of current
regulation.

On the other hand, it is very advantageous to equip the rectifier with a voltage limiting
regulator which, as a superimposed control loop or as a parallel regulation, is responsible
for preventing the maximum rated voltage of the d.c. transmission line from being
exceeded. Such voltage limitation regulation is particularly important in HVDC systems
with cables.

- 338 -
5.1.4 Control Characteristics of an HVDC System

Interactions of the converter substations in an HVDC two-point system can be presented


most simply by the control characteristics of the two substations plotted as a current-
voltage diagram. In this type of presentation, it has become customary to allocate the same
sign to the same voltage polarity for the rectifier and the inverter.

Fig. 5.1 depicts the control characteristics of an HVDC two-point system which is
operated according to the so-called current margin method. The point of intersection A of
the two characteristic curves indicates the operating point of the system. It is determined
by the current regulation of the rectifier with the set point IdGR and the extinction angle
regulation of the inverter with the set point gmin.

Figure 5.1:
Control characteristics of an HVDC
a=0
two-point system
g= g

Id* = set point of the current


regulation
D DI = current margin

[Note: WR = inverter
* * GR = rectifier]

When the rectifier cannot achieve its current reference, I*dGR, the inverter regulates current
as shown by the set point Id*WR, which is smaller by the current margin DI than the set
point of the rectifier current regulation.

The voltage characteristic curve of the rectifier is indicated by the maximum level (a = 0).
If this curve drops, e.g., because the ac busbar voltage has undergone a reduction, the
result is an intersection at A'. Here the inverter controls the direct current and the rectifier
determines the dc voltage by its lowered characteristic.

- 339 -
Fig. 5.2 depicts the control characteristics of an HVDC two-point system in which both
substations are equipped with electronic voltage regulators. The set point of the voltage
regulation of the rectifier is greater than the set point of the voltage regulation of the
inverter by the voltage margin DU.

Figure 5.2:
*
Control characteristics with the use
* of electronic voltage regulation

D
Ud* = set point of the voltage
D regulation
DU = voltage margin

If the sign of the voltage margin is changed, the rectifier will automatically take over
voltage regulation and the inverter will take over current regulation.

An improvement of the transient behaviour can be obtained by introduction of the so-


called "voltage dependent current order limitation" VDCOL. A drop in the system dc
voltage Ud caused by ac system problems, automatically results in a reduction of the
current set points.

Figure 5.3:
Modified control characteristics *
*

VDCOL = Voltage Dependent D

Current Order
Limitation
CEC = Current Error
Compensation

- 340 -
In this way, an increase in reactive power demand and increased valve stresses are
prevented in the substation not affected by the disturbance. In addition, fault clearing is
made easier in the substations affected by the fault.

Fig. 5.3 also depicts an approximation, labeled CEC (current error compensation), of the
current control characteristics of the rectifier and inverter in the vicinity of the operating
point. This prevents an abrupt reduction of DI in the transmission current when system
voltage is gradually reduced.

The use of the current margin method is also applicable to HVDC multiterminal systems.
Fig. 5.4 depicts the control characteristics of such a system. Here, the characteristics of all
current controlling substations (CC) are combined, with the current set points of rectifiers
being counted positively and those of inverters negatively.

Figure 5.4:
Control characteristics of an HVDC
*
* multiterminal system

D VC = Voltage Controlling Station


CC = Current Controlling Station
D

* S *

The point of intersection with the control characteristics of the one voltage controlling
inverter (VC) yields the operating point A of the system. Here again, the current set point
of the voltage regulating inverter is smaller by the current margin DI than the sum of the
current set points of all current regulating converters. The voltage set points of the current
regulating converters are larger by the voltage margin DU than the voltage set point of the
voltage regulating converter.

- 341 -
There are many suggestions for controlling an HVDC multiterminal system, differing
essentially in the design and combination of regulation characteristics. Space does not
permit an individual presentation and discussion of these variations, especially since the
current margin method depicted in Fig. 5.4 was used in the planning of the actual HVDC
multiterminal system, the East-West High Power Link in Europe.

5.1.5 Regulation Functions of the Transformer On-Load Tapchanger

Usually, an HVDC converter has, in addition to the control angle a or g a second control
variable, the transformer conversion ratio, which can be changed by the tap switch (on-
load tap changer) of the converter transformer. Thus, independent of the voltage on the ac
busbar, the valve voltage Uv or the rated no-load dc voltage Udi can be adjusted as is
required for the respective load point and the desired reactive power behaviour. Control
functions independent of the valve gate controls are used to determine the converter
transformer taps.

5.1.5.1 Udi Regulation

The function Udi = const has proved to be a particularly suitable regulating process for
HVDC back-to-back links both on the rectifier and the inverter sides. The result is that the
converter valves are always acted on by the valve rated voltage UvN . The system always is
loaded with the highest permissible dc voltage Ud, and thus the system carries the least
possible direct current Id for a specified transmission power. This results in minimization
of losses. With a partial load, dc voltage exceeds the nominal value UdN, but in a back-to-
back link, this has no negative consequences. Fig. 5.5 shows the operational characteristics
of the Etzenricht back-to-back link, an installation using Udi regulation.

- 342 -
Figure 5.5: Operational characteristics of the Etzenricht back-to-back link

Below 220 MW, the electronic reactive power regulation increases the control angle and
thus reduces the dc voltage Ud.

This regulation function Udi = const cannot be used in long-distance transmission systems,
even if electronic dc voltage regulation by the inverter prevents exceeding the rated
voltage of the HVDC line. At partial load, the rectifier would inevitably have a large
control angle a and a correspondingly large reactive power demand. This may even be
considered an advantage since the reactive power surplus, often present with light load in
the ac network, is compensated. However, higher harmonic currents occur as a result of
the increased control angle. These currents require the connection of more filter circuit
units, resulting in higher losses. In addition there are higher valve losses as a result of the
large a.

- 343 -
5.1.5.2 Control Angle Regulation

For the reasons given above, it is more advantageous to operate with a constant control
angle, a or g, for long distance transmissions. It is possible to allocate the regulation
functions a = const and g = const to the transformer tap switches, if the inevitable
tolerance of approx. ± 2o el, corresponding to roughly a voltage change of 2 steps will be
accepted. Greater deviations, due to the slow tap operation may occur dynamically. Since
the major variables of HVDC, direct current and dc voltage, are precisely and quickly
regulated electronically via the converter itself, the deviations in control angles described
are completely acceptable. Of course, the inverter must be prevented from falling below
the minimal extinction angle by the g limiting regulation described previously.

5.1.5.3 DC Voltage Regulation

When the inverter is operated with electronic extinction angle control, as was described in
Section 5.1.2, the dc voltage regulation must be transferred to the transformer tap switches
of the inverter. The regulating tolerance as well as the slow mode of operation must be
accepted. However, this regulation principle has proved itself in many HVDC systems.

- 344 -
5.2 Master Control Functions

While the converter regulation functions assure the coordination of rectifiers and inverters,
there is a group of master control functions which determine the actual transmission tasks
of an HVDC system as well as the interfaces between the HVDC system and the ac
networks.

5.2.1 Power Regulation

The primary task of most of the HVDC systems built to date has been the transmission of
a specific power, independent of the variables of the associated ac networks at a given
time. The use of power regulation seems obvious. However, direct power regulation is
seldom used. Power control is used in which the set point of the transmission power is
divided by the measured actual value of the dc voltage and used as the set point to the
current regulator. Accurate voltage measurement and current regulation is required. The
outstanding control dynamics of current regulation are used advantageously to determine
the dc transmission power.

Since both, current regulation and voltage measurement are pole-based, the power control
of a bipolar system is also pole-based. Power compensation, by which an eventual power
limitation of one pole is compensated by increased transmission power of the other pole,
must be specially designed.

Changes in power transmission must be carried out slowly, in consideration of the


connected ac networks. The set point processing of the power regulation or control must
consequently include a ramping function with an adjustable rate of change, usually
between 1 and 100 MW/s. This ramp must not be effective for the power compensation
between the poles referred to above. The goal is to achieve a shift between the two poles
which does not disturb the ac network.

- 345 -
5.2.2 Frequency Regulation

When an HVDC system links two asynchronous ac networks, then, and only then, can it
be used to control the frequency in one of the networks, usually the weaker of the two. The
transmission power can be controlled according to a frequency-power characteristic, as
shown in Fig. 5.6. The HVDC system then participates exactly like the power plants of the
network in the regulation of the frequency, whereby the value DP usually equals only a
few percent of the nominal power.

Figure 5.6:
D Frequency-power characteristic of an
HVDC system used for frequency
regulation

However, if the HVDC is the only power supply of an island or the predominant one, it
may assume the function of frequency regulation alone. The output of the frequency
regulator is then the set point of power regulation or control.

It is necessary to remember that transmission power is a system variable, i.e., always


occurs with the same size and same shape in both networks coupled by the HVDC system.
Consequently, compatibility with the requirements and limits of the respective networks
must be verified.

- 346 -
5.2.3 Reactive Power Regulation

While active power is the same on both sides, the reactive power flow between an HVDC
substation and an associated ac network can differ and substantially determines the
interface behaviour. This reactive power flow may be controlled independently on both
sides of an HVDC two-point system, or at each substation of an HVDC multiterminal
system. This independence also applies whether or not the ac systems of the substations
are synchronously coupled to each other.

Two correcting variables are available for influencing the flow of reactive power: (1)
generation of reactive power in switchable filter circuits and capacitor banks (more rarely,
by means of continuously controllable static compensators or synchronous condensors)
and (2) reactive power consumption of the converter, which can be influenced by changing
the nominal control angle. The switching operations are in most cases slow
noncontinuously operating adjustments. The accuracy of the adjustments is limited by the
size of the switchable reactive power units or the stepped progression of the transformer
tap switch. Regulating speed is 10 to 20 seconds per step.

Figure 5.7: Reactive power regulation of the Etzenricht back-to-back link

- 347 -
Fig. 5.7 serves as an example of the regulated reactive power flow at the interfaces
between the HVDC system and the ac network. It shows the conditions in the Hradec
substation which is connected by a 160 km long 380 kV line with the Etzenricht HVDC
back-to-back link. It is obvious that the reactive power over the entire power range up to
110% of the nominal power falls within the specified tolerance band.

In the middle and upper range of the transmission power, two switchable reactive power
units serve as correcting variables: 1) at approx. 550 MW, a 100 Mvar shunt reactor is
switched on or off; 2) at 420 MW a 120 Mvar filter circuit is switched on or off.

In the power range below 200 MW, electronic reactive power regulation determines the
firing angle and thus determines the dc voltage. This type of regulation which acts on both
sides of an HVDC system can only be used with back-to-back links because of the strong
reduction of the dc voltage.

- 348 -
5.2.4 AC Voltage Regulation

Occasionally, an HVDC substation is required to regulate the voltage on its ac busbar.


Since the voltage in one node of an ac network is determined essentially by the reactive
power balance, the above-described reactive power regulation can be used to regulate the
ac bus voltage on the line side of the converter transformer. The limitations relative to
accuracy and dynamics mentioned for reactive power regulation also apply to ac voltage
regulation.

If dynamic high-quality ac voltage regulation which acts in response to transient events


and contributes to voltage stability is required, it is essential to employ the principle of
electronic reactive power regulation. It is a prerequisite that the network of the opposing
substation tolerate the reactive power fluctuations caused by this regulation and both
networks tolerate the likewise unavoidable fluctuations in active power.

The only other solution is the use in the HVDC substation of static compensators or
synchronous condensers to which the function of voltage regulation is then allocated.

5.3 Dynamic Regulation Functions

The high regulating speed of converter control variables (dc voltage and direct current)
permits the use of the HVDC system for dynamic regulation functions. This can clearly
improve the interactions of the HVDC system and the ac network.

5.3.1 Damping of Electromechanical Oscillations

If an HVDC system is incorporated into a synchronous ac system, it can be used for


damping of electromechanical oscillations between network sections onto which the two
converter substations are connected, or which appear in ac overhead lines running parallel
to the HVDC system. By modulation of the transmission power -- most only a few percent
of the nominal power of the HVDC system -- effective damping of power system
oscillations can be achieved. However, it is a prerequisite that the modulation power be
correctly defined in advance according to

• frequency
• phase relationship
• amplitude

- 349 -
Critical to such damping control is the derivation of the correct modulation signal from the
network parameters measured. The modulation of the transmission power itself is not a
problem in the frequency range of interest from 0.1 to 2.0 Hz, even if telecommunications
between the two substations must be employed.

Also when the HVDC connects two asynchronous networks, its use for damping of
electromechanical oscillations in one of the networks is conceivable, providing, the other
network tolerates the power modulation. In this case, the modulation power enters into the
generator/load balance. The derivation of the modulation signal required for damping is
particularly problematic and critical here.

In HVDC multiterminal systems, the damping of electromechanical oscillations in the


associated system can be achieved by suitable modulation of the power of one substation.
For this, both local control of the power set point and central control through the set-point
computer of the system are conceivable. In the case of local control, the substation
regulating the voltage must accept the modulation power. In the case of central control, the
modulation power can be distributed on all substations through telecommunication.

5.3.2 Damping of Subsynchronous Oscillations

Subsynchronous oscillations have been observed in ac systems when the output of a very
remote steam power plant must be transported via series-compensated overhead lines
whose natural frequency falls in the range of a critical torsional frequency of the power
generation unit. In some cases, serious damage including shaft failure has occurred.

A similar effect, although without any damage, has been experienced with a steam power
station in the North of the USA, from which power is predominantly transported to the use
site via an HVDC system. Control loop amplification of the current regulation had a value
>1 in the range of the lowest mechanical natural frequency of the power generation unit.
Vibrations of this frequency were amplified by the HVDC system.

In this case, the situation was remedied by introducing a notch filter into the current
regulation loop of the rectifier, so that in the range of the subsynchronous resonant
frequency -- 7 Hz in this case -- the loop amplification was reduced to <1. This effect may
be called passive damping. It is noncritical, entails no expense, and affects regulation
dynamics only negligibly.

- 350 -
In more recent specifications, the requirement for active damping of subsynchronous
oscillations occasionally appears. This means oscillation damping by well-conceived
power modulation. This requirement is problematical. The regulating speed of the
converter is approached in this frequency range. Nevertheless a sinusoidal power
modulation needs to be implemented and frequency, phase relationship and amplitude
must be controlled very accurately. This works only with local set-point control.
Activation by use of telecommunications is out of the question.

Modulation of the current reference of the rectifier current regulator is possible. A


prerequisite is that a large enough current margin DI has been selected. It may be that, as a
guideline value, the current margin must be approximately double the maximum
anticipated modulation amplitude.

With the inverter, use of the electronic voltage regulation would be conceivable. In this
case, a prerequisite is that both, the voltage margin DU (see Fig. 5.2) and the extinction
angle margin (g - gmin) are adequately large.

5.3.3 Suppression of Low-Order Harmonics

The suppression of low-order harmonics, e.g., the second harmonic, by the use of HVDC
transmission technology would seem even more problematic than the active damping of
subsynchronous oscillations. With a very low short-circuit ratio at the connection point of
the HVDC station, e.g., SCR = 2.5, parallel resonance between network inductance and
filter circuit capacitance often develops in this frequency range. Excited by transient
events in the network, in particular by temporary overvoltages which result in transformer
saturation, resonant voltages can occur on the ac side at n = 2. These oscillations
correspond to harmonics on the direct current side at m = 1, i.e., network frequency. This
in turn can result in a dc component (n = 0) in the alternating currents of the converter
bridges, leading to further transformer saturation, thus closing the active loop.

The proven remedy for the phenomena described is the installation of appropriately tuned
filter circuits, so-called "low-order filters", which require significant expense. It is also
possible to attempt to remedy the situation by using HVDC control technology. This was
completely successful with the Blackwater back-to-back link. However, it was necessary
to bypass the regulators and use the trigger equipment directly in order to convert a
damping signal of twice the network frequency into appropriate control angle patterns.

- 351 -
5.4 Regulation Hierarchy

Fig. 5.8 shows the major components of an HVDC substation and their allocation to

• converter group
• substation pole
• HVDC substation

Figure 5.8:
Functional organization of a
converter substation

According to this organization, it is


possible to arrange the various
regulation functions according to
the actuator elements used:

• converter group regulation


• pole regulation
• substation regulation

Regulation functions which use the major variables of the HVDC system -- direct current
or dc voltage -- as control variables can also be assigned to a higher system level. Their
hardware implementation is, however, always assigned to the control devices used in the
station. For example, current regulation of an HVDC station with two converter groups
per pole connected in series is hierarchically assigned to the pole level (upper half of Fig.
5.8) but to the group level in an HVDC station with 2 parallel converter groups per pole
(lower half of Fig. 5.8).

- 352 -
5.4.1 Converter Group Regulation

5.4.1.1 Trigger Device

The regulation variable of converter regulation processes, such as current regulation and
electronic voltage regulation, is the control angle a. The function of the trigger device is to
convert the control angle into the 12 trigger pulses for a 12-pulse converter group. For this,
synchronization to the network voltage and a reference to the phase relationship of the
associated valve voltage are required. The trigger pulses, however, should not follow every
rapid phase change of the bus voltage with no delay. An appropriate transitional function
is required. Even the reference to the phase relationship of the associated valve voltage
requires restriction. To prevent noncharacteristic current harmonics, an equal spacing of
30 + 0.1° el of the 12 trigger pulses of a converter is required, even with an asymmetric ac
system. The admissible deviation of 0.1° el from the equal spacing is required for the
entire chain of action including the converter valve itself; consequently, for the trigger
device itself, a maximum deviation of about 0.02° el from the equal spacing, which
corresponds roughly to 1 ms, is required. To meet these extremely high and sometimes
contradictory requirements, a radically new invention, the "phase-locked oscillator", a
resonant circuit tuned to 12 times the network frequency, synchronized to the network
voltage via a phase regulator with adjustable transition behaviour, is employed..

This resonant circuit is used in the formation of a saw-tooth voltage supplied by an


integrator. The saw-tooth voltage is compared to a trigger level for each valve to generate
the actual trigger pulses as shown in Fig. 5.9.

- 353 -
Figure 5.9: Trigger pulse generation

Additional requirements for trigger devices are (1) adjustable limitation of the control
angle so that amin < a < amax is always maintained, and (2) the emission of special trigger
pulse patterns so that a valve auxiliary path is established or released again in the shortest
possible time.

With modern digital trigger devices, in which the functions described are implemented by
software, the trigger pulse equidistance required for steady-state operation can be
transiently replaced by equiangular control. In the event of asymmetric network errors, this
has the advantage that the theoretically possible dc voltage is always formed, and the
effects of the fault on the HVDC itself and on the ac system of the opposite side are
minimized.

5.4.1.2 Extinction Angle Regulation

The extinction angle is actually a variable of each individual valve. In the case of a
symmetric ac system and with identical transformer stray reactances, both, with respect to
the three phases as well as to the wye and delta groups (conditions which are adequately
met in the steady-state operation of an HVDC system), the extinction angle can be
considered a common variable of a 12-pulse converter group. Accordingly, extinction
angle regulation is allocated to the group level, i.e., each converter group has an extinction
angle regulator of its own.

- 354 -
The actual value is taken from the valve arm which has the smallest extinction angle in
steady-state operation. This provides a margin of safety in the event of any asymmetries
which might exist.

The extinction angle is measured by detection of the valve current zero crossing (due to
the carrier storage effect of the thyristors, the valve current transiently swings from
positive to negative) and the zero crossing of the valve voltage from reverse into the
forward direction.

This way of actual value detection means a dead time of one cycle. In order to counter an
inverter commutation failure in case of transient events, the regulator is immediately
switched to the extinction angle of that valve arm which shows the smallest value.

5.4.1.3 Current Regulation

If in a substation pole, two (or more) converter groups are connected in series, direct
current is a common variable for them. Current regulation is on the level of pole
regulation. On the other hand, when two (or more) groups are connected in parallel, each
group must perform its own current regulation, which is then a component of the group
regulation.

5.4.1.4 DC Voltage Regulation

Observations similar to those for current regulation also apply to electronic dc voltage
regulation. DC voltage regulation normally belongs to the level of pole regulation but is a
component of group regulation in the case that two or more groups per substation pole are
connected in parallel (e.g., in an HVDC multiterminal system), however, since the dc
voltage is the common variable of all converter groups of a substation pole, only one of
them can regulate the voltage.

- 355 -
5.4.1.5 The Regulation Functions of the Tap Changer

The function of control angle regulation assigned to the transformer tap changer is
designed as part of group regulation. In this way the independence of the group is achieved
even with regard to special tap changer movements upon connecting and disconnecting of
the group.

It is different when the function of dc voltage regulation is allocated to the tap changer,
then it belongs to the pole level.

5.4.2 Pole Regulation

When a substation pole includes only one converter group--and that is the case in most of
the newer HVDC systems--the hierarchical level of group regulation is eliminated. The
regulation functions mentioned in the previous section belong, together with the functions
described in the following section to the level of pole regulation.

5.4.2.1 Current Regulation

Each substation pole is equipped with current regulation. In most systems, in normal
operation current regulation is performed by the rectifier and determines the transmission
behaviour of the system. The current regulation of the inverter substation is used only with
transient drops in dc voltage caused by faults.

The set point of the current regulation is usually supplied by a master control function or
the power control. In the event of a fault, the current set point is adapted through VDCOL
or CEC (see Section 5.1.4 and Fig. 5.3).

The output of the current regulation is fed to the trigger device of the associated converter
group(s), which determines the control angle a.

In principle, the current regulations of the two poles of a bipolar HVDC system are
independent of each other. However, additional functions can be provided, such as to
increase the current set point, in the event of a fault on one pole. Of course, this is limited
by the temporary overload capacity of the system. In the event of a fault on one line pole,
another protective function may set the current of the good pole to zero after the expiration
of the time permitted for earth return.

- 356 -
Because of the importance of current regulation for the operation and protection of the
HVDC system, this regulating circuit must be implemented with full redundancy including
the formulation of set points and detection and processing of measured values. In the event
of any kind of failure, a monitoring and switching device handles the switchover to the
redundant circuit.

5.4.2.2 DC Voltage Regulation

Apart from the special case dealt with in Section 5.4.1.4, the dc voltage regulation is
assigned to the pole level. In most cases, regulation of the transmission dc voltage is
assigned to the inverter. Either the transformer tap changer is used as the control unit or--
with newer systems--the converter valves are used as control units in electronic regulation.

To ensure that the rated dc voltage of the HVDC line is not exceeded, and that the
admissible line voltage is fully used to minimize the transmission direct current, keeping
losses as small as possible, the dc voltage on the terminals of the rectifier must be kept
constant. When the inverter regulates the dc voltage, the set point must be adjusted so that
the voltage drop along the line is compensated (Ud*WR = UdN - Id.RL.).

To compensate for the inaccuracies due to the temperature and wind dependencies of the
line resistance RL, transmission of the voltage measurement value of the rectifier
substation via telecommunications for calibration of the set point processing is
recommended.

There are some additional functions which are executed by dc voltage regulation and
which the set point processing must take into account:

• operation with reduced dc voltage (prevention of contamination flashovers of insulators


with unfavourable weather).
• cable relief regulation (prevention of cavitation in the dielectric of solid cables with
power reduction)
• execution of the voltage ramps predefined by control sequences (shut-down and start-
up of the system, changing the direction of energy)

- 357 -
5.4.2.3 Power Control

When an HVDC system includes no power regulation but utilizes power control, this must
be allocated at the pole level. The set point of the transmission power is fed in equal parts
to the two pole control devices. The transmission power divided by the measured dc
voltage of the pole is used as a set point of the current regulation.
The allocation of the power set points can be changed on both poles by a "permissible load
computer" belonging to the substation level, if, for example, in a bipolar system with
serially connected converter groups, one group must be disconnected.

5.4.3 Substation Regulation

5.4.3.1 Power Regulation

If an HVDC is to be run with power regulation, the regulation belongs to the substation
level regulator hierarchy. For this reason, the power regulator should always be located in
the particular substation in which the current regulator is active in normal operation, i.e.,
normally in the rectifier substation. This applies even if the power request comes from the
inverter side. Set points may be transmitted by telecommunications, control variables must
not.

When the system is designed to transmit power in both directions, both substations should
be equipped with a power regulator whereas the regulator of the rectifier substation takes
over the regulation of the transmission power.

5.4.3.2 Frequency Regulation

As was presented in Section 5.2.2, the HVDC can assume the frequency regulation of a
small ac network, as for example, on an island, or it can assist in the maintenance of the
frequency of a larger ac system.

Although the active power is the variable which determines the frequency, it is possible to
do completely without power regulation in the first case above. The frequency regulator
acts directly on the current regulator, benefitting regulation dynamics.

In the second instance (assisting in frequency regulation of a relatively large ac system),


the specified frequency-power characteristic acts on the set point formulation for power
regulation or control.

- 358 -
5.4.3.3 Reactive Power Regulation

In general, all main components of an HVDC substation (converters, filter circuits,


capacitor banks) are connected to a common ac busbar. The reactive power flow to the
interface between the HVDC and the ac system is a common variable of the substation,
therefore there is only one reactive power regulator and it is allocated to the substation
level.

Only in exceptional cases, mostly with very high substation rating, the components of an
HVDC substation may be divided into two unconnected ac busbar sections. The outgoing
ac lines are coupled at a relatively great distance, after passing through several network
nodes. In this case, it makes sense to have two reactive power regulation circuits. In this
situation, the regulators can be allocated to the pole level.

5.4.3.4 AC Voltage Regulation

The regulation of the busbar voltage by controlling the set point of the reactive power
regulator is also done on the substation level -- with the exception of the above-described
special case.

5.4.3.5 Dynamic Regulation Functions

The dynamic regulation functions dealt with in Section 5.3 are, without exception,
allocated to the ac system of one side of a 2-point HVDC system. The associated
regulation circuits are a component of the substation regulation. Reference is made also
here to the exceptional case dealt with in Section 5.4.3.3.

5.4.3.6 Current Compensation Regulation

Current compensation regulation assumes special significance when there are very strict
requirements for the avoidance of current flow through earth. By measuring the pole
current difference or the ground current itself, the regulator detects even slight deviations
from the zero set point. The current regulators of the two poles are supplied with an
additional set point of the same size but with a different sign to eliminate the ground
current. Only one current compensation regulator is used per substation and it has no
effect on other variables common to the substation.

- 359 -
5.5 Control Functions

In addition to the regulation functions discussed in Sections 5.1 through 5.4, HVDC
systems have many control functions for various processes and parameters. These
functions are essential for the performance of the HVDC system. They are especially
important for the transition from one operating condition into another.

5.5.1 Start-up and Shut-down of the System

With the exception of back-to-back links, the great distance between converter stations
presents a number of coordination problems in the start-up and shut-down of the system.
In addition, the conductor-to-ground capacitance of HVDC lines and the requirements of
ac systems require carefully designed, well coordinated procedures. (Conductor-to-ground
capacitance is especially important in relatively long submarine cable systems.) An
automatic control sequence is required for system start-up and shut-down. The sequence
consists of a series of steps with testing and monitoring procedures between the steps.
These steps include the following:

• Establish ready status in both stations by activation of power supply and all other
auxiliary systems; activate all control, regulating, and protective systems and run test
routines; position the transformer tap changers at the lowest position.

• Connect the converter transformers in sequence; check the valves (thyristor


electronics); enable the valve control system of the inverter station (limit position ßmin);
send "ready status" message.

• Enable valve control system of the rectifier station and bring current to the established
minimum direct current by "fast ramp" with direct voltage at the nominal value and
simultaneously connect the minimum filter circuit combination; position the
transformer tap changers into operating position; report commencement of operation.
(In bipolar systems, the ramping processes of the two poles can be separated from each
other by a few seconds.)

• Bring the system up to the transmission power selected or ordered by the controlling
regulator while complying with the specified ramps; operate in accordance with
activated regulation functions.

- 360 -
In many existing HVDC systems, the connection of the minimum filter circuit
combination takes place prior to the start-up process. One disadvantage to this is that a
higher temporary overvoltage occurs on the ac busbar due to excess reactive power
because no reactive power is consumed by the converter before the start-up.

In critical network relationships, a control sequence may be selected which stops the fast
ramping of direct voltage at 70% to 80% of nominal voltage. Operation is initiated with
reduced direct voltage to increase the reactive power demand of the converter. Transition
to nominal voltage then takes place as the transmission power is being ramped-up. Such a
control sequence requires optimization and testing in a simulator.

Shut-down takes place in reverse order. Power is reduced at the specified rate of change to
the minimum direct current and to the minimum filter combination, and the system is
placed in "ready status" or total deactivation, as required by the control sequence.

5.5.2 Connection and Disconnection of a Converter Group

If a station pole consists of two or more series-connected converter groups and an


individual group must be connected or disconnected during power transmission, there
must be a by-pass switch parallel to the direct current terminals of each group.

If a group is to be disconnected, the first step is the reduction of system power at the
specified rate of change to a value which can be transmitted without the group which is to
be disconnected. In some cases, the capacity for continuous overload may be utilized.
Following this, the control system brings the direct voltage of the group to zero by a fast
ramp. At the same time, system current is increased by corresponding amounts so that
power remains constant. Immediately afterwards, a by-pass is initiated in the group to be
disconnected. This is shown schematically in Fig. 5.10. Now the by-pass switch can be
closed and the group can be disconnected on the ac side.

- 361 -
Figure 5.10:
Valve by-pass and by-pass switch of a 12-pulse
group

It is important that the same control and switching


actions be taken in both stations of a two-point
system. If a group is disconnected in one station, it
is not acceptable to make the adjustment for the
reduced direct voltage by increasing the control
angle in the other station. This would not be
permitted neither by the reactive power balance nor
by the valve stresses. To this extent, this control
sequence encompasses both stations. The
individual steps require coordination of timing
through telecommunications.

When the group is to be connected again, the sequence can be run in reverse. The by-pass
switch must have a large commutation capacity in order to commutate the direct current of
the system into the valve by-pass chain. The by-pass switch must overcome both, the
threshold voltage of the thyristors and the inductivity of the unsaturated valve reactors.

Another process starts with the enabling of valve control at a = 90° while the by-pass
switch is still closed. The control angle is moved so far into the rectifier range that a direct
current flows in the short-circuit loop which corresponds to the system current which is to
be commutated. When the contact is opened, the by-pass switch does not have to
commutate any direct current at all. Instead, it must interrupt a harmonic current, primarily
the 12th harmonic. A detailed discussion of this, in particular concerning switch stress, is
presented in Section 2.7.1.

A completely different set of problems requiring fundamentally different process control is


posed by the connection or disconnection of one of several parallel-connected converter
groups of one station pole. For the rectifier station, this does not present a problem. The
group to be disconnected is brought into zero-voltage operation (a = 90°) and is then
disconnected in zero-current state. The group to be connected is enabled at a = 90° and
then is slowly brought up to the rectifier range until it has assumed its share of station
current.

- 362 -
In contrast, connection and disconnection of a group in the inverter station is quite
difficult. One possibility is to reduce the voltage in the remaining groups of the station
pole temporarily to such a degree that there is no current in the group to be disconnected
and it can be blocked. The reverse procedure is followed in the process of connection.
Inverter voltage is transiently reduced and the group to be connected is unblocked in this
condition at ßmin. Then all groups are brought into the steady-state operating condition.

5.5.3 Change of Energy Direction

A change of energy direction in a two-point HVDC system generally means the reversal of
the polarity of the direct voltage while the direct current retains its direction. This is true
for both, long-distance transmission systems and back-to-back links. To make the change
of energy direction for the ac network compatible for both sides, a special process control
system is required which essentially influences the set values of the current and voltage
regulators.

Fig. 5.11 shows the sequence of events for change of energy direction in a two-point
HVDC system with rate of change of transmission power held constant.

Figure 5.11:
Change of energy direction in a two-point
HVDC with constant dP/dt

The desired dP/dt value is generally so small that reactive power and control angle
regulation is able to follow with the appropriate switching processes of reactive power
units and transformer tap changers. As Fig. 5.11 demonstrates, the direct current is first
reduced linearly to a value between 0.2 p.u. and 0.4 p.u. It is then held constant while the
direct voltage at constant dU/dt reverses polarity. After achieving the nominal voltage with
reversed polarity, the current is then brought up with the same di/dt to the required value.

- 363 -
In addition to scheduled energy reversal, sometimes an emergency energy reversal is
specified. In this case, there is no time for switching processes of reactive power elements
and control angle regulation. The process depicted in Fig. 5.12 can then be used to make
the stress on the ac networks compatible in this kind of situation. Id and Ud are controlled
in such a manner that, at constant rate of change of active power, reactive power remains
constant. An energy reversal from nominal power of one energy direction to the nominal
power of the other is depicted. The values aN = 15 ° and dxN = 0.06 were assumed as
nominal values. In a procedure of this kind, the rate of change of power can be set almost
at will. Maintenance of voltage at the ac busbars is ensured by holding the reactive power
demand constant.

Figure 5.12:
Emergency energy reversal in a
two-point HVDC system with
QHVDC = const

The processes described for energy reversal are applicable only in HVDC two-point
systems. In multiterminal systems, dc voltage remains constant. Therefore the energy
reversal of one converter station can be accomplished only by change of current direction.
Since double converters (with valves for both current directions) are ruled out for cost
reasons, polarity reversing switches have to be provided and a zero-current pause must be
inserted. This is shown schematically in Fig. 5.13.

- 364 -
Figure 5.13:
Energy reversal in a station of an
HVDC multiterminal system

The length of the zero-current pause is determined by the polarity reversing switches used.
When poles of modern ac breakers are used as rapid isolators, the pause only needs to be a
few hundred milliseconds. Such a pause is of little significance for normal energy reversal
processes. An emergency energy reversal as depicted in Fig. 5.12, however, is not
possible.

5.5.4 Modification of System Configuration

In order to continue operation during a failure of a subsystem of an HVDC such as a


failure of one pole of a line or station when ground return is not permitted, a change in
system configuration is necessary on the dc side. Switching actions are required which
must be coordinated between the stations. By further configuration modifications the
transmission losses in monopolar emergency operation can be reduced and, if long-term
conduction of current through the ground is permitted, the loss in transmission capacity
can be almost completely avoided. To accomplish this, complex process control steps are
necessary. Three examples of system configuration changes are shown below.

5.5.4.1 Metallic Return

Fig. 5.14 shows the transition into the metallic return operating mode, which allows
continuation of transmission at half power upon the loss of a station pole when earth
return is not permitted. In both stations, the GRTS (Ground Return Transfer Switches) are
closed and then the MRTB (Metallic Return Transfer Breaker) is opened and commutates
the system direct current into the line of the disturbed pole. This transmission line now
assumes the function of a neutral conductor.

- 365 -
This operating mode can also be used during service or repair work on a ground electrode
when connection of the converter neutral point to the HVDC station ground mat is not
permitted or desired and a reduction in power is considered acceptable.

The return into the bipolar operating mode takes place in reverse order. First MRTB
(Metallic Return Transfer Breaker) is closed. Then the residual current flowing in the
metallic return path (spill current) is commutated through the GRTS into the valves which
are ready for by-pass operation. During this process, both of the procedures described in
Section 5.5.2 are possibilities. Transition into normal operation follows.

Figure 5.14:
Transition into the metallic return
operating mode.

5.5.4.2 Line Paralleling

If continuous earth return is permitted, the metallic return mode is not necessary. This may
even be desirable because metallic return operation is accompanied by higher transmission
losses than operation with earth return. Transmission losses can be reduced still further if
the two transmission line poles are connected in parallel, as shown in Fig. 5.15. This kind
of operating mode is also possible in bipolar cable transmission systems.

- 366 -
Figure 5.15:
Transition into the "line-paralleling"
operating mode

Switching over into this operating mode does not present problems. This is particularly
true if poles of ac breakers are used as switching devices. Switching back into bipolar
operation requires cancellation of the line-parallelling configuration. If this switching
process occurs during transmission, the switching device must possess a significant
commutation capacity. This requires the use of HVDC breakers. The other possibility is a
brief current-zero of the system to enable disconnector opening.
This mode of operation does not require modifications of the regulating system. Only the
line protection system must be modified.

5.5.4.3 Station Pole Paralleling

If continuous operation with earth return is permitted, it is possible to switch over to


"station pole paralleling" operation upon loss of a line pole. Monopolar operation with
doubled nominal current allows transmission of almost the full capacity (see Fig. 5.16).
An overhead line designed for economic current density allows continuous operation with
doubled current. The earth electrodes and electrode lines must be sized for this kind of
operation. In addition, at least one pole in the HVDC stations must be equipped with
polarity reversing switches. Normal disconnectors are adequate for this purpose since the
switch-over is always performed under zero-current condition.

- 367 -
Figure 5.16:
Transition into "station pole paralleling"

For the start-up and shut-down of the now parallel-connected converter groups, the same
applies as was stated in Section 5.5.2: It is no problem in the rectifier station, but is
possible in the inverter station only with a transient reduction of voltage.

- 368 -
5.5.5 Tap Changer Control

In an HVDC station, the transformer tap changer is primarily activated in regulation


functions such as

• Firing angle regulation

• Direct voltage regulation

• Udi = constant regulation

A few control functions also need to be provided, as described in the paragraphs below.

5.5.5.1 Synchronization Control

To avoid non-characteristic harmonics, it is important for all tap changers of a 12-pulse


converter group to always be on the same tap. Even if measurement or determination of
angle in control-angle regulation is independent in the two 6 pulse bridge circuits, the
command for shifting the tap changer must come from one point and must be directed to
all transformers in the 12-pulse group.

5.5.5.2 Switch Delay

To avoid excessive switching activity in tap changers, a delay should be inserted. The tap
change command is not forwarded until the signal has persisted for a certain period of time
such as 10 seconds. After a completed switching operation, an opposing switch command
(i.e. one which would cancel the command which had just been executed) is transmitted
only after a certain waiting period. In this way, tap changer "pumping" is avoided.

5.5.5.3 Pre-selected Settings

In cases of a relatively weak network connection (SCR £ 3), it is useful to bring the tap
changer to the minimum tap before making the connection to the ac system. This will
avoid a switching rush and keep transformer saturation as low as possible. This is
important in avoiding the generation of low frequency harmonics, primarily the second
harmonic.

- 369 -
The same placement can be used for operation with reduced direct voltage. To disconnect
one of several parallel-connected converter groups, however, the tap changer must be
moved to the extreme opposite position (a limit is set by the rated voltage of the valves). It
is very important that movements of the tap changers of a 12-pulse group be synchronized,
but the switch delay should be disabled to avoid unnecessary delay.

5.5.6 Switching of Reactive Power Units

In general, reactive power units (filter circuits, capacitor batteries, and sometimes shunt
reactors) are switched automatically as a part of reactive power regulation or control. In
the latter case (reactive power control), the units are activated at discrete values of
transmission power and are deactivated again at a transmission power which deviates by a
"switching hysterisis" value.

In addition, there are systems which have no reactive power regulation or control system at
all. In these systems, reactive power units are switched by hand. In every case, however,
special control and latching functions are needed to prevent unacceptable operating
conditions and processes.

5.5.6.1 Switching Sequence

The relationship of harmonic currents of differing ordinal numbers is a function of


transmission power. For this reason, it is often necessary to determine a specific sequence
in which the filter circuits of different designs or tuning are connected as the transmission
current rises so that compliance with the specified distortion limits is assured throughout
the entire power range. The minimum filter combination necessary at minimum load must
be determined. Disconnection of the minimum filter combination during transmission
operation must be avoided, even when reactive power units are switched manually.

5.5.6.2 Switch Delay

As in tap changer control, a switch delay must be provided when reactive power regulation
or control is used. This switch delay is designed to prevent excessive switching
(“pumping”).

- 370 -
5.5.6.3 Switch Priority

When the HVDC station has several reactive power units of the same kind at its disposal,
their use should be controlled so that operating periods and switching frequency are
equalized. For this reason, a switch priority may be determined by a special control
function.

5.5.6.4 Controllable Sources of Reactive Power

If continuously controllable sources of reactive power in an HVDC station are provided


they are usually utilized for regulation or control of quantities of the ac network
independent of the HVDC. These reactive power sources can also be incorporated into
reactive power regulation or control of the HVDC station. The controllable elements
intercept the reactive power jump which occurs upon the switching of a filter circuit F or a
capacitor bank C (jump function). The controllable elements then pass through their
operating range on a continuous basis until the next switching of a reactive power element
becomes necessary. Fig. 5.17 shows this process for the employment of a symmetrically
controllable static compensator (SVC). The curve would also be quite similar in the case
of a synchronous condenser.

Figure 5.17:
Combined reactive power generation
S

- 371 -
5.6 HVDC Protection Functions

5.6.1 Principles of Protection

An HVDC system is made up of very different components: two converter stations,


usually located at a great distance from each other; a connecting line consisting of an
overhead line or cable or both, and two ground electrodes and the associated electrode
lines. In a back-to-back link, most of the components listed do not exist and the two
converters are combined within one station. But even this is a very complex system. The
protection functions must take into account the complex nature of the system. Careful
coordination of the numerous necessary protection functions is required. It is prudent to
follow the customary philosophy in ac network technology of overlapping zones of
protection. This makes it possible to achieve coverage of the entire system without gaps
and to also realize selectivity so that adequate protection functions can be activated for any
given event.

Figure 5.18: Protection functions of an HVDC back-to-back link

Fig. 5.18 shows the protection zones of an HVDC back-to-back link. The overlapping of
the individual zones can be clearly seen in the figure. Within each protection zone, an
entire series of protection functions is provided. In addition, there are also higher-order
protection functions which affect several protection zones. These higher-order protection
functions ensure an optimal response by the HVDC system to a fault event.

- 372 -
Figure 5.19: Protection zones of an HVDC long-distance transmission system

Fig. 5.19 shows the protection zones of a long-distance transmission system. The system
illustrated is a monopolar system. In a bipolar transmission system, the number of
protection zones would be doubled to achieve a high degree of independence of the two
poles. Only the exit of the electrode line is common to the two poles.

Protection of the numerous devices, components, and subsystems of an HVDC system


requires an extensive, well coordinated protection system. The converters offer many
possibilities for fast and reliable protection functions. Use of such functions helps in many
fault events to avoid a shut-off and also achieves a very fast, "soft" restoration of
transmission operation.

While equipment protection and protection of ac busbars and lines are usually realized
today through the use of digital protective relays, it is possible for the dc-side protection to
be combined with the converter regulation and control functions. For most of the
protection functions, parameters are utilized which are also needed for the regulation
functions. This provides a continuous self-monitoring of the transducers and processors of
measured values. In general, dc-side protection benefits from the high requirements for
reliability which are placed on all components of the regulation functions including the
philosophy of redundancy.

- 373 -
There are two requirements for use of the (n-1) principle in the protection system: (1) the
full extend of system protection be ensured in the event of outage of any given component
and (2) an unjustified shutdown of the system or a part of it must not result from such an
outage. It is easy to achieve the first requirement by arranging redundant circuits or by
employing back-up protections. The second requirement entails far greater problems. In
some places this may necessitate the installation of 3 parallel protective loops with a "2-
out-of-3" selection routine.

Of even greater importance are the possibilities for monitoring all essential protection
functions and modification of parameters such as tripping values, time delays, reset values,
etc. If possible, monitoring and modification of parameters should be enabled during
operation of the system.

Among the most important tasks at the beginning of project planning for process control
of an HVDC system is a protection coordination study. With the help of a computer
simulation of the entire system, the setting parameters of the various protection functions
are mutually coordinated and optimized. These settings serve as initial values in the start-
up of the system.

5.6.2 Protection Functions of Current Regulation

Current regulation determines the steady-state and dynamic transmission performance of


an HVDC system. In addition, it assumes essential protection functions by limiting the
load current of the most essential system components — valves, converter transformers,
smoothing reactors, and HVDC lines — to acceptable values. Permissible overload
currents and their duration may sometimes be calculated for the individual components of
the system and incorporated into the processing of current-limitation target values taking
into account the ambient conditions. In this way, target values can reflect the transient
overload capacity which depends upon ambient conditions, operating status, and bias. A
permissible load calculator of this type is very complex. Considerable costs and time are
required for proper definition and sizing of the individual functions.

- 374 -
Additional protection functions of current regulation are:

• Avoidance of drops below minimum dc current to prevent current gaps (non-


continuous current) particularly during dynamic regulation processes. The gap limit is
dependent upon many parameters such as the control angles.

• Reduction of dc current at disturbance-related drops in system dc voltage by means of


the VDCOL function (see Fig. 5.3); avoidance of valve overloads in an inverter which
is not able to commutate; facilitation of resumption of commutation; and avoidance of
excessive drain of reactive power from the ac networks, etc.

• Interruption of current in the event of a line-to-ground fault on the HVDC overhead


line. In this way the extinction of arcing is facilitated (see also Section 5.6.4, travelling-
wave protection).

5.6.3 Protection Functions of g and Ud Regulation

Whether the converter-related (electronic) control of an inverter is performed as extinction


angle regulation or as dc voltage regulation, it can assume two important protection
functions.

• Prevention of the exceeding of rated dc voltage as a result of temporary overvoltages in


the ac network on the inverter side. In Ud regulation, this protection function is
inherent. In g regulation, the superimposed protection function temporarily enlarges g
until the tap changers have reached their required positions.

• Prevention of inverter commutation failures during dynamic events such as lowering of


ac network voltage or elevating of system dc current through preventative actions. In
the case of g regulation, g * is temporarily increased. In electronic Ud regulation, Ud* is
temporarily reduced. This also leads to an increase of g at an even more rapid rate.

- 375 -
Two additional protection functions are provided by Ud regulation, regardless of whether it
makes use of converter valves or of tap changers:

• Operation with reduced dc voltage to prevent flash-overs of contaminated line


insulators during unfavourable weather conditions such as fog or drizzle.

• Cable protection by reducing dc voltage when transmission power is reduced. The


accordingly increased current helps prevent the formation of hollow spaces in the
dielectric of paper-insulated, solid-type cables as a result of pressure reduction, which
can result in corona and damage to insulation.

5.6.4 Travelling-Wave Protection of Transmission Lines

In HVDC systems using overhead transmission lines, line-to-ground faults can be


expected relatively frequently. Generally, the faults are transient. The clearing of such
faults, however, requires a brief interruption of power transmission in the affected system
pole and the insertion of a zero-current pause of 100 to 200 ms. During this time, the
arcing channel should be sufficiently deionized (restabilized).

A transmission line ground fault must be detected as quickly and as reliably as possible.
The line fault must be distinguished from voltage dips resulting from ac network faults or
from inverter commutation failures. Travelling-wave protection of transmission lines
serves this purpose. It detects travelling waves which emanate from arcing points and enter
the rectifier station. Thus voltage dips, the rise of current, and the rate of change of both
are detected and measured. This provides a reliable criterion for identifying low
impedance line faults which can occur during contamination arcing of a line insulator or
during a direct arcing between a line and ground resulting from fire beneath the
transmission line.

The affected rectifier station pole reacts with a rapid forced retard of the delay angle,
bringing dc current to zero within about 30 ms. After the zero-current pause, the rectifier
brings up system dc voltage using the specified ramp.

The inverter need not be involved in this process if it is reliably prevented from shifting its
control pulses into the rectifier range (a < 90°). If its control pulses were shifted into the
rectifier range, current regulation of the inverter would maintain the flow of current across
the fault site and would prevent the zero-current pause. Consequently the arc would not be

- 376 -
extinguished. Transmission line protection in the inverter station, however, can assume the
back-up function. In this event, telecommunications are inserted into the protection loop.

If arcing reoccurs as the system voltage is being brought up, line protection must also
detect this fault reliably, even if the arcing event occurs at a relatively low voltage.
Generally, a longer zero-current pause (for example, 400 ms) is inserted before the second
re-start attempt.

Transmission line protection must generally allow an adjustable number of re-start


attempts each with a longer pause period than the preceding one. The option of operating
with reduced dc voltage after the second or third attempt should also be provided.

5.6.5 Differential Protection

This very fast and reliable method of protection is frequently employed in the ac
technology but can also be applied in the converter area and on the dc side of HVDC
systems. The system is divided into overlapping protective zones so that no area is without
protection. At the same time, a high degree of selectivity can be achieved so that in any
fault event, the adequate protective measure can be activated and unnecessary shut-downs
can be avoided.

5.6.5.1 Bridge Differential Protection

A comparison of valve alternating currents and system direct current facilitates recognition
of short circuits within the 6-pulse bridge, the clearing of which require switching off the
converter group with the ac breaker. The short-circuit current usually flows through one or
several valves. Due to the limited heat capacity of thyristors, the converter group must be
switched off as quickly as possible.

Fig. 5.20 depicts the measuring principle of bridge differential protection.

- 377 -
D( ) D( )

DC-
Filter

Figure 5.20: Measuring principle of bridge differential protection


In back-to-back links, it is also possible to utilize the rectified valve currents of the
opposite side for this purpose rather than the directly measured dc current of the link.

Triggering criteria for the short-circuit protection function are as follows:

I~ > I=; DI > 3 · IdN; tv = 0

where: I~ = rectified valve alternating current of a bridge, normally equivalent Id


I= = system direct current Id measured by dc transformer (transducer)
DI = difference between incoming and outgoing current of a bridge as
evaluated in differential protection
and
tv = trigger delay

5.6.5.2 Group Differential Protection

As is shown in Fig. 5.21, the maximum value of the rectified valve ac currents of the two
bridges is evaluated as the incoming current of a 12-phase converter group. System direct
current measured by dc transformers is deemed to be output current.

- 378 -
D

MAX

DC-
Filter

Figure 5.21: Measuring principle of group differential protection

The special function of group differential protection is to recognize faults in the converter
area which do not cause short-circuit current and thus are not detected by bridge
differential protection. Such faults include short circuits to ground in an inverter group.

- 379 -
Therefore triggering is delayed and is often provided with two response levels. The
following are examples of triggering criteria:

I~ ¹ I=; DI > 0.4 IdN; tv = 200 ms


DI > 0.2 IdN; tv = 1000 ms

5.6.5.3 Station Pole Differential Protection

Usually the direct current of a station pole is measured both by a dc transformer in the
valve hall bushing and by a free-standing dc transformer in the HVDC transmission line
leaving the station, as depicted in Fig. 5.22.

Figure 5.22: Station pole differential protection

Short circuits to ground on the dc side can be reliably detected by differential protection
inserted between the two dc current transformers. Equipment faults on the dc side of the
station (faults of the smoothing reactor, dc filter circuits and dc switch devices) are also
covered by this protection, to the extent they result in a short circuit between the station
pole and ground or station neutral bus. To prevent false triggering by discharge currents of
the dc filter circuits during rapid changes of dc voltage, a response delay must be provided.

Triggering criteria (example):

Ids ¹ IdL; DI > 0.4×UdN; tv = 200 ms

where Ids = direct current measured on the converter side of the station pole dc bus
IdL = system direct current measured on the line side of the station pole dc bus

- 380 -
5.6.5.4 Transmission Line Differential Protection

Longitudinal differential protection is an appropriate means of detecting high impedance


line-to-ground faults in an HVDC overhead line which cannot be detected by travelling-
wave protection. It is necessary, however, to utilize telecommunications in the protection
loop. In this special case this is considered acceptable because a high impedance fault does
not present any acute danger and relatively long reaction delays can be accepted in the
clearing of such faults.

Triggering criteria (example):

Idrect.>Idinv.; DI = 0.2IdN; tv=1000 ms

5.6.5.5 Protection of Electrode Lines

A short circuit to ground on the electrode line does not cause any appreciable current or
voltage jumps and therefore cannot be detected by travelling-wave protection. There are
two types of differential protection which may be applied:

• Longitudinal differential protection. This very reliable protection function requires a


measuring station with power supply at the earth electrode site and telecommunications
with the HVDC station. Thus it is relatively expensive.

• Parallel differential protection. If the electrode line is divided into two separate systems
(configured as a homopolar line), the currents of the two systems can be measured and
compared. In the event of a short circuit to ground in one of the systems, a current
difference occurs due to the parallel-connected ground path. This current difference can
be utilized for detecting the short circuit to ground.

In both cases, the differential protection will function only in the presence of flow of
current through the electrode line. In bipolar systems, current does not flow through the
electrode line during normal operation. In order to be certain there is no short circuit to
ground in the electrode line, a test method may be employed in which a transient
unsymmetry is intentionally introduced into the two system poles from time to time. One
means of introducing unsymmetry is intervention in the current balancing regulator. The
pulse-like current through the electrode line created in this method makes it possible to
monitor for short circuits to ground by longitudinal or parallel differential protection.

- 381 -
Recently, measurement of impedance using a high-frequency current fed into the line has
been suggested for monitoring the electrode line. This testing method is independent on
the operation of the HVDC.

5.6.5.6 Differential Protection of Equipment

The equipment on the ac side of an HVDC station also must be adequately protected by
the use of differential protection. This is especially true with transformers, but also applies
to filter circuits, capacitor banks, shunt reactors, etc.

5.6.6 Overcurrent Protection

Even when current regulation prevents an unpermissible overcurrent in most components


in an HVDC station, overcurrent protection will nevertheless be provided as a back-up
protection. This also applies to components whose load current is independent from the
regulated system direct current. An example of such a component is an ac filter circuit. In
this instance, the overcurrent protection serves as a back-up for the filter differential
protection.

Fig. 5.23 depicts an additional example of overcurrent protection for a converter group.
Maximum rectified valve ac currents of the two bridges are evaluated. Overcurrent
protection is of multiple level design and is scaled according to fault intensity. The
response values and trigger times are adapted to the load capacity of the valves.

Figure 5.23: Measurement principle of overcurrent protection

- 382 -
Examples of protection triggering of the individual levels of a four level overcurrent
protection system:

Level 1: 1.2 x IdN / 20 s

Level 2: 1.35 x IdN / 10 s

Level 3: 1.6 x IdN / 100 ms

Level 4: 2.5 x IdN / 50 ms

Overcurrent protection initiates disconnection by triggering of the ac breaker.

5.6.7 Other Special HVDC Protection Functions

Disturbances of converter functions do not always result in operating conditions which can
be detected and identified by overcurrent or differential protection systems so that
countermeasures may be taken. Therefore, a series of HVDC-specific protection functions
have been developed.

5.6.7.1 Detection of Misfires

The periodic loss of the triggering pulse of an individual valve in a rectifier station results
in a network-frequency component in the direct voltage of the affected rectifier group, but
operation continues. In addition to increased stresses on valves, this operating condition
results in disturbances in the vicinity of the HVDC overhead line. In the case of an
existing network-frequency resonance, dangerous overvoltages can occur. This is
especially true in the case of a submarine cable transmission. Therefore misfires must be
detected. A convenient method of accomplishing this involves filtering out network-
frequency components from the measured direct voltage. It is necessary to ensure by
appropriate filtration that transient or dynamic events do not result in false trips.

In an inverter, misfires always result in commutation failures which require much quicker
reaction. For this reason, a different method of fault detection is needed.

- 383 -
5.6.7.2 Detection of Commutation Failures

Inverter commutation failures are unavoidable. Depending upon the extinction angle at a
given time, a voltage dip at the inverter station ac busbar of 10% or 15% will result in a
commutation failure. If the commutation failure persists, it will result in a collapse and
may even cause a transient reversal of polarity of the dc voltage of the affected bridge or
group. It is important to detect an inverter malfunction quickly and to initiate
countermeasures which will cause a resumption of commutation before a short-circuit
current can grow up. This is especially true for relatively slight voltage reductions caused
by a remote short circuit in the inverter-side ac network. A relatively slight voltage
reduction will permit inverter operation if the necessary extinction angle will be
maintained.

Bridge differential protection as shown in Fig. 5.24 represents one possibility for rapid
detection of an individual commutation failure. In this event a transient valve by-pass is
formed so that the system direct current no longer flows through the transformer phases
and the differential protection D can detect the current difference.

Figure 5.24: Detection of an individual commutation failure using the bridge differential
protection
K = Commutation failure
N = By-pass operation

- 384 -
Recently, preventive commutation failure protection has been advocated. This involves the
detection of rapid changes of operation parameters of an inverter which may cause a
commutation failure. In the event of a rapid increase of direct current or reduction of ac
busbar voltage, the angle of advance ß is increased spontaneously. Despite increased
commutation overlap (resulting from increase in current and/or voltage drop), a
sufficiently large extinction angle g can be achieved.
The increase in ß can be produced by intervention in converter regulation — extinction
angle regulation or direct voltage regulation — or it can be brought about directly in the
trigger device. Even if the first commutation following the disturbance does not succeed,
the commutation conditions for the following valve can be improved to such an extent that
commutation will succeed and the disturbance can be intercepted (Fig. 5.24).

5.6.7.3 Protection of Filter Circuits

The protection of ac filter circuits or capacitor banks against short circuits and line to
ground faults can be easily provided by overcurrent and differential protection. In addition,
the detection of gross unsymmetries by phase-comparison protection or by measuring the
flow of unsymmetric current to ground may be considered conventional methods of
protection.

Fig. 5.25 depicts one phase of a double-tuned highpass filter circuit with surge arresters for
overvoltage protection and current transformers T1 through T43 which are necessary for
comprehensive protection. A configuration is depicted in which the primary capacitor C1
is subdivided into four partial banks connected in an H-circuit. The current transformer T2
in the transverse branch can detect even very small unsymmetries.

Figure 5.25:
Protection system in
ac filter circuits

- 385 -
Detection of the loss of an individual capacitor unit or the loss of an individual coil within
a capacitor unit with interior fuses is an HVDC-specific task. In sharply tuned series
resonance circuits, even slight deviations from the desired capacity cause significant
impairment of filtration.

In addition, the loss of an element in one row of the capacitor matrix means an increase in
capacitive resistance of the row and assumption of a higher proportion of total voltage.
The remaining elements of the row are overloaded with respect to voltage and current. The
smaller the number of parallel-connected elements, the greater the degree of overload.
This must be considered when the filter connection voltage is high (³ 400 kV) and the
filter capacity is relatively small so that the capacitor matrix consists of many rows, each
with only a few parallel elements.

If ac connection voltage is not too high and the filter circuit has a relatively high
fundamental frequency capacity, capacitor units (cans) with external fuses, which will
deactivate a complete unit from the matrix of the bank in the event of defects, can be
economically feasible. H-circuit configuration and current transformer T2 are then not
necessary because the changes in the event of a can fault are large enough and can be
detected with unsymmetry detection in “model filter" monitoring.

This monitoring system requires the voltage of the ac busbar and the total filter current as
measured values. The current flowing through the ac filter circuit is calculated from the ac
voltage. This calculated current has precisely the same magnitude in zero-fault condition
of the capacitor bank as the measured current. In the event of a unit fault, an increased
amount of total current flows through the ac filter due to the declining capacitance. This
increased current will flow until the unit fuse has isolated the defective element. Busbar
voltage, in contrast, remains constant. Thus there is a difference between calculated and
measured current which can be evaluated. If the difference exceeds a certain value, a unit
fault can be assumed.

Following the melting of the external fuse, a stationary current is established which is
slightly smaller than the current in zero-fault condition of the capacitor bank.

- 386 -
The monitoring system usually has three levels:

1) Defect of one unit Alarm

2) Defect of a second unit Alarm and shutdown of the filter circuit after approx.
two hours

3) Defect of a third unit Shutdown of the filter circuit without time delay

Recently, foil capacitors with internal coil fuses have become prevalent for use in ac filter
circuits. In the event of a defect, only the damaged coil is switched off. Total capacitance
is only slightly changed and the voltage overload of the remaining coils and units in the
same row is slight, even in the case of high connection voltage of the filter circuit and
relatively small fundamental frequency capacity.

Figure 5.26:
H-circuit configuration of the capacitor
bank of a filter circuit (one phase)

The H-circuit configuration depicted in


Fig. 5.26 has proven effective for the
detection of coil losses. The loss of even a
single coil will cause a measurable current
in the transverse branch.

Customarily, three monitoring steps are defined:

• In the first step, the loss of a number of coils is reported which causes a noticeable but
still tolerable impairment of filtration effect and/or a significant but still permissible
voltage overload.

• In the second step, an alarm is triggered when the filter effect is impaired to such an
extent that the specified tolerance of voltage distortion is no longer met in all operating
conditions and/or the voltage stress of individual elements can reach levels which are
only temporarily acceptable.

- 387 -
• In the third step, protective shutdown is triggered.

Filter circuit protection is not perfect. For reasons of safety, it must be assumed that all
coil defects occur in one row of the matrix. If the defects are distributed across several
rows of the same block, as would be expected statistically, the protective steps would be
activated earlier than necessary. On the other hand, an equal number of defects in each of
the two parallel blocks or in the blocks arranged in series on one side would not be
detected. On the whole, however, this configuration of filter protection has proved useful
and reliable.

5.6.7.4 Arrester Protection

The metal oxide surge arresters generally used today in HVDC systems are not immune
against aging. This can result in a gradual increase in arrester residual current in steady-
state operation, eventually resulting in thermal instability and the destruction of the
arrester. This danger is particularly great for the arresters in a converter group if unusual
operating conditions such as operation with high control angles results in temporarily
elevated arrester voltages. Therefore it is important to monitor arrester residual current.
But in the case of valve arresters, this is very difficult since there is no fixed relationship to
ground potential. If the valve arresters are incorporated into the mechanical structure of
valve towers, as in the SIEMENS design, the monitoring of the arresters can be combined
with the valve electronics and integrated in the valve monitoring.

Figure 5.27:
Surge arresters in a 12-pulse converter
group

- 388 -
5.6.8 Protection of Equipment

Proven protection systems should be provided for all conventional equipment in an HVDC
station. These protection systems should be provided with redundancy or with back-up
protection.

If the full utilization of temporary overload capacity available at any given moment is
specified, thermal models of the individual devices or pieces of equipment must be
provided or must be incorporated by software into a central set value calculator.

Special attention must be given to fire protection. If converter transformers are to be


placed directly adjacent to the valve hall and their bushings penetrate into the hall, it is
especially important that adequate heat damping and an effective fire extinguishing system
be provided. The valve-side transformer bushings should be oil free, i.e., they should be
designed as dry bushings. The necessity of having reliable fire alarm and fire fighting
systems in the valve hall itself has been dramatically demonstrated several times in recent
years.

- 389 -
5.7 Measurement

5.7.1 Requirements on the Measuring System

Regulation and protection functions of a system can be only as reliable as the system
quantities with which they work. Therefore special consideration should be given to the
measuring system.

Based on technical or economic considerations, certain quantities may be derived or


calculated from other quantities rather than being measured directly.

Measured quantities needed for the regulation, control, and protection functions must be
available on a redundant basis. This can be accomplished by having duplicate transducers
or by processing different, independently measured quantities which contain the same
information. Thus bridge dc current is derived in Fig. 5.19 from the rectified valve-side ac
currents.

Measuring systems, like regulation systems, require continuous monitoring for equipment
failures, self-tests for system faults, and automatic switchover to a redundant system. The
switchover can be independent in the regulation and measurement systems or, if there is a
one-to-one correspondence, it can be performed jointly.

In addition, there are quantities which are not needed for control, regulation, and
protection, but which provide a quick grasp of the condition and operation of the system
and therefore should be measured continuously and displayed on demand. The display can
be either in the form of instruments in an overview instrument panel or as images on the
computer screen in the control room.

5.7.2 Quantities to be Measured

In project planning for an HVDC system, it is useful to prepare a list of all quantities to be
measured and to indicate the following data for each:

• Measurement range and unit


• Purpose
• Transducer (type, site of installation, precision)
• Measurement converter (type, conversion units, range, precision, installation site)
• Redundancy or back-up measurement available

- 390 -
• Monitoring of measured value (how, where)
• Display (where, how), transmission (where to, how)
• Recording (where, how)
• Accessibility for tests, diagnosis, simulation

In addition, it is helpful to summarize the measured quantities and condition reports which
characterize the operating condition of the system and which therefore should be displayed
on a continuous basis.

5.7.3 Fault Recorder

The measurement system also includes fault recorders which are indispensable in HVDC
stations. These fault recorders continuously record the important operating variables, store
them temporarily and then erase them. In the event of a disturbance, the curves shortly
before and during the disturbance are printed and are available for fault analysis.

It can be useful to install two units, each covering a different time period. One of the units
would cover a period of 0.5 to 1 second. It would be used primarily to record converter
variables at a relatively high resolution with respect to time. The second unit should cover
5 to 10 seconds and would record primarily ac network variables.

- 391 -
5.8 Telecommunications

5.8.1 Functions of the Telecommunication System

Except for back-to-back links, the converter stations of an HVDC system are located long
distances apart. These distances may be on the order of many hundreds of kilometers,
sometimes more than a thousand kilometers. Since all converters are involved in the
control functions of the HVDC system, a telecommunication system between the stations
is necessary. The telecommunication system is not necessarily a component of the HVDC
system. Transmission capacity could, for example, be leased from public tele-
communication companies. Nevertheless, an integrated system is preferable.

Exchange of information between the converter stations and the ac network control centers
or load distributors is also necessary.

In incorporating telecommunications between the converter stations into the control


system of the HVDC, signal transmission times as well as the general susceptibility of
telecommunications in the form of information errors and transient interruptions must be
taken into account. The transmission of test signals and the installation of system
redundancy with automatic switchover are adequate remedies.

5.8.1.1 Transmission of Set Values

As described in Section 5.1, all converters of an HVDC system are equipped with current
regulation. In addition they are usually equipped for dc voltage regulation, giving them the
ability to perform a change of energy direction, and more importantly, to buffer system
operations during transient events. At any given time, one station performs the regulation
of a quantity while the other is standing by. The result achieved by this means is that their
regulator set values deviate by a marginal amount (DI or DU) from the set value of the
active regulator. It is important that set value changes, whether they are issued manually or
by a higher-order regulation or control function, are always undertaken simultaneously in
all converter stations. This simultaneousness, which is limited by the properties of
telecommunications, requires special flow plans.

This can be illustrated using current regulation in a two-point HVDC system as an


example. An ordered increase in set value is immediately converted by the current
regulating rectifier and thus becomes effective. At the same time, the command for change
of the set value is transmitted to the inverter, which then carries it out with corresponding

- 392 -
delay. Thus the temporarily increased current margin is returned to the original value. A
set value reduction is first transmitted to the inverter which executes it. This increases the
current margin. At the same time, it confirms by telecommunication the execution of the
order to the rectifier. The rectifier now reduces its set value and thus brings the
transmission current to the desired value and at the same time reducing the current margin
to the original value.

In general terms, all set values transmitted by telecommunications are stored in the
receiving station so that even a brief loss of the telecommunication system does not impair
the continuation of transmission operation.

The steady-state operation of the HVDC does not place any special demands on the signal
transmission speed of the telecommunication system. The rate of change of transmission
power which is selected (with consideration given to the ac systems) allows sufficient time
for the transmission of set value changes, even if they are implemented in small steps.

5.8.1.2 Transmission of Dynamic Control Signals

If an HVDC system is to be employed for damping of electromechanical oscillations and if


a modulation power which exceeds the nominal power by more than a few percent is
required, rectifier and inverter control must be coordinated. The current set values of both
stations must be modulated. The modulation frequency on the order of 1 Hz determines
the required transmission rate of the telecommunication system. Many of the known
transmission media are ruled out for this task.

5.8.1.3 Transmission of Control Signals

The start-up and shut-down of an HVDC system, connection and disconnection of a


converter group, or modification of dc-side configuration require coordination of
processes in both stations. This requires transmission of control signals through the
telecommunication system. The execution of the commands will employ the step-by-step
principle described above. In this procedure, the station which merely provides
preparedness always makes the first step and only after receipt of the report of completion
the other station can take the second step which executes the command.

- 393 -
5.8.1.4 Transmission of Disturbance Signals

The converter stations of an HVDC system are fundamentally autonomous with respect to
their protection functions. Except for longitudinal differential protection for clearing a
high impedance line-to-ground fault in an HVDC overhead line, no telecommunications
are necessary for the triggering or execution of the protection functions of the converter
stations. Nevertheless, the fastest possible transmission of information concerning actions
taken for clearing faults, such as the disconnection of a converter group, is important. This
will permit the other station to perform corresponding actions to allow continuation of
orderly transmission of power, i.e., transmission which conforms to the specified
secondary conditions, even if the transmission is at reduced power.

The requirements for transmission of disturbance signals in a multiterminal HVDC system


are quite different. The loss of a converter group must be reported very rapidly to the
central set-value calculator which computes modified current set values for the new
system configuration and transmits them to the converter stations. This exchange must be
completed as rapidly as possible to avoid overload of individual converters or a system
crash. Therefore the transmission medium with the highest currently available
performance (transmission by light signals through fiberoptic conductors) may be
necessary.

5.8.1.5 Transmission of Status Signals

The transmission of status signals ensures the exchange of information on the actual status
of the other station. Included in this category are the following:

• Position of disconnectors, earthing switches and breakers


• Readiness for activation or operation
• Non-availability of subsystems or equipment effected by faults
• Measured values of important operating parameters
• Available overload capacity

The large number of signals and measured values (in digitized form) are usually grouped
into information blocks and transmitted cyclically in a multiplex process to the other
station where they cause stored status signals to be reset.

- 394 -
5.8.1.6 Voice and Image Communications

In most cases, one telephone and one telefax channel will be integrated into the
telecommunication system. Using these channels, technicians can inform each other of
planned or executed actions and of special circumstances. Moreover, voice and image
communications facilitate the reaching of understanding by system engineers in analyzing
disturbances and planning changes in process control structure or parameters. In addition,
printouts of disturbance recorders (to the extent such recorders are integrated into the
HVDC control system) can be automatically transmitted to the other station.

5.8.2 Operation without Telecommunications

As previously noted, continuation of power transmission of an HVDC system in the event


of a temporary loss of telecommunications between converter stations is generally
included among performance requirements. This is true even if the telecommunication
system is configured in a redundant manner or if a back-up system is available. In
conjunction with this requirement, the provision may be made that the loss of
telecommunications and the temporary nonavailability of telecommunications in the
HVDC stations be reported so that changes in control processes may be initiated.

5.8.2.1 Steady-state Operation

Since transmission of set values is not possible, operation continues with stored set values,
i.e. with constant transmission power. If transmission power is to be changed in the event
of relatively extended loss of telecommunications, automatic adjustment of the margins
may be used. If for example, an inverter station recognizes that dc current deviates by 2%
from the stored desired value for a certain minimum period of time, the set value of the
inverter current regulator is reset by 2%. This causes the current margin to return to the
original value. In this way a change in transmission power desired by the rectifier side can
be implemented. If the desire for change comes from the inverter side, the inverter can
modify the transmission dc voltage by 2%, for example. When the rectifier registers this
deviation for a certain minimum period of time, it adjusts its current set value by 2% in the
same direction. The inverter then returns dc voltage to the original value.

In this way almost unlimited operation is possible throughout the entire power range,
though with slower rates of adjustment.

- 395 -
5.8.2.2 Dynamic Operation

During the time when telecommunications are lost, it may be acceptable in most cases for
the HVDC system to be without dynamic regulation functions. If, however, these
functions are essential for stable operation of the ac systems, a different solution must be
found. If the ac system of the rectifier side needs a power modulation of ±20% for the
damping of oscillations, the inverter must increase its current margin to approximately
30% upon loss of telecommunications. The problem will be more difficult if the power
modulation is needed from the ac system of the inverter side. One method may be to
employ the electronic dc voltage regulation of the inverter for this purpose. The dc voltage
will follow the modulation signal arround a lowered mean value, while the rectifier holds
the system dc current constant, tolerating dc voltage modulation.

5.8.2.3 Control Sequences

Since transmission of control signals for the coordination of processes in the converter
stations is not possible, important control sequences, such as shut-down of the system,
transition to metallic return, etc., must be triggered by certain pre-determined special
operating conditions maintained for a defined period of time.
Other, less important control sequences may be blocked until telecommunications are
restored.

5.8.2.4 Information Loss

Since disturbance and status signals in general contain information which is not necessary
for either orderly operation or for protection of the system, it is possible to do without their
transmission for an extended period of time. The same applies to voice and image
communications.

5.8.3 Telecommunication Systems

Various telecommunication systems employing various transmission media are available


for the transmission of information between HVDC converter stations. The quantity of
information to be transmitted and the requirements of transmission speed, reliability,
availability, and cost are determinative in the selection of the system.

- 396 -
The following distinctions are made with respect to the requirements for information
transmission:

• Time-critical signals for regulation and protection


• Less time-critical signals for control, operation, and observation (status signals)
• Information which is not time-critical, such as speech, fault recorder data (fax), fault
location etc.

The selection of a system is also determined by project-specific data and local conditions;
not every desirable solution is feasible.

In existing ac networks, the utilities usually operate their own telecommunication systems
in the form of:

• Power line carrier


• Directional radio link
• Fiber-optic conductors in the shield wire of overhead lines or in buried cables

In addition to these transmission systems owned by the utility, third-party dedicated


telephone or data lines may also be leased. If an HVDC is imbedded in an ac meshed
network, such a system can be shared, provided sufficient transmission capacity of the
required quality can be made available. In all other cases, a dedicated telecommunication
system must be installed for the HVDC.

In the following sections, telecommunication systems suitable for HVDC systems are
presented with their characteristics, advantages, disadvantages, and limitations.

Figure 5.28 gives an overview of the capacity of such systems. The bit rate and the
distance which can be spanned without intermediate amplification are presented. The
values given are typical values. Thus, for example, the distance which can be spanned by
power line carrier systems depends largely on the carrier frequency, as will be
demonstrated later. In optical transmission using fiber-optic conductors, capacity is far
above the values needed by an HVDC, considering the fact that a lightning shield wire will
always contain several fiber-optic conductors.

- 397 -
Distance
km

500

2
Mbit/s
8
400 Mbit/s 34
Mbit/s
140
Medium Mbit/s

300
Fiberoptics
Radio VHF/UHF
64 kbit/s 2 8 34 140
Mbit/s Mbit/s Mbit/s Mbit/s
200
directional radio link
Telephone
64 kbit/s
100
PLC digital
64 kbit/s
PLC analog
1200 - 9600 bit/s
Bit-Rate
64 kbit/s 2 Mbit/s 140 Mbit/s

Figure 5.28: Capacity of telecommunication systems

5.8.3.1 Power Line Carrier (Carrier Frequency Transmission on High Voltage Lines)

Carrier frequencies in the range between 25 and 500 kHz coupled onto high voltage lines
are an information transmission medium which has proven effective over a period of
decades. Each system covers several channels within a fixed frequency range. Factors
which limit the number of channels include the band width of the coupling elements
which can be achieved cost effectively and the transmission properties and sensitivity to
interference which are dependent upon the carrier frequency. There can also be a lack of
carrier frequencies since frequencies used in nearby high voltage lines must be avoided to
reduce the possibility of mutual interference. Certain frequency ranges cannot be used with
respect to resonance with the lines.

The coupling elements, particularly the coupling capacitor, insulate the power line carrier
equipment against the line voltage. Wave traps at the ends of the line prevent high
frequency energy from flowing into the stations or to ground in the case of grounded lines.

It is also possible for the power line carrier system to use the lightning shield wire of a
high voltage line. This can make coupling capacitors and wave traps much more
economical. This is offset by the costs for the insulation of the lightning shield wire. In

- 398 -
monopolar HVDC lines, this represents the only possibility for the use of a power line
carrier system, since the conductor-to-ground coupling demonstrates such a high degree of
damping that only distances less than 100 km can be spanned. This solution is practical
only for short-distance overhead line sections of sea cable HVDC transmission.

Transmission quality is weather-dependent. Atmospheric disturbances (thunderstorms)


and intensified corona (for example during the formation of frost) can cause signal
disturbances.

Transmission speed is usually 2.4 kbit/s, i.e., is very limited, and only six to ten full duplex
channels can be accommodated in a system. In special cases and at distances up to approx.
150 km, transmission speeds of 9.6 kbit/s can be achieved.
The transmission of a full duplex channel allowing the simultaneous transmission of
signals in both directions requires a frequency band of 2 x 4 kHz. Such a frequency band
can be utilized for several services:

• One speech channel 0.3 — 2.0 kHz


• Two protection signals
• One 1200 bit/s data channel

Digital data transmission with power line carrier will be possible in the near future. With
the use of a frequency band of 8 kHz instead of 4 kHz per channel, up to 64 kbit/s can be
achieved.

The distance which can be spanned is highly dependent upon carrier frequency due to
frequency-dependent damping of the lines. At 300 kHz, approximately 300 km of
overhead line can be spanned; at 60 kHz, approximately 800 km can be spanned. For
longer distances, intermediate amplifiers must be employed which require coupling
elements, wave traps, and of course a reliable power supply.

Fig. 5.29 shows the single-pole equivalent circuit diagram of a power line carrier
intermediate amplifier station. A diesel plant is provided as a standby in addition to an
infeed from the local power network.

- 399 -
Figure 5.29: Power line carrier intermediate amplifier

If a local power supply network is not available, two or three diesel generators must be
provided (as in the Cabora-Bassa Project) in order to achieve the required 100%
availability.

Other methods of providing power have been suggested, but not yet utilized. As an
example, it has been suggested that the harmonics be separated from the dc line by moving
the dc filter circuits from the HVDC station into the power line carrier amplifier station.
But in this arrangement, the purpose of the dc filter circuits — avoidance of telephone
disturbances — would be lost in at the least the section of line between the HVDC and
power line carrier amplifier stations. In addition, such a method of power supply would
function only during operation of the HVDC system and would be dependent upon the
operating parameters.

If an insulated ground wire is available for power line carrier purposes, it could also be
used for the transmission of network frequency alternating current of a few kilovolts.
Protection of the associated equipment against atmospheric overvoltage, however, appears
to be especially difficult.

Depending on the specific climatic conditions, one interesting possibility for the supply of
current for intermediate amplifier stations could be the use of photovoltaic generators in
combination with maintenance-free batteries.

In cable connections, the damping of the high frequency is up to 10 times stronger than in
overhead lines. For this reason, a power line carrier can be used with HVDC sea cable
connections only at relatively short distances (< 100 km).

- 400 -
5.8.3.2 Radio Communications

If a power line carrier cannot be used, e.g. in relatively long submarine cable transmissions
or if the installation of intermediate amplifier stations is not feasible in very long overhead
line transmission systems, directional radio communications in the VHF range can be used
for distances up to 200 km. The high degree of dependency upon weather conditions and
sensitivity to interference from outside signals, however, make this type of system
relatively unreliable. Therefore a system of this type cannot be considered for the
transmission of time-critical protection and regulation signals.

Transmission quality and freedom from interference are more favourable in carrier
frequencies in UHF ranges. Line-of-sight contact between transmitter and receiver is
necessary for UHF communications. Therefore application of this technology is only
feasible in relatively short submarine cable transmission systems. The assignment of
suitable frequency bands, however, may become difficult.

5.8.3.3 Directional Radio Links

Directional radio links are characterized by high transmission speeds (64 kbit/s or higher)
and high transmission capacity. The high transmission frequencies utilized allow
communication only with line-of-sight contact between transmitter and receiver. Therefore
an HVDC system with the typical transmission distances requires the installation of a large
number of relay stations. The distance which can be spanned is around 40 km, depending
upon the specific topographical conditions.

The transmitted radio waves are sensitive to interference. The error rate can be reduced by
a factor of 10 to 100 by the use of frequency diversity (two channels with different
frequencies are used for the same message) and spatial diversity (the message on one
channel is received by two different receivers).

The cost for the implementation of a directional radio link is very high, but the cost of an
additional channel is low. Often there are free channels available in an existing directional
radio system.

- 401 -
5.8.3.4 Fiber-optic Communications

Optical transmission of data via fiber-optic conductors is characterized by high


transmission speeds (2 Mbit/s) and high transmission capacity. The information is
transmitted digitally. The fiber-optic conductor can be incorporated into the ground wire
of overhead transmission line systems. In the case of cable transmission systems, the fiber-
optic conductor can be integrated into the energy cable or can be laid parallel to it. Due to
its insensitivity to electromagnetic fields, optical transmission of data is destined for wide-
spread use in electrical energy systems.

The costs are still high. For systems spanning great distances, the primary costs are for the
fiber-optic conductor and the amplifiers, which are still necessary today every 100 to 200
km. This field is undergoing rapid development. It may be that in the foreseeable future,
distances typical for HVDC systems may be spanned without intermediate amplifiers.

For HVDC systems with high dynamic regulation requirements and for HVDC
multiterminal systems, fiber-optic conductor transmission is the solution of choice even
today.

5.8.3.5 Satellite Communications

Transmission of information via satellites is very powerful. Due to the high lease costs,
these systems are economical only when large quantities of data must be transmitted over
great terrestrial distances. The long transmission distances to geostationary satellites and
back require a transmission time of at least 240 ms. Thus time-critical requirements cannot
be fulfilled. Less critical signals may be transmitted via satellite, but the cost may be
feasible only in special cases.

5.8.3.6 Leased Data Circuits

Leased lines or direct data connections are dedicated physical transmission paths. In many
countries, they can be leased from public telephone service agencies. These
communication systems do not entail any construction costs, but there are fixed monthly
lease costs regardless of the quantity of data transmitted and the length of time used.
Normally, capacity and transmission quality can be specified for the particular application.
Technical and financial dependency are always given. The availability of the data
exchange requires special consideration. This should always be cleared with the local
authorities.

- 402 -
5.8.3.7 General Requirements for the Signal Transmission System

Time-critical analog signals, such as current set value, actual voltage and power
modulation value (if applicable), are transmitted serially in binary code in a data telegram.
Binary signals which must be transmitted quickly are transmitted either as bit information
in the same telegram or as binary signals with almost no delay. When data are time-
critical, they must be transmitted as rapidly as possible and at short time intervals. This can
be a problem with the relatively slow power line carrier system on high voltage lines.

An important parameter for regulation and protection functions is the delay in data
communications between the converter substations. This delay consists of the following:

• Data transmission time, which depends upon the data transmission rate of the channel
and the length of the data telegram. For example at 2400 bit/s and a telegram length of
96 bits, data transmission time is 96 bits/2400 bit/s = 40 ms.

• Modem or multiplexer delay, which depends upon the type of device will range from 2
to 10 ms total delay for both ends of the transmission

• Transmission medium delay, which depends upon the medium itself, the length of the
transmission, and the number of intermediate amplifiers (2 to 15 ms)

• Processing time for the processors at the sending and the receiving end (1.5 ms each)

In communication systems with low data rates such as power line carrier systems, analog
signals are transmitted serially at 2400 bit/s. Transmission of data from the control system
in one station to the control system in the other station takes a maximum of 50 ms. The
binary signals are transmitted simultaneously over the protection data transmission
facilities directly to the power line carrier device. The transmission time of the binary
signals is small (10 to 15 ms).

In addition to the few time-critical regulation and protection signals which must be
transmitted rapidly, there is a relatively large quantity of data required for the control of
the system and for remote control. No special requirements are made for these data. They
are transmitted cyclically approximately every half-second.

- 403 -
To increase the availability of information transmission, the equipment is configured
redundantly, two communication systems being provided on the same transmission
medium. In addition to equipment redundancy, the information can also be transmitted by
a back-up communication system. Thus both, medium and equipment redundancy are
provided.

5.8.3.8 Telecontrol

It is possible to operate and monitor the system remotely from a load distributor by
telecontrol interfaces. A telecontrol facility is characterized by reliable transmission of
information with predetermined content. It acquires, transmits, and delivers the data which
are necessary for operation management. All conventional communication paths can be
used for transmission of the information. The use of telecommunication systems owned by
the utility itself is particularly feasible for this purpose.

- 404 -
5.9 General Requirements for HVDC Control Systems

There are several general requirements for all components and subsystems of control
technology of an HVDC system. These requirements must be met in order to ensure
reliable, state-of-the-art operation of an HVDC system.

5.9.1 Functionality

Control technology must ensure that the specified requirements for operating performance
as outlined below are met and that the system performs as a state-of-the-art HVDC system.

• Steady-state range: compliance with the specified operating parameters within the
specified tolerance range and conformity with guaranteed limits for various interactions
and disturbances emanating from the HVDC system

• Dynamic range: compliance with maximum and minimum rates of change for
operating parameters, change of energy direction, effective damping of
electromechanical or subsynchronous oscillations in one of the ac systems, etc.

• Transient range: clearing of faults and restoration of transmission operation within the
specified period of time.

These requirements apply insofar as network and environmental conditions are within the
specified operating or normal range. With respect to network conditions, this pertains
primarily to voltage and frequency but also to network short-circuit power at the
connection point of the HVDC station. With respect to environmental conditions, it
pertains to temperature and also to humidity, wind, etc., if relevant.

In most projects, an expanded operating range or emergency range was specified for
network and environmental conditions. In these ranges, power transmission must be
continued and reliable protection of the system and its components must be ensured, even
if not all specified data are achieved and a few limit values for interactions are exceeded.
If network or environmental conditions fall outside of the extended ranges, the control
system must ensure a reliable and soft shut-down of the HVDC system.

- 405 -
5.9.2 Flexibility

A high degree of flexibility is required of the control system of an HVDC system. This
means that all control, regulation, and protection functions can be adapted as required even
after commissioning of the system. It should be possible to modify these functions or
replace them with other functions. Preferably, it should be possible to carry out parameter
changes and, as far as possible, structural changes for control functions during operation of
the system without endangering the transmission process or the protection of the HVDC
system.

Hardware and software of the control system must be designed so that an electronic
engineer of average experience can make the interventions needed. Appropriate
programming aids and diagnostics must be provided.

5.9.3 Reliability

The most important criterion for evaluating the operating performance of an HVDC
system is its reliability. Reliability means above all availability (= the ability of the system
to transmit a certain quantity of energy in a defined period of time in relation to the rated
transmission capacity). This is referred to as energy availability. It is most important that
the control system of an HVDC system not contribute in any significant degree to non-
availability. If one wishes to quantify this requirement, the contribution of the control
system to non-availability of an HVDC system pole should be limited to < 0.1% that is to
< 8 hours of downtime per year. The control system should not be permitted to cause any
non-availability of the bipole at all. From this is derived the requirement for complete
independence of the control facilities for the two poles.

The requirement for limitation of the number of forced outages is significantly more
difficult to fulfill. In recent times, this parameter has usually been specified at 10 to a
maximum of 20 forced outages per system pole per year. Of these forced outages, no more
than one or two may be attributed to the HVDC control system.
Not only system disturbances or outages caused by the control equipment must be
considered but also unjustified shut-downs of the system caused by malfunctions of the
control system. This places high demands on the reliability of the protection systems.

- 406 -
5.9.4 Redundancy

The requirements for reliability which are placed on HVDC control systems can only be
fulfilled if a fully effective redundancy is provided for all control facilities including the
measuring systems and telecommunications. If that is not possible or is not economically
feasible, an equivalent back-up function must be provided for every function.

The requirements are still higher for converter and dc-side protection systems which are
integrated into the HVDC control system because there is no continuous self-testing as is
the case with the regulation loops. Whether a protection system is functioning properly
does not become apparent until there is a fault event. For this reason, three-fold
redundancy topped with “two-out-of-three voting” has been suggested and implemented.
This solution has the clear advantage that an unfounded shut-down signal given out by a
defective protection circuit does not become effective. Thus a forced shut-down is
avoided. Therefore, it is also possible to test all protection functions of any one of the
redundant systems during operation without risking a shut-down.

5.9.5 Automatic Switch-over

The installation of redundant systems makes sense only if a fault is recognized quickly and
reliably and results in automatic switch-over to the stand-by system. Means of monitoring
and self-testing must be provided for both, hardware and software in the form of

• Monitoring systems for devices and/or components for the detection of hardware
failures

• Test routines which run in cycles for the detection of software failures

• Computational operations as plausibility tests for the detection of system failures

Failures should be recognized in < 1 ms so that the switch-over can take place smoothly,
without detectable or, at a minimum, without harmful reactions on the ac system. This of
course assumes an on-going tracking of control and regulation circuits by the stand-by
system.

Each instance of a switch-over means a loss of redundancy and must be reported by an


alarm in the control room (in remotely operated stations, in the remote control room). A

- 407 -
detailed display, which will facilitate the identification of the defective module or the
module to be changed, should be provided in the electronics room of the station affected.

Following the replacement of a module, its successful repair must be confirmed through a
test-run by means of a specialized testing program. Tests of function are performed with
manual blocking of the function output by a test switch. Afterwards, the system is released
for ready status. Manual return switch-over to the original system by the electronics
engineer is advisable as an additional testing step.

5.9.6 Telecommunications

The telecommunication system must also be configured in a redundant manner or a back-


up system must be available. Facilities must be provided that an outage of transmission of
information will be detected immediately. A distinction should be made between an
outage of the telecommunication system and loss of the message.

Loss of a message is usually characterized by the fact that information is temporarily


distorted or destroyed. Such a temporary disturbance can occur as a result of
electromagnetic fields emanating from high-voltage systems or as a result of atmospheric
disturbances. A disturbance of the transmission of signals is detected by signal testing such
as comparison of level, frequency, or phase position at the receiving end. A disturbance of
the transmission of data is also recognized at the receiving end by transverse or
longitudinal parity in character-oriented protocols. Upon detection of error, the new value
is discarded and the last valid value is retained.

In an outage of the telecommunication system, there is usually no information transfer.


This condition can be recognized at the sending end. The receiver will be able to recognize
the outage with certainty if, for example, no information is received after an adjustable
monitoring time. An additional criterion for the detection of an outage of the
telecommunication system is the reception of contaminated data on a regular basis. If
several telegrams are recognized in sequence as being flawed, a disturbance of the
communication system will be reported.

Upon recognition of the outage of the telecommunication system, an automatic transfer to


the second transmission channel of the redundant transmission system will be made. In
this way, a fault in data-end equipment or in data transmission facilities will be eliminated.
If the medium-dependent transmission path is down, the fault cannot be eliminated by

- 408 -
switching over to a redundant system. In this case, a switch-over to the reserve system
(back-up system), if available, will be made.

All signals and measured values transmitted by telecommunications must always be saved
so that in the event of interruption of telecommunications, the operation of the HVDC
system will continue unchanged for a time.

In the event of a relatively long total outage of the telecommunication system, various
switch-overs must be made in control facilities, in particular in the generation and
processing of reference values. At the same time, the load dispatchers of the networks
should be informed that the HVDC system is available only to a limited extent for
dynamic operations.

5.9.7 Power Supply and Auxiliary Facilities

The principle of redundancy should be used for power supply as well as for auxiliary
facilities of the HVDC control system. If possible, two independent feeds for power
supply should be available. At a minimum, two uninterruptable power supply systems
should be available along with appropriate automatic switch-over devices.

These requirements also apply to the relay stations of the telecommunication system. If no
network power is available, a diesel station must be constructed with two or more
generator sets which should be operated in alternating cycles.

If air conditioning of the electronic rooms is necessary for the operation of control devices,
redundancy must also be provided with respect to the air conditioning system.

The fire alarm system must be absolutely reliable. Whether the fire extinguishing or
smothering system should be configured with redundancy is an open question. A
significant increase in security can be achieved by housing redundant control systems in
separate electronic rooms with fire retarding walls. This may, however, entail
technological complications and significantly increased costs.

- 409 -
5.10 User Interface

5.10.1 Operating System, Operating Location

The operator of an HVDC system should be able to operate and observe the system at any
time. In addition to conventional facilities for operation and observation such as operating
panels or switch boards, it is becoming increasingly common today that powerful
computers such as workstations with graphical user interface be utilized. In addition,
display of alarms and of the diagnosis systems are being incorporated into the overall
interface concept.

There are various places where service and monitoring functions can take place. Usually a
user interface inside the control room of the HVDC station (local operation) and at a
remote operating station, such as in the load dispatcher station, are installed. The HVDC
system can be operated from these sites using varying degrees of automation. In HVDC
long-distance transmissions, the individual converter stations as a rule are operated
independently, either from the local operating site or from the remote site. The functions
which determine the transmission of power (such as start of power transmission, changes
in set values and regulation modes) can be performed from only one site. This will be
either from the rectifier or from the inverter side. This higher level of operation for both
stations should be available from both the local operating site and from the remote site.
The operating site selected (local operation, remote operation; rectifier side, inverter side)
is then assigned the appropriate operating priority. It should, however, be possible to
observe the entire system at any time from either operating location.

5.10.2 Operating and Monitoring Functions

Customarily, an order for an appropriate transmission of energy is given by the load


dispatcher. The order specifies the essential settings. These are:

• Energy direction (e.g. west to east)


• Power level (e.g. 1000 MW)
• Configuration (e.g. bipolar)
• Ramp slope (e.g. 30 MW/s)

After these specified settings have been entered into the operating system, sequence
control carries out the orders independently in the correct sequence. This concerns
primarily the following functions:

- 410 -
• Setting the dc side connections (configuration)
• Checking of auxiliary operations (power supplies, cooling system)
• Energization of converter transformers
• Checking of the thyristor electronics
• Coordination of both stations prior to release of the trigger impulses to the valves

During the course of the sequence, processing information is displayed on the screen.
Essential processing information includes positions of switches, actual values of voltage,
current, power, reactive power, transformer step settings, and control angle. In addition to
this information (which is associated with energy transmission itself), the computer screen
monitoring system provides a large amount of additional information from the converter
station. This may include status information on the cooling system (valve cooling,
transformer cooling) or on the power supply systems. Computer screen images are
provided for the various kinds of additional information in the graphical user interface
which can be accessed and displayed. Normally, fully automatic control is in operation.
The fully automatic control mode provides the highest degree of automation and is the
preferred mode for normal operation. The operator can initiate changes in operating
condition, whereupon all necessary actions are performed automatically and in the proper
sequence with consideration to all necessary latching procedures by the control and
regulation systems. During the start-up phase or during servicing, it is possible to switch
over to manual operation. In this operation condition, the individual components such as
individual breakers can be operated directly from the operating system. Fig. 5.30 shows as
an example a system overview of the Etzenricht back-to-back link.

- 411 -
VALVE AUTOMATION SHUTDOWN NO-LOAD AVAILABLE AVAILABLE FOR TRANSMISSION
HALL TRANSMISSION
GROUNDER VALVE COOLING ON CONVERTER BRANCH P = 648 MW
VENTILATION ON
JA = + 20°C zul
REMOTE TRENNERWEG ON
TRANSF. COOLING ON ETZ 401 JVH = + 23°C P = -200 MW
OFF ON soll
OWN NEEDS ON ETZC 401 ON JKW = + 29°C P = -199 MW
ist
12 . 05 . 93 13 : 33 : 53

L1 = 11 L1 = 8
L2 = 11 L2 = 8
Q= + 0 Mvar Q= + 0 Mvar V
471 L3 = 11 L3 = 8 471

a = 18°
V Q= + 106 Mvar g = 23° Q=+ 106 Mvar V
401 401

V Q= -122 Mvar I = 1206 A Q= - 132 Mvar V


472 d 472
Ud = 169 kV
441
HRA
Q= + 0 Mvar V
P Soll = 200 MW 478
RAMP = 10 MW/sec

P Ist = 199 MW

ETZ SPECIFY DESIRED VALUES ETZC


P Soll 200 MW
QETZ = - 21 Mvar RAMPE 10 MW/sec Q ETZC = - 31 Mvar
UETZ = 401 kV F1 ETZ ETZ C U ETZC = 415 kV
f ETZ = 50.00 Hz F2 ETZ ETZ C f ETZC = 50.00 Hz
F12 TAKE OVER
F6 RETURN
12/05-93 13:33:40

Figure 5.30: System overview screen of the Etzenricht back-to-back link

The system overview screen is the primary tool for operation of the HVDC system. From
this screen, the operator can initiate most commands and operating sequences. The status
line in the upper portion of the system screen is subdivided into six zones in which the
most important preselections and status indications are given. Four of the zones are
allocated directly to operating conditions of the back-to-back link. (Switched off, no-load
available, available for transmission, transmission). The zone highlighted in color
indicates the present operating condition. The other two fields give information about the
switching status of valve hall earthing switches and of the preselected operating mode and
location. The Etzenricht back-to-back link is represented in the display portion of the
screen in the form of a single-line block diagram. The present status of important
measured values and switch positions are displayed in real-time.

The program is operated using function keys and input windows.

5.10.3 Signal System

In addition to the information displayed in the system image on the screen, there are event
signals which are displayed on a separate screen in an appropriate sequence. The sequence
of displays is determined by various sorting criteria. The most commonly used sorting

- 412 -
criterion is the chronological display. This binary information (for example, changes in
switch positions) is displayed on the screen and also is printed in the control room. Event
signals are subdivided into three different categories:

• Changes in condition (example: change in a disconnect switch position)


• Warnings (example: limiting temperature is exceeded)
• Alarms (example: forced shut-down by the protection system)

If warnings or alarm signals occur, they are also reported to the operator by audible signal.
The operator must acknowledge the signal in order to shut it off.

Figure 5.31 shows an example of an excerpt from the event recorder of the Etzenricht
system. All events detected are listed on the event recorder which is integrated into the
operating and monitoring system.

Each event is presented on one line containing signal number, acknowledgment, date,
time, branch designation, voltage level, description, remote/on-site operation symbol, and
coming/going index.

The operator may scroll forward and backward through the event list using the function
keys. In addition, a warning signal list can be displayed.

EVENT LIST SWITCH PRIORITY: REMOTE

12.05.93 13:42:43
Message Q Date Clock Branch Voltage Message Text F
No. Level N

197 12 . 05 . 09 : 08 : 52 : 63 GKK REG CONTROL VALUE DIFFERENCE 1/2 GOING


197 Q 12 . 05 . 09 : 09 : 21 : 43 GKK REG CONTROL VALUE DIFFERENCE 1/2 COMING
197 12 . 05 . 09 : 09 : 21 : 69 GKK REG CONTROL VALUE DIFFERENCE 1/2 GOING
642 12 . 05 . 09 : 21 : 04 : 41 401 ETZ STEP ACTUATOR RUNS TO 10
828 12 . 05 . 09 : 22 : 34 : 17 472 ETZC BREAKER ON
829 12 . 05 . 09 : 29 : 13 : 62 472 ETZC BREAKER OFF
1190 12 . 05 . 09 : 42 : 06 : 93 472 ETZC SEPARATOR PATH NORMAL ON
1914 12 . 05 . 09 : 42 : 25 : 01 472 ETZC AVAILABILITY ON F COMMAND
1409 12 . 05 . 09 : 42 : 29 : 18 472 ETZC AVAILABILITY COMING
827 12 . 05 . 09 : 43 : 01 : 25 471 ETZC BREAKER OFF
636 12 . 05 . 09 : 43 : 01 : 75 472 ETZC BREAKER ON F COMMAND
828 12 . 05 . 09 : 43 : 02 : 04 472 ETZC BREAKER ON
1408 12 . 05 . 09 : 43 : 02 : 83 471 ETZC AVAILABILITY GOING
1912 12 . 05 . 09 : 43 : 57 : 84 471 ETZC AVAILABILITY ON F COMMAND
1408 12 . 05 . 09 : 44 : 02 : 04 471 ETZC AVAILABILITY COMING
644 12 . 05 . 09 : 45 : 27 : 68 401 ETZC STEP ACTUATOR RUNS TO 7
644 12 . 05 . 10 : 15 : 21 : 64 401 ETZC STEP ACTUATOR RUNS TO 8
832 12 . 05 . 10 : 32 : 11 : 41 489 ETZC BREAKER ON
1201 12 . 05 . 10 : 32 : 15 : 09 441 ETZC BREAKER OFF
1201 12 . 05 . 11 : 08 : 35 : 16 441 ETZC BREAKER ON
1201 12 . 05 . 11 : 08 : 40 : 86 441 ETZC BREAKER OFF
642 12 . 05 . 11 : 12 : 41 : 85 401 ETZ STEP ACTUATOR RUNS TO 11
642 12 . 05 . 11 : 50 : 20 : 05 401 ETZ STEP ACTUATOR RUNS TO 10

F3 WARNMELDELISTE
F8 AKTUELLE MELD.
F9 MELDUNGSARCHIV F12 MENU

Figure 5.31: Event signals, Etzenricht back-to-back link

- 413 -
5.10.4 Signal Archive, Process Data Archive

The archive functions of the workstation systems can be used for statistical purposes such
as evaluation of switching frequency of breakers and transformer tap changers,
determination of energy transmitted and calculation of outage and availability data in
accordance with CIGRE-protocol, etc.

- 414 -
5.11 Hardware and Software Design

HVDC control systems up to the 1980's almost exclusively used analog technology.
Advancing development of computer technology made the use of digital technology
attractive for HVDC control systems. As early as 1980, a memory programmed system
control (SIMATIC S5-110A) was employed for the ACARAY back-to-back link
(Paraguay/Brazil).

Dynamic requirements are much more stringent in regulation and protection technology.
Thus, it took a few more years before powerful processors made the transition to digital
technology possible. In the meantime, several HVDC systems were outfitted by SIEMENS
with the SIMADYN D system. These systems are the Etzenricht (Germany/Chechnya),
Vienna-Southeast (Austria/Hungary) and Welsh (Texas/Eastern network in the USA). All
of these back-to-back links have a transmission capacity of 600 MW.

The Cabora Bassa HVDC long-distance transmission system (1920 MW transmission


capacity) will be returned to operation in 1997. SIEMENS is replacing the complete
control panel in the control room with a modern computer screen interface using
workstations. The previous control system, which employed relay technology
(approximately 35000 relays per station), will be replaced by a redundant digital control
system (SIMATIC S5). Both changes will result in a significant increase in system
availability.

5.11.1 HVDC Control Design


Fig. 5.32 shows the HVDC control structure of a long-distance overhead line transmission
system. The individual hardware components are described below. The control hardware
is based on the demands made of it and on the tasks which will be performed by the
hardware. There is, however, a hierarchy of HVDC control devices assigned to the system
elements, as is described in detail in Section 5.4.

All control components are configured in a redundant manner. The operation and
monitoring interface in the control room communicates with the control components
through the bus system (LAN). This bus system is based on ethernet with transmission
rates up to 10 Mbit per second. All commands from the control room or from remote
operation are transmitted serially through the LAN. Process data such as actual quantities
in the monitoring system or alarm signals are transmitted in the reverse direction.

- 415 -
GPS Load Dispatchter
Operating and Monitoring System

HVDC DSA Station Remote


HVDC+DSA Clock Interface
B&B B&B Event

LAN 1
LAN 2

E E E
HVDC+DSA E R Bridge HVDC Control + E HVDC Control + E
R R
Control+Reg. R Regulation Pole 1 R Regulation Pole 2 R

FPE FPE
B&B => Operating and Monitoring Pole 1 Pole 2
System
LAN => Local Access Network
FPE => Base Electronics
GPS => Global Positioning System
PLC => Power Line Carrier
RF => Directional Radio Link HVDC HVDC
ER => Event Register Prot. Prot.
RF
DSA => AC Switchgear
n n
Optical
Field Bus

I/O Module I/O Module I/O Module I/O Module


PLC PLC
AC Switchgear n HVDC Switchgear n n
Pole 1 Pole 2

Figure 5.32: HVDC control system of a bipolar station

- 416 -
The pole control system and the station control and regulating systems are connected to
the LAN. These systems are implemented with the multiprocessor system SIMADYN D
(see Section 5.11.3). The binary exchange of signals on the field level is carried out by
means of a redundant optical field bus system (PROFIBUS). The optical telegrams are
converted into electrical signals in the ac system or in the HVDC switchgear in order to
perform such functions as triggering of a breaker. In the same way, valve triggering
initiation is transmitted through fiber-optic conductors from the valve base electronic
system to the thyristor electronics.

Protection and regulation signals are exchanged in HVDC long-distance transmission


systems through telecommunication links between the stations. Exchange of less time-
critical system condition data is often ensured by use of directional radio links. Redundant
protection systems are connected directly in parallel at the secondary side of the current
transformers. A central station clock provides all process control components with the
same clock time over the LAN. This central clock is synchronized by the GPS system (see
Section 5.11.5). The remote control interface provides the station with commands from the
load dispatcher over the LAN and forwards station data to the load dispatcher.

5.11.2 Operating and Monitoring System

The fully graphical operating and monitoring system is implemented using modern RISC
workstations (UNIX operating system). Complete redundancy of individual functions is
ensured with integrated error monitoring and automatic switch-over in the event of errors
(master/stand-by principle). The software is based on international software standards for
GUI's (OSF/MOTIF, XII,TCL/TK).

5.11.3 Control and Regulation System, Valve Base Electronics

The control and regulation functions are implemented with the standard regulating system
SIMADYN D. In addition to the HVDC control application, this parallel computer system
is utilized primarily in drive technology.

Capacity can be expanded as needed by the addition of processor components and can thus
be adapted to the stage of development needed. Optimum scan times can be selected
through the selection of processors. Scan times can be freely selected in the range from
100 ms to the upper millisecond range. In this way, all HVDC-relevant control and
regulation requirements can be met.

- 417 -
All essential control, regulation, and protection elements and the valve base electronics are
designed in a two-channel structure. Proper functioning is monitored by hardware,
software, and plausibility controls. If there is a failure in one of these monitoring systems,
this results in a switch-over from the currently active system to the hot stand-by system
(for example from System 1 to System 2). This redundancy switch-over does not have any
effect on the ongoing process of energy transmission since all regulator signals are
continually tracked. Therefore no angle difference occurs during the switch-over process.

Fiber-optic conductors have been used for the transmission of signals between the valve
base electronics and the thyristor electronics for more than 15 years. Triggering signals are
transformed in the valve base electronics system into optical signals and are transmitted to
the thyristor electronics. In the thyristor electronics, these optical signals are converted
back into powerful electrical signals feeding the thyristor gate.

5.11.4 Signal Exchange with Switchgear, Event Detection, and Recording of Analog
Values

The exchange of signals with the ac and dc switchgear is carried out over fiber-optic
conductors in serial form (field bus). The optical telegrams are converted in the switchgear
into binary electrical information (for example switching commands). Supplemental time
stamping functions are incorporated into these transformer components which log each
event (binary condition changes) with a resolution of 1 ms. These event signals are sorted
in the control system and are transmitted to the monitoring system for display.

Actual values such as current or voltage are currently transmitted in conventional form. In
the future there will be optical solutions also in this area. Thus the entire exchange of
signals (binary information, digitized analog information) will take place in purely optical
form.

Within an HVDC station, time series of data, for example analog measured values, are
needed after fault events along with binary event signals. These process parameters are
registered with digital recording devices (Oszillostore P561). Results of measurements can
then either be printed and evaluated in the station or they can be accessed via modem by
the central station, for example by the load dispatcher. Telecommunications was discussed
in Chapter 5.8.

- 418 -
5.11.5 Fault Location (DC Transmission Line), Station Clock

In overhead line transmission systems it would be beneficial if in the event of a line-to-


ground fault the fault location is being determined. With recently developed dc line fault
locators it is possible to obtain an indication of the fault location through evaluation of the
arriving travelling waves and their exact time relationship in the two stations within ± 0.5
km. To accomplish this, an exact time base is necessary which is equally valid for both
stations. This can be achieved with the worldwide GPS system. This is a satellite
supported radio location, navigation, and time transmission system. Using GPS reception
and 3-D control (position determination by reception from at least 4 satellites), the
accuracy of time determination is ±100 ns. Detection of the travelling wave by the fault
locator causes a retention of the exact point in time. A comparison of the time points at
each end of the line makes it possible to determine the difference in travel time and thus
the localization of the fault.

This highly accurate station clock is also used for the synchronization of all process
control systems within the station (event recorder, oszillostore). This makes fast, accurate
diagnosis possible during event classification.

5.11.6 HVDC Protection Equipment

In addition to the protection of the dc zone (dc transmission lines, dc filter, convertors,
earth electrodes), converter transformer protection and ac filter protection also are
included in the HVDC protection system. Standard digital protection devices are utilized
for converter transformer protection. For the remaining protection functions, the
SIMADYN D system described above is employed. Normally, a two-channel structure is
used for these protective functions. Archiving functions are implemented in the protective
devices which permit a diagnosis of the fault and of the cause of triggering.

5.11.7 Remote Diagnostics

Remote diagnostics are available for all SIMADYN D via modem. This interface allows
to link control, regulation, and protection equipment of the system with the service center.
In the event of a disturbance, immediate diagnostic support by service experts is available.
Thus, for most repairs to be performed by local personnel, significantly reduced
downtimes are achieved.

- 419 -
5.11.8 Standard Software Module Blocks, Graphical Planning and On-line
Documentation

In the SIMADYN D system, not only is a standard hardware system available, but also a
library of standard software modules. More than 300 standard software modules are
available which cover all important control, regulation, and protection functions.

These modules may be accessed as needed and configured on the monitor (STRUC G). In
this way, the desired function structures can be generated very quickly with the desired
scanning times per block or zone and with appropriate parameter settings.

After the structure is generated on the screen, this configuration is downloaded to


EPROMS and the submodules are plugged into processor modules. Structural changes and
changes of parameters such as regulator settings can be executed later by means of project
planning devices.

Documentation of the functions implemented is generated from the EPROMS and is


converted into a drawing. These circuit diagrams contain all essential information. Along
with the structure and parameters, the scanning times of the different functions are given.
The source and destination of the signals are automatically generated.

- 420 -
5.12 Project Planning and Testing of the Control System

In recent specifications for HVDC systems, precise requirements are usually given for
steady-state, transient and dynamic behaviour (see Chapter 6). Often documentation is
demanded as early as the tendering stage to demonstrate that these requirements can be
met. Therefore even in the tendering stage, the control system is not only planned in all of
its essential aspects, but also designed and tested either by means of a real-time simulator
or with a digital simulator (NETOMAC or EMTDC). The customer obtains a feeling for
the capability of the technology being proposed and can gain assurance that the HVDC
system will fulfill its mission. In many cases, the customer will also obtain indications of
critical areas or operating conditions of his ac network.

When the contract is obtained, these initial designs of regulation, control, and protection
algorithms are refined and further checked to determine wether or not all requirements are
fulfilled. These verification procedures are the basis for actual selection of the hardware
and for software generation. Since the functions in the simulation computers are identical
with SIMADYN D function algorithms, transition to system software does not present a
problem.

The basis for project planning for software and hardware are the preliminary trials referred
to above and the customer approval of the final design. Software development and
planning for hardware including the design of electronic cabinets are performed in a
parallel manner. In order to obtain reliable proof of the dynamic properties of the control
system, the original software with all regulation, control, and dc protection functions is
checked in the real-time simulator.

The basis for setting the parameters of the protection algorithms is the protection
coordination study. This study is also performed with digital simulation tools (EMTDC) at
the beginning of a project.

When all control cabinets have been constructed, proper coordination of all components is
checked in an integration test. During this process, the original cabling is also tested. An
important check in this regard is overall communication, both within each converter
station and between the stations and for the remote service level. The check includes
communications up through the thyristor electronics. This means that the fiber-optic
connections are also checked. All major components are simulated by separate simulation
computers in real time. Thus the interface and the dynamic behaviour is checked as a

- 421 -
whole. Reactions of the control system to the outage of a power supply system or, for
example, to switch-overs to redundant devices are also integral parts of this test.

While the operator interface is completed and tested, the operating personnel of the
customer should become familiar with the operation of the system during the test period.
When all tests are passed, all control components can be dismantled and shipped to the
station site.

- 422 -
6. HVDC System Performance

In this chapter, the HVDC will be discussed as an entire system. The object will be to
examine the performance of an HVDC as an integral part of a large ac network or as the
linking element for two or more separate ac networks. The following types of system
performance will be distinguished:

• Steady-state performance
which primarily concerns the specified energy transmission but also the reactive power
balance and every effect of the HVDC on its environment, and the system losses

• Transient performance
which concerns the response of the system to faults and similar events, the limiting of
their effects, and the restoration of transmission

• Dynamic performance
which concerns defined transitions from one steady-state condition to another and the
utilization of the excellent regulating properties of the HVDC for supporting the ac
network such as damping of electromechanical oscillations

The exact definition of the interfaces between the HVDC and its surroundings and of all
external factors which impact upon the HVDC are a prerequisite for designing the
specifications for the performance of the system.

The criterion for judging the overall performance of an HVDC system is its reliability as
measured by

• Energy availability
• Number of forced outages
• Transient reliability

- 423 -
6.1 System Environment

Although it is easy to describe the interfaces between an HVDC system and its
environment, the definition of the external factors which affect the HVDC and of the local
conditions which should be considered present difficulties. Only the most significant
factors will be introduced in the following sections. For other factors, see IEC Publication
919 "Performance of High Voltage DC (HVDC) Systems."

6.1.1 The AC Networks

The interface to the ac network is the ac busbar to which an HVDC station is connected.
For project engineers to be able to plan an HVDC system, the quantities and data
discussed below must be known or must be determined for each station. This applies to
every step in the system development. Future foreseeable changes should be taken into
consideration.

Generator data of nearby power stations are also important, especially if their power will
be transmitted predominantly by the HVDC system, since the possibility of
subsynchronous oscillations must be considered.

The results of load-flow and short-circuit calculations are important for project planning. If
an HVDC system will be used to damp electromechanical oscillations, a stability study of
the of the ac network is mandatory.

Load-flow and stability studies should generally be performed if the rated power of the
HVDC system is approximately as large as that of the largest power plant block in the ac
network or larger so that the consequences of an HVDC outage can be evaluated.

- 424 -
6.1.1.1 AC Network Voltage

The characteristic quantity of a high voltage ac system is the network voltage. The
following voltage parameters must be known:

• Rated value
This is the highest busbar voltage permissible on a continuous basis. It determines the
design of converter transformers and reactive power elements.

• Steady-state voltage range


The HVDC system must be capable of functioning within this range. All specified
guarantee values must be met with (unless exceptions have been explicitly agreed
upon).

• Expanded voltage range


Within this “emergency range,” the HVDC system must be capable of functioning, the
protection of all system components must be ensured. A reduction in transmission
capacity is normally permitted, and the exceeding of limit values for interactions is
tolerated.

• The negative sequence system


The negative sequence system is important for the calculation of noncharacteristic
harmonic currents on the ac side and noncharacteristic harmonic voltages on the dc side
of the HVDC station and hence for the design of filter circuits.

- 425 -
6.1.1.2 Network Frequency

The network frequency — 50 or 60 Hz — has a certain influence on the achieveable


power of a transformer unit when transport weight is limited. In the event of a direct valve
short circuit, the frequency dependent relationship îk / ò ik2dt must be considered in sizing
the thyristors. Otherwise, network frequency has little influence on project planning and
dimensioning of an HVDC system component. In contrast, temporary deviations from
nominal network frequency are very important.

• Steady-state frequency range


This range usually covers only a few hundred millihertz. Within this range, all specified
limit values (in particular, voltage distortions on the ac and dc sides) must be met. Loss
tolerance limits must not be exceeded.

• Expanded frequency range


This covers emergency situations in the ac network. It is important for the HVDC
system to continue functioning in order to render assistance. Although exceeding of
specified interaction limits may be tolerated, no dangerous parallel resonance between
the network and ac filter circuits must be permitted. To evaluate this phenomenon, the
duration of the frequency deviation must be known.

• Load rejection overfrequency


If an HVDC rectifier station is fed by a hydropower plant which is isolated
continuously or intermittently from the ac network, overfrequency associated with a
partial load rejection (such as a long-lasting fault of an HVDC line pole) must be
accepted. It must not result in shut-down or 100% load rejection. The disconnection of
filter circuits may be tolerated in order to avoid parallel resonance.

• Emergency disconnection
If the frequency deviation exceeds the limits described above, the HVDC must be
capable of reducing transmission power “smoothly” to zero.

- 426 -
6.1.1.3 Network Impedance

Network impedance at the fundamental frequency as seen from the ac busbar determines
the short-circuit current in the event of faults on the ac side of the HVDC station and in the
valve zone. In addition, it is an important criterion for reliable functioning in inverter
operation. For this reason, both the minimum and maximum network impedances or
alternatively the maximum and minimum network short-circuit power at the ac busbar are
needed. All operating conditions of the network, (i.e. heavy load and light load) and
changes of network configuration must be considered. Foreseeable future changes in
network structure must also be considered.

Network impedance in the frequency range from the 2nd to the 50th harmonic, for all
operating conditions, is required for the design for the ac filter circuits. Since these
impedances are usually not known and are difficult to calculate reliably, other procedures
must be used, as discussed in Section 2.2.3.

Surge impedances of the positive sequence system and of the zero sequence system of all
ac and dc lines emanating from the HVDC station are needed to design blocking filters in
the frequency range of power line carrier systems.

6.1.1.4 External Sources of Harmonics

Often limits are specified for voltage distortion at the ac busbar of an HVDC station
without consideration of whether or not a voltage distortion already exists at this point in
the ac network. Therefore the harmonic spectrum of nearby sources of harmonic currents
(static compensators, high power converter-fed drives, and the like) should be known in
order to consider them in dimensioning ac filter circuits.

Alternatively, a certain amount can be added to the harmonic currents generated by the
HVDC converters to represent the influence of outside sources of harmonics. There is a
danger, however, that in the presence of noncharacteristic harmonics (for example, the
5th) on the ac busbar of the HVDC station, an externally caused unacceptably high
individual distortion may occur.

- 427 -
6.1.2 Environment of the HVDC Station

In an invitation for bids, the location of the HVDC stations will generally be fixed. Many
of the data and ambient conditions described below which are necessary for project
planning will be given or can be easily obtained. If, however, an optimal location for a
station is to be determined in a preliminary stage of the project, the points discussed below
can be used in preparing an evaluating matrix for the various sites under consideration.

6.1.2.1 Geographical Data

Geographical data includes the following data and information about the station location:

• Geographical position
In addition to the longitude and latitude, there should be some information about the
nature of the landscape, preferably in the form of precise maps on a large scale with
contour lines, similar to a topographical map.

• Elevation above sea-level


The elevation of the site above sea-level is a determining factor for minimum
clearances of life parts. In addition, the elevation must be considered for the design of
the air cooling system.

• Ground characteristics
This pertains primarily to the suitability of the substrate for construction, building of
foundations, etc. The location of subsurface rocks and boulders and the frost line are
also important. If construction services are to be included in the proposal, on-site
studies including test bores are necessary.

- 428 -
• Water table
This information is important not only for construction planning, such as for
basements, but also for the siting of deep wells for station water supply. Water quality
and the inflow rate and its variation through the year are of interest.

• Susceptibility to contamination
Particularly, dust content of the air including aerosols from plant sources is significant
for the design of filters for air cooling and air conditioning systems. Contamination of
insulator surfaces, particularly those to which dc voltage is applied is even more
critical. For this reason, the conductivity of the aerosols and particles must be known.
Strongly recommended is the use of a test setup for at least a year to determine the
contamination of insulators, i.e. the ESDD value (equivalent salt deposit density) in
milligrams per square centimeter. This value along with the expected frequency of
wetting from dew or fog can be used to determine the required creepage distance (see
Section 3.1.6).

• Seismic conditions
If the site is in a region in which there is a risk of earthquakes, the maximum horizontal
and vertical accelerations, and the frequency range of vibrations and their duration must
be known. If the converter station is in the immediate vicinity of a new or future
hydroplant on a reservoir it may be necessary to take into consideration earthquakes
which are triggered by the mass of impounded water.

- 429 -
6.1.2.2 Climatic Conditions

An understanding must be reached whether extreme values which may occur only once
every few years will be used as the basis for the design, or whether more moderate values
will be used and reduction of transmission power and other limit values will be permitted
on the occurrence of extreme values.

• Temperature
In addition to the absolute maximum and minimum values, maximum and minimum
average values for 24 hour periods are needed. If evaporation cooling towers will be
used, wet bulb temperatures are needed in addition to dry bulb temperatures. In many
systems, an increased transmission power is required in the lower range of ambient
temperature. In this case, temperature values for this range must also be specified.

• Humidity
This primarily has to do with the determination of periods of dampness (how often the
relative humidity reaches values close to 100% during a year), causing the wetting of
contaminated insulator surfaces, and how long the periods of dampness last. (Complete
wetting of insulator contamination occurs only after several hours).

• Insolation
The maximum insolation (intensity of sun rays) for vertical and horizontal surfaces is
needed for the dimensioning of all open-air equipment (transformers, smoothing
reactors, capacitor cans). It is even more important for the design of cooling and air
conditioning systems of buildings.

- 430 -
• Wind speed
In addition to maximum continuous wind speed, the maximum wind speed in storms is
important. Maximum wind speed at temperatures below freezing is important in
connection with the maximum ice coating on spans and insulators and also on
equipment such as disconnecting and earthing switches. If there is a prevailing wind
direction, this can be important in the orientation of evaporation cooling towers and
also of intensive sources of noise.

• Precipitation
Precipitation measurements are important for the design of the drainage system. They
include annual precipitation, maximum quantity during one hour, maximum quantity
during five minutes, and maximum depth of snow on the ground if applicable.

• Keraunic level
Keraunic level, i.e. the number of lightning strikes per square km per year, is important
for the design of lightning protection systems.

- 431 -
6.1.2.3 Water Supply

If direct water cooling is not accepted or impractical, recooling using an evaporation


cooling tower will be required. In this case, the following paramters must be known:

• Type of water supply


Use of an existing public water supply system will be unusual. Ordinarily, a pipeline
will have to be built or a well will have to be drilled.

• Afflux quantity
A reliable flow quantity during the year is important, especially during dry periods.
Temporary fluctuations during the day can be buffered by the use of a holding tank.

• Water quality
Water temperature, conductivity, pH value, nature and quantity of soluble and insoluble
materials are required quantities.

6.1.2.4 Access Roads

The nature of access roads to the station site can impose limitations on the weight and
dimensions of plant components. These limitations can be critical factors in project
planning for an HVDC station. This can determine, for example, whether converter
transformers will be configured as three-winding or two-winding units. Often it is more
economical to circumvent limitations of this type by the use of alternate routes,
reinforcement of bridges, etc.

- 432 -
6.1.2.5 Requirements Imposed by Governmental Bodies

Governmental requirements can be of varied nature. They often involve expensive


measures, many of which are unnecessary from a technical perspective. A good
understanding of such requirements should be acquired prior to submitting a proposal.
Governmental requirements may include

• Aesthetic considerations
In addition to architectural design of buildings and structures, this can include
avoidance of overhead connection wires in the switchyard etc.

• Limitation of height of buildings


This can present problems in the design of valve halls. “Sinking” of the hall is very
expensive and often almost impossible. This requirement may necessitate selection of a
modified valve configuration. (Avoidance of quadrivalve towers. )

• Nature conservation,
This item also sometimes involves very significant cost.

• Noise emission
Particularly strict requirements can be expected if the site will be near residential areas,
protected natural areas, or local recreation centers.

- 433 -
6.1.3 Location of Earth Electrodes

Selection of suitable sites for earth electrodes of an HVDC system is more dependent upon
environmental conditions than is the case for converter stations. The environment to be
considered in siting earth electrodes extends up to ten's of kilometers in radius and several
hundred meters in depth. The determinative factors, especially the selection criteria were
detailed in chapter 4.

6.1.4 The Environment of HVDC Overhead Lines

The routing of lines and rights of way of an HVDC system will not be discussed here. This
section assumes that the route of the lines has already been determined. The items to
follow should be considered as given conditions of the environment of the lines.
Knowledge of these conditions is also important for the design of HVDC stations. They
play a very important role in the performance of the HVDC system as a whole.

6.1.4.1 Geographical Data

A detailed description of the route of the lines is required. This description should be in
the form of maps of adequate scale. Other important information includes:

• Line route
Information about parallel ac lines or telecommunication lines on the same or
neighbouring rights of way is especially important. Both, the length of the parallel
routing and the distance between the lines are needed to determine mutual influences
and interference. In the case of high voltage ac lines, the main electrical and mechanical
data are needed; in the case of telecommunication lines, limits for disturbance voltages
are needed. Intersecting angles are important at points where lines are crossing.

- 434 -
• Landscape
Operating behaviour of an overhead line system depends to a significant degree on the
landscape over which the line passes: mountains, plains, river valleys, swamp land,
coast land. The length of the portion of line passing across each of these landscape
types should be known.

• Vegetation
The range of possibilities extends from sand or boulder deserts to tropical rain forests.
Knowledge of high-growing grasses or shrubs is important, as is knowledge of
agricultural use. Sugar cane plantations, for example, customarily follow the practice of
burning off sugar beets. This is particularly crucial. The same applies to the burning of
grasses on steppes prior to sowing as in the southern part of Africa.

6.1.4.2 Geological Structures

Geological structures near the surface along the route of the line are important in two
respects:

• Specific resistance of soil


This is important for the footing resistance of the line towers. It should not exceed 20 to
25 W at any point. Within the first 10 km from converter stations, it should be less than
10 W in order to prevent back flashes during lightning strikes on towers or the
immediate vicinity. If necessary, the grounding of towers can be improved by a bare
copper wire connecting the grounding points of the towers, called a counterpoise.

• Conductivity of dust particles


A heavy incidence of dust must be taken into consideration, particularly in areas of
sparse vegetation and strong winds. The dust will tend to coat the surfaces of dc voltage
insulators (electrofilter effect). The nature of the dust, conductivity, salt content,
particle size, solubility in water, etc., play a critical role in flashover behaviour of
insulators.

- 435 -
6.1.4.3 Climatic Conditions

An HVDC system using overhead lines will always extend over a great distance. The line
will usually pass through different climatic zones. Weather conditions along the line will
not be uniform. Therefore it is important to define the various relevant climatic zones. The
length of each and the weather conditions within each climatic zone should be described
by means of minimum and maximum values as well as statistical averages.

• Temperature
Minimum and maximum temperatures of the different climatic zones must be
determined. These minimum and maximum temperatures, combined with minimum
and maximum continuous direct current, maximum and minimum wind speed, and
minimum and maximum sun radiation, determine the range of line resistance to be
considered in system design and in calculating transmission losses. For this purpose,
the fact that the extreme values of climatic conditions will not occur simultaneously
throughout the length of a line can be taken into account by additions or subtractions to
the extreme values.

• Sun radiation,
Sun radiation always results in the elevation of conductor temperature. Its influence
should not be neglected, especially in subtropical and tropical areas.

• Wind speed and direction


In addition to the mechanical design of lines, which will not be discussed here, the
wind influences a series of characteristics of the line. It always results in the lowering
of conductor temperature. Corona effects are intensified. Distribution of ion currents
are strongly influenced by the wind direction, and thus space charge and the electrical
field underneath the lines are also influenced.

- 436 -
• Precipitation
Occasional heavy rains are especially important. Heavy rains provide self-cleaning of
contaminated insulator surfaces. Rain in general has the effect of reducing radio
interference and acoustic noise disturbance. It does, however, tend to increase corona
losses. Snow and ice have a lesser impact.

• Fog and Dew


Fog and dew result in the wetting of contamination layers on insulator surfaces. This
increases the danger of contamination arcing. Determination of the number and
duration of expected moisture periods annually for the various sections of lines is
important. Since a complete wetting requires at least 2 to 3 hours, periods of less
duration should be discounted.

6.1.4.4 Insulator Contamination

As presented above, contamination flashover is critical in the behaviour of HVDC


overhead lines. Insulator surface contaminants include dust; all kinds of aerosols, of both
plant or industrial origin; combustion products and in coastal areas, salt mist. The dc field
in the vicinity of the line results in the charging of such particles. Therefore precipitation
of the particles onto insulators corresponds to the pattern of field strength (electro-filter
effect). This results in a nonuniform deposit along the insulator chain, which has a
considerable influence on flashover performance. In addition, the washing off of the
contamination layer by rain only effects the upper side of insulator covers.

Since the sum of the parameters and effects cannot be summarized in a dimensioning rule
for line insulators, it is strongly recommended that the “equivalent salt deposit density”
(ESDD) be empirically determined. Prior to the construction of the line, a number of test
sites should be established along the intended route and should be operated for at least a
full year. In these open-air sites, shortened original insulators to which dc voltage
corresponding to the specific stress of the future line is applied, should be suspended.
After a year, the ESDD can be determined in the lab, providing the most important
dimensioning parameter for specific creepage distance and chain length. Setting up test
sites of this type is most important where a relatively high degree of contamination and
frequent periods of wetting are expected.

- 437 -
6.2 Steady-state Performance of the HVDC

6.2.1 Nominal Data of the HVDC

The nominal transmission power of the HVDC system is of primary importance. In a two-
point HVDC system, this generally is understood as the power measured at the dc-side
terminals of the rectifier station. In long-distance transmissions, nominal transmission
voltage is in most cases also specified, while in back-to-back links, the selection of dc
voltage is usually left to the bidder.

6.2.1.1 Nominal Power

In many cases, the nominal power is not the same as the rated power. Often an overload
capacity is specified for relatively long periods of time. This overload capacity determines
the rated power for many system components such as the valves. For many HVDC
systems, two nominal powers were specified, one for high (maximum) ambient
temperatures and the second, higher, nominal power for lower ambient temperatures (for
example, <10° C).

Additionally, the nominal power for the two energy directions is sometimes specified
differently or nominal power is specified only for the main energy direction and it is left to
the bidder to indicate transmission capacity of the system in the opposite direction. In such
a case and in systems with only one fixed energy direction, the rated power of the inverter
station can be selected smaller than that of the rectifier station. Care must always be taken
that the inverter dc voltage at partial load will be higher than at nominal load due to the
reduced voltage drop along the dc line. As the voltage profile of a two-point HVDC
system as presented in Fig. 6.1 indicates, the relationship of the voltage drop along the
transmission line to the internal voltage drop of the inverter (dx-dr) determines, whether
the value Udio which is critical in the voltage rating of the inverter should be higher or
lower at partial load than at nominal load.

- 438 -
a

Figure 6.1:
Voltage profile of a two-point HVDC system at nominal load ______; at partial load ------

Glossary:
GR rectifier
L transmission line
WR inverter

6.2.1.2 Nominal Voltage

In HVDC terminology it is customary to designate the voltage between line pole and
ground as nominal voltage. On this basis, insulation between the poles of a bipolar HVDC
system must be sized for twice the nominal voltage. This applies not only for the distance
between conductors in a bipolar overhead line system, but also for all switching devices in
the dc switchgear of an HVDC station which are arranged between the station poles (see
Section 2.7).

The IEC defines the voltage of the rectifier station — measured between station pole and
station neutral bus — as nominal voltage. The voltage drop of the rectifier-side earth
electrode and associated electrode line is not included in this definition. For monopolar
HVDC systems, this causes the rated voltage of the line to be smaller than the nominal
voltage as is illustrated in Fig. 6.2.

- 439 -
Figure 6.2: Voltage profile of a monopolar HVDC system

Glossary:
GR rectifier
MP neutral bus
EE earth electrode
WR inverter

This relations will only apply if dc voltage regulation insures that conductor-to-ground
voltage is not increased at partial load.

This problem does not occur in bipolar systems. Nominal voltage of the rectifier station is
also rated voltage of the dc line. However, there are a few HVDC systems (for example
Nelson River I, James Bay-Boston) in which the dc voltage of the inverter station or an
intermediate value was declared as the nominal voltage. If this relationship is not clearly
defined in the specification, a clarification must be obtained prior to the beginning of
project planning.

6.2.1.3 Nominal Current

The nominal current of the system is derived directly from nominal power and nominal
voltage. It applies for both stations and for the line. In contrast, the rated current can be
different for the various components of the converter station, corresponding to the required
overload conditions.

For the HVDC overhead lines, other conditions apply. The cross-section is usually
selected according to the economic current density. Nominal current is then far below the
thermal limit value which, however, can be the rated value of the line if operation over one
conductor pole with parallel-connected station poles (emergency operation with earth
return in the event of the loss of one line pole) is specified.

- 440 -
6.2.2 Limiting Data of the HVDC System

This section refers to the maximum or minimum values of transmission power, dc voltage
and direct current permitted on a continuous or transient basis.

6.2.2.1 Overload

In defining overload conditions, the extent to which alleviating circumstances may be used
must be specified:

• Availability and operation of redundant cooling systems

• Lower ambient temperatures

• Overtemperature of devices, combined with a shortening of anticipated useful life.

In addition, it must be clear whether the nominal direct voltage must also be achieved in
the event of overload. This requirement will result in the following relationships:

• The converter must be designed for a higher ideal direct voltage Udio, i.e., the valves
must be sized for higher voltage.

• The tap changer range of converter transformers must be sized larger and hence the
transformer rated power must be selected higher.

If direct voltage of the rectifier station at overload is permitted to drop below the nominal
value, these expensive measures become unnecessary. However, a higher percentage of
overcurrent is needed to achieve a desired percentage of overpower. This not only has
consequences for current design, but also causes higher transmission losses in the overload
range.

- 441 -
In addition to the above considerations, the length of time a particular overload shall be
accommodated must be defined. Three typical characteristics can be defined:

• Temporary overload in which a power Pü = (1 + ü) × PN is carried for a time tü

• Decaying overload in which a momentary overload decays linearly (or in accordance


with a specified function) to nominal value during a time tü

• Oscillating overload in which an oscillating modulation power of amplitude ü and


frequency fm is superimposed on nominal power for a period of time tü

Figure 6.3: Temporary overload

Figure 6.4: Decaying overload

Figure 6.5: Oscillating overload

- 442 -
Occasionally it is required that the bias of the system be considered. Thus, a higher
overload ü or a longer overload period tü will be permitted if the system was previously
operated at partial load. In this event, knowledge of the available overload capacity of the
individual system components is necessary.

6.2.2.2 Overload Capacity of the Devices

Devices almost always have a certain overload capacity even without assuming a
shortening of expected useful life because the maximum ambient temperature on which
the design was based is present only for short periods of time. Additional overload
capacity can be gained by using redundant cooling systems and by considering the lower
bias if applicable. The relationships of overload capacities of various components,
however, are quite varied:

• For valves, the maximum permissible junction temperature of thyristors must not be
exceeded. Overload capacity for periods greater than one minute will exist only if the
temperature of the coolant is lowered on the feed side. This provides an increased
temperature differential and thus a greater amount of heat can be dissipated.

• In transformers and smoothing reactors, the windings have a thermal time constant of
approximately 15 minutes, while the oil circuits have a thermal time constant of 1 hour
or more.

• In filter circuits, the elevated harmonic currents which occur at overload usually do not
cause special problems with respect to stress of the filter components. However, they
result in increased voltage distortion at the ac busbar. It is necessary to search
agreement whether increased distortion at overload is permissible or whether the filter
design must cover the overload range.

• The increased reactive power demand of the converter at overload may be critical since
it can result in an unacceptable drop of the ac busbar voltage. Therefore it must be
clarified whether it is permissible for the specified reactive power range to be exceeded
at overload or whether the installation of additional reactive power generators such as
capacitor banks is necessary for maintaining the reactive power balance.

With knowledge of the overload time functions of individual components in relation to the
external conditions, it is possible to determine the overload capacity of the entire system

- 443 -
using a permissible load calculator which determines the lower limiting curve of the
individual overload time functions.

6.2.2.3 Minimum Power

Because of the unavoidable ripple of the direct current, which is a function of the control
angle and of dc-side inductances, there is a danger at low transmission power of current
discontinuities. This not only results in increased stress on valve snubber circuits; it also
periodically interrupts the coupling of rectifier and inverter by the common value Id which
can result in regulation instability.

In order to avoid the current discontinuities, a minimum direct current is defined which is
above the discontinuity limit by a generous margin of safety. Therefore, even dynamic
current or control angle fluctuations do not result in dc current zeroes.

The minimum transmission power of the system is derived from the minimum current.

6.2.2.4 Operation with Reduced Voltage

There is a danger of contamination flashovers of dc voltage insulators in the event of


wetting by fog or dew. In order to be able to continue energy transmission even in extreme
conditions, though at reduced power, operation with reduced dc voltage is often specified.

Another application of this operation mode is cable-dependent control which reduces


voltage and by the same percentage increases the current in the event of a drop in
transmission power in order to avoid excessive release of pressure in the dielectric which
would result in the formation of hollow spaces and corona discharges.

In both cases, a reduction of transmission voltage to 70-80% of the nominal value is


customary. The following measures may be used:

• Placement of the transformer tap changer on the lowest level and, if necessary, design
of the transformers with an expanded range of adjustment

• Operation with increased control angle of the converters

- 444 -
The latter measure means increased losses in valve snubber circuits, higher harmonic
currents, and greater reactive power demand. For these reasons, transmission current also
must be reduced by perhaps 10%. This, of course, reduces the transmission power further.

If operation at reduced dc voltage is required, the related constraints must be precisely


defined and agreed upon.

6.2.2.5 No-load Ready State

No-load ready state is a system condition in which immediate assumption of energy


transmission is possible.

• The converter transformers are switched on, or if the operator wants to avoid no-load
losses, are prepared for being switched on. This means that the grounding switches are
open, the disconnect switches are closed, and the oil circulating pumps and cooling fans
are in operation at the lowest level. After the converter transformers are switched on,
the switching inrushcurrent must be allowed to decay before the converter valves are
deblocked.

• The ac filter circuits and capacitor banks are switched off. If the load dispatcher wants
to use them to maintain voltage in his network, their losses are not considered in
calculating the total loss of system no-load ready state.

• The converter valves are in blocked condition; the pumps of the primary and secondary
cooling circuits and the fans of the re-cooling system are in operation at the lowest
level; the same applies for the valve hall cooling system.

• The dc-side disconnect switches for HVDC lines, electrode lines, and dc filter circuits
are closed; oil circulating pumps and coolers of oil-cooled smoothing reactors are
running.

• The auxiliary power supply and all control devices including telecommunications are
fully functional.

- 445 -
6.2.3 Reactive Power Behaviour

In contrast to ac transmission, reactive power behaviour of an HVDC in steady-state


operation is not influenced by the transmission line, but is determined exclusively by the
reactive power demand of the converters and the reactive power generating units which
are used for compensation. The reactive power balance at the interfaces between HVDC
stations and the ac networks is of great significance for the system behaviour.

6.2.3.1 Reactive Power Demand of Converters

In Section 1.4, basic principles of reactive power demand of HVDC converters and its
calculation were presented along with the relationships between reactive power demand
and transmission power and other parameters.

On the basis of these fundamental principles, the reactive power demand of the converters
can be determined during project planning for an HVDC system for various specified
operating conditions. This can be done independently for the two stations. The following
parameters have to be considered.

• Energy direction
Reactive power demand of rectifier and inverter is different. If both energy directions
are required, two complete scenarios must be computed for each station.

• Range of power
Reactive power demand of the stations must be computed over the entire specified
power range including the required overload. If electronic reactive power regulation is
to be used in the weak-load range (i.e., increasing reactive power demand by enlarging
the control angle) reactive power behaviour of the two stations is no longer independent
of each other because this procedure influences the parameters Ud and Id, which are
common to both stations.

- 446 -
• Operating modes
The specifications often require consideration of other operating modes in addition to
normal bipolar operation. For example, monpolar operation with earth return, with
metallic return, parallel-connected line poles at one station pole, parallel connected
station poles at one line pole, etc. Because of the various voltage relationships in the dc
circuit, the reactive power demand also differs in various operating modes.

• Operation at reduced dc voltage


Depending upon the method used for the reduction of voltage, significant deviations in
reactive power demand compared with normal operation may need to be considered.
The extent to which the power range is limited in this mode of operation is important.

Consideration must be given in these computations to the fact that control angle a of the
rectifier and — depending upon regulation concept — the extinction angle g of the
invertor or system dc voltage must be regulated with the assistance of tap changers of the
converter transformers. Because of the stepped operation and the dead range necessary for
the avoidance of instability, reactive power demand will be set somewhere between two
limits. The two limit curves must be calculated and must be considered in project
planning, as was shown in Section 2.2.2.

6.2.3.2 Reactive Power Balance

The reactive power balance at the interface between the HVDC station and the ac network
is important for maintaining the ac voltage. This balance is determined by the reactive
power demand of the converter and by reactive power generation by filter circuits and
capacitor banks. This reactive power balance can be specified in various ways:

• The reactive power curve Q = f(P)


A certain reactive power curve, usually linear, over the entire load range is specified.
Due to the stepped operation of reactive power units, a large tolerance range is provided
for the reactive power balance. In many systems, the curve is specified simply as
Q = f(P) = 0 ± DQ

This means that the HVDC system does not intervene in the reactive power balance of
the ac network. The uneconomical transport of reactive power over ac lines is avoided,
aside from the tolerance, DQ.

• Displacement factor cos j

- 447 -
The requirement for a constant cos j in the entire load range — lagging for the
rectifier, leading for the inverter- is based upon the concept of the HVDC system as
consumer (rectifier) or as generator (inverter). This kind of reactive power behaviour
requires a very great expenditure for reactive power units — especially on the inverter
side — and impairs inverter stability due to the decreasement of the ESCR (effective
short-circuit ratio). In addition, the transport of reactive power in the ac network on
both sides is disadvantageous.

• AC busbar voltage
Specification of regulation of the ac busbar voltage, usually to held it constant
throughout the entire load range, primarily ignores reactive power balance. The results,
however, are about the same. In this approach, the HVDC may have to connect
additional reactive power units to compensate for a reactive power deficit or, in the
event of excess reactive power in the network, it may have to disconnect reactive power
units. In the latter case, the necessary filter effect may not be assured. In any case, the
limits and areas of responsibility must be carefully defined.

If the rectifier station is connected directly with a power plant, voltage regulation and
compensation of reactive power balance can be left to the generators. At light loads, there
may be a danger of self-excitation by the excess reactive power supplied from the filter
circuits which are still connected.

6.2.3.3 Reactive Power Generation

Reactive power which needs to be generated is derived from the specified reactive power
balance and the converter demand. The filter circuits contribute a significant portion of it.
The dimensioning of the filter circuits, however, is determined in large part by the
specified filter effect. Demand beyond the reactive power provided by filter circuits is
most economically covered by capacitor banks. Only when very narrow tolerance limits
are specified or voltage stability problems are feared, continuously controlled static
compensators — or more rarely — synchronous condensers may be used. Occasionally a
shunt reactor may be employed as a supplemental reactive power consumer in order to
handle light-load or no-load conditions.

The maximum size of switch-operated reactive power units is derived from the
permissible tolerance of reactive power balance and from the maximum voltage change
upon switching of a unit which is usually also specified. The influence of the busbar
voltage within the specified band of fluctuation (operating range) upon the reactive power

- 448 -
output of the reactive power units must be considered. All of these questions have been
discussed in detail in Section 2.2.2.

6.2.4 Harmonics on the AC Side

Converters generate harmonics. In relation to an ac network, the converters of an HVDC


system act as sources of harmonic currents. In addition to the characteristic harmonics
determined by the pulse number, as consequence of all kinds of unsymmetries, non-
characteristic harmonics and, under certain conditions, non-harmonic oscillations may
occur. The theoretical foundation is presented in Section 1.5.

6.2.4.1 Distortion Criteria

To the extent they are not absorbed by the filter circuits, the harmonic currents cause
voltage drops in the reactances of the ac network which result in distortion of the ac busbar
voltage of the HVDC station. The harmonic currents which enter the network over ac
overhead lines can also cause telephone interference in communication lines which are
parallel or which cross the overhead lines.

As discussed in Section 2.2.3, the distortion phenomena can be described by various


criteria:

• Individual distortion Du,


defined as the ratio of a harmonic voltage Uu to fundamental voltage U1:

Un
Dn = · 100 [%]
U1

• Total distortion Dtot, defined as the geometrical sum of all individual distortions:

¥
D tot = å n=2
D 2n [%]

Whereby the range up to u = 50 is customarily evaluated.

• Telephone interference factor TIF or, alternatively,

- 449 -
• Harmonic form factor THFF, defined as the ratio of the geometric sum of all harmonic
voltages multiplied by a psophometric weighting factor to the total value of system
voltage. Equations and weighting factors are given in Section 2.2.3.1.

• IT product,
defined as the geometric sum of all harmonic currents multiplied by the associated TIF
factor Fn, which emanate from the busbar of the HVDC station and flow into the
connected ac lines.

å (I × Fu )
¥ 2
IT = u =1 u [ A]

6.2.4.2 Distortion Limit Values

Usually not all distortion criteria are specified. The guarantee values for HVDC systems
vary for the time being in wide limits. IEC 919 lists the following typical limit values as
examples:

• Individual distortion Dv =1%


(a very detailed new recommendation of the CIGRE WG 36-05 is found in Section
2.2.3.1)

• Total distortion Dtot = 2...5%

• Telephone interference factor TIF = 25...50

• Harmonic form factor THFF = 0.6...1.25%

• IT product IT=25000...50000A per conductor

In order to avoid later disputes, a precise definition must be agreed upon in the early stages
of a contract with respect to distortion limits, range of applicability, and external
circumstances to be considered.

- 450 -
6.2.4.3 Filter Effect

The definition of distortion limits determines the required filter effect. The following
factors must be considered in the configuration of filter circuits:

• Harmonic currents generated by the converters in steady-state operation over the entire
load range up to the overload limit for which the limit values for distortion are
applicable according to specifications

• Harmonic currents generated by extraordinary operating modes, in particular operation


with reduced dc voltage and electronic reactive power regulation, in each case in the
power ranges provided for such operation

• The assumed unfavorable curve of network impedance versus frequency (see Section
2.2.3.3)

• The anticipated range of fluctuation of network voltage and frequency in steady-state


operation (operating range)

• Deployment strategy of switch-operated filter circuits and capacitor banks for the
realization of the specified reactive power behaviour over the entire load range, for both
energy directions if applicable

• Operation in the event of loss of any particular filter circuit (maintaining of the (n-1)-
condition)

• Consideration of initial detuning of filter circuits; loss of a specified number of


capacitor units; temperature-dependent changes in capacitance with consideration of
specified extreme temperature values, unequal sun radiation and wind.

Deviations from the specified filter effect should be permitted in the event of extraordinary
operating conditions such as:

• Network voltage and/or frequency outside of the operating range

• Temporary overvoltages in the ac network as a result of load rejection or fault clearing


or voltage distortions caused by transformer saturation

- 451 -
• Transient overload

6.2.5 DC-Side Harmonics

Harmonic voltages are superimposed upon the dc voltage of a converter. These harmonic
voltages are dependent upon control and overlap angles, as was presented in Section 1.6.
A distinction is made between the characteristic harmonics determined by the pulse
number of the converter and the noncharacteristic harmonics caused by unsymmetries of
the ac-side and by leakage capacitances in the converter. These harmonic voltages cause ac
currents in the HVDC lines superimposed upon the dc current which may cause
interferences. The harmonic currents undergo a phase rotation along the line and form
standing waves, whereby the currents of the same frequency which are fed from the two
sides are combined vectorially. If the ac networks are not synchronous on both sides,
fluctuations and non-harmonic oscillations arise.

6.2.5.1 Interference Criteria

In themselves, the harmonic voltages on the dc side of the converter, which can be easily
calculated, do not cause any interference. Also the harmonic currents caused by them,
which can be reliably determined with the knowledge of station and line data and the ac
network structure, are not in themselves a criterion for possible interferences. It is only
through the inductive coupling of the harmonic currents into telecommunication lines and
other metallic structures that interference voltages and currents arise. In addition to the
coupling mechanisms which are dependent upon local conditions and circumstances,
shielding is of decisive importance: HVDC systems using cable do not cause any dc-side
interference and telecommunications utilizing cable are not interfered with by HVDC
transmissions.

Two interference criteria are recognized today, and these criteria are associated with
specific harmonic frequencies or ordinal numbers:

• Dangerous touch voltages m = 1...6

• Telephone interference m = 7...48

Although interference currents in the frequency range below 300...360 Hz (m =6) may
cause telephone interference and inductive interference voltages in the frequency range
above 300 Hz may be dangerous, this delineation has gained general acceptance.

- 452 -
6.2.5.2 Touch Voltage

Ordinal numbers from m = 1 to m = 6 designate noncharacteristic harmonics in HVDC


systems. As was demonstrated in Section 2.6.4, the occurrence of the third harmonic and
its multiples must be expected even in the case of completely symmetrical relationships on
the ac side and exact equidistance of the control pulses. This phenomenon is caused by
stray capacitances to ground of the series-connected 3-pulse commutation groups of a 12-
pulse converter.

These “low order harmonics” can induce voltages in metal structures close to an HVDC
overhead line such as fences, pipes, etc. which may present a danger to human beings and
animals and may damage sensitive instruments.

6.2.5.3 Telephone Interference

The frequency range of the 7th to the 48th harmonic comprises the essential characteristic
harmonics of an HVDC converter. The resulting amplitudes of the individual harmonic
currents must be limited to such an extent that the voltages induced in telecommunication
lines which run in parallel to the HVDC line or which cross the line, remain below the
interference limit for the particular harmonic.

Vin ( x ) = Z · I eq [ mV / km]

According to this equation, the interference voltage occurring in location (X) depends
upon the coupling impedance Z between HVDC and telecommunication lines and upon
the equivalent interference current Ieq. The equivalent interference current comprises all
harmonic currents weighted with a psophometric weighting factor (the “C-message”) at
point (X), and in addition, the frequency-dependent coupling factors. These factors
include, among other things, the specific ground resistance of the particular location (see
Section 2.6.2).

6.2.5.4 Limit Values for Interference Voltage and Current

The national regulations of a particular country are determining the limits for the touch
voltages induced by “low-order harmonics.” IEC 919 cites the limits established by CCIT
and AT&T as follows:

60 V at 50 Hz

- 453 -
50 V at 60 Hz

To date there are no rules or standards covering other frequencies. With respect to
telephone interference, usually limit values have been established by telephone authorities
and companies for induced interference voltages in telephone circuits. Precise knowledge
of local conditions is necessary for the calculation of induced voltages. Therefore a limit
value of this type is of limited usefulness as a basis for project planning. For this reason,
the equivalent interference current flowing in a fictitious telecommunication line of one
kilometer length running in parallel to an HVDC line at a distance of one kilometer has
frequently been specified in recent projects.

Ieq £ 500 mA in bipolar operation

Ieq £ 800 mA in (temporary) monopolar operation with ground return

6.2.5.5 Filter Effect

The harmonic current permitted to flow in a given section of an HVDC line is derived
from the specific interference limits as discussed above. From this, the harmonic voltages
can be calculated which are permitted at each end of the line and in turn the amount by
which the dc-side harmonic voltages generated by the converter must be reduced. This is
the design criterion for the dc filter circuits which, in conjunction with the dc-side
inductances, must bring about this reduction.

In addition, the following items must be considered in the design of filter circuits:

• Amplitudes and relative phase position of the individual harmonic voltages at both
ends of the line throughout the entire load range up to the overload limit at which the
interference limits must be complied with according to the specifications.

• Harmonic voltages occurring in extraordinary operating modes, in particular in


monopolar operation with ground return, in operation with reduced dc voltage, and in
operation with electronic reactive power regulation. In each case, the power ranges and
increased permissible interference limits (if applicable) for the particular operating
mode must be taken into consideration.

• Frequency-dependency of smoothing reactor inductance

- 454 -
• Range of fluctuation of network frequencies in steady-state operation

• Consideration of initial detuning of the filter circuits, of the loss of a specified number
of capacitor units or coils, of temperature-dependent changes in capacitance values, etc.

- 455 -
6.2.6 Power Line Carrier Disturbances

HVDC converters emit high frequency interference in the frequency range of power line
carrier systems, i.e. between 20 and 500 kHz. They are caused by voltage jumps during the
firing and extinguishing of high voltage valves which excite localized resonance circuits
of stray capacitances and inductances in the vicinity of the converter valves. Due to the
large number of such resonance circuits, the entire frequency band is covered. The
interference intensity decreases rapidly with the frequency.

Fig. 6.6 shows the typical curve of the high frequency interference voltage on the dc side
of an HVDC station measured at 3 kHz band width, indicated in dBm, with 0 dBm
corresponding to 0.775 volts (1 mW in 600 W).

Figure 6.6: Frequency spectrum of the high frequency interference voltages emitted by an
HVDC station in dBm (0 dBm = 0.775 V).

- 456 -
Interference energy is dependent upon the amplitude of voltage jumps. Thus it is only
slightly dependent on load current, namely over the overlap angle, but in contrast, it is
strongly dependent on control angles. This must be considered for specified operation with
reduced dc voltage.

6.2.6.1 Interference Mechanisms

High frequency interference in the power line carrier range generated by the converter is
propagated almost exclusively through conductors, both in the direction of the ac network
(i.e., through the converter transformers) and in the direction of the dc line (i.e., through
the smoothing reactors). Therefore power line carrier systems on ac overhead lines which
extend from the HVDC station as well as the power line carrier systems using the dc line
can be subject to interference.

Power line carrier systems on ac lines which use the same right of way as the HVDC line
may also be subject to interference. In this case, the interference frequencies are
transmitted by electromagnetic radiation. However, high frequency interference currents
drawn by the lines are very quickly damped. Therefore the emanation of the interference
may present a problem only in the first 100 km from the HVDC stations.

High frequency disturbances caused by corona depend primarily on the surface field
strength of the HVDC line but also on weather conditions. Typical values for this type of
interference are -30 to -40 dBm in the entire frequency range considered here. Therefore
they are generally far less significant than interference generated by the converter.

6.2.6.2 Interference Criteria

Signal-to-noise ratio at the receiving end of the power line carrier system is considered the
decisive criterion for fault-free signal transmission. Tolerated interference voltage is then
derived from this signal-to-noise ratio. Generally a value of -20 dBm or less is specified.
As can be seen in Fig. 6.6, interference of a neighbouring power line carrier system or of
the power line carrier system on the dc line must always be expected if the carrier
frequency is below 150 kHz. As far as possible, neighbouring power line carrier systems,
or at least the own power line carrier system should be configured with carrier frequencies
> 150 kHz.

The power line carrier system of the 1400 km long Cabora-Bassa HVDC can serve as an
example. As can be seen in Fig. 6.7, the power line carrier system is subdivided into three

- 457 -
sections. At the transition points, the signal is amplified and the carrier frequency is
changed.

Figure 6.7: Power line carrier system of the Cabora-Bassa HVDC

In this way, sections of the carrier system close to the converter stations use carrier
frequencies which are quite immune to interference. In the middle section in which no
interference is to be feared, very low carrier frequencies are used so that spanning the
distance of 800 km is possible without additional amplification.

6.2.6.3 Power Line Carrier Filters

If there are existing or planned power line carrier systems with carrier frequencies < 150
kHz in the vicinity of the HVDC stations and a change of the carrier frequency is not
practicable, special power line carrier filters must be employed on the ac and dc sides of
the HVDC stations. They must be sized so that the interference voltage in the frequency
bands used by the power line carrier systems is reduced to between -30 dBm and -20 dBm.

6.2.7 Radio Interference

As described in the preceding section, voltage jumps caused by converter valve firing and
extinction processes are the most important source of high frequency interference
extending from an HVDC station, at least in the lower frequency range. In the upper
frequency range, however, interference caused by corona may predominate.

Radio interference emanating from an HVDC line is influenced by processes in the


converter station only in its immediate vicinity. Beyond this area, interference triggered by
line corona predominates.

- 458 -
6.2.7.1 Interference Mechanisms

Propagation of high frequency interference generated in an HVDC station takes place


primarily along HVDC overhead lines and in ac lines which extend from the HVDC
station. The two following mechanisms should be distinguished:

• Conductor-to-ground mode which undergoes a rapid damping effect with distance and
can be neglected after approximately 15 km.

• Conductor-to-conductor mode which can propagate over several hundred km.

In addition, radio interference caused by HVDC line corona which is dependent on HVDC
line polarity should be taken into consideration. This interference occurs practically with
the same intensity along the entire line since the voltage drop is normally a few percent of
the conductor-to-ground voltage only. As was demonstrated in Section 3.1.4, however,
corona is strongly dependent upon weather conditions.

6.2.7.2 Interference Criteria

Radio interference considered in this section covers a frequency range of 0.15 MHz to
30 MHz. The signals which may be affected are primarily:

• Amplitude-modulated middle wave and short wave radio

• Nondirectional radio beacons for aviation

Less effected are:

• Frequency-modulated ultra-short wave radio

• Microwave communication connections

To avoid interference with radio reception, interference field strengths occurring in the
vicinity of residential areas and public transportation routes should not exceed certain limit
values, usually specified in mV/m. These values can vary widely in different regions and
are dependent upon the incoming signal strength of the transmitter.

- 459 -
For nondirectional radio beacons within a radius of 30 km around an HVDC station, a
permissible interference field strength should be specified for a bandwidth of ±10 kHz
around the transmission frequency.

In the case of communication links, a specified signal-to-noise ratio should not be


exceeded.

In order to avoid difficulties in the specification of many individual interference limits,


IEC 919 proposes a typical limit value of 100 mV/m for the total of all radio interferences
emitted by an HVDC system. This limit value must not be exceeded outside of a reference
line. As is schematically depicted in Fig. 6.8, the reference line is defined as follows:

• A line around the HVDC station at a distance of 500 meters from all voltage carrying
parts

• 150 meters on each side of any overhead line extending from the HVDC station
measured at the intersection with the 500 m line.

• Linear reduction of this distance from 150 m to half of the right-of-way width of the
line within a distance of 5 km.

The limit values for radio interference must be met over the entire load range. For
extraordinary operating modes (monopolar operation with ground return, operation with
reduced dc voltage, etc.), higher limit values may apply in a defined load range. These
must be clearly defined and delimited in the specifications. In addition, the specifications
should indicate under what weather conditions (for example fair weather) the specified
limit values apply.

Figure 6.8: Reference line around an HVDC system for definition of radio interference

- 460 -
6.2.7.3 Interference Suppression

In current HVDC engineering practice, converter valves are usually located in valve halls.
Propagation of high frequency interference, generated by firing and extinction processes,
can be effectively suppressed by shielding of the valve hall. Due to their relatively slow
current rise at the moment of firing, thyristor valves place far lower demands for shielding
than did the earlier mercury-arc valves. An all-steel hall in which the plates are welded to a
steel framework fulfills the requirements. The requirements are also fulfilled by a concrete
hall in which the rebar is welded at all intersection and contact points. When the
transformer bushings of the valve side windings project into the hall, the openings in the
hall wall should have a shield through which high frequencies cannot pass. The same
applies to hall doors, ventilation openings, etc.

If it is necessary to damp radio interference passing out through the lines, high frequency
chokes (wave traps) can be installed in the ac- and dc-side outgoing lines.

Unacceptably high corona-caused interference of an HVDC station can only be reduced by


lowering the surface field strength on conductors and devices carrying high voltage. This
can be accomplished either by increasing the diameter of conductors or by using bundled
conductors or multiple conduits. Shielding of the dc switchgear, as in the Sylmar HVDC
station (south end of the Pacific Intertie), is no longer accepted.

Excessive corona-caused radio interference of the HVDC line also requires a reduction of
conductor field strength. This can be accomplished by a modification of the cabling, by
increasing the line-to-line spacing in bipolar lines, and by increasing average line elevation
above ground.

6.2.8 Acoustic Interference

Noise from an HVDC station can produce acoustic interference in the immediate vicinity.
Since remedial action after system commissioning is very expensive and is often
ineffective, great attention should be devoted to this problem in the project planning stage.

The requirements for suppression of sound emission from an HVDC station are varied
depending upon the population of the area and type of landscape. All of the factors should
be considered in the design and arrangement of the components of an HVDC station.

- 461 -
In addition to the avoidance of public acoustic interference, certain noise level
requirements must be met within HVDC stations, particularly in service rooms and
general purpose areas.

Acoustic interference can also emanate from an HVDC overhead line. It is caused by
corona and displays similar interrelationships to those of radio interference (see section
3.1.4.3).

6.2.8.1 Noise Sources

In calculating sound emission, the following noise sources must be considered at least:

• Converter valves

Effective noise damping can be achieved by appropriate configuration of valve halls if


converter valves will be situated indoors.
Outdoor valves require special consideration.

• Valve cooling system

Considerable noise emission can be expected especially in the case of air cooling
towers. This problem can be diminished by the selection of slow-running, large
diameter ventilators; by optimized spatial arrangement and use of sound-damping air
filters.

• Converter transformers

In addition to the noise sources and levels which are comparable to those of ac network
transformers, the harmonic currents and a dc premagnetization caused by control pulse
asymmetries generate additional noise, especially in the higher frequency range. The
total noise level may be 10 dB higher due to such phenomena. This type of noise level
cannot be measured in the test field. Therefore the transformer manufacturer must
calculate the anticipated sound intensity level for the specified operating conditions and
the associated frequency spectrum. Measurement results from running HVDC systems
with comparable transformers should be reflected in these calculations.

- 462 -
• Smoothing reactors

The primary cause of noise are the dc-side harmonic currents. Harmonic currents
absorbed by dc filter circuits also pass through the smoothing reactors. Test field
measurements are also of no help for this source of sound. The manufacturer must
calculate the anticipated noise intensity level, taking into consideration all specified
operating conditions such as operation with reduced dc voltage.

• Filter circuit reactors

Appreciable contribution to acoustic interference of an HVDC station can be expected


from the filter circuit reactors if they are placed close to the station fence and if the
other main sources of noise are well shielded by structural or other means. Problems
due to sound from filter reactors can be diminished by sound-absorbing hoods.

6.2.8.2 Total Sound Emission

The method described in IEC 551 is suitable for the determination of acoustic interference
in a particular location in the vicinity of the HVDC station. The noise level of each
individual source is used as a starting point which is the mean value of measurements on a
line around the source. These values are converted to an acoustic power level of an
equivalent noise source. The acoustic intensity level from each individual sound source
can be calculated for the location under consideration. The total acoustic intensity level is
calculated by logarithmic summation of the individual acoustic intensity levels. If this total
acoustic intensity level exceeds the specified limit value, noise reducing or damping
measures must be provided. This is most effective at the source or sources which make the
highest contribution to the overall level.

6.2.8.3 Limit Values for Acoustic Interference

A reference boundary can be named in the specifications for which a limit value for the
permissible total acoustic intensity level, measured in dB (A), is specified for a defined
location. Different values can be specified for day and night, for example, 50 dB (A) for
day and 40 dB (A) for night. Another possibility is to define such a limit value for a
reference line around the HVDC station at a distance of 100 m from the active
components. All specified limit values are customarily valid with a tolerance of +3 dB
(A).

- 463 -
In most cases, the specified limit values must be met over the entire power range,
including the overload range. If ac voltage and frequency is outside of the operating range,
a higher level such as +5 dB(A) should be permitted. The same should apply for
extraordinary operating modes such as operation with reduced dc voltage.

In addition, it must be clear that the specified limit values do not include noises emitted by
foreign sources. The total acoustic intensity level should be measured both during HVDC
operation and without HVDC operation. The contribution of the HVDC system can then
be determined mathematically.

Within the work area of the HVDC station, higher limit values apply such as 85 dB(A) in
the service area and 70 dB(A) in general purpose rooms. When areas with high sound
levels such as pump rooms are entered for monitoring or servicing, hearing protection
must be used.

6.2.8.4 Interference Suppression

If a noise study of the converter stations in the project planning stage indicates that the
specified limit values cannot be met with reasonable costs with the given sizing of station
area and arrangement of equipment, structural shielding measures must be taken such as
providing an enclosure for the converter transformers. The effects of such a measure in
other aspects must also be taken into consideration. Thus the specified time required for
the changing of a transformer unit may be influenced by the housing provided.

6.2.9 System Losses

System losses are a very important factor for the economic efficiency of an HVDC system.
Therefore they are often specified as a guarantee value. They should be verified in a
suitable manner by the manufacturer. Losses above the specified guarantee value are
subject to monetary penalties.

System losses include:

• Total losses in the two HVDC stations


• Loss in the HVDC line
• Losses in the two earth electrodes and associated electrode lines occurring in
monopolar operation

- 464 -
Usually statement and guarantee for losses is required for several operating conditions:

• Nominal power in normal operation


• Partial load (for example, 0.7 or 0.5 p.u.)
• Minimum power
• No-load stand-by condition

Occasionally a statement of system losses during abnormal operating conditions is


required but they are not usually required as guarantee values. Examples of such abnormal
operating conditions are operation with reduced dc voltage and monopolar operation of a
bipolar system with ground return and with metallic return.

Direct measurement of system losses is not possible due to inadequate precision of


available measuring instruments and methods. For this reason it is customary and is
accepted as verification to determine total losses through addition of individual losses of
system components which are easier to determine. Even the individual losses of the most
important components of an HVDC system can only be determined with some degree of
uncertainty by using a combination of field test measurements, mathematical adjustments
for actual conditions and circumstances, and consideration of individual measurements in
running HVDC systems.

6.2.9.1 Environmental Conditions

The specification must clearly define all relevant environmental conditions of stations and
line for which the system losses are to be determined. Extreme values should not be used.
Instead, averages over a period of many years or prevailing conditions as listed below
should be used:

• Temperature
Depending upon the cooling method used, wet bulb temperature as well as dry bulb
temperature must be given. For the transmission line, the temperature distribution along
the route must be specified.

• Insolation
Insolation or sun-radiation is needed not only for estimation of the power demand for
cooling and hall ventilation systems, but also for the heating of line conductors which is
an important factor in determining the resistance of transmission lines.

- 465 -
• Wind speed and direction have similar though opposite effects as sun radiation.

• Weather
Weather conditions are important for the determination of corona and leakage losses of
HVDC overhead lines. Thus it might be specified that fair weather is to be assumed for
the entire length of the line or fair weather for 80% of the line and rain for 20%.

In general it is assumed that the environmental conditions prevail long enough for all
components to reach their end temperature.

AC network conditions on both sides must also be established. For purposes of


determining losses, it is usually assumed that the busbar voltage corresponds to the rated
value, network frequency corresponds to nominal value and that the negative sequence
system has the amount zero.

6.2.9.2 Load Cases

In determining system losses for the specified load cases, the following assumptions
should be made in addition to the above listed ambient conditions:

• For quantities regulated through transformer tap changers (control angle, dc voltage, if
applicable) the mean values between the limits which trigger switching should be used.

• AC filter circuits and reactive power units should be considered activated to the extent
they are needed in the particular load to establish the specified reactive power balance
and to meet the distortion limits.

• If electronic reactive power regulation is used at minimum load, the enlarged control
angles, the transformer tap setting and the resulting modified dc-side parameters must
be taken into consideration in calculating line losses.

• For no-load stand-by condition (the system is prepared to assume power transmission),
unless otherwise specified, the following assumptions apply: Converter transformers
energized, valves under voltage but blocked, ac filter circuits and reactive power units
disconnected, all auxiliary systems active, ventilators and pumps running at the lowest
level, hall ventilation and air conditioning system activated.

- 466 -
• In monopolar systems and bipolar systems for which loss determination is required for
monopolar operation with ground return, losses of earth electrodes must be determined
using the transition resistances applicable in continuous operation, or at the end of half
the time specified for emergency operation. For electrode lines, end temperatures
corresponding to the current and ambient conditions are assumed.

6.2.9.3 Loss Sources

For the specified load cases under the assumptions given above for specified
environmental and operational conditions, the losses of all significant loss sources of the
HVDC system are to be determined. Often, individual losses are required. The guarantee,
however, should always apply to the total losses of the system which are to be stated in the
bid or the contract. Significant loss sources should include:

• Converter valves
In converter operation, losses occur mainly in the thyristors. Losses due to their
differential resistance are proportional to the square of the current; losses due to their
threshold voltage are proportional to current, switching-through losses occur at gating
and losses due to the carrier storage effect during extinction (see Section 2.4.2).
Additional losses occur in in the RC snubber circuits and in saturable valve reactors.
The determination of all these different losses is a very complex task. In practice,
conversion methods have proven to be useful which are based on calometric loss
measurements performed in a module test circuit (i.e. in actual converter operation).
Original elements are used but are limited to one module. Thus, six to ten thyristors in
series are used per valve branch. Additional loss sources are current heat losses in valve
buses. According to IEC 919, the power of the valve cooling system is to be included in
valve losses to the extent it is needed for the load case under consideration.

• Converter transformers
No-load losses are verified in customary test field measurements. Increase of losses as a
result of dc premagnetization, particularly at minimum power, can only be determined
mathematically. For load losses, the additional losses caused by harmonic currents must
be taken into consideration by means of selection of a higher fundamental current for
test field measurements. The earlier IEC 146 recommended that the valve current shall
be assumed to be rectangular (neglecting commutation overlap) leading to an increased
fundamental r.m.s. value. This method has proven to be inadequate for HVDC
transformers. Until a new IEC publication is available, the transformer manufacturer
must perform the calculation of an adequate test current incorporating the

- 467 -
considerations of CIGRE-WG 14-12. The cooling system power needed for the
contemplated load case must also be included in the transformer losses.

• Smoothing reactor
Here also the additional losses caused by dc-side harmonic currents can only be taken
into consideration by mathematically increasing the test dc current by an appropriate
amount. With forced cooled reactors the inclusion of cooling system power is
necessary.

• AC filter circuits and capacitor banks


In loss determination, it is assumed that 100% of the characteristic and
noncharacteristic harmonic currents generated at a particular load by the converters are
flowing into the ac filter circuits and capacitor banks which are connected in the
particular case, and that no additional harmonic currents flow in from the ac network.

• DC filter circuits
In addition to capacitor coils, the interior and/or exterior discharge resistors should be
considered for direct voltage-caused losses. With respect to losses caused by dc-side
harmonic currents in capacitors, reactors and resistors, in contrast to the ac filter
circuits, only the harmonic currents which actually flow into the filter circuits have to
be considered. This includes, however, currents flowing from the other station. Thus
non-harmonic frequencies may also be included.

• Other components of HVDC stations


In addition to auxiliary energy demand (reduced by the cooling system power included
in the equipment losses), losses in measuring transformers, bus and cable connections,
etc., have to be estimated.

• HVDC transmission line


In addition to current heat losses, corona losses are important. Leakage current losses
only play a role if a significant portion of the line will be subject to bad weather
conditions and significant insulator contamination is to be expected.

- 468 -
6.3 Transient Performance of HVDC Systems

6.3.1 Definition

The transient performance of an HVDC system as described here refers to the course of
transient events such as

• Switching processes in normal operation


• Disturbances caused by malfunctions
• System faults caused by short circuits, etc.

The course of an event includes the event itself, the “natural” reaction of the HVDC
system caused by the event, and finally the intervention of protection functions to limit the
effects of the event.

Transient performance extends to the point of time at which a quasi-stationary condition is


achieved, including fault clearing and restoration or continuation of transmission.
Optimization of operating data, reconfiguration of high voltage circuits and similar actions
are part of the dynamic performance which is described in Chapter 6.4.

The following discussion is limited to HVDC two-point systems. Aspects of the transient
performance of multiterminal HVDC systems are addressed in section 6.3.8.

6.3.2 Switching Processes in Normal Operation

In this section, all switching processes on the ac and dc side of an HVDC station which are
not triggered by disturbances or faults are discussed. The transient performance of an
operating HVDC system during and directly following the switching process and the
measures which can be taken to improve the transient performance, particularly avoidance
of consequential failures, are of particular interest.

6.3.2.1 Connection and Disconnection of AC-Side Components

During the operation of an HVDC system, components on the ac-side such as ac filter
circuits, capacitor banks, shunt reactors, and static compensators are occasionally
connected and disconnected. In the case of reactive power regulation, some components
are frequently connected and disconnected. The excessive current and voltage stresses
caused by such switching processes must be considered in the design of components, but

- 469 -
they are not the subject of this discussion. Converter transformers are also occasionally
connected or disconnected during transmission, particularly in the start-up or shut-down of
an individual converter group.

Such switching processes cause changes and distortion in the ac busbar voltage of an
HVDC station, effected by energization, transient responses, saturation processes, or
disturbance of the reactive power balance. They can have a negative effect on the
operation of an HVDC, particularly the commutation processes of converter valves. Such
influences are particularly critical when there is a low short-circuit ratio (SCR) at the
station busbar.

If possible, the surge arresters used for overvoltage protection should not intervene at
switching actions in normal operation. If necessary, the breakers can be equipped with
closing resistors for the limitation of switching-in transient overvoltages. Another
possibility is synchronous switching. This technique involves the control of phase-correct
closing moments in the three separately controlled poles of the breaker. By such control,
switching-in overvoltages can be largely avoided. It is furthermore mandatory that ac filter
circuits and capacitor banks be switched off without flashback, since hazardous
overvoltages can otherwise be generated.

In each case, the switching procedures of ac-side components of an HVDC station must be
carefully studied during the planning of a system. This study should take into
consideration the most unfavorable network configuration and the various operating
conditions of the HVDC system. The connection and disconnection of ac lines extending
from the ac busbar of the station must also be studied with respect to the transient
performance of the HVDC system. The changes in the short-circuit ratio (SCR), harmonic
impedance, and phase angle of the applied ac voltage associated with switching operations
must be taken into consideration.

The results of the investigations must be incorporated in the insulation coordination of the
system (see Chapter 7). A cost optimization between the expense for protective measures
against overvoltage on the one hand and the cost of designing components with respect to
voltage on the other hand should be made.

6.3.2.2 Connection and Disconnection of a Converter Group

In addition to the switching processes of converter transformers discussed above, the


activation and deactivation of a converter group can cause transient processes on the ac

- 470 -
and dc sides of an HVDC station which are very similar to those of switching actions. This
is especially true when an HVDC station has two or more series-connected converter
groups for each pole. The procedures for activation and deactivation of a group during
transmission operation, that is, under conditions of flowing direct current, are described in
Section 5.5.2.

In one of the processes, a changeover is made directly from normal commutation to by-
pass operation or vice versa. As a result, a reactive power jump occurs at the ac busbar in
the amount of the reactive power demand of the converter group which was connected or
disconnected. The disturbed reactive power balance then must be re-established by
disconnection or connection of reactive power units. The other process provides for
continuous change from normal to zero-voltage operation or vice versa. In zero-voltage
operation, the converter assists with the commutation of direct current into the by-pass
switch or vice versa. This process, however, causes a change in the reactive power demand
of this group of between zero and approximately double the normal demand. This process
continues until normal operation is achieved. It is nevertheless advantageous that the
changeover occurs gradually so that it can be synchronized with the switching of reactive
power units.

In comparison, the activation or deactivation of one of several parallel-connected


converter groups does not cause any significant transient processes since the blocking or
deblocking always takes place under conditions of zero or very low current.

6.3.2.3 Load Rejection and Inverter Separation (Islanding)

The transient events associated with activation and deactivation of a single converter
group are much more severe in the event of a sudden load rejection of an HVDC pole or
the entire HVDC system. The causes for a load rejection of the HVDC system in normal
operation can be:

• Unintended triggering of an ac breaker in the HVDC station


• Blocking of the converter as a result of a malfunction in the control system
• Events in the other station or its vicinity

Though such events may be rare, they must be considered in the design of the protection
system. Operation of surge arresters will be tolerated, but no thermal overstressing of the
arresters. Therefore temporary overvoltages triggered by momentary excess reactive power
in combination with transformer saturation and mutual reactions with the filter circuits

- 471 -
must be reduced to safe levels in the shortest possible time. The measures available for
this purpose will be discussed in Section 6.4.2.4.

A special case of load rejection is the separation of the inverter from its ac network the so-
called islanding. In most cases, the inverter will be able to continue operation with the
commutation reactive power being supplied by the ac filter circuits. The energy
transmitted has no path except into the capacitors of the filter circuits and capacitor banks.
The voltage of these capacitors rises rapidly. The overvoltage arresters would soon be
destroyed if the energy flow is not immediately being interrupted. The fastest means by
which this can be accomplished is to force the converter valves into by-pass operation.

The cause of the inverter separation can be the unintended triggering of the ac breaker(s)
of the line(s) extending from the busbar of the HVDC station within the station itself, but
also can originate at the far end of the line(s). While the first case can be reliably detected
without any delay, the latter case is problematical since transmission of information over
the available telecommunication system is often too slow. In this case, the event must be
detected on the basis of quantities available at the station, primarily the ac busbar voltage.

6.3.2.4 Switching Actions on the DC Side

Depending upon the purpose of an HVDC system and its configuration, the converter
stations may have a more or less extended switchgear. To handle the great variety of
switching actions in normal operation, which primarily serve to make changes in the
configuration of the dc circuit, a series of dc switching devices, as was described in
Section 2.7, are available:

• Disconnect switches
• Earthing switches
• Commutation switches
• Breakers

Disconnect and earthing switches must often have certain switching-in and commutation
capabilities which are called upon during connection and disconnection of dc filter
circuits, connection of an inverter pole, grounding of a line pole, etc.

Commutation switches must commutate the entire system direct current from a current
path which is to be disconnected into an parallel current path. The best known application
is the by-pass switch.

- 472 -
Breakers have to perform current interruption and current commutation functions with
considerable levels of energy dissipation. Their primary area of application are
multiterminal systems with redundant dc lines. They are also needed, however, in simple
bipolar two-point systems as metallic return transfer breakers (MRTB).

All dc-side switching actions during normal operation are part of automatic control
sequences. To the extent necessary, they are coordinated with the regulation functions of
the converter. In this way, abrupt changes of operating quantities of an HVDC are largely
avoided. An exception to this was noted in Section 6.3.2.2.

All switching actions on the dc-side, to the extent they do not take place under the
condition of both, zero current and zero voltage, are accompanied by transient effects.
These effects, however, are scarcely noticeable in the ac networks or along the HVDC
transmission lines. Therefore they do not influence transient performance of an HVDC
system.

6.3.2.5 Start-up and Shut-down of the HVDC System

The discussion above also applies to the processes of start-up and shut-down of the HVDC
system. These processes take place in the framework of an automatic control sequence in
close coordination with regulation functions of the converters. The changes in power occur
gradually so that these processes are better classified as steady-state rather than dynamic
performance of the system.

This does not apply for the range of power around P=0. After transmission power has been
reduced to Pmin, the converters are blocked. This results in a power jump which should be
classified as a transient event. In the same way, a power jump from zero to Pmin occurs
during the start-up process at the moment of the unblocking of the converters. Since these
power jumps are relatively small, they do not present any particular problems or stresses.

- 473 -
6.3.3 AC Network Faults

The transient performance of an HVDC system during a fault in one of the ac networks
and the recovery following fault-clearing are important criteria for the design of system
components and the optimization of control and protection functions.

6.3.3.1 Fault Categories

AC network faults can be classified according to place of occurrence and type of fault
which materially determine the transient performance of the HVDC system. These
categories include:

• Fault in the network at the rectifier side or the inverter side


• Fault directly at the busbar of the HVDC station or at some distance (usually expressed
in terms of residual voltage in p.u. measured at the station busbar).
• 3-phase network short circuits with or without ground contact
• 2-phase network short circuits with or without ground contact
• Single-phase-to-ground faults
• Duration of fault until clearing
• Fault clearing by means of permanent disconnection of the fault location or by 3-phase
or single-phase auto-reclosure
• Mutual effects of ac and dc overhead line faults (when the lines use the same right of
way or that they intersect at a small angle); extreme case is the galvanic contact upon
the breaking of a cable

6.3.3.2 Criteria of Transient Performance

Various criteria can be defined for the evaluation of the transient performance of an
HVDC system during an ac network fault and immediately following fault clearing. These
criteria can usually be found in the specifications for HVDC systems.

• Transmission capacity of the HVDC during the fault


• Recovery time following fault clearing
• Reactive power performance during the fault and during recovery
• Performance in the event of fault-caused load rejection
• Harmonic performance during the fault and the period of resumption of operation

- 474 -
In applying these criteria to the transient performance of a planned HVDC, the full range
of operating and environmental conditions should be considered.

• Network conditions
The admittance of the ac network as seen from the busbar of the HVDC station should
be used as the criterion rather than the better known short-circuit ratio (SCR). One very
important parameter associated with this is the impedance angle.
• Busbar voltage and network frequency
• Transmission power of the HVDC system from minimum power to maximum
continuous overload
• Special operating conditions such as monopolar operation or operation with reduced dc
voltage

In simulation runs to ensure compliance with the specified transient performance, all
criteria should be considered under all operating and environmental conditions. Time and
cost constraints, however, require a well-thought limitation to cases which are critical in
actual practice. Experience and good documentation of earlier projects are valuable means
in making this selection.

6.3.3.3 Power Transmission During the Fault

The operation of an HVDC converter is relatively sensitive to ac network faults, even if


these faults occur electrically far away from the HVDC station. Voltage drops at the ac
busbar of a rectifier station cause a corresponding drop in dc voltage. Due to the transition
of current regulation to the inverter this generally results in a more than proportional
reduction in transmission power. This effect is drastically amplified if the dc voltage drop
results in an intervention of the VDCOL (Voltage Dependent Current Order Limitation)
function.

Voltage drops at the ac busbar of an inverter station of 10% or 15% (depending upon the
extinction angle g) lead unavoidably to commutation failures which mean a transient
interruption of power transmission. Measures for rapid resumption of commutation are
required so that permanent commutation failure is avoided. These measures may consist of
a momentary increase of the angle of advance b so that the extinction angle g will be
increased to ensure commutation of current to the following valve despite the rise of dc
current Id caused by the fault and despite reduced valve voltage Uv.

- 475 -
This measure, however, entails a material reduction of system dc voltage and (through the
action of the VDCOL function) of direct current. Thus transmission power is greatly
reduced.

Transmission power of the HVDC system during an ac network fault can be maximized
by optimizing the measures described above including establishment of appropriate
activation criteria for the VDCOL function.

During the uncommon 2-phase and the common single-phase network faults, a
pronounced unsymmetry of voltage at the ac busbar of the HVDC station occurs. The
achievable transmission power of the HVDC system then depends upon the principle of
control pulse generation utilized by the triggering device of the converter. If the principle
of equidistant control pulses is retained, the fixed control pulse pattern aligns itself with
the furthest lagging valve voltage in the rectifier and with the furthest leading valve
voltage in the inverter. These two effects cause an unnecessary enlargement of the control
angle of all other valves and thus an unjustified large drop of dc voltage.

This dc voltage drop can be reduced and thus transmission capacity can be increased by
abandoning the principle of equidistant control pulses for the duration of the fault in favor
of the “equal angle principle,” in which the same control angle a or b is set for each
individual valve in relation to the associated valve voltage without regard to the time
interval between individual control pulses.

In the inverter, this procedure is particularly difficult because the control angle ß necessary
for reliable current commutation must take into consideration the commutation overlap
which is strongly dependent upon valve voltage.

It is possible using the above measures to increase the transmission capacity of the HVDC
system during an ac network fault. The robustness and simplicity of the control system of
the HVDC, however, will suffer through the use of these measures. Therefore, they should
only be used if essential requirements of a specification cannot otherwise be fulfilled.

- 476 -
6.3.3.4 Recovery

Because of the sensitivity of an HVDC system to ac network faults, rapid recovery of


transmission operation following fault clearing is particularly important.

Pd Pd
1
Pd*

0,9 * P d* 2
a) b)
0,9 * Pd*

tw tw1
t tw2 t

Figure 6.9: Resumption of power under conditions of


(a) strongly damped transition behaviour
(b) weakly damped transition behaviour

tW = recovery time

For this reason, there will be requirements in the specifications for all HVDC systems
concerning the maximum permitted recovery time.

Recovery time usually means the time from the moment of the clearing of the fault
through attainment of 90% of the transmission power prior to the fault. If the recovery
process is strongly damped, as presented in Fig. 6.9(a), definition of recovery time tW is
simple. In weakly damped transition behaviour, as presented in Fig. 6.9(b), the first
instance of exceeding the 90% value is considered the end of the recovery time only if
there will be no subsequent transient drops below the 90% value. Otherwise, the second
(final) rise above the 90% value is considered the end of the recovery time (in Fig 6.9b,
tW2 rather than tw1).

If an HVDC system with short dc lines or a back-to-back link is connected on both ends to
an ac network with a high short circuit ratio (SCR > 5), recovery times of 50 to 100 ms
will be achievable. If the network connection on one or both sides has a short-circuit ratio
of SCR < 3, and/or the HVDC system has a long dc cable or a very long dc overhead line,
the achievable recovery times will be a multiple of the above value.

- 477 -
During the last 10 years, permanently shortened recovery times have been specified. This
is also true with HVDC systems with long HVDC lines and low short circuit ratios.
Conforming to such requirements often necessitates the use of special or even tricky
control techniques which are usually optimized for the operating conditions which are
defined in the specification as verification. This does not serve either the general transient
system performance nor the system reliability.

In establishing or monitoring recovery times, the stability of the ac networks should also
be considered, both in the case of rapid fault clearing (first protective level) and in the case
of delayed fault relaying (back-up protection). In some circumstances, very rapid recovery
of transmission power can result in intensification of fault-caused electromechanical
oscillations to the point of violation of the stability limits. (See Section 6.4.3).

It is an absolute requirement that the recovery process does not lead to an inverter
commutation failure and thus leading to a renewed interruption of power transmission.

Maintaining of the flow of direct current and uninterrupted generation of control pulses for
the converter valves during the fault are very important measures for the achievement of
short recovery times. This is true even if valve voltage has dropped to very low values or
even to zero. By this means, undelayed commutation of current is realized upon the return
of ac voltage. Thus the reactive power demand in the converter is also reestablished
immediately. A load rejection overvoltage and associated transformer saturation with the
resulting phenomena of ferro-resonance and voltage distortion are largely avoided. The
prerequisite for this is that the energy storage capacity of the thyristor electronics will be
adequately sized.

6.3.3.5 Reactive Power Behaviour in the Event of AC Network Faults

Rapid, successful re-establishment of stable power transmission following an ac network


fault depends in no small part upon the reactive power behaviour of the HVDC stations
during the fault and the recovery process. The goal for the station not affected by the fault
is to keep the reactive power balance somewhat constant. If the VDCOL function is
suitably designed, this goal can be achieved by means of retention of the system direct
current at a reduced valve which is to be ensured by this station, whether it functions as a
rectifier or an inverter.

- 478 -
In the station affected by the fault, immediate re-establishing of commutation and thus
absorption of reactive power by the converter is critical for an acceptable reactive power
balance during the recovery process.

A special fault-clearing strategy is necessary in the case of distant ac network faults on the
inverter side in which the voltage drop has resulted in commutation failures but which still
permits inverter operation at reduced dc voltage. In this case, buffering action by the
inverter, that is the resumption of commutation and the prevention of further commutation
failures has absolute priority over the maintenance of a particular reactive power balance.

On the whole, however, the attainment of an acceptable reactive power behaviour of the
HVDC stations during ac network faults and at recovery is among the most important
tasks in the framework of system optimization in which the goal is a satisfactory transient
system performance.

6.3.3.6 Load Rejections as a Result of AC Network Faults

If the recovery process following an ac network fault cannot be initiated or if it fails, the
entire system is confronted with an abrupt interruption of energy transmission. This is
referred to as "fault-caused load rejection." Among the reasons for this type of fault are the
following:

• The ac busbar of the HVDC station continues to be separated from the associated ac
network (“islanding”).

• The inverter does not correct itself (does not resume commutation) and passes into the
condition of continuous commutation failure.

• The transient events have resulted in an overload-caused disconnection of important


components of the HVDC station.

• Consequential faults or multiple faults have caused a permanent blocking of the


converters as a fault reaction.

Depending upon the short circuit ratio at the connection point of the HVDC station, very
significant temporary overvoltages can arise as a result of the excess reactive power
caused by load rejection. This results in a saturation of converter transformers. Ferro-

- 479 -
resonance can occur which causes significant voltage distortions which in turn can lead to
operation of surge arresters or to false tripping of protection relays.

Load rejection overvoltage is to be incorporated into the insulation coordination and


therefore must be considered in the sizing of components. For this reason, means of
avoiding these overvoltages or, at a minimum, reducing them are extremely important.

Measures employed may include the following:

• Disconnection of all or some of the reactive power units (ac filter circuits, capacitor
batteries)

• Activation of shunt reactors

• Installation of overvoltage limiters (surge arresters with a low residual voltage but high
energy absorption capacity) which are activated by means of breakers.

The first of the measures listed above depends, however, upon the ac breakers being able
to reliably switch off the increased capacitive power which is due to the overvoltage. Here,
“reliably” requires that the possibility of strike-backs is eliminated.

In the case of the last measure listed, it must be noted that the overvoltage limiter may be
switched on only for brief periods, namely until its energy absorption capacity has been
reached. Permanent reduction of the overvoltage must be achieved by other means.

6.3.3.7 Non-characteristic Harmonics Resulting from AC Network Faults

Multiple commutation failures of an inverter resulting from an ac network fault and also
unsymmetries of the ac voltage system during single-phase network faults result in
noncharacteristic harmonics of considerable amplitude. The low frequency harmonics in
particular can excite resonance circuits on the ac and the dc sides which in turn contribute
to a reinforcement of the phenomenon to the point of harmonic instability (see Section
6.4.5.2).

These interactions are particularly critical if there is a series resonance for the network
frequency on the dc-side. Harmonic currents of the ordinal number m = 1 cause harmonic
currents on the ac-side with the ordinal numbers v=0 and v=2.

- 480 -
The harmonic current v=0 is a dc component which causes pre-magnetization of the
converter transformer which in turn causes the occurrence or the amplification of even-
numbered low frequency harmonics. The cycle of effect is closed for the harmonic of the
ordinal number v=2, and the danger of harmonic instability is present.

6.3.3.8 Measures for Improvement of the Transient Performance in the Event of AC Network
Faults

In addition to the measures described in the preceding sections, it can be useful or even
necessary to temporarily discontinue the functioning of power regulation or control during
an ac network fault and the subsequent recovery process so that the transient performance
of the HVDC system is determined by the current regulation of the rectifier. It may also be
necessary to switch from extinction angle regulation in the inverter to constant dc voltage
regulation.

The appropriate regulation strategy for optimization of the transient performance during
and after ac network faults must be found or at least confirmed in simulator tests.

It will often be found that the optimal regulation circuit for normal operation will not
produce a satisfactory system performance for all kinds of ac faults. If a temporary change
of regulator parameters is necessary for optimization of transient performance, clear
criteria for making the switch-over must be found. Under some circumstances, self-
adapting regulators can be an elegant solution.

6.3.4 Converter Faults

Here the term “converter” means the area of an HVDC station between the bushings on
the network side of the converter transformers and the converter-side connections of the
smoothing reactors and the neutral bus connections of the converters. Faults will include
both, short circuits in the converter area and converter malfunctions.

The discussion of the transient system performance will be concentrated on the effects of
such faults on the HVDC operation and on the ac networks. Measures for limiting the
effects of faults and for the rapid resumption of transmission, though perhaps at reduced
power, will also be considered.

- 481 -
6.3.4.1 Short Circuits in the Converter Zone

The term “short circuit” will mean in this context the breach of external or internal
insulation, i.e. flashover or puncture of insulators and also the galvanic short circuit due to
an unintentionally closed switch.

Fig 6.10 shows seven different short circuit locations in a 12-pulse HVDC converter
group.

Figure 6.10: Short circuit locations in an HVDC converter group

Fault location #1 in the figure can involve either an external valve flashover or a valve
break-through, that is, the loss of blocking capability of all thyristors of the valve.
Flashover or penetration of the parallel-connected valve arrester would also fall into this
fault category. Fault location #2 includes internal transformer insulation faults of the
valve-side neutral connection, and also exterior arcing of the neutral connection with
single-phase units, as well as penetration of the neutral bushing. The other fault locations
can be interpreted analogously. Fault location #7 also includes the unintended closing of
the by-pass switch of one of two or more series-connected converter groups of a station
pole.

Short circuits in the converter zone require immediate disconnection and — since there is
always a possibility of damage or destruction — visual inspection. The disconnection is
generally limited to one station pole. If a station pole has two or more converter groups, it

- 482 -
depends upon the short circuit location whether the selective disconnection of one group
will clear the fault.

As soon as the short circuit has been detected by the converter differential protection, the
valves are immediately blocked. The effectiveness of this method depends among other
things upon the location of the short circuit. Since there is no certainty that the short circuit
current can be suppressed by the blocking of the valves, the ac breaker of the converter is
triggered simultaneously. Since in many cases the short circuit current flows through one
or several valves, the disconnection must take place so quickly that the blocking capability
of the thyristors is not lost due to current-related elevation of the junction temperature.

The consequences of a short circuit in the converter zone are thus always the loss of
transmission capacity of one converter group or more, in most cases of one station pole,
for a considerable period of time. Only the fault itself and the transition into transmission
operation with reduced power are part of the transient system performance.

6.3.4.2 Converter Malfunctions

The primary function of the converter is cyclical commutation of the system direct current
between the valves of a commutation group (Section 1.3.1). Prerequisites for proper
commutation are a sufficiently high valve voltage Uv and control pulses synchronized to
the valve voltage with a delay angle a set by the control system.

The causes of malfunctions can be missing or insufficient valve voltage as a result of ac


network faults as presented in detail in the previous section and also lack of control pulses
as a result of faults in the control system, although these systems are configured with
redundancy.

The effects of missing or reduced valve voltage are relatively minor in rectifier operation.
Transmission power is reduced or completely stopped but returns immediately when the
valve voltage returns. Effects of periodic loss of individual control pulses, however, can
have particularly unpleasant consequences. An ac current component of network
frequency is superimposed upon the direct current. In the presence of resonance, this ac
component can become very large. As was mentioned in the previous section, this results
in amplified even-numbered harmonic currents on the ac side and transformer saturation.
This can result in harmonic instability. A fault of this kind must be quickly detected and
the converter group must be disconnected.

- 483 -
In inverter operation, a converter malfunction always means a commutation failure. This
results in a partial or even a total short circuit of the dc circuit. The short circuit current
must be limited and brought down by the rectifier, thus energy transmission comes to a
halt. Therefore quick countermeasures which facilitate re-activation of the inverter, that is,
resumption of commutation, are important. In order to prevent a persistent commutation
failure, an increase in the inverter angle of advance ß combined with a reduction of direct
current (automatically effected by the VDCOL) are appropriate countermeasures. If the
inverter cannot be re-activated, for example, because there are no control pulses, the
disturbed converter group must be disconnected from the dc circuit, because it could
otherwise cause an ac component in the direct current circuit with the effects described
above.

The first step in disconnecting a converter group is always activation of a by-pass in the
converter group. By this measure it is ensured that the ac-side circuit breaker of the group
to be disconnected has not to interrupt a dc component.

With careful project planning, adequate sizing of clearances in the converter zone, and
complete redundancy in the control system, converter faults will be extremely rare.
Nevertheless, they must be considered in system design. They should be incorporated into
the protection coordination.

6.3.5 DC-Side Station Faults

The dc side of an HVDC station is connected directly to the converter zone and extends to
the terminals of the dc line and of the electrode line. It includes smoothing reactors, dc
filter circuits, the complete dc switchgear, the dc-side measuring transformers, power line
carrier coupling elements, etc. DC-side station faults which determine the transient
performance of the HVDC system include the following:

• Station pole-to-ground faults


• Station pole-to-neutral bus faults
• Neutral bus-to-ground faults
• Equipment faults
• Malfunctions of dc switching devices

The first three fault categories include interior and exterior faults of insulation, that is,
arcing and penetration, which usually occur as transient or permanent short circuits,
respectively, and accordingly require different fault clearing strategies.

- 484 -
6.3.5.1 Protection Zones

In harmony with the protection philosophy generally applied in ac systems, the dc side of
an HVDC station is subdivided into protection zones as is illustrated in Fig. 6.11 showing
a bipolar HVDC station which is a portion of a two-point long distance transmission
system. Since dc current transformers usually have only a single measuring system, it is
not possible to achieve an overlapping of protection zones. Nevertheless, protection of the
zone without any gaps is insured by overlapping protection functions. Each device and
each section of bus bar is protected by at least one, and usually by several protection
functions.

Figure 6.11:
H V D C Li n e Protection zones of the dc side of
a bipolar HVDC station
DC-
Fi l t er GRTS

M RT B El e c t r o d e Li n e s

One peculiarity should be remembered. The protective functions to be activated are


dependent not only upon fault location and fault category, but also upon the operating
mode of the converter. An inverter cannot execute a protective intervention in the event of
a station pole-to-ground fault. Instead, it must depend upon the assistance of the rectifier,
thus of the other station. The protection function activated in the other station, however, is
autonomous. That is, it does not depend upon a triggering signal transmitted from the
inverter by telecommunications. To this extent, there is compliance with the basic rule that
telecommunications may be used for optimization of fault clearing, and thus for the
recovery process, and possibly also for more precise fault diagnosis, but they must never
be a prerequisite for reliable protection.

- 485 -
This protection function assumed by the rectifier for the dc-side of the inverter station is
initiated by travelling wave line protection. In this way, the consideration is given to the
possibility that the fault is of transient nature, i.e., an exterior arcing , which can be cleared
by suppressing the system direct current and maintaining a zero-current pause prior to the
ramp-controlled recovery. In the case of faults on the dc side of a rectifier station, this type
of protection and fault clearing function must be specially implemented since travelling
wave line protection does only react to faults outside of the station.

6.3.5.2 Station Neutral Bus Faults

There is one necessary exception to the philosophy of the complete independence of the
two poles of a bipolar HVDC system. There is only one earth electrode per station. Thus
there is one small zone on the dc side of the HVDC station which is common to both
poles. This zone is the exit of the electrode line with the associated switching devices,
including the metallic return transfer breaker (MRTB). However, since in normal
operation of a bipolar HVDC system the earth electrode is not needed, a fault in this zone
can be eliminated by closing the disconnecting switch to station ground (see Fig. 6.11) so
that transmission can be continued without limitation. In this mode of operation a
permanent fault in one pole will of course force the shutdown of the entire system, since a
continuous direct current through the ground mat of the HVDC station cannot be
permitted. This measure does present the possibility that a shutdown of the entire system
to eliminate the fault can be deferred to a suitable point in time (scheduled maintenance as
defined in Section 6.5).

6.3.5.3 Smoothing Reactor Faults

If oil-insulated smoothing reactors are used, most of the protection functions and devices
associated with transformers can be applied:

• Differential protection
• Buchholz relays
• Oil level and pressure monitoring
• Temperature monitoring

Smoothing reactor bushings permit the installation of dc current transformers without


additional insulation cost. This means an economical use of differential protection is
enabled which may be supplemented by overcurrent time protection as a back-up.

- 486 -
The situation is different in the case of air-insulated dry-type smoothing reactors. Usually
there is no special device protection for this type of reactors. Instead, full reliance is placed
upon the station pole differential protection which compares the currents in the converter
output (measured by the dc transformer in the valve hall bushing) and in the line exit by a
free-standing dc current transformer.

6.3.5.4 DC Filter Circuit Faults

A short circuit in the zone of the dc filter circuits will be detected by the station pole
differential protection. Current transformers provided in the neutral bus connections of the
filter circuits (see Fig. 6.11) facilitate a classification of filter circuit faults. External
insulation faults (i.e. arcing toward ground) can be cleared at this point by interrupting the
transmission current and recovery following a zero-current pause. Permanent faults
(including the response of the triggering step of filter circuit monitoring system) require
disconnection of the affected filter circuit. This generally is accomplished by means of
disconnecting switches in zero-voltage condition. The prerequisite for continuation of
transmission in the affected pole is a redundancy in the dc filter circuits or permissibility
of temporarily increased harmonics in the direct current lines.

6.3.5.5 Malfunctions of DC Switching Devices

Despite the integration of switch activation into monitored, redundantly configured control
functions and supplemental latching, the possibility of a malfunction of one of the dc
switching devices can never be completely avoided. The peak values of a fault current
occurring upon unintended closing of a dc switching device are limited to 2.4 to 4 times
IdN due to the inductance of the smoothing reactor combined with the protective function
of rectifier current regulation. Thus these peak values do not present a danger to the
components of the HVDC station. Such an event, however, will cause the loss of the
affected pole until the fault is eliminated.

In contrast, unintended opening of a dc switching device can result in destruction. For this
reason, special attention should be devoted to reliable latching of such switching
processes.

The dc-side of an HVDC station, particularly the dc switchgear, should be planned and
executed in such a manner that no malfunction and no fault event, not even a conductor
break, can result in a short circuit of both poles. For this reason, the function of a switch

- 487 -
located between the poles, such as a polarity reversal switch, must be split between two
series-connected switching devices with provision for intermediate grounding at the
connection point. As an example, Fig 6.12 depicts the dc switchgear of a bipolar station of
a multiterminal HVDC system with polarity reversal switches for the change of energy
direction.

Figure 6.12: DC switchgear of a bipolar station in a multiterminal HVDC system

This ensures that the malfunction of a single switching device such as arcing across the
gap will never result in a bipolar short circuit and thus in a total loss of the HVDC station.

6.3.6 HVDC Line Faults

Faults of the HVDC overhead line are of special significance for the transient performance
of an HVDC system. This is due to the great length of the line which is subject to the
environmental influences and to the special sensitivity of the line insulation against dc
voltage.

HVDC sea cable are endangered by ship anchors and drag nets of fishing vessels, as has
been demonstrated in the operation of many systems. Cable faults, however, are not
transient events. They represent the permanent loss of the cable with very long repair
times. Therefore, only HVDC overhead lines will be considered in this section.

- 488 -
6.3.6.1 Transient Line Faults

The following are possible causes of transient faults on HVDC overhead lines:

• Direct lightning strike


Although HVDC overhead lines are probably always equipped with a lightning shield
wire, the possibility of a direct lightning strike in a line pole cannot be completely
excluded. The direct lightning strike can be considered a rare event, and it involves
only one pole of the line. The reason for this is that the potential difference between the
two poles of an HVDC line of approximately 1 million volts is on the order of
magnitude of atmospheric charges. A discharge will therefore always take the path to
the line pole with the opposite polarity, as experience has confirmed.

• Back flashover
Depending upon climatic conditions along the line, particularly the keraunic level
(number of lightning strikes per year/ per square km), it may be necessary to deal with a
large number of lightning strikes every year on the lightning shield wire, the towers, or
the soil in the immediate vicinity. Unless the footing resistance of the tower is very low,
the tower potential can increase to such an extent that the lighting surge withstand
voltage of the line insulators is exceeded. This results in a so-called back flashover.
Due to the reasons given above, this fault can only affect one line pole.

• Contamination flashover
If the HVDC line passes through a region with natural contamination (sand, dust,
aerosols of plant origin) or of man-made contamination (agriculture, industry,
transportation), it is likely that contamination will constantly accumulate on insulator
surfaces. In the event of a thorough wetting of the contamination layer by dew, fog, or
drizzle, contamination flashovers can occur. DC voltage insulators are especially at risk
because the contamination layer accumulates more rapidly in the electric dc field
(electro-filter effect) and because a partial discharge which begins along the surface of
the insulator is not repeatedly interrupted by the periodic voltage zeroes.

• Overvoltages
When there is an arcing event of one pole to ground, the other pole of a bipolar line
experiences an overvoltage, which can be 1.4 to 1.9 times UdN due to the capacitive
coupling. Line insulation must be sized such that this type of overvoltage will never
result in arcing, because this would result in a bipolar fault. Other causes for

- 489 -
overvoltages can be the startup of the HVDC system with an open line end or a
resonance overvoltage in the event of impressed ac voltages (for example, due to a
missing control pulse of a valve in the rectifier station.

• Ionization caused by fire


If the right of way of an HVDC overhead line is not constantly kept free of high
growing grass and shrubs, especially in tropical regions, a fire in the right of way can
result in an intense ionization of the air so that a flashover from the line to ground can
occur, most commonly in the middle of the span. This situation represents an acute
danger of a bipolar fault.

• Line intersection short circuit


Occasionally, protection against a short circuit of the HVDC line with an intersecting
ac line is required. This kind of short circuit may occur due to a line or insulator break
of the higher positioned line. A transient fault, and thus one which must be quickly
cleared, however, is conceivable only in the case of a broken, falling conductor.
Otherwise, it is a permanent fault and it remains only the disconnection of both
systems.

6.3.6.2 Consequences of Transient Line Faults

Line flashovers are transient events. They do not cause a material loss of transmitted
energy. Since the magnitude and duration of the short circuit current is limited by current
regulation of the rectifier, damage need not be feared either to the conductors nor to the
insulators.

In the event of flashovers which are caused by lightning strike, by back flashes, or by
overvoltage, it is generally adequate to shut off the current in the affected pole for 100 to
200 ms in order to deionize the arcing channel sufficiently so that the first attempt to bring
up the voltage will be successful and transmission will be resumed within an additional
100 ms.

During a contamination flashover, the affected insulator will be sufficiently dried by the
flashover arc so that it will be able to sustain the returning voltage, but there will be many
other insulators with similar conditions of contamination and wetting. Therefore,
additional flashovers must be anticipated. After the second or third flashover, the system
should be brought up to a reduced pole voltage (between 70 and 80% of nominal voltage),
so that at least a portion of the transmission capacity can be retained.

- 490 -
If a flashover is caused by bushfire, the first recovery attempt can be successful, but the
next high flame can trigger another flashover. Even operation with reduced dc voltage
does not help in this situation. Either operation with a large number of flashovers in quick
succession must be accepted or transmission must be ceased for 10 or 20 minutes until the
fire in the right of way of the line has burned out.

6.3.6.3 Fault Sensing

The prerequisite for rapid fault clearing is quick detection and clear identification of the
fault as a line flashover.

Travelling-wave line protection has proved to be a suitable system for detection and
identification of transient line faults, and it has come into general use. The current and
voltage travelling waves emanating from the fault point are analyzed as they reach the
converter stations. In this way, it is possible to distinguish between a line fault and a
disturbance within a station or a fault in the ac network in the other station. Such a
distinction is easily possible even in the case of lines of 1000 km in length thanks the
smoothing reactors in the station poles. If operation with reduced dc voltage is provided,
however, the response values of the travelling-wave protection may have to be adjusted in
order to assure the selectivity.

If an HVDC system comprises line redundancy (i.e. if two parallel bipolar lines or a
double line connect the two stations), it will be necessary that a line fault be assigned to
one of the two line systems.

If no HVDC breakers are available, in the event of a line fault the affected pole of the
system will be shut down as described above and the recovery process will be initiated
after a zero-current pause. It is not necessary to associate the fault with one of the two
systems, but if the ramping is not successful (i.e. there is a permanent fault), assignment of
the fault to one of the two systems is necessary in order to separate out the faulty line
section. As shown in Fig. 6.13, such selectivity is not always assured. E.g. the travelling
wave protection of station A cannot assign a fault close to station B to either line 1 or line
2. The impedance to the fault location is practically the same over both line sections and
the period of time of the travelling waves is practically the same over both paths. In this
case, a discriminating signal must be sent from station B, where assignment is no problem,
over the telecommunication system to station A.

- 491 -
Figure 6.13: Sensing of line faults in parallel lines

This method of assignment in the event of line faults close to a station is also necessary
when HVDC breakers are used. In this case, shut-down of transmission and the zero-
current pause are not needed but clear association is necessary to determine the switches
which will cut out the line system containing a fault. In practice, it is possible in 80% to
90% of faults to immediately associate the fault with one of the parallel line systems, and
the tripping command can be issued immediately. Only in the remaining faults, close to a
station, it will be necessary to wait for the discriminating signal transmitted over the
telecommunication system.

In recent specifications, a fault location system, preferably with tolerances corresponding


to the distance between two line towers, has been occasionally required. In this way, the
evaluation of repeated transient line faults is facilitated. Flashovers which always occur on
the same tower, for example, may indicate a damaged insulator. Flashovers which always
occur in a specific area may indicate extreme weather conditions. As described in Section
5.11.5, determination of fault location is possible by means of highly accurate
determination of the arrival time of the travelling waves in the two stations and evaluation
of the difference in time. The fault location can be calculated from the time difference.
The requirements for accuracy are enormous, since a time tolerance of one microsecond is
the equivalent to a distance tolerance of 300 meters.

- 492 -
6.3.6.4 Design of Overhead Lines with Respect to Required Transient Performance

As demonstrated in Chapter 3, it will be possible to determine the flashover performance


of a planned HVDC system in advance if all environmental conditions along the planned
route of the line are known.

If the keraunic level (number of lightning strikes per square kilometer per year) is known,
the number of direct line strikes expected per year can be determined taking into account
the protective angle of the lightning shield wire.

By specifying the footing resistance of the tower grounding, the statistically expected
number of back flashovers can also be determined if the lightning surge withstand voltage
of the insulators is known.

Calculation of the number of anticipated contamination flashovers per year is more


difficult. Knowledge of the insulator contamination under the influence of direct voltage
expressed as the equivalent salt deposit density (ESDD) in mg/cm² and the number of
wetting periods per year must be known. These factors together with the specific creepage
distance provide the flashover probability of the insulator. By applying this figure over the
entire line lenght, the statistically expected number of contamination flashovers per year is
obtained.

Due to control, regulation and protection functions of converters and appropriate back-up
functions, flashovers as a result of overvoltages should be excluded from consideration.
The same applies to flashovers caused by fire under the lines. With careful maintenance of
the line route, they can be excluded.

Thus the fault events which are to be considered in the transient performance of an HVDC
overhead line concern only flashovers limited to one line pole. This means that the energy
transmission continues on the other pole. For this to occur, the neutral points of the
converter station must be connected with their earth electrodes so that the current of the
undisturbed pole can return through ground. If the system has a short-time overload
capability, the effects of the short-time interruption of a pole can be reduced. In
combination with control of the dynamic reactive power behaviour of the converter
stations (to be discussed in greater detail in Section 6.4), the interactions with the ac
network can thus be almost eliminated. Although the flashover of an HVDC line pole is a
relatively frequent fault event, it can be almost neglected with respect to the transient

- 493 -
system performance, at least for those fault events in which the first restart attempt is
successful, and thus the total duration of the fault is limited to < 300 ms.

6.3.7 Electrode Line Fault

Because the electrode line is a component associated to both poles, its fault performance is
very significant. It should be configured in such a manner that, at a minimum, permanent
failures are practically excluded. This means the following for the design of electrode
lines:

• The insulators must be configured such that no damage is caused even by frequent
flashovers ( caused by direct or indirect lightning strikes).
• A continued flow of direct current (caused by pole unsymmetry) in the range of < 1 A
to several tens of Amperes must not damage the insulator surface with its arc and must
extinguish itself. The design and material prerequisites for this are presented in Section
4.7.3 and in Fig. 4.15.
• Switching surge withstand voltage of insulators should be high enough that the voltages
induced in the electrode line by transient processes in the HVDC overhead lines will
not cause a flashover. This applies in particular for cases in which the electrode line is
mounted on the towers of the HVDC overhead line over great distances.
• Specific creepage distance of the insulators in relation to potential to ground at the
station neutral point during monopolar operation must be selected so generously that
contamination flashovers can be reliably excluded.
• Interruption of the connection between station neutral bus and earth electrode must be
avoided under all circumstances, with the exception of operation with metallic return.
Therefore configuration of the electrode line with two separate conductors, each with
its own insulators (i.e. as a homopolar line) is indicated. This configuration also
facilitates simple monitoring by differential protection.

Continuous monitoring of electrode lines for ground-faults and for interruption is


absolutely necessary, but is not at all easy to realize. Recommendations in this regard are
discussed in Section 5.6.5.5.

6.3.8 Line Faults in Metallic Return Operation

To facilitate continued transmission operation of a bipolar HVDC system upon loss of a


station pole, even if the current return through ground is not permitted or is permitted only

- 494 -
for a very limited time, the operating mode “metallic return” is often specified. This type
of monopolar operation can be realized in two methods.

6.3.8.1 Metallic Return Through Neutral Conductor

Fig 6.14 shows a bipolar HVDC system equipped with a neutral conductor. The overhead
line has a third, weakly insulated wire which is connected to the neutral buses of the two
stations. Earth electrodes and MRTB are not absolutely necessary, but they provide
favourable behaviour during transient faults of a line pole. For monopolar operation, the
MRTB is opened and commutates the ground current into the neutral conductor.

Figure 6.14: Metallic return through a neutral conductor.

It is also possible to install a lightning shield wire with weak insulation and to dimension it
such that it can assume the function of the neutral conductor (making use of the thermal
capability). In this case, however, the statistically frequent lightning strikes with insulator
flashover must be accepted.

6.3.8.2 Metallic Return through Cleared Line Pole

Fig. 6.15 shows a bipolar HVDC system in which a line pole, which is not being used due
to a station pole fault, is cleared and is used as neutral conductor for the metallic return
operating mode. After the circuit is established, the MRTB opens and commutates the
system current from the ground path into the line pole which then is being used as neutral
conductor.

- 495 -
Figure 6.15: Metallic return through cleared line pole

6.3.8.3 Insulation Failures of the Metallic Return Conductor

In the case of use of a cleared line pole as metallic return line, insulator flashovers caused
by direct or indirect lightning strikes are to be expected with the same frequency as in
normal transmission operation. Due to the high level of insulation, i.e. the long insulator
chain, self-extinction of an arc sustained by continued flow of direct current should not be
a problem. Contamination flashovers are also excluded for the same reason. The situation
is different in the case of the use of a weakly insulated neutral conductor. During relatively
long lasting monopolar operation, frequent flashovers must be expected, in particular in
combination with the function as lightning shield wire. Contamination flashovers,
depending upon the specific creepage distance, are also not completely excluded. Self-
extinction of the arc sustained by continued flow of direct current is important. This
current is determined by:

• System direct current


• Location of the insulation fault
• Arc resistance
• Footing resistance of the tower
• Transition resistance to remote earth of the grounded station neutral bus
Figure 6.16 shows the relationships schematically.

- 496 -
Figure 6.16: Distribution of the fault current during an insulation failure of the metallic
return conductor

As is shown in Fig. 6.16, the fault current through ground can also influence the
neighbouring ac systems. The portion In of the total ground current IE flowing through the
rigidly grounded transformer neutral points can result in saturation of current transformers.
This can cause false triggering of protective relays. Therefore it is important for such
ground faults to be eliminated as quickly as possible, preferably by self-extinction of the
flashover arc. When this is not assured, they should be eliminated by fault sensing and
clearing through protection functions.

6.3.8.4 Fault Diagnosis and Clearing

An insulation failure can easily be detected by measuring the current in the connection
between neutral bus and ground mat in the station which is solidly connected with its earth
electrode. Shutting-off the system direct current for perhaps 200 ms is a relatively sure
method of fault clearing, but it means a total interruption of energy transmission and
corresponding load rejection overvoltages in the ac networks.

Figure 6.17: Drying out of fault current by brief closing of the MRTB

- 497 -
In contrast to this, Fig. 6.17 shows a method for fault clearing in which energy
transmission is not impaired. A low resistance parallel path to the metallic return
conductor is established by brief closing of the MRTB and thus the fault current “dries
out.” This method, however, will not work in the case of a fault located close to the
grounded station, and the first method described must be used.

The operating mode “metallic return” is an emergency operating mode during loss of a
station pole. Statistically viewed, it has only slight relevance for the transient performance
of a bipolar HVDC system.

6.3.9 Peculiarities of HVDC Multiterminal Systems

IEC Publication 919-2, upon which the above discussion is based, expressively excludes
HVDC multiterminal systems. A few aspects of the transient performance of such
multiterminal systems will be given below. This will give the project planning engineer
guidance on the extent to which knowledge and experience with respect to HVDC two-
point systems can be transferred to multiterminal systems and at what point a fresh start is
required.

6.3.9.1 Assumptions

A constant voltage system in which any number of converter stations, but more than three,
are parallel-connected over a meshed network of direct current lines will be considered. It
is assumed that there is full line redundancy, that is, at least two lines extend from each
converter station. In the simplest case, such a network consists of a double line which
connects all converter stations or a ring formed by a single bipolar line. Fig. 6.18 shows
such a ring shaped network with four stations in single-line representation.

Figure 6.18: HVDC four-terminal system in single-line representation

- 498 -
All line and station terminals are equipped with HVDC circuit breakers. The regulation of
the converters in the four stations follows the current margin method with the reference
values of direct current and dc voltage regulators being generated by a central reference
value computer. The regulation characteristic of such an HVDC multiterminal system is
shown in Fig. 5.4.

6.3.9.2 Switching Actions in Normal Operation

HVDC circuit breakers allow connection and disconnection of sections of the line system
at will, providing such actions do not rupture the dc network, that is, as long as line
redundancy is available. The switching actions take place without the assistance of
converter regulation. The direct current in each of the stations remains unchanged and the
direct voltages change only insignificantly. Thus these switching actions are not at all
apparent in the connected ac networks.

The situation is different in the case of planned connection or disconnection of converter


stations. Such actions are controlled by the central reference value computer. It provides
for maintaining the power balance in the HVDC multiterminal system, causing the
transition from one steady-state transmission scenario into another one at prescribed rates
of change of power. The switching action itself takes place at zero station power, and
therefore does not appear at the outside. The HVDC breakers allow the reliable
disconnection of an inverter station without having to bring about in advance the rather
tricky blocking of control pulses.

6.3.9.3 AC Network Faults

Fundamentally, there are no differences between HVDC two-point systems and HVDC
multiterminal systems with respect to the transient performance and the effects of ac
network faults. These faults have to do with mutual effects between the converter and its
ac network. However, in two-point systems, the remote effects of a fault are limited to the
other station and the connected ac network, while in a multiterminal system, all stations
and their ac networks are affected.

Therefore, a special fault strategy such as given below must be developed:

• If the ac network fault is cleared by the first protective level (for example, in less than
80 ms) and if the converter station connected to the affected ac network is functional,

- 499 -
the dc network connections will be maintained and the recovery process will be carried
out by all stations at the same time.
• If the second level (back-up protection) is required to clear the fault or if the converter
station is not functional after a rapid clearing of the fault, this station will be
disconnected on the dc side. The HVDC system then continues energy transmission
with modified distribution.

In the latter case, the same ramifications on the entire system occur as in the converter
faults discussed below.

6.3.9.4 Converter Faults

In general, a converter fault affects a single group. After the affected converter is
disconnected on the dc and ac sides, transmission operation of the HVDC multiterminal
system can continue with an appropriately modified power distribution scenario. The other
group(s) of the affected station can participate as long as current through ground is
permitted or a neutral conductor is available. In other cases or after expiration of the time
period which is provided for ground return, a group of the other station pole must also be
disconnected. If the affected station has only one converter group per pole, this means
separation of the associated ac network from energy exchange through the HVDC
multiterminal system.

The disconnection of a converter group causes a disturbance of the power balance in the
HVDC system which may lead to overload of individual converters or to a collapse of dc
system voltage. Therefore the central reference value calculator must immediately
compute a set of compatible reference values and send them to the converter stations. A
high-performance telecommunication system is necessary for this purpose.

The now existing deficiency in the power balance must be divided in a suitable manner
among the other converters. There are various strategies by which this may be done. A
converter fault in an HVDC multiterminal system is a particularly critical event with
respect to the transient system performance. Special care must therefore be devoted to the
establishment of regulation and protection functions.

6.3.9.5 DC-Side Station Faults

The same considerations apply as in the case of converter faults, except that in this case an
entire station pole will always be affected even if more than one converter group per

- 500 -
station pole are available. The care necessary in the design and disposition of the dc
switchgear was noted in Section 6.3.5 and Fig. 6.12.

6.3.9.6 DC Line Faults


Due to the meshed configuration and redundancy of lines, the length of the lines in an
HVDC multiterminal system is significantly longer than the geographical distances
involved. Therefore line faults have special significance in the evaluation of the transient
system performance.

The use of HVDC breakers in combination with converter regulation following the current
margin methode is especially beneficial. DC voltage collapses to zero during a line-to-
ground fault in the dc line system are causing the power transmission to drop briefly to
50%. Direct current continues flowing in all converters of the affected pole. Therefore
there is no interruption of current. At the moment when the HVDC breakers of the section
of the faulty line open their contacts, the voltage returns. After approximately 100 ms
(measured from the occurrence of the fault), energy transmission is restored in the affected
system pole with the original distribution scenario.

Even though dc line faults may be a frequent occurrence in an extensive HVDC


multiterminal system, they have little effect on the transient performance of the overall
system.

6.3.9.7 Faults of Electrode and Neutral Lines


Basically the same considerations apply as for HVDC two-point systems. The effects of
faults of electrode lines and neutral lines on the transient system performance are
negligible. Nevertheless, they cannot be allowed to last for a long period of time due to
possible influences on neighbouring ac systems by stray ground currents.

- 501 -
6.4 The Dynamic Performance of the HVDC

6.4.1 Definition of Terms

The term “dynamic performance” of an HVDC system shall refer to the phenomena in the
range of time between steady-state and transient which result from the mutual interaction
between the HVDC system and the ac networks.

The dynamic performance under this definition results from the transition performance as
determined by the HVDC control functions, in particular, transition performance during
switching actions and network events. Dynamic performance can have either a positive
effect (stabilization) or a negative effect (destabilization) on network voltage and
frequency or phase angle.

The described “natural” dynamic performance of the HVDC can be influenced and
improved through modification and expansion of the control functions, such as functions
to damp oscillations of the network quantities.

The use of the HVDC system to determine the load flow within the ac network or for
regulation of frequency shall not be discussed in this section. In contrast to IEC
Publication 919, these applications are assigned to the steady-state range. The same
applies to the incremental modification of the reference value of the transmission power of
the HVDC (step power change), even if the commands are derived from events in the ac
network. Changes in transmission power are usually made so slowly in accordance with
the specified ramp that there is no danger of oscillations being triggered which would
endanger stability.

- 502 -
6.4.2 Voltage Stability

6.4.2.1 Statement of the Problem

Changes in reactive power flow at the interface between the HVDC system and the ac
network cause voltage changes on the ac busbar of the HVDC station. The lower the short-
circuit ratio (SCR), the more pronounced these voltage changes are. Each change in
transmission power of the HVDC is accompanied by a change in the reactive power
absorbed by the converters. This effect cannot be fully eliminated by reactive power
regulation, since normally the only adjustable variables available are switched reactive
power units (ac filter circuits and capacitor banks), and transformer tap changers, the
function of which involves also incremental steps. For this reason, the permissible voltage
change upon switching of a reactive power unit is usually specified.

Another very important phenomenon are voltage changes caused by events in the ac
network and the reactions in the reactive power behaviour of the HVDC station.
Depending upon the regulation method selected, these reactions can reduce or reinforce
the voltage change. Special attention should be given to this effect in the case of ac
networks with a tendency to voltage instability. In critical cases, measures must be taken to
enhance voltage stability. Such measures can be in the form of modifications to HVDC
control or installation of continuously controllable reactive power devices.

6.4.2.2 Reactive Power Characteristics of HVDC Stations

If consideration is limited to the time range of dynamic performance, the effect of reactive
power and ac voltage regulation, to the extent they make use of switchable elements, can
be ignored. The same also applies for the regulation of transmission power, although in
this case, regulation functions which use valve control are inserted. Changes of
transmission power occur so slowly with respect to the connected ac networks that the
changes in reactive power balance which they cause do not have any influence on the
dynamic system performance.

- 503 -
The following discussion will therefore be limited to the effects of external interferences
and will concentrate on changes of the ac busbar voltage (since changes in network
frequency and phase angle have almost no effect on converter performance).

Id

Ud

P d = U d* I d

S´´
k
SCR =
P dN Qc
q = P
c dN
Q
q = P st
st dN

Figure 6.19: Circuit diagram (upper part) and equivalent circuit (lower part) of an HVDC
station

The upper portion of Fig. 6.19 is the circuit diagram of a bipolar HVDC station with two
12-pulse groups. The lower part is a simplified equivalent circuit in which the converters
are presented as a variable reactive power consumer with related power qSt. The reactive
power units are represented as a capacitor with related power qc.

First, it will be assumed that this is an inverter station with the data relevant to the
discussion as follows:

gN = 20°

dXN = 0.0825

Full compensation at nominal operation, i.e. qcN = 0.6, was assumed.

At nominal power and nominal voltage of the ac network, the reactive power demand of
the converter is qSt N=0.6. At temporary overvoltages on the ac-side, but with dc voltage
held constant, the reactive power demand of the inverter increases steeply, as is shown in
Fig. 6.20. Naturally, the reactive power generated by the reactive power units also
increases. This increase is with the square of the overvoltage factor.

- 504 -
1.6

1.4

1.2

1.0

0.8

0.6

1.1 1.3 1.5 1.7 1.9

Figure 6.20: Reactive power demand of the converter in relation to valve voltage Uv.

The effects of a temporary overvoltage of 1.25 p.u. occurring at the ac busbar of the
HVDC station on the reactive power balance will be considered below for three different
operation modes of the converter.

a) Inverter with dc voltage regulation

The inverter operating with the extremely fast electronic Ud regulation will keep its
terminal dc voltage constant, even in the event of ac-side overvoltages. This means that
the rectifier and its ac network does not react to this event. The busbar voltage has
increased by 25% and thus the valve voltage Uv. This forces an increase of the extinction
angle from 20° to 41°. Although the inductive dc voltage change is reduced because of the
increased valve voltage to dx=0.066, the reactive power demand increases to qSt = 1.06,
this means by 75% compared with qStN. On the other hand the output of the reactive power
units increases to 1.25², representing an increase of 55%. There remains a pronounced
excess of reactive power demand in the reactive power balance of the HVDC station. In
this case, the HVDC contributes to the stabilization of the ac network voltage. In Fig. 6.21,
the changes in reactive power balance as a function of the overvoltage factor are presented
with the nominal degree of compensation qcN as the parameter. It is obvious that a higher
degree of compensation has a destabilizing effect.

- 505 -
q
0.4 q : 0.8
cN
0.2
0.6
0 1.1 1.3 1.5 1,7 1.9 U v (p.u.)
-0.2
0.4
-0.4

-0.6
0.2

Figure 6.21: Change in the reactive power balance of the HVDC station as a function of
temporary overvoltage (inverter with Ud-regulation)

b) Inverter with extinction angle regulation

Here the interactions are much more complicated. If the rectifier would be able to hold the
dc current constant, despite increased countervoltage, the reactive power absorbed by the
inverter is being reduced to qST = 0.56. At the same time, the reactive power generated
would climb to qC = 0.94. The reactive power balance reflects a strongly destabilizing
positive deviation. This relationship is valid at least for several tens of milliseconds, as
long as the dc-side inductivities are able to maintain the direct current at a value above the
inverter current reference value. Afterwards, the direct current is regulated by the inverter
to a value of (I d - DI). In the present example, this value is 0.9 IdN . Now the dc voltage is
determined by the rectifier which has reached its limit at amin. The quasi-stationary
reactive power balance which evolves depends upon the system parameters. In our
example, a 1200 km long, 500 kV overhead line, Id = 0.9 Idn, Da = 10° was assumed. This
resulted in an inverter extinction angle of 38° and a reactive power consumption of qST =
0.98. This is slightly above the reactive power generation of qc = 0.94, and thus does
hardly contribute to stabilize the ac voltage.

- 506 -
c) Rectifier with current regulation

The behaviour is analogous to that of an inverter with Ud-regulation. With the same
assumptions and aN = 15°, the reactive power demand rises in the presence of a temporary
overvoltage of 1.25 p.u., from qStN = 0.53 to qSt = 1.0. This represents a 90% increase
compared with an increase in reactive power generation of 55%. The contribution to
voltage stabilization is even more pronounced than in the case of the inverter with Ud
regulation.

Note:
At this point, an erroneous interpretation must be avoided. A differential quotient dq/du can
be read from the curves of Fig. 6.21. This might be taken as the criterion for voltage
stability. This would be a mistake because the sign of the quotient d u/dq which is critical
for the voltage stability must include all elements connected to the busbar including the
elements of the ac network. However these elements are not contained in the reactive
power balance of the HVDC station presented in Fig. 6.21. Nevertheless, this does not
negate the important indication of the zones in which the HVDC station contributes to
voltage stability of the ac network and in which zones it has a destabilizing effect.

6.4.2.3 Reactive Power Characteristic of the AC Network

In this context, reactive power characteristic of an ac network shall mean the curve of the
reactive power balance at the interface between HVDC station and ac network which is
necessary to hold the busbar voltage constant in the presence of changing active power.
Fig. 6.22 shows a typical reactive power characteristic for a network point with low short-
circuit ratio (SCR » 2.5) with the busbar voltage as a parameter. The first quadrant of the
graph indicates the range of active and reactive power input from the ac network, and thus
rectifier operation of the HVDC station. For an actual project, the reactive power
characteristic of the ac connection points is obtained by means of load-flow calculation.

- 507 -
0.95 p.u.
1.0 p.u.
0.5 1.05 p.u.

-1 -0.5 0.5 1

-0.5

Figure 6.22: Reactive power characteristic of a network with SCR » 2.5


Parameter: ac busbar voltage of the HVDC station

In operation, it is almost impossible to maintain the required reactive power balance


exactly. This is especially true in the dynamic range. Upon changes of quantities, changes
of voltage occur. The “weaker” the ac network at the connection point of the HVDC is, the
more pronounced these changes are.
To make the interrelationships more clear, the extreme case of a 100% load rejection will
be considered, in which the HVDC transmission power is reduced from 1 p.u. to 0, while
the reactive power units are still connected.

Fig. 6.23 shows the load rejection overvoltage (dynamic overvoltage DOV) computed by
application of the given rough formula in relationship to the short-circuit ratio (SCR) of
the ac network at the connection point of the HVDC. The reactive power compensation of
the HVDC station is used as parameter.

1.9 0.2 0.4


0.6 0.8 1.0

1.7

1.5

1.3

1.1

1 2 3 4 5 6

Figure 6.23: 100% load rejection overvoltage in relation to short-circuit ratio (SCR)

- 508 -
The conclusion might be drawn that at SCR < 2.5, a compensation level of qC=0.6 is not
permissible. But in a bipolar HVDC, the 100% load rejection is in actuality a catastrophic
event which should be beyond consideration for dynamic performance of the system. If
this situation does occur, it will be handled by the surge arrester in combination with a
rapid switching off the reactive power units.

The 50% load rejection is of much more interest. Fig. 6.24 shows the load rejection
overvoltage which occurs in this case.

0.2
1.9
0.4

1.7 0.6
0.8
1.0
1.5

1.3

1.1

1 2 3 4 5 6

Figure 6.24: 50% load rejection overvoltage in relation to short-circuit ratio (SCR)

An inverter station using electronic Ud regulation, operating with Pd =1 p.u., was assumed.
Upon loss of a pole, all reactive power units continue to be connected. The difference in
the reactive power balance Dq, which is determinative for the load rejection overvoltage
(DOV), is derived from the increased generation of the reactive power units less the
increased reactive power demand of the pole which continues to be in operation. At a
compensation level qcN=0.6, the overvoltage remains below 1.3 down to SCR=1.5p.u.

- 509 -
6.4.2.4 Influence of the HVDC on Voltage Stability

Voltage changes, in particular, voltage jumps in ac networks may cause oscillations. In


critical cases they can cause voltage instability. In the present context, the important
question is whether well-conceived intervention in the regulation of an HVDC can prevent
voltage instability and can damp voltage oscillations. This question is especially important
at low short-circuit ratio (SCR) at the connection point of the HVDC station, because in
this case, even slight changes in the reactive power balance can result in significant
fluctuations of ac busbar voltage. The options for interventions through the control system
depend upon the nature of the reactive power units used.

a) Switchable reactive power units

The switching of reactive power units always results in a voltage jump which triggers
corresponding oscillations in the ac network. The subdividing of the total compensation
reactive power needed into many elements of small nominal power does reduce the
voltage jumps. However, the economic burden of this process is rather high. Synchronous
switching of a reactive power unit with a shunt reactor of half the power reduces the
voltage jump, but is also very expensive.

Therefore the technique of “cushioning” assumes an important role. This technique


reduces reactive power jumps through interventions in the converter regulation, or
completely avoids jumps and replaces them through continuous changes. Prior to
disconnection of a reactive power unit, the reactive power consumption of the converter is
increased through gradual increase of the control angle a or g. At the instant of switching,
the control angle is then suddenly reset to the initial value. In the other direction, at the
instant of connection of a reactive power unit, the control angle is suddenly increased and
then is gradually brought back to the initial value. Since the reactive power consumption
of a converter depends upon a series of parameters in addition to control angle, a control
computer is needed in order to determine the change of control angle needed in a
particular case. By this means, the dependance of the switched reactive power of the
busbar voltage can also be taken into consideration.

- 510 -
In using this method, one must be aware that a change in control angle means a change in
the main quantities of the HVDC system, direct current or dc voltage. While the minor
fluctuations of active power caused by such a modification of the control angle (for several
hundred ms) can be accepted, the effects on the reactive power balance of the counter
station must be carefully studied to determine wether they are acceptable.

The control procedures described above are not the only means for intervening in the Id- or
Ud- regulation. An electronic ac voltage regulation can be superimposed upon the Id- or
Ud- regulation which is active at a given time with the goal of compensating voltage
deviations which occur, and damping voltage oscillations which arise. In this case, the
influence upon voltage deviations and voltage stability at the ac busbar of the other station
must be given particular consideration.

b) Static compensators

The bridging of reactive power jumps upon the switching of a reactive power unit by
means of a continuously controllable supplemental reactive power source can be achieved
through the use of a static compensator (static var compensator SVC). Its nominal power
should be somewhat higher than the nominal power of the largest switchable unit. In this
context, nominal power means the entire controllable range of the compensator.

In this case, too, a simple control of the reactive power in accordance with a preselected
characteristic or a regulation of the ac busbar voltage with damping of oscillations can be
provided. In the latter case, problems can arise from the interactions of the regulation
processes of both, the converters of the compensator and of the HVDC station which are
of approximately equal speed. It is possible that a decoupling with respect to time or the
insertion of special control functions for the suppression of power oscillations between the
converters may be necessary. Detailed simulator studies are essential to assure a
satisfactory dynamic performance of the total system.

- 511 -
c) Synchronous condenser

In comparison with static compensators, rotating synchronous condensers are expensive,


require heavy foundations, have higher losses, and require much more maintenance.
Therefore they are used only in situations where an HVDC feeds into a low inertia ac
network with predominantly passive load (i.e., with low rotating mass). But, of course,
synchronous condensers also increase the short-circuit ratio (SCR), which can be very
desirable.

A synchronous condenser can perform basically the same control and regulation functions
as a static compensator. The time constant depends upon the design of the exciter system
and the ceiling voltage of the machine, i.e. from the acceptable overvoltage of the exciter
winding. The time constant is on the order of magnitude of 100 ms. This provides a
decoupling with respect to time from the converter regulation of the HVDC. On the other
hand, the occurrence of subsynchronous torsional oscillations will be quite possible if the
machine has an unfavourable rotating mass natural frequency. Therefore detailed
simulator studies should also be performed in this case to insure trouble-free operation and
satisfactory dynamic performance.

d) Opposite station of the HVDC

In Section 6.4.2.3, the 100 % load rejection is excluded from consideration of the dynamic
performance of an HVDC system as a catastrophic event. This naturally does not apply for
the 3-phase network short circuit in the vicinity of the station. This possibility must be
taken into consideration in any event. It is one of the most important cases for the
evaluation of the transient performance of the HVDC system (6.3.3.1). For the ac network
of the opposite station which is not affected by the fault, such event means a 100% load
rejection of active power, but not necessarily of reactive power. If a by-pass is initiated in
the station affected by the fault and the direct current is maintained by the unaffected
station with reduced value by means of the VDCOL function, the reactive power
consumption on the unfaulted side can be kept. The reactive power balance can be
maintained to a certain degree by suitable selection of the VDCOL function which can be
matched still better to the operating parameters by means of a control computer. In this
way, large voltage changes can be avoided.

This performance can be improved by a superimposed dynamic ac voltage regulation


which controls the current reference value in such a manner that the ac voltage remains
constant.

- 512 -
6.4.3 Steady-State and Transient Stability

6.4.3.1 Introductory Comments

The possibility of regulating the active power of an HVDC system and (with limitations)
also the reactive power balance at the interface between the HVDC station and the ac
network, provides the possibility of using the HVDC for the improvement of steady-state
and transient stability of ac networks.

An ac system is designated to be steady-state stable if it returns to steady-state operation


after a small disturbance.

An ac system is designated to be transient stable if it returns to the condition of steady-


state stability following a large disturbance or a series of large disturbances after they have
been cleared.

In both cases, the phenomenon of electromechanical oscillations between individual or


groups of synchronous machines or different areas of a large system are involved.

Oscillations of this kind can arise as a result of faults or disturbances and in the case of
instability, they can grow up until finally individual synchronous machines fall out of step
or the connection lines between network areas disconnect due to overload. But even if
steady-state stability is still maintained, such oscillations may be damped so weakly that
network operation is disturbed for an extended period of time.

The frequency range of electromechanical oscillations is 0.1 to 2 Hz. In this range, an


HVDC system can precisely control quantities determined by Id- and Ud-regulation. That
always includes the transmitted active power. Therefore an HVDC can excellently be
employed for the damping of electromechanical oscillations.

- 513 -
6.4.3.2 Power Modulation

When electromechanical oscillations are to be effectively damped by power modulation,


the three parameters of the modulation signal must be properly selected and be precisely
maintained. These parameters are:

• Frequency
• Phase angle
• Amplitude

For this reason, damping regulation should intervene directly in the converter regulation,
for example, the current regulation of a rectifier or the dc voltage regulation of an inverter.
Any power regulation or higher order control function should either be deactivated or
slowed down to such an extent that it influences only the mean value but not the
instantaneous values of the modulated transmission power. The same applies for control
angle regulation by means of transformer tap changers and reactive power regulation by
switchable reactive power units.

Depending upon the triggering event and the condition of the ac network, either a lightly
modulated power may suffice for damping of the oscillations or a strongly modulated
power may be necessary. The two situations require different interventions in the HVDC
regulation.

a) Small signal modulation

For this purpose, intervention in the regulation of the HVDC station connected to the ac
network to be stabilized is sufficient. Transmission of the modulation signal to the other
station is not required. The modulation signal must not exceed the limits of the margins of
the regulator characteristics. Thus, if the rectifier is modulating the transmission direct
current, the modulation signal must not exceed the current margin, DI, and if the inverter is
modulating the transmission direct voltage, the modulating signal must not exceed the
limit of the voltage margin, DU, as is shown in Fig. 6.25.

- 514 -
Ud

U*d GR
U*
d WR
A
U
Umod I

Imod Id

Figure 6.25: Regulation characteristics of rectifier (GR) and inverter (WR) with
modulation ranges of DI mod and DU mod for small signal modulation

In many cases, the small signal modulation will be sufficient for damping of
electromechanical oscillations which pose a risk to the steady-state stability. This will be
particularly true if the HVDC carries a significant portion of the total power of the network
or of the network part.

Increasing the current modulation range in an upward direction is usually permissible and
is possible to the extent the rectifier has adequate adjustment margin (amin) to regulate the
increased instantaneous values of the direct current. Increasing the voltage modulation
range in a downward direction is likewise possible, but must remain above the value at
which the VDCOL function is initiated.

- 515 -
b) Large signal modulation

This requires a coordinated intervention in the regulation of both HVDC stations. In


particular, the current reference values in both stations must be varied in the same manner
because the current margin DI must always be maintained. Telecommunications between
the two stations are needed for this. In addition, the VDCOL functions must be deactivated
if the direct voltage is to be modulated to a greater degree.

For operation with large signal modulation, which is necessary only for short periods of
time in order to re-establish the transient stability following relatively large interferences,
use can be made of the short-time overload capacity (see Section 6.2.2.1 and Fig. 6.3-6.5).

In back-to-back links, which usually have a relatively small rated power in comparison to
the generation power of a network, the power modulation may even include the periodic
change of energy direction. The strict coordination of the regulation of the two converters,
necessary for this, does not impose any difficulties in this case.

6.4.3.3 Reactive Power Modulation

A modulation of the active power of an HVDC system will unavoidably be accompanied


by a modulation of the reactive power balances at the interfaces to the ac networks of the
two HVDC stations if not the procedures described in Section 6.4.2 are being used. In
many cases, however, a reactive power modulation is necessary or desirable along with
active power modulation to ensure or restore voltage stability. This is basically achievable.
It is necessary to involve the converter regulation of both HVDC stations. Both, direct
current modulation and direct voltage modulation, affect active power and reactive power.
The reference values for direct current and direct voltage regulation can be determined by
calculation so that the desired active and reactive power modulations are achieved in one
of the ac networks. In the ac network of the other station, the same active power
modulation will unavoidably occur and also a reactive power modulation which has the
same frequency but is in the opposite phase. This will be acceptable only if the ac network
of the other station has significant inertia and there is a high short-circuit ratio SCR at the
connection point.

- 516 -
In this respect, the reactive power behaviour in the two networks resulting from small
signal modulation carried out by one HVDC station is of interest. In Figs. 6.26 and 6.27,
the reactive power characteristics of a network connection point with SCR= 2.5 and
busbar voltages of 0.95, 1.0, and 1.05, respectively, are presented, similar as in Fig. 6.22.
In addition, the Q/P characteristic of an HVDC converter as explained in Section 1.4.4.5 is
provided.

1 1

-1 1 -1 1

Figure 6.26: Figure 6.27:


Power modulation through current Power modulation through direct
regulation by the rectifier voltage modulation by the inverter (WR)
(GR) (explanation in text)
(explanation in text)

In Fig. 6.26 the modulation of direct current by the rectifier GR is presented. The reactive
power curve of the rectifier tracks the characteristic of the ac network almost exactly. In
contrast, the reactive power curve in the inverter runs in opposite direction to the needs of
the ac network.

In Fig. 6.27, the modulation of dc voltage by the inverter WR is presented. The reactive
power curve in the inverter corresponds, at least in its direction, to the network
characteristic even if the change is far greater than is required by the ac network. In the
rectifier, however, the reactive power curve is in the opposite direction to the needs of the
network.

These relationships must be observed in planning of damping control. Acceptability must


be tested by computer and simulator studies.

- 517 -
6.4.3.4 Classification of Networks to be Stabilized

The achievable degree of improvement of stability not only depends on the relationship of
modulation power to installed generation power, on the network characteristic and on the
feed-in point, but especially on the arrangement of the HVDC in relation to the network
configuration. Two fundamentally different cases can be distinguished.

• The HVDC is running in parallel to ac lines; it interconnects two synchronized system


parts
• The HVDC connects two asynchronous networks which are not linked on the ac side

The derivation of suitable modulation signals and the effectiveness of power modulation
are different in these two situations.

a) HVDC running in parallel to ac lines

This arrangement depicted in Fig. 6.28 is particularly suitable for damping control by an
HVDC.

A B

Figure 6.28: Arrangement of the HVDC running in parallel to ac lines

There is always the danger of instability in very long ac lines. In some cases, a full load is
not possible on the lines without reaching the limit of stability. In such cases, a small
power modulation of a parallel connected HVDC represents an effective remedy. The
1,360 km long Pacific Intertie HVDC system, which runs in parallel to 500 kV ac lines, is
a convincing example of this application.

- 518 -
The most suitable damping or modulation signal would be the phase difference between
the busbar voltages designated in Fig. 6.28 as A and B. However, it is scarcely possible
over such great distances to measure this difference. In place of this, the frequency
difference between A and B can be determined and used in a suitable manner. In principle,
the measuring of oscillations of power or of current in the parallel ac lines is simpler. In
this case, the transmission of measured quantities would not even be necessary. The results
of measurement, however, are not always unambiguous. In the case of very large phase
angles, an increase in angle can even result in a reduction of power and vice versa. For this
reason, the use of frequency difference is more reliable in such cases.

b) HVDC between asynchronous ac networks

This arrangement, shown in Fig. 6.29, is much less suitable for damping of
electromagnetic oscillations in one of the ac networks. It is usually used to damp
oscillations between network portions or, as shown here, between generators.

Figure 6.29: HVDC for damping control in one of two asynchronous ac networks

The effectiveness of HVDC damping control for the improvement of stability depends
essentially upon the availability of a suitable modulation signal. The signal may have to be
transmitted over telecommunication systems from a distant point of the network. In this
case, determination of phase and amplitude of the modulation signal is critical.

Detailed simulatorstudies using Netomac or EMTP programs are necessary for


determining the required modulation signal and power. In these studies, the HVDC and its
regulation functions must be accurately simulated. The studies should be supplemented by
real time simulator tests to ensure that the HVDC control system will convert the derived
modulation signal into the required modulation power in a precise manner.

- 519 -
If the specifications do not clearly indicate who is responsible for the selection and
definition of the modulation signal and of the transition functions of the damping control,
clarification must be obtained prior to signing of the contract.

6.4.3.5 Requirements on the Telecommunication System

Use of large signal modulation for damping control places special requirements on the
telecommunication system with respect to transmission speed and reliability. The
telecommunication system and the automatic modulation procedures must be monitored to
ensure that a power modulation in the wrong phase is reliably prevented, even in the case
of loss of telecommunications.

6.4.4 Subsynchronous Stability

6.4.4.1 Statement of Problem

Practice has shown that turbogenerators in the vicinity of an HVDC rectifier station, the
power of which is transmitted largely by the HVDC and which have a relatively weak
connection to the ac system, are susceptible to subsynchronous rotor oscillations. Large
turbogenerators in nuclear power plants can show subsynchronous oscillations down to
5 Hz.

Rotor oscillations of turbogenerators can be excited by a large number of events. The


inherent damping of such subsynchronous rotor oscillations is relatively weak. These
oscillations decay slowly. In the event of continuously repeated excitation or resonance
with another structure capable of oscillations, they can continue or even build up to the
point of subsynchronous instability. Such a condition presents an acute danger to the shaft
of the turbogenerator. Immediate shut-down is required. Structures capable of oscillations
which can come into resonance with subsynchronous rotor oscillations of large
turbogenerators include series compensated ac high voltage lines. An HVDC rectifier
station with constant current regulation can cause a similar effect. In the relevant
frequency range up to approximately 25 Hz, current regulation generally has a positive
circuit amplification. Depending upon the time constants of the regulation circuit and upon
phase shifting of the regulation signals an amplification of the subsynchronous rotor
oscillations is possible.

- 520 -
By using a band filter, it is possible to reduce regulation circuit amplification for a
particular frequency to below 1. This simple but very effective method of “passive”
subsynchronous oscillation damping will not be discussed further. This discussion will be
concentrated on “active” subsynchronous oscillation damping by converter regulation.

6.4.4.2 Criteria for the Occurrence of Subsynchronous Resonances between HVDC and
Turbogenerators

The power transmitted by an HVDC with constant current regulation is independent of


network frequency and therefore power does not contribute to the damping of frequency
oscillations of a turbogenerator and thus to the damping of subsynchronous torsional
oscillations. This by itself is not enough to create resonance oscillations or
subsynchronous instability. A series of criteria must all occur for this to result. These
criteria include:

• Turbogenerator and HVDC rectifier station located close together


• Weak connection of the turbogenerator to the ac meshed network
• Rated power of the HVDC and of the turbogenerator of the same order of magnitude

Torsional oscillations of the turbogenerator cause periodic changes in amplitude and phase
angle of the generator voltage. The shorter the distance to the HVDC station, the more
these changes appear at the station busbar. The resulting changes in valve voltage Uv and
in control angle a influence converter direct voltage and, as a result, direct current and
transmission power of the HVDC. This in turn means a change in the electrical torque for
the turbogenerator. Thus the cycle of effects is closed. If the phase angle between the
oscillating electrical torque and the torsional oscillation of the mechanical masses becomes
larger than 90°, the oscillation can become unstable.

The current regulator of the HVDC rectifier tries to counteract the current deviations and
influences control angle a. Depending upon the time constant of the regulation circuit and
phase shift of the regulation signal which changes with the frequency of the
subsynchronous rotor oscillation, current regulation can have a damping or an amplifying
effect upon the subsynchronous rotor oscillation.

For this reason, the “weakness” of the connection between turbogenerator and the ac
network, expressed by the impedances of the connecting lines, plays an important role.
Conventional consumers with their frequency-dependent loads cause a damping of the
torsional oscillations of a turbogenerator. This damping, however, is made largely

- 521 -
ineffective by weak connections to the ac system. In addition, these impedances cause the
active power drawn by the HVDC to be overwhelmingly supplied by the nearby
turbogenerator. Thus power fluctuations also occur essentially between the HVDC and the
nearby turbogenerator. The prerequisite for these effects is that the HVDC and the nearby
turbogenerator have approximately the same rated power. Care must be taken with respect
to special operating conditions since the circuit amplification of the current regulation
increases with increasing control angle a (despite cos-linearization of regulator
amplification). The tendency to subsynchronous resonance, therefore, also increases. This
must be considered when operation with reduced dc voltage is specified.

Based on past experience, hydropower plants do not tend to have subsynchronous rotor
oscillations, even if they are in close proximity to HVDC stations. This is probably
attributable to the very large rotating masses and the resulting very low natural frequencies
which easily can be suppressed by current regulation.

Turbogenerators in the vicinity of an inverter station are also not endangered by possible
mutual interactions with the HVDC. They do not supply any power to the HVDC.
Operating in parallel with the inverter station, they supply convential, frequency-
dependent consumers. In addition, an inverter, at least when it operates with dc voltage
regulation, contributes to damping to the extent that each increase in voltage triggers an
increased reactive power demand and vice versa (see Section 6.4.2.2).

It should also be noted that damping control for suppression of electromechanical


oscillations (0.1 to 2 Hz) may interact mutually with one of the natural torsional
frequencies of a turbogenerator, thus contributing to destabilization.

- 522 -
6.4.4.3 Calculation of the Tendency for Subsynchronous Oscillations

It is relatively easy to estimate whether the occurrence of subsynchronous resonance must


be expected in a planned HVDC system for which several of the above criteria are fulfilled
and, therefore, in-depth subsynchronous resonance studies should be undertaken to
identify whether or not subsynchronous damping controllers are required.

IEC 919-3 has developed a quantitative screening tool for this, showing the approximative
relationship between the magnitude of the expected interaction and the ac system strength:

2
S æ SC i ö
UIFi = HVDC ç 1 - ÷
Si è SC TOT ø
where:

UIFi = Unit Interaction Factor of ith generating unit


SHVDC or Si = Rating in MVA as per subscript (HVDC or ith unit)
SCi = Short circuit capacity at HVDC station a.c. bus excluding ith unit
(excluding a.c. filters)
SCTOT = Short circuit capacity at HVDC station a.c. bus including ith unit
(excluding a.c. filter)
It is said that a Unit Interaction Factor less than approximately 0.1 suggests that no
significant interactions are to be feared and, therefore, no in-depth subsynchronous
resonance studies are required.

6.4.4.4 Subsynchronous Oscillation Damping Control

If the subsynchronous oscillation study has shown that the occurrence of subsynchronous
resonance between the HVDC rectifier and nearby turbogenerators is to be expected (and a
“passive” subsynchronous oscillation damping by means of a narrow-band reduction of
regulation amplification to below 1 does not promise to be successful), the use of
subsynchronous oscillation damping control should be considered. Subsynchronous
oscillation damping control must modulate the direct current and consequently the active
power absorbed by the HVDC in such a way that it becomes effective at the nearby
turbogenerator as a damping electrical torque for the torsional oscillations of the rotating
masses. The frequency and its rate of change and/or a signal derived from the current and
voltage of the converter may be used as control signals for the damping control of the
HVDC.

- 523 -
Determination of the control parameter is difficult. Damping control should be effective in
the right sense under all operating conditions and network configurations which may
occur. Its influence must be strong enough to effectuate at least a certain degree of positive
damping on the strongest torsional oscillation of the turbogenerator which may be
triggered in the worst case by a disturbance.

Basically, only the small signal modulation per Section 6.4.3.2a can be considered for
subsynchronous oscillation damping control. Involving the other station is not a possibility
because of the significant signal telecommunications transmission time between the
HVDC stations in relation to the frequency of the modulation signal. HVDC back-to-back
links are an exception to this statement.

The reliability of subsynchronous oscillation damping control should be comparable with


other HVDC regulations. It must be configured in a redundant and fault-tolerant manner.
Nevertheless it should not be considered a protection system according to IEC 919-3. The
subsynchronous oscillation protection relays for turbogenerators are not made unnecessary
by subsynchronous oscillation damping control.

Subsynchronous oscillation damping control should not have a negative influence on the
steady-state, the transient, or the dynamic performance of the HVDC. The overall
performance of the HVDC and the ac system with inclusion of subsynchronous oscillation
damping control, should therefore be carefully investigated in the project planning stage
by means of computers and HVDC simulators. Later, confirmation should be obtained
through well-conceived operating start-up tests.

6.4.5 Higher Frequency Stability

6.4.5.1 Introduction

In this discussion, the term “higher frequency stability” means the dynamic performance
of an HVDC system in the frequency range above approximately 1/2 the network
frequency. As was shown in Sections 1.5 and 1.6, HVDC converters generate
characteristic and non-characteristic harmonics of the fundamental frequency and under
some circumstances, non-harmonic oscillations which have no relationship to the network
frequency. However, converters not only generate harmonics; they also react to them. The
behaviour is in large measure dependent upon ac- and dc-side impedances, in particular
resonance points in the impedance curve.

- 524 -
If several unfavourable circumstances coincide, instabilities can occur making system
performance no longer acceptable. Such instabilities at higher frequencies may „grow
from nothing“, even in perfectly symmetrical systems. Generally, however, certain
unsymmetries in the ac systems or in the operating quantities of the HVDC and a low
short-circuit ratio (SCR) are prerequisites for their development.

6.4.5.2 Types of Higher Frequency Instability

Four different types of higher frequency instability can be distinguished based on the
triggering factors and the effects on system performance.

a) Harmonic instability

Harmonic instability has occurred in some older HVDC systems which were not equipped
with triggering devices for the generation of equidistant control pulses. It is caused by the
mutual interaction between the ac-side harmonic currents generated by the converter (these
harmonic currents always have a degree of unsymmetry in the three phases) and the
voltage distortions caused by them (the voltage distortions manifest themselves in angle
shift of the voltage zeroes and as a result they cause a control puls shift which may lead to
amplification of the phase unsymmetry of the harmonic currents). The danger that this
effect will build up into instability is greatest for frequencies for which a resonance exists
on the ac or dc side.

In HVDC systems which are equipped with trigger devices for the generation of
equidistant control pulses (e.g. the phase-locked oscillator), harmonic instability does not
occur. At least it has not been observed in newer HVDC systems.

b) Regulation circuit instability

The dynamic performance of an HVDC system is essentially determined by the


inductances and capacitances of the ac network and of the dc circuit. To improve dynamic
system performance, an attempt might be made to fully utilize the great control speed of
the converters by increasing the regulator amplification or by reducing their integration
time. This measures, however, could result in oscillations of regulation variables and
finally in instability of the regulation circuit. The danger is especially great when the
natural frequency of a regulation circuit corresponds to that of the ac system or is close to
it.

- 525 -
Regulation stability often can only be achieved by reduction of regulator amplification
and/or increase of integration time, i.e. by relaxation of requirements on the dynamic
performance of the HVDC system. If that is not acceptable in view of the requirements of
the specifications, one can try to take remedial actions by changing the parameters of the
regulator circuit in relation to the load or other operating parameters, or by the use of self-
adapting regulators.

c) Transformer saturation instability

A harmonic component of network frequency (m = 1) in the direct current of an HVDC


causes harmonics of the second order (v = 2) and a direct current component (v = 0) in the
currents of valve-side windings of the converter transformer. If the direct current
component reaches approximately 50% of the amplitude of the magnetizing current, a
noticeable saturation of the transformer core occurs. This causes the magnetizing current
to be distorted very strongly in a half-wave unsymmetrical manner. The second harmonic
becomes the dominating component. The ac busbar voltage also becomes distorted. The
second harmonic in it causes an additional component of the network frequency (m = 1) in
the direct current. Thus, the cycle of effects is closed. If the ac network has a parallel
resonance point for the second harmonic, and/or the dc circuit has a series resonance point
for network frequency, instability can develop.

If the existence of such resonance conditions is determined during the planning of an


HVDC, one can attempt to shift the resonance point on the dc-side by changing the
smoothing reactor inductivity, and on the ac-side by the use of specially tuned filter
circuits (low-order filters). In many cases, a special control circuit is also promising as a
remedy, but one must first determine that the transient performance will not be
unacceptably worsened.

d) Harmonic mutual reaction

Harmonic mutual reactions are processes in which harmonics from foreign sources are
impressed on the dc circuit of an HVDC and enter into mutual reaction with the converters
and their ac network.

An example of such a source is an ac high-voltage line which runs over a relatively long
distance on the same right of way with the HVDC overhead lines. Depending on the
configuration of the conductors and the distance from the HVDC, a significant network
frequency component can be impressed on the direct current in the presence of large

- 526 -
currents in the ac line (in particular in the presence of imperfect phase symmetry). The
phenomena, including instability, which are triggered by this were described under
paragraph c.

The opposite station of the HVDC can also be considered a source of foreign harmonics,
in particular if the two ac networks are not synchronous. In this case there are two patterns
of harmonics of slightly different frequency fed from the two converters and superimposed
on the dc circuit. As a result, non-harmonic oscillations arise in a broad frequency band, in
particular low frequencies. Such oscillations are very likely to find resonances. However,
their amplitude is usually so small that there is no danger of instability.

6.4.5.3 Stabilizing Actions

Several measures are available for avoiding the types of higher frequency instability
described and of assuring satisfactory dynamic performance of the HVDC system in the
higher frequency range.

The most cost effective measure, and thus the first choice, is intervention in HVDC
regulation. The intervention may either be in the converter regulation functions or,
bypassing the regulator, may be a direct intervention in the triggering device input signal.
In many cases, modulation of dc voltage or direct current may be successful. The goal is
the damping of higher frequency oscillations of converter quantities through activation of
a suitable modulation signal.

Only one quantity is available for this damping control in each station, namely the control
angle a or ß. Incompatibilities can occur if too many regulation or control signals are
switched in parallel to achieve very different functions, all of which require a modification
of a or ß. The effectiveness of the damping control and its compatibility with the other
regulation functions must be ensured by intensive simulator studies. The use of real time
simulator components with a realistic transition behaviour up to the kHz range is
particularly important. This applies especially to the saturation behaviour of converter
transformers.

If interventions in the HVDC control system do not help or if they are not compatible with
specified system performance, measures must be taken in the high current circuit. These
measures include increasing smoothing reactor inductivity, use of a network frequency
blocking filters in the dc circuit, use of a supplemental filter for low harmonics on the ac
side, the decoupling of foreign sources of harmonics, etc. Here too, the effectiveness of the

- 527 -
measures must be carefully studied in the project planning stage by simulator or computer
studies. Measures added later to a completed system are always very expensive.

6.4.6 Dynamic Interactions with Near-by Power Plants

6.4.6.1 Introduction

The endangering of turbogenerators located close to an HVDC through subsynchronous


instability has already been pointed out in Section 6.4.4. Additional problems may arise
from the proximity of an HVDC to a power plant which can be grouped into the category
“dynamic interactions”. The direct connection of a power plant and an HVDC rectifier
station in the form of unit connection will not be discussed at this point since the degree of
freedom with respect to torsional speed results in a completely different dynamic system
performance.

The assumptions for this discussion are that an HVDC station is located in the vicinity of a
power plant, the nominal power of both are of the same order of magnitude, and that the
power plant is connected through ac high voltage lines with the meshed network. In
contrast to Section 6.4.4, the operating mode of the HVDC station is left open. Both,
rectifier and inverter operation are considered. The type of power plant is also left open;
both, hydro and thermal power plants are considered.

6.4.6.2 Frequency Deviations

Disturbances in the HVDC system may result in a 100% load rejection of the HVDC as
the extreme case. Depending upon the operating mode of the HVDC station, the initial
frequency deviation will be positive or negative. The power deficit must be absorbed by
the ac network. The generators are brought back into the steady-state operating condition
in the form of dampened oscillations. A sufficiently strong coupling of the power plant by
connecting ac lines of low impedance is necessary for this. It must be demonstrated by
means of stability studies that this catastrophic event will be withstood without loss of
synchronism. For the much more probable case of a 50% load rejection, the HVDC can,
by utilizing the short-time overload capacity of the remaining pole, reduce the power jump
and, by power modulation, effectively damp the electromechanical oscillations triggered
from the disturbance.

Also if the 100% load rejection lasts only a short period of time (for example, fault
clearing in the ac system of the opposite station within 100 ms), the HVDC can handle the

- 528 -
effects of the disturbance and restore steady-state operation of the power plant quickly by
means of damping control.

If the power deficit is caused by events in the ac network of the power plant, the HVDC
will be available with its full power for the suppression of effects of the disturbance. An
additional catastrophic event, but one which is practically unimaginable, would be a
separation of the system consisting of power plant and inverter station from the meshed
network. The power fed in through the HVDC would add to the load-rejection acceleration
of the power plant machines. The high rate of change of frequency can be used as a
criterion for fast shut-down of the HVDC (for example, by-pass control within 10 ms). It
may be doubtful whether the much more effective fast change of energy direction of the
HVDC can be tolerated by the ac network of the opposite station because the power jump
will achieve up to twice the transmission power of the HVDC. If such a fault event is
conceivable at all, it should be carefully investigated with respect to its effects and
countermeasures.

A hydroplant which is far remote from the consumer center and whose power is almost
exclusively transmitted over an HVDC system but which also supplies a small local ac
network can be considered a special case. In this case, it may make sense in the context of
total system optimization to tolerate the frequency deviation due to a 50% load rejection,
that is to continue energy transmission with one pole of the HVDC, but to temporarily
disconnect the local network until restoration of the frequency in the operating range. This
approaches the dynamic performance of unit connection.

The requirements for reliability of energy transmission are especially high in the case of a
nuclear power plant. In the above scenario, the frequency deviation due to a 50% load
rejection of the HVDC would have to be held very small. That requires a high short-time
overload capability of the remaining HVDC pole.

The possibility of 100% load rejection must be completely excluded. This requires that the
two poles of the opposite station must be connected to two separate ac busbar sections
which are linked only at such a distance that a 3-phase short circuit will never cause
commutation faults in both poles of the inverter station.

- 529 -
6.4.6.3 Problems of Frequency Regulation

This problem will only occur if the near-by HVDC station is operating as a rectifier and is
receiving a significant portion of the power generated by the power plant. The cause is the
frequency-independence of the power of the HVDC. The HVDC does not provide any
damping of torsional speed fluctuations of the machinery. This makes high demands on
the turbine governor. If these demands cannot be fulfilled, the HVDC must intervene by
damping control of electromechanical oscillations as described in detail in Section 6.4.3.

6.4.6.4 Overvoltage Problems

The load rejection processes described in Section 6.4.6.2 also have effects on the voltage
in the ac network. Especially if the HVDC station is operating in rectifier mode and is
receiving a significant portion of the power plant output, high temporary overvoltages will
occur upon the blocking of the converters if no filter circuits and capacitor banks are
disconnected. Even worse, self-excitation of the generator may occur since it is now only
slightly loaded.

Therefore the blocking of the converters must always result in immediate disconnection of
the majority of the ac filter circuits and of all capacitor banks. The breakers for the ac filter
circuits and capacitor banks must always have the ability to reliably disconnect the
increased capacitive power caused by the increased ac busbar voltage.

Preferably a load rejection of the HVDC should not be accompanied by blocking of the
converters. Disturbances in the opposite station or in its ac network may lead to
interruption of transmission of active power. The absorption of reactive power by the
rectifier will nevertheless be maintained if dc current is allowed to continue flowing at
reduced magnitude. In this way, load rejection overvoltages and generator self-excitation
are avoided.

DC-side faults, the clearing of which requires shutting off the direct current for more than
several hundred ms, are an exception. Such faults, however, always involve only one pole
of the HVDC and therefore result in 50 % load rejection. If necessary, the reactive power
demand of the remaining converter groups can be increased by the temporary lowering of
dc voltage in the unaffected pole to such an extent that the reactive power balance at the
interface to the ac network is essentially maintained.

- 530 -
By these measures, the disconnection of ac filter circuits and capacitor banks can be
completely avoided. This facilitates and speeds-up the restoration of the original power
transmission following the clearing of the fault.

Other events such as the clearing of a network fault close to the power plant can also result
in an ac overvoltage. Due to existing resonances for lower frequencies, these events can be
associated with electromagnetic oscillations to the point of voltage instability. The HVDC
can contribute to remedy the consequences of the fault as described in detail in Section
6.4.6.2.

6.4.6.5 Harmonic Stress of Generators

Harmonic stress of generators may be viewed as a problem when the power plant and the
HVDC station are located close together and a generous specification permits incomplete
absorption of the ac-side harmonic currents generated by the converter. The harmonic
currents flowing in the windings of the generators cause a heating of the stator windings
and of the damper windings of the rotors. In addition, mechanical vibrations and ripple
torques could develop.

Such concerns may be completely justified in the case of six-pulse converters. In


particular, currents of the fifth and seventh harmonics could cause a ripple torque of the
sixth harmonic which could coincide with a super-synchronous resonance of the rotating
masses in a turbogenerator. On the other hand, investigations of the CIGRE Working
Group 11/14.09 “HVDC Unit Connected Generators” have shown that harmonic loads
and stresses can be tolerated by a conventionally designed generator connected directly
with a 12-pulse converter (i.e., without any ac-side filter circuits) without significant
reduction of its rated apparent power.

This leads to the conclusion that the physical proximity to a normally configured HVDC
station with 12-pulse converters and ac filter circuits will cause no significant harmonic
stressing of the generators.

- 531 -
6.4.7 Dynamic Performance of HVDC Multiterminal Systems

6.4.7.1 Introduction

IEC publication 9 19-3, which deals with the dynamic performance of HVDC systems,
expressly exclude HVDC multiterminal systems. Accordingly, only the dynamic
performance of two-point HVDC systems was discussed in the above Sections 6.4.1
through 6.4.6. In many cases, coordinated actions of both stations were necessary to
achieve a desired or specified performance. Therefore the results cannot be automatically
transferred to HVDC multiterminal systems.

Nevertheless, some essentials can be said about the dynamic performance of HVDC
multiterminal systems which can be expected or achieved. The prerequisit for this
discussion will be the typical HVDC multiterminal HVDC system as described in Section
6.3.9, the stations of which are regulated by the current margin method, with current and
voltage reference values determined by a central reference value computer.

6.4.7.2 Voltage Stability

The phenomena described in Section 6.4.2 pertain only to the ac network of one side and
the HVDC station connected to it. This applies for the reactive power characteristic of the
ac network, the HVDC station, to the measures investigated for maintaining the reactive
power balance, and to the voltage control. This also applies to interventions in converter
regulation and to the use of supplemental, continuously controlled sources of reactive
power. In all cases, only one side of the HVDC is involved. The results and conclusions of
Section 6.4.2 are therefore fully transferable to an HVDC multiterminal system.

6.4.7.3 Steady-State and Transient Stability

The phenomena described in Section 6.4.3 of electromechanical oscillations and the


associated endangering of steady-state and transient stability always were in reference only
to one ac network. Both stations, however, can be affected by it if the HVDC is integrated
as a whole into an ac meshed network covering a large area. More importantly, in many
cases, interventions are made in the control systems of both stations to damp
electromechanical oscillations and to ensure steady-state and transient stability. The
following generalizations can be made:

- 532 -
• Results and conclusions on phenomena for which only small signal modulation is used
without the involvement of the opposite station or the use of telecommunications can
be applied without limitation to HVDC multiterminal systems where they fall in the
category “local control.”

• Results and conclusions on phenomena for which large signal modulation is used with
intervention in the control system of the opposite station and thus telecommunications
between the two stations is involved, cannot be applied to HVDC multiterminal
systems. Such phenomena fall into the category “central control.”

In this discussion, central control means the damping of electromechanical oscillations in


the ac network (or network portion) of one partner by well-conceived signal modulation of
its station with the involvement of several other stations. The central reference value
computer generates modulated current reference values which are transmitted to the
involved stations. The modulation signal is transmitted to the central reference value
computer from the affected station or from a measuring site in the associated ac network
or network portion. There are three methods by which the other stations can participate in
the modulation power:

• Only stations of the opposite operating mode participate in providing the modulation
power. (If the affected station is an inverter, only the rectifier stations are involved in
the modulation power, and vice versa). The degree of involvement must be agreed
upon in advance.

• Only stations of the same operating mode participate in providing the modulation
power in accordance with a previously agreed key. Their power is modulated in
counterphase to that of the affected station.

• All stations participate in the modulation power. Stations of the same operating mode
are controlled in phase opposition, while the stations of the opposite operating mode
are controlled in phase with the affected station.

In the last case, the power modulation affects all stations, but it results in the smallest
possible amplitude for each of them.

With central control, all effects described in Section 6.4.3 with strongly modulated power
can also be achieved in a multiterminal system. The only limit to power modulation is at
the threshold of energy direction change in one station which in a multiterminal system

- 533 -
can be accomplished only with the help of polarity reversal switches. Thus it is not an
option for a modulation even at frequencies around 0.1 Hz.

6.4.7.4 Other Dynamic Performance

For the other types of dynamic performance of two-point HVDC systems discussed in
Sections 6.4.4 through 6.4.6, the applicability to multiterminal HVDC systems is as
follows:

• Measures for the suppression of subsynchronous instabilities are always limited to one
station namely the station which is connected to the affected ac network. Therefore
there are no essential differences to the local control in a multiterminal system.

• Measures for the suppression for higher frequency instabilities are also limited to the
station associated with the affected ac network, at least to the extent an intervention in
the converter regulation or the triggering device is supposed to take place. In the case of
main circuit measures such as modification of smoothing reactor inductivity, all
possible changes of the dc-side network configuration of a multiterminal system will
have to be considered.

In summary, it can be stated that the dynamic performance of a multiterminal HVDC


network can be predicted just as well as that of a two-point system and that the appropriate
measures for the improvement of the dynamic performance of a multiterminal HVDC
system are also available.

- 534 -
6.5 System Reliability

The required reliability of an HVDC system is expressed in the permitted number of


forced outages during a year and the allowed nonavailability of the system.

Devices and components should be sized and subsystems should be structured in such a
manner that the specified reliability requirements will be fulfilled. Since these
requirements are expressed as statistical parameters, compliance can be monitored only
over a relatively long period of time, theoretically only over the entire useful life of the
system. In the project planning stage, only a comprehensive reliability study can provide
an indication of the system reliability to be expected. This study should consider the useful
life and outage rates of components as well as the redundant structures provided and the
spare parts and maintenance philosophy.

In many specifications, substantiation of reliability, based on calculations is required. If,


on the other hand, a guarantee of such values is required which extends over a reference
period of one or two years, appropriate reservations must be made in view of the statistical
nature of outage rates including their typical development in the form of a “bathtub curve”.

6.5.1 Definitions

The definitions and designations of quantities of the CIGRE 14-89 (WG 04)-21 “Protocol
for Reporting the Operational Performance for HVDC Transmission Systems” shall be
used in this section. The reliability statistics of the HVDC systems in operation worldwide
which are published every two years by CIGRE are compiled in accordance with this
protocol.

6.5.1.1 Reliability

This should be understood as the ability of the HVDC system to transmit a certain power
— within the rating — for a specified period of time without interruption or limitation and
to do so under specified system and ambient conditions.

- 535 -
6.5.1.2 Outage

This is an event related to the HVDC equipment in which the ability to transmit energy in
accordance with 6.5.1.1 is lost temporarily or permanently. The outage can be intended or
unintended. The terms scheduled and forced outage are used.

6.5.1.3 Scheduled Outage

This is an intended outage, planned at least one week in advance or which could have been
deferred if necessary for at least one week (deferred maintenance).

6.5.1.4 Forced Outage

This refers to the state in which an equipment is unavailable for normal operation and
includes every automatic protective outage and every intended outage which does not
correspond to the criteria of the scheduled outage under 6.5.1.3.

Transmission outages due to network faults or other external events are not considered
forced outages to the extent the HVDC reacts in accordance with specifications and,
following the clearing of the fault, energy transmission is restored.

In addition, outages due to operating errors are not considered forced outages.

6.5.1.5 Period Hours (PH)

This refers to a calendar year with 8760 or 8784 (leap year) hours.

6.5.1.6 Outage Hours

This is the sum of actual outage durations within the reporting period expressed in hours
from the beginning of an outage to the point in time at which energy transmission was
resumed or could have been resumed. The following distinctions are made:

• Actual outage hours (AOH)

• Equivalent outage hours (EOH)

- 536 -
Equivalent outage hours are computed as follows:

EOH = ODF × AOH

Where ODF (Outage Derating Factor) is the loss of transmission capacity caused by the
outage expressed in p.u. In the event of loss of one pole of a bipolar HVDC system, for
example, ODF = 0.5.

6.5.1.7 Energy Unavailability (EU)

This represents the quantity of energy which cannot be transmitted during the period hours
as a result of outages. EU is usually given as a percent and is calculated as follows:

EOH
EU = ×100 [%]
PH

A distinction is made between

• Scheduled Energy Unavailability (SEU) and

• Forced Energy Unavailability (FEU)

depending upon whether the unavailability is caused by scheduled or forced outages.

The relationship of scheduled and forced energy unavailability is as follows:

EU = SEU + FEU

6.5.1.8 Energy Availability (EA)

This refers to the energy quantity related to the rated transmission capacity which the
HVDC system could have transmitted during the period hours, taking into account the
energy unavailability.

EA=100 - EU [%]

6.5.1.9 Energy Utilization (U)

- 537 -
This represents the relationship of energy actually transmitted during the period hours
(regardless of the energy direction) to the product of rated power times period hours. U is
customarily given as a percent.

6.5.1.10 Transient Reliability

Transient reliability is the relationship of the number of ac network faults in which the
HVDC system behaves according to specifications to the total number of ac network faults
during the period hours. Only such faults are counted in which the voltage at the ac busbar
drops in one or more phases to £ 85% of the voltage before the disturbance and increases
after the fault is cleared to at least the value required by the specifications for emergency
operation of the HVDC.

6.5.1.11 External Events

If the causes of outages or malfunctions of the HVDC system are beyond the scope of the
contract, such events and their consequences shall not be considered in the evaluation of
the degree of reliability achieved. This should be clearly stated in the contract.

Note:
In HVDC two-point systems, it is customary to relate the reliability indices to the entire
system. In HVDC multiterminal systems, and also in two-point systems with different
parties awarding the contracts for the two HVDC stations, reliability indices should be
determined separately for each HVDC station. In this case, outages which were caused in
another station are also considered “external events.”

6.5.2 Severity Code

The consequences of outages on the ac networks, especially of forced outages, are in large
part determined by whether the energy transmission was completely interrupted or was
only reduced by a certain percentage (ODF). Especially in the case of bipolar systems
which at the time of the fault were not fully loaded or which have a significant temporary
overload capacity, the effect of an outage limited to one system pole or to one converter
group can be largely cushioned. Accordingly, the CIGRE-Protocol provides for the
classification and designation of forced outages in three severity code categories:

BP = Bipolar outage

- 538 -
P = Monopolar outage

C = Converter Group Outage

In monopolar systems with only one converter group per station, there is of course only the
category P. Therefore the designation C is not necessary.

If in a bipolar system, two monopolar outages overlap which are attributable to events
independent from each other, they are treated as two outages of category P. If, however,
the second outage is caused by the first outage or by its consequences, the disturbance is
treated as a category BP outage.

The definitions and distinctions must be clearly stated in the specifications.

6.5.3 Reliability Requirements

There is currently no uniform practice for the specification of requirements for reliability
of HVDC systems. In general however, it can be stated that the requirements are
continually becoming more strict. This goes hand in hand with the continually increasing
reliability of newer HVDC systems as borne out in the CIGRE statistics.

6.5.3.1 Forced Outages

Category BP forced outages are permitted, if at all, one to two times per year as the
maximum. Category P forced outages may be permitted 10 to 20 times per year, and less
in the case of back-to-back links.

It is essential during project planning to assign a certain number of permitted forced


outages to the individual system parts and to supply this limit to the manufacturers or
subcontractors. The numerical values for this are derived from the reliability calculation
described in the next section.

6.5.3.2 Forced Energy Unavailability

This value, if it is given at all in the specifications, is usually specified at <= 1%. This
means less than 90 hours per year. It should be noted that the replacement of a converter
transformer which has failed by a spare unit is included in this time.

- 539 -
6.5.3.3 Energy Availability

This value is nearly always specified. Depending upon system configuration and spare part
philosophy, it can be specified at 95% to 98%. All unavailabilities must be taken into
consideration whereby the shortest possible time for service and repair work, using a
reasonable number of personnel, shall be taken into account.

6.5.4 Calculation of Anticipated Reliability

A reliability study is an essential component of project planning for an HVDC system for
which specific reliability values are specified. The following data must be known for all
devices and components which go into the calculation:

• Failure rate, possibly as a function of the system operating years (bathtub curve)

• Average repair-free time

• Duration of a failure (time required for fault identification, repair or replacement, and
resumption of operation)

• Time required for and frequency of regular maintenance

Redundant structures of facilities for control, regulation, protection, and measurement and
redundancies provided in converter valves and cooling facilities are to be considered in the
calculation.

Spare parts provided, especially for heavy equipment such as transformers and smoothing
reactors are also important elements of the study.

The calculation should extend over the theoretical life of the system such as 30 years. The
beginning and the end of the life of the system are of special interest.

With respect to the results of the calculation the following applies:


• The possibility of a forced outage of category BP as a result of a single event must be
absolutely excluded; the (n-1) criterion must be strictly maintained.
• A forced outage of category P as a result of a single event will be permitted in the case
of failure of a heavy equipment. In all other cases, the (n-1) criterion must apply here as
well.

- 540 -
7. Insulation Coordination and System Studies for HVDC

7.1 Introduction

Insulation coordination of an HVDC system cannot be limited to the HVDC stations


themselves. The configuration, parameters, and behaviour of the ac networks on both sides
of the HVDC and the dc line which connects the two stations must be considered.

HVDC insulation coordination is a system question!

What is insulation coordination? The following definition is found in IEC Publication


71.1: "Insulation coordination is the selection of the electrical ability of a device with
consideration to the voltages which can occur in the system provided for the device and to
the characteristics of the available protective devices. The goal is to limit, to an
economically acceptable level, the likelihood that the resulting voltage stress will damage
the insulation or the device or impair its operation." The first consideration is thus to
define the possible overvoltages in the system. They are characterized by amplitude, wave
shape, energy content, and frequency of occurrence. Then it must be decided if and how
the device under consideration can be protected against damage from these overvoltages.
The result is a configuration containing surge arresters. These, of course, must not be
damaged by either the voltages applied in normal operation nor by the amounts of energy
which must be absorbed in performing their protective function. The remaining voltages
are determining the voltage and/or insulation rating of the device.

The insulation coordination of an HVDC station differs in a number of essential aspects


from insulation coordination of an ac switchgear or substation. As a result of the converter
arrangements and the associated valve-side transformer windings being connected in
series, the surge arresters employed for their protection do not have any direct connection
to the protective ground. To this extent, overvoltages can only be limited through the
combined action of several arresters connected in series. In addition, the seemingly
insignificant task of defining continuous operating voltage for arresters in the converter
area presents quite significant difficulties. These difficulties are only exceeded by the
difficulty of establishing a test voltage adequate to the operating voltage which can be
simulated in the high voltage laboratory.

- 541 -
Adding to the difficulty is the fact that the converter control system has an essential
influence on the makeup of the operating voltage consisting of
• DC voltage
• AC voltage of network frequency
• AC voltages of higher (harmonic) frequencies
• High frequency (commutation) oscillations

The control reaction on transient events is often so fast that the waveshape of the resulting
overvoltage is significantly determined by it.

For this reason, it is necessary in considering the total system, also to include the functions
of the converter control. This makes the determination of initial data of insulation
coordination difficult and time consuming. In addition, it places high demands on the
flexibility and modelling accuracy of the tools employed.

The establishment of arrester data is an iterative process. The current-voltage characteristic


of metal oxide arresters - which are exclusively assumed - determines the energy
absorption in a given overvoltage event and thus the required number of arrester columns
which must be connected in parallel. This yields the residual voltage which occurs in the
presence of the "coordination current" at the terminals of the arresters. This residual
voltage in turn is critical for the dimensioning of the device insulation.

An important aspect of this process is the optimization between the cost of overvoltage
arresters for achieving a low residual voltage and the costs associated with voltage rating
and/or insulation dimensioning of the device.

The information in the following sections is based essentially on the results of the
following CIGRE Working Groups:

33-05: "Application Guide for Insulation Coordination and Arrester Protection of


HVDC Converter Stations"
33/14-05: "Application Guide for Metal Oxide Arresters without Gaps for HVDC
Converter Stations"

The first study appeared in "Electra," No. 96 (1984). The second was published as a
CIGRE brochure.

- 542 -
7.2 Arrester Protection Configuration

7.2.1 Historical Development

At the end of the 1960's, the companies in the HVDC Working Group (AEG, BBC, and
Siemens) the so-called Arbeitsgemeinschaft HGÜ were faced with the task of developing
the insulation coordination for the first HVDC long-distance transmission system using
thyristor valves. Never before there had been a transmission voltage of more than a million
volts (± 533 kV); never before a series connection of four 6-pulse bridges per station pole
had been implemented. Comparable HVDC systems such as the Pacific Intertie (± 400
kV) used mercury-arc valves which themselves provide overvoltage protection. For
thyristor valves, however, overvoltage protection had to be created. At that time, only
ordinary overvoltage arresters with air gaps were available. In addition, knowledge of the
nature and magnitude of overvoltages to be expected in HVDC systems was very limited.

Without exaggeration, the development of the insulation coordination in the Cahora Bassa
project can be described as pioneering work by the HVDC Working Group. This solution
has proved itself magnificently and in addition has become the prototype for insulation
coordination and overvoltage protection of all HVDC systems which have followed.
Figure 7.1 shows the arrester protection scheme developed for Cahora Bassa. For purposes
of simplification of the drawing, only one station pole of the first stage of the project is
presented with two 6-pulse bridges. The 6-pulse bridges added in the second and third
stages of the project, however, are identically configured.
A B C DC1 D
DC7

DC2
SC1
DC8

DC3

DC4

DC9

SC2 DC5
DC10

DC6

Figure 7.1: Arrester protection scheme of the converter stations in the Cahora Bassa
HVDC Project (first stage of construction)
A, B, C, D = Types of arresters
Designations of the switching devices in accordance with CIGRE 13/14.08

- 543 -
The A-arresters in delta configuration on the valve side of the converter transformers have
not proven to be necessary. Today they are connected on the network side of the converter
transformers, as close as possible to the bushings. In combination with the B-arresters, this
assures adequate overvoltage protection for the converter transformers. The key item of
the arrester scheme is the direct parallel connection of the B-arresters to the thyristor
valves. In the oil-insulated, open-air valves used in Cahora Bassa, they are directly
adjacent to the valve bushings. The C arrester limits the voltage across the dc side
terminals of the bridge to a low level when the valves are current-less and therefore two
series-connected B-arresters are effective across the dc side terminals.

The energy-absorbing capacity of silicon carbide gap type arresters was very limited, and a
parallel connection of arresters was not possible. Therefore it was accepted as inevitable
that an arrester had to "sacrifice" itself in certain fault events to perform its protective
function. This occurred only once, however, during test operation in the case of a C
arrester in the Apollo inverter station. During five years of commercial operation of this
system, not a single arrester has had to "sacrifice" itself. The arresters have performed
reliably, and there has been no overvoltage damage in the system.

The protective scheme of the Nelson River II HVDC system - configured in the same way
with the same arresters, though adapted to indoor valves - has also proven its effectiveness
in almost twenty years of operation.

7.2.2 Customary Arrester Schemes of Today

As mentioned above, the arrester protection scheme developed for Cahora Bassa can be
considered the prototype of the arrester arrangement customary in HVDC stations today.
Figure 7.2 shows the arrester protection scheme of an HVDC station pole with one twelve-
pulse group taken from IEC Publication 919-2.

- 544 -
R

V CB DB DL

M
FD

E1 E2

Figure 7.2: Arrester protection scheme for an HVDC station consisting of one twelve-
pulse group per pole with arrester designations per IEC 919-2

IEC already assumes the use of metal oxide, MO, arresters without air gaps. Here the C
arresters of Fig. 7.1 are replaced by arresters CB and M which are connected to ground. In
addition, the smoothing reactor is protected by the parallel-connected arrester R. This is a
somewhat questionable measure, because it at least partially undermines one of the
essential protection functions of the smoothing reactor, namely the prevention of the entry
of steep waves of surge voltage into the converter area. Therefore in most newer HVDC
systems, the R arrester is omitted. In addition, in pure cable transmissions, arresters DB
and DL are superfluous since overvoltages on the dc side caused by atmospheric events
can be eliminated by effective lightning protection of the station. Note that the IEC arrester
arrangement depicted in Fig. 7.2 is intended only as an example. Deviations from it
certainly appear in publications and in completed HVDC systems, both with respect to
arrester arrangement and designation of the arresters. Thus, for example, Siemens -
following the Cahora Bassa scheme depicted in Fig. 7.1 - uses a C arrester parallel to each
twelve-pulse group which is not connected to ground.

- 545 -
7.2.3 Arrester Protection Scheme for a Back-to-Back-Link

There is a significant simplification for the arrester scheme of an HVDC back-to-back


link. As shown in Fig. 7.3, the dc-side arresters can be completely omitted.

V V

A A

V V

Figure 7.3: Arrester scheme of an HVDC back-to-back link with one twelve-pulse group
per side

7.2.4 Filter Protection Scheme

The ac filter circuits are not depicted in the arrester protection schemes of Figures 7.1
through 7.3. Overvoltage protection of the ac filter circuits was discussed in detail in
Section 2.2.6. Fig. 7.4 shows the arrester arrangement of one pole of a double-tuned filter
circuit. The filter circuit arresters are not intended to provide protection against external
overvoltages. Instead, their function is to prevent the full busbar voltage from being
applied to the reactors and resistors at the moment when the filter circuit is energized. In
the case of a busbar ground fault the arresters prevent the voltage of the main capacitor
from being applied to the filter reactors and resistors during the fault.

- 546 -
The dc filter circuit depicted in Fig. 7.4 follows the same principle (see also Section 2.6.7).

+
-

Figure 7.4: Arrester arrangement of a double-tuned dc filter circuit

7.2.5 Selection of the Arrester Protection Scheme

CIGRE Working Group 33-05 has formulated the following principles for the selection of
a suitable arrester protection scheme:

• Overvoltages originating on the ac side should be limited as far as possible by arresters


on the ac side of the HVDC station, with the busbar arresters playing the major role.

• Overvoltages originating on the dc side should be limited by arresters on the dc side,


primarily by arresters at the point of entry of the line, at the dc busbar, and at the neutral
bus of the station.

• Critical components should be protected by arresters assigned specifically to them and


located in their direct vicinity. This applies, for example, to thyristor valves, the line-
side of converter transformers, etc. Overvoltage protection of valve-side transformer
windings can be left to series-connected arresters on the dc side.

- 547 -
7.2.6 Functions of the Arresters in an HVDC Station

As in Fig. 7.2 the arrester designations per IEC 919-2 are used in the following listing. In
some cases, these designations deviate from the CIGRE nomenclature and from that used
by Siemens.

7.2.6.1 Valve Arrester (V)

The main function of the V-Arrester is to protect the associated thyristor valve from excess
voltage stresses regardless of origin. Its protection level and the valve protective firing,
effective in the conductive direction of the thyristors, determine the voltage dimensioning
of the valve. For this purpose, redundant thyristors are ignored.

Since valve costs and losses are roughly proportional to the voltage dimensioning, the
protection level of the V arrester should be selected as low as possible.

At this point, a brief historical consideration may be appropriate. In 1970, the author gave
a lecture titled "HVDC Systems and Their Planning," which was published in the ETZ in
1971. The following is an excerpt on the subject of insulation coordination and voltage
dimensioning of thyristor valves.

"The critical factors influencing the voltage dimensioning of an HVDC converter valve
and the insulation coordination are represented schematically in the figure. Starting with
the nominal dc voltage of the converter, the various factors which lead to the maximum
crest voltage of the valve which can be expected in operation are represented by arrows.
These factors, combined with the use of an overvoltage arrester, are the basis of voltage
dimensioning of the valve. The lower the individual factors can be set, the lower the total
voltage dimensioning factor of the valve and thus the less expensive the system will be.

- 548 -
U n max / U n N , R n x Û RRL
Û Steh
Û A max
Arrester data
Û A min
Û Sp max
Cs , D

S"k Û lo
1,05 Û lN
dx U di
U do
a, g U dN
1 2 3 4 5 6 7

Figure 7.5: Voltage dimensioning of the HVDC valve (See text for explanation)

(1) The control angle a in rectifier operation determines the control margin and thus the
dynamic control response of the system. The extinction angle g determines the safety of
the system against commutation failures in inverter operation. A reasonable compromise
must be found between technical performance and economy of the system.

(2) The internal voltage drop of converter dx is essentially determined by the transformer
reactance which, however, also determines the maximum current in the event of an
internal short circuit. Therefore, a change in transformer reactance has an opposite effect
on the voltage and current dimensioning of the valve.
(3) The ac busbar voltage rise in the event of load rejection is determined by the short-
circuit power S"k. This can be beneficially influenced by the use of synchronous
condensers. An optimal overall solution must be found which also takes into consideration
the problem of providing the required reactive power. Special control procedures, which
have not yet been developed to the point of practical application, can significantly reduce
the load-rejection factor.

(4) The commutation oscillations inherent to converter operation can be influenced in the
project-planning stage both by suitable measures with the resonance circuit (e.g. limiting
the stray capacitance Cs of the transformer valve side connections) and by improving the
damping D of the resonance circuit (dimensioning of the snubber circuits). This involves
again a cost optimization.

(5) Starting with the maximum valve voltage USp max as determined above, the minimum
response voltage UA min of the overvoltage arrester which will be arranged parallel to the

- 549 -
valve must be determined. Here the project-planning engineer is faced with the task of
selecting a reasonable margin of safety. (Comment: a separate consideration of the arrester
extinguishing voltage will have to be made as well.)

(6) It may be necessary to establish another margin of safety between the protection level
of the arrester UA max and the valve withstand voltage Usteh.

(7) In establishing the dimensioning factor resulting from the withstand voltage, the
primary consideration in addition to the design-dependent internal distribution of voltage
Un max/UnN is the valve redundancy R, i.e., the percentage of valve elements connected in
series which are allowed to fail in the course of time without impairing the operability of
the valve. Redundancy determines the average operating time which can be expected
between valve overhauls. Here too, an optimization should be made between device costs
on one hand and maintenance and down-time costs on the other.

As a result, the voltage dimensioning factor fu is gained, which is the ratio of the product
of the number of thyristors connected in series n and the maximum acceptable periodic
peak blocking voltage of a thyristor UDRL or URRL to the nominal voltage of the converter
bridge UdN."

Of course, the above considerations apply for gap-type arresters with minimum and
maximum spark-over voltages and the regulating process for limiting the load-rejection
overvoltage has been introduced in practice since many years. Nevertheless, the statements
still apply and indicate the many ways, the project-planning engineer can influence the
voltage dimensioning of valves and consequently influence the cost factor of the HVDC
station. In addition, the V arrester determines the insulation dimensioning of the valve-side
transformer windings, both, along the winding and, in combination with other V arresters
and the station neutral bus arrester (E), against ground.

7.2.6.2 Bridge Arrester (M)

This arrester, which is connected to ground, has the same function as the C arrester in Fig.
7.1. It prevents the protection level of the lower bridge from climbing at no-load to the
value of two series-connected V arresters, while in converter operation, at least one valve
per commutation group is conductive and thus bypasses the associated V arrester.

- 550 -
7.2.6.3 Converter Group Arrester (CB)

The purpose of the CB arrester is to protect the converter arrangement of one pole of a
station against incoming travelling waves from the dc side. In many cases, the CB arrester
will be dispensable, specifically if the smoothing reactor is close to the valve hall, one
bushing projects into the hall, and no R arrester is connected in parallel to the smoothing
reactor.

7.2.6.4 DC Bus Arrester (DB)

The purpose of the DB arrester is to protect the dc-side equipment of the HVDC station
against overvoltages. In the case of dc switchgear which is spread out physically, several
DB arresters, spatially separated, should be provided in order to ensure a reliable lightning
protection on the dc-side of the station. The expense is held in bounds, because all DB
arresters can be considered as connected in parallel with respect to energy-absorbing
capacity for switching-surge overvoltages which is determinative for dimensioning.

7.2.6.5 DC-Line Arrester (DL)

The DL arrester protects the dc part of the HVDC station from incoming steep-surge
voltage travelling waves resulting from direct lightning strikes on the HVDC overhead
lines or from back flashovers when a lightning strikes a tower or the immediate vicinity.
The assumption is made that the first few spans next to the station are protected against
high current lightning-strikes by double ground wire. The wave fronts of lightning-strike
overvoltages which come from a relative long distance are flattened by the time they reach
the station to such an extent that the line arrester combined with smoothing reactor and dc
filter circuits guarantees an effective limitation of the lightning-surge level.

7.2.6.6 Station Neutral-Bus Arrester (E)

The E arrester protects all devices and components which are connected to the station
neutral bus or to the entry of the electrode line against overvoltages of varied origin.

Direct lightning strike to the electrode line must be considered a frequent occurrence,
particularly if it is used in the first kilometers of the right of way as the shield wire for the
HVDC overhead line.

- 551 -
Another fault situation which causes overvoltages of the station neutral bus is the ground
short circuit of the dc busbar of a station pole or a flashover within the valve area. Since
the electrode line and the earth electrode represent a high impedance connection to the
effective ground for events in the millisecond range, the potential of the neutral bus may
swing to high values if it were not protected by the E arrester. Thus its residual voltage
determines the insulation level of the devices and plant components which are connected
to the station neutral bus.

The voltage on the E arresters in normal operation is very low, even in the case of
"metallic return"; but the energy which must be absorbed in a fault event is often quite
significant.

As with the CB arrester, the E arrester may be subdivided into several spatially separated
units to protect against steep-front surge waves.

7.2.6.7 DC Filter Circuit Arrester (FD)

The FD arresters protect filter-circuit elements such as reactors and damping resistors
against overvoltages when a filter circuit is energized from the dc busbar or in the event of
a busbar short circuit to ground. In the latter case, the energy stored in the main capacitor
discharges into the FD arrester(s).

7.2.6.8 Smoothing Reactor Arrester (R)

In some early HVDC systems, an R arrester was employed to limit the voltage across the
winding in the event that transient voltages of opposite polarity occur on both sides of the
smoothing reactor. However, because this impairs the protection against in-coming
travelling waves from the HVDC overhead line, most systems do not employ R arresters.
In this case, however, the insulation across the winding of the smoothing reactor must be
dimensioned for the sum of the protective levels of DB and CB arresters.

7.2.6.9 AC Busbar Arrester (A)

Surge arresters must be provided where ac lines leave the station and as close as possible
to the network-side transformer bushings. They provide protection against lightning-strike
and switching-surge overvoltages. They also protect against "fast transients," occuring
particularly in encapsulated SF6 switchgear.

- 552 -
7.3 Characteristics of Metal Oxide Arresters

Since overvoltage arresters play a decisive role in the insulation coordination of an HVDC
system, the characteristics of arresters will be presented in greater detail below. The
discussion will be limited to MO arresters without air gaps which are used exclusively
today in HVDC systems.

7.3.1 Arrester Material and Mechanism of Conduction

The active material of the MO arrester consists of a ceramic, produced in a sintering


process. It is composed essentially of ZnO with small admixtures of other metal oxides
such as Bi2O3, Cr2O3, MnO, etc. The particles of metal oxide with a grain size of
approximately 10 mm have a very low specific resistance, but are surrounded by an oxide
layer, approximately 0.1 mm thick, which has a very high specific resistance. This
intergranular layer determines the arrester resistance in the range of low electric field
strength.

Above a critical field strength, the specific resistance of the oxide layer decreases from
around 1010 Wcm to values below 1 Wcm, and the arrester passes into a low-impedance
condition.

This effect can be explained by the conduction mechanism. The intergranular oxide layer
represents an energy barrier for the charge carriers, i.e., the electrons. Only a few have the
thermal energy to overcome this barrier. This also explains the temperature-dependence of
the arrester current at low field strengths.

When the electric field strength in the blocking layer reaches a value of about 106 V/cm,
electrons can overcome the energy barrier through tunnel effect. With increasing field
strength, the blocking effect of the intergranular layer decreases. Finally, essentially only
the internal resistance of the grains of ZnO remains.

Besides this conductive mechanism which is dependent on field strength, and only when
an ac voltage is applied, there is a capacitive current determined by capacitances between
the grains. This capacitive current, however, can only be noticed in the lower voltage
range.

- 553 -
The active material of the arrester is pressed into disks which may be 100 mm in diameter
and 20 mm thick. The arrester columns are composed of such disks, they are, alone or with
other columns, imbedded in suitable porcelain or polymer housings which serve as
protection against external influences. At the same time, the housings provide, by means
of ribs, the creepage path necessary for external insulation.

7.3.2 Current-Voltage Characteristic

The field-strength-dependent conduction mechanism of the MO arrester described above


becomes evident in its current-voltage characteristic. Fig. 7.6 shows the current-voltage
characteristic of an MO arrester element 80 mm in diameter and 20 mm thick.

Figure 7.6: Current-voltage characteristic of an MO arrester element 80 mm in diameter


and 20 mm thick. UA = arrester voltage; IA = arrester current

Three regions can be clearly distinguished in the characteristic of Fig. 7.6:

• In Region 1, below an arrester current of 1 mA, the temperature-dependency is clearly


evident. When ac voltage is applied, capacitive arrester currents also come into play in
this region.

- 554 -
• In Region 2, from around 1 mA to 1 kA, the arrester voltage undergoes little change. In
this range, the dependency of the arrester current on the voltage applied can be
approximated with the formula

IA = k × UAa

where 10 < a < 50.

• In Region 3, above around 1 kA, voltage increases sharply with increasing arrester
current. This is the range in which the intergranular layer has lost its blocking effect.
Here, only the resistance of the ZnO grains is effective, and the arrester behaves as an
ohmic resistance.

The current-voltage characteristic of an MO arrester unfortunately remains not always the


same. There are influences which can cause a permanent deviation:

• Excessively high continuous operating voltage


• Temporary overvoltage applied too long
• Excessive surge current stress
• Excessively high ambient temperatures

In addition, there is also a polarizing effect under conditions of dc voltage application.


After an abrupt change of polarity, the arrester current may become a multiple of the
original value, which returns only after approximately 200 hours. A deformation of the
blocking layer caused by ions migrating into the layer is made responsible for this effect.

When the loss heat generated in the arrester due to one of the above effects becomes so
great that the temperature continues to increase while continuous operating voltage is
applied, thermal instability will result.

For this reason, precise determination of the stresses on the arrester which occur in
operation and in the event of faults is of great importance. For thermal stability, arrester
temperature must return to the initial level while it is subjected to continuous operating
voltage, following the greatest stress which is to be expected. Aging effects must be taken
into consideration as well.

- 555 -
7.3.3 Arrester Data

For the insulation coordination of an HVDC system, reliable data are needed on the
behaviour and characteristics of MO arresters during continuous operation and for all fault
events which can be expected. Below, the most important arrester data are defined and
explained considering the special conditions in an HVDC system.

7.3.3.1 Arrester Reference Voltage Uref

This parameter, which must be established by the manufacturer, describes the crest value
of arrester voltage at a defined current, around the transition from Region 1 to Region 2 of
the current-voltage characteristic.

7.3.3.2 Arrester Reference Current Iref

The arrester reference current is the crest value of the arrester current established for the
determination of arrester reference voltage. It should be selected such that neither the
capacitive current of the arrester nor that of the measuring apparatus have a significant
influence.

7.3.3.3 Arrester Nominal Surge Current (Nominal Discharge Current)

The arrester nominal surge current has an impulse shape of 8/20 ms. Its peak value is used
for classifying the arrester.

- 556 -
7.3.3.4 Arrester Residual Voltage

Arrester residual voltage is the peak value of the voltage which appears at the arrester
terminals when a defined current such as the arrester nominal surge current is passing
through the arrester.

Since MO arresters have certain tolerances, the maximum arrester residual voltage, which
is critical for insulation coordination, is usually specified and guaranteed. However, since
the quantity of energy which must be absorbed by the arrester in the event of a fault
usually rises with decreasing residual voltage, the minimum arrester residual voltage
should also be specified.

7.3.3.5 Continuous Operation Voltage

With ac voltage arresters, this is the r.m.s. value of a sinusoidal ac voltage which may be
applied on a continuous basis even in unfavourable ambient conditions without resulting
in overload, thermal instability, or deterioration.

With dc voltage arresters, it is the highest dc voltage acceptable on a continuous basis at


the terminals of the arrester.

In contrast to this simple rules, the definition of the permissible continuous operating
voltage is very difficult for arresters in the area of the converter valves. For this reason,
three different characteristic values are defined:

a) Peak continuous operating voltage (PCOV) which describes the peak value of voltage
occurring at the arrester terminals in normal operation, including commutation
overshoot.

b) Crest value of the continuous operating voltage (CCOV) which describes the crest
value of the voltage occurring at the arrester terminals excluding commutation
overshoot.

c) Equivalent continuous operating voltage (ECOV) which describes a voltage of any


form causing the same losses, or resulting in the same heating as the continuous
operating voltage which is actually applied to the arrester terminals.

- 557 -
The equivalent continuous operation voltage is important for monitoring the thermal
stability of an arrester in the laboratory. Usually an "equivalent" ac voltage or a dc voltage
is selected, or occasionally a combination of the two. The proof of equivalence, i.e.,
determination of the correct ECOV is a challenging task, as will be explained in greater
detail in Section 7.4.

7.3.3.6 Arrester Continuous Current

This refers to the arrester current which is actually present while the continuous operating
voltage is applied. With arresters in the converter area, it also includes higher frequency
capacitive components. Therefore it is not necessarily identical with the arrester current
which occurs during the test with equivalent continuous operating voltage.

7.3.3.7 Coordinating Current

This is an arrester current with a pre-defined impulse shape for which the arrester voltage
is to be determined. As an approximation of the typical overvoltage shapes which occur in
high voltage systems, four different impulse shapes of the coordinating current are
distinguished:

a) The steep surge coordinating current (1/20 ms) is typical for ground short circuits within
the converter arrangement, such as flashover of a valve-side transformer bushing,
particularly at the upper bridge. A fault of this type results in a steep surge overvoltage
of the associated thyristor valves, limited by the parallel-connected V arresters. The V
arresters must absorb a steep surge current with a front time of around 1 ms. The falling
edge of the coordinating current defined for this fault event should not take more that
20 ms to drop to 50% of the peak value. The required crest value of the coordinating
current must be determined on a case by case basis by circuit analysis.

b) Lightning surge coordinating current (8/20 ms) is typical for overvoltages which are
triggered by a direct lightning strike on an overhead line. Although the HVDC station is
effectively shielded against lightning strike overvoltages between converter
transformers and smoothing reactors, this stress - due to its general significance (BIL
value) - is also used for insulation coordination and arrester testing within the HVDC
station.

c) Switching surge coordinating current (30...100/60...200 ms) is typical for the most
frequent form of arrester stress in an HVDC station. Sometimes it is a switching surge

- 558 -
on the ac side which is transferred cumulatively to the dc side. Sometimes it is a
lightning-surge travelling wave from the ac or dc overhead lines, which in combination
with transformer or smoothing reactor inductance and ac or dc filter circuit capacitance
is transformed into the form of a switching surge voltage. The crest value of the
coordinating current is usually selected at 2 or 3 kA. At this level, it will cover all the
stresses which can be expected for the arrester.

d) The slow-front surge coordinating current (1000/2000 ms) is typical for travelling
waves coming in from very great distances which can cause a significantly longer
overvoltage than is the case with a standard switching surge. Therefore, it is advisable
to test the arresters of the HVDC station with the slow-front surge coordinating current
and to determine the residual voltage.

7.3.3.8 Energy Withstand Capability

It is not so much the peak value of the arrester current as its duration that determines the
quantity of energy which must be absorbed and changed into heat. Reliably determining
the required energy withstand capability is one of the major tasks of an insulation
coordination study.

In contrast to the air-gap arresters used earlier, MO arresters offer the significant
advantage that practically any number of them can be connected in parallel. In this way, it
is possible to achieve any energy withstand capability required.

7.3.4 Arrester Protection Level

The protection level of an arrester which is defined by the continuous operating voltage
and the energy withstand capability is the maximum arrester residual voltage as
determined by testing the arrester with the coordinating current defined in the preceding
section. This maximum arrester residual voltage is different for the different types of
overvoltage. The crest value of coordinating currents must be determined for a specific
system in an insulation coordination study.

In the case of protection levels for steep-surge and lightning-surge stresses, the separation
distance between the arrester and the object to be protected must also be considered. The
decisive parameter is the relationship between the front time and twice the travel time of
the travelling wave. The greater the distance and the shorter the front time, the higher the
reflection overvoltage can become. Therefore, the V arresters are attached directly at the

- 559 -
valve tower or are set up in the immediate vicinity, just as the A arresters are attached on
the line-side bushings of the converter transformer. One possibility for lowering the
protection level consists in oversizing the arrester with respect to its energy withstand
capability (Fig. 7.7).

Figure 7.7: Protection level as a function of the number of parallel arrester columns n.
Parameter: disk diameter D

Whether this measure is worthwhile is determined by comparison of the added cost of


arresters with the costs saved by reduction of the voltage and/or insulation dimensioning.

- 560 -
7.4 Continuous Operating Voltage

The difficulties involved in establishing the equivalent operating voltage for the arresters
in the converter area of an HVDC system have already been pointed out. In the following
sections, the voltages which appear on the arresters during operation will be described and
analysed. Next the dependance on operating data and parameters will be discussed and
conclusions will be drawn on the basis of these interrelations with respect to arrester
dimensioning.

7.4.1 The Valve Arrester (V)

Fig. 7.8 shows the voltage across a thyristor valve in a twelve-pulse converter group.

a t

PCOV CCOV

(a +u)

Figure 7.8: Continuous operating voltage of a V arrester in rectifier operation


Angle of delay a = 20° el
Angle of overlap u = 24° el
Overshoot k = 1.15

This continuous operating voltage curve which also applies for the V arrester is composed
of a dc voltage component, a fundamental frequency ac voltage component, and
components of higher frequencies. The commutation overshoots impressed on the voltage
curve at discrete points deserve special attention. These are highly damped high-frequency
oscillations, the amplitude of which depends to a large degree on a and u. They must not
result in any unacceptable heating of the V arrester.

- 561 -
The crest value of the voltage CCOV is proportional to the ideal no-load dc voltage of the
six-pulse bridge and is computed by the following formula:

p
CCOV = · Udiomax
3

The maximum bridge voltage which occurs in operation must be taken into consideration.
Usually this will be in rectifier operation at maximum transmission direct current.

The dc voltage component is essentially dependent upon the degree of control of the
æ uö
converter, i.e., upon the angle of delay a. It fluctuates between 0 at ç a + ÷ = 90° and
è 2ø
around 0.45×CCOV at a = 0.

The commutation oscillations are dependent upon the leakage capacitances of the
commutating circuit and the RC damping (snubber circuits) of the thyristor valves as well
as the operating parameters (a, u). The maximum value is achieved at (a + u = 90°). In
general the following equation applies:

PCOV = (1.1.....1.25) × CCOV

The equivalent continuous operating voltage (ECOV) can be computed from the actual
voltage curve occuring at the most unfavourable operating parameters by application of
the current-voltage characteristics of the arresters intended for use.

7.4.2 The Bridge Arrester (M)

Fig. 7.9 shows the voltage at the dc voltage terminals of a six-pulse bridge; ignoring the
potential of the station neutral point, which in normal operation is very low, this is also the
voltage of the M arrester.

- 562 -
K

CCOV PCOV

Figure 7.9: Continuous operating voltage of the M arrester in rectifier operation


Operating parameters: a = 20°, u = 24°, k = 1.15

The crest value of continuous operating voltage is about the same as in the V arrester. The
voltage curve, however, is fundamentally different. It is a dc voltage with superimposed ac
voltage components of higher frequencies. There is no fundamental frequency component.
Commutation overshoot K occurs at the same amplitude as in the V arrester.

7.4.3 The Converter Group Arrester (CB)

In HVDC stations with two (or more) converter groups per station pole connected in
series, MO arresters are connected in parallel to the dc terminals of each group. If the
station consists of only one twelve-pulse group per station pole, the converter group
arrester (CB) can also be connected to the station ground, as shown in Fig. 7.2. Fig. 7.10
shows the voltage curve on the CB arrester.

The higher frequency ac voltage components and the commutation overshoots are the
same as those at the bridge arrester (M), but related to the dc voltage they are only half the
magnitude and they occur twice as often.

- 563 -
K

PCOV CCOV

Figure 7.10: Continuous operating voltage of the CB arrester (designations and operating
parameters the same as in Fig. 7.9)

7.4.4 Other Arresters

The ac busbar arresters (A) or the arresters parallel to the network-side transformer
bushings see more or less sinusoidal ac voltages in normal operation. They are
dimensioned in accordance with the rules of ac technology or following the practice of the
responsible power company.

The filter-circuit arresters (FA) on the ac side or (FD) on the dc side of the HVDC station
are exposed to only very low terminal voltage in normal operation. They are dimensioned
according to the requirements of transient stresses. Determination of a continuous
operating voltage is of no significance.

The same applies for arresters at the station neutral bus (E). Their continuous operating
voltage is negligible and has no significance for dimensioning.

The situation is naturally quite different in the case of dc-line arresters (DL) and of dc-bus
arresters (DB). Their continuous operating voltage can be considered as a pure dc voltage
the magnitude of which is determined by the dc voltage regulation of the converter. In the
case of the arrester parallel to the terminals of the smoothing reactor (R), a multi-
frequency ac voltage occurs in operation which is dominated by the twelfth harmonic. It is
small compared with the transient voltage (Ldi/dt), which occurs during fault events and
determines the dimensioning. Nevertheless, because of the capacitive arrester currents

- 564 -
caused by the relatively high-frequency harmonics, the definition of a continuous
operating voltage may be significant. For this purpose, the various operating conditions
acceptable on a continuous basis must be analyzed in order to determine the harmonic
voltages which will occur at the smoothing reactor under the most unfavorable conditions.
These harmonic voltages result from the interaction of the converter which acts as a
voltage source and the harmonic impedances which result from the combination of reactor
inductivity, the dc filter circuits, and the dc line.

- 565 -
7.5 Temporary Overvoltages

Temporary overvoltages in ac networks are triggered by switching events, but mainly


through faults and their clearing and particularly during load-rejection events. Their
magnitude depends upon various parameters. In the case of very low short circuit ratio
(SCR) at the connecting point of an HVDC station, temporary overvoltages as a result of
HVDC load-rejections are particularly predominant. Overvoltages of this type are damped
only slightly. They can last many cycles, even seconds, causing transformer saturation and
as a result strong voltage distortions which in turn can be significantly amplified in the
presence of system resonance such as for the second harmonic.

Temporary overvoltages can also occur on the dc side of HVDC stations as a result of
faults and in combination with malfunction of converter control, as will be illustrated at a
later point.

Because of their long duration, temporary overvoltages cannot be effectively limited or


reduced by surge arresters alone. Even multiple-column MO arresters do not have the
energy withstand capability necessary for this task. Therefore, unless other measures of
reduction are available, temporary overvoltages are to be fully included in voltage or
insulation dimensioning. This causes considerable additional costs and losses, particularly
in the case of thyristor valves.

The events which lead to temporary overvoltages on the ac and the dc side will be
considered in greater detail below. Measures to avoid them or at least to reduce them
significantly will be discussed. Temporary overvoltages which are critical for the
dimensioning of MO arresters can thus be defined.

- 566 -
7.5.1 Temporary Overvoltages on the AC Side

Temporary overvoltages at the ac bus of an HVDC station reach their highest value in the
event of a 100% load rejection of the HVDC with blocking of the converters (when no
commutation takes place and there is no dc current). Excess reactive power provided by ac
filter circuits and capacitor banks causes - particularly at low short circuit ratio - a high
overvoltage, resulting in the phenomena of transformer saturation and voltage distortion
referred to above.

With HVDC control systems of modern design, however, this situation only occurs as
result of a malfunction of the HVDC. Three cases can be distinguished:

• If the cause of the load rejection is in the other station or in the ac network connected to
it, the station which is not affected should maintain the flow of direct current, possibly
at reduced magnitude as a result of the voltage dependent current order limit (VDCOL)
function of the current regulation. With a properly selected current reduction (in the
event of direct voltage zero), the reactive power demand of this station remains about
the same; no significant overvoltage occurs.

• If the cause of the load rejection is in the dc circuit of the HVDC, there should always
be only one pole of a bipolar system which is affected. Thus there is a 50% load
rejection, the effects of which can be further reduced or even avoided by temporarily
increasing the direct current and/or reducing the direct voltage in the unaffected pole.
This may nearly double the reactive power demand of the remaining pole.

• If the cause of the load rejection is in the ac network of the station under consideration,
the danger of a temporary overvoltage will not exist until the moment of clearing the
fault. Therefore it is important for the converter to resume commutation operation and
begin drawing reactive power from the ac network without delay upon return of ac
network voltage. This assumes, of course, that the flow of direct current was
maintained during the fault event by the other station.

In all three cases, the temporary overvoltage is avoided or reduced to a moderate residual
value by the intervention of HVDC converter control. This effect may be taken into
consideration in insulation coordination and in the dimensioning of MO arresters to the
extent that a double fault, especially the malfunction of converter control in the critical
moment, can be excluded or the temporary overvoltage in such an event can be limited by
other means.

- 567 -
a) AC busbar voltage

b) Arrester current in one phase

c) Energy absorption of the arrester

Figure 7.11: Clearing of a three-phase network fault with blocked converters;


disconnection of the ac filter circuits after four cycles

As an example, Fig. 7.11 shows an oscillogram of the clearing of a three-phase system


fault, as presented by CIGRE-WG 33/14.05, in which the converters were blocked. As
back-up protection, the filter circuits are disconnected after four cycles. The MO arresters
at the ac busbar limit the strongly distorted voltage during the first four cycles after return
of voltage. In course of this process, they absorb considerable amounts of energy. After the
filter circuits are disconnected, the busbar voltage drops to such a degree that the arresters
no longer carry any discernible current and do not absorb any more energy.
In this way, costly voltage dimensioning for the full load-rejection overvoltage can be
avoided even in the case of high reliability requirements.

To the extent temporary overvoltages at the ac side must be reflected in insulation


coordination, they apply not only for the dimensioning of A arresters but also for the V

- 568 -
arresters and the arresters on the dc side of the HVDC station. Three cases must be
distinguished:

• Converter commutation continues. In this case, the voltage of the valve-side


transformer windings is applied in periodic alternation to the V arresters and the
temporary overvoltage of the ac side gets through to the bridges. The same applies to
the dc-side arresters of the HVDC station to the extent that the return of the direct
voltage to its pre-fault value is not accomplished by means of the converter control.

• The converters are transferred into bypass operation. In this case, the phase-to-phase
voltage of the valve-side transformer windings is applied continuously to two V
arresters of each commutation group and the temporary overvoltage is fully applied to
these arresters but not to the dc-side arresters of the HVDC station.

• The converters are blocked and there is no flow of direct current. In this case, the three
arresters associated with a commutation group are exposed to the transformer voltage
as a wye circuit with a 1/Ö3 share of the phase-to-phase voltage on each arrester. The V
arresters, as well as the thyristor valves, are not in danger, even in the event of very
high load-rejection overvoltage. Transmission of the overvoltage to the dc-side
arresters is not possible.

7.5.2 Temporary Overvoltages on the DC Side

Temporary overvoltages on the dc side of an HVDC station can primarily occur upon
malfunction of converter control. This applies at least in cases in which each station is
equipped with a dc voltage regulation, acting through the converter control, which is either
operating as a regulator or is held ready for limiting regulation.

Two examples of malfunction of converter control will serve as examples:

• Rapid ramping up of the dc voltage by the rectifier station while the inverter valves are
still blocked. Especially in the case of cable transmissions, an overvoltage > 2 p.u. can
occur due to the interaction of smoothing reactor inductance and cable capacitance.
Such an overvoltage is damped only very slowly by the insulation resistances. This
overvoltage can be limited by the dc-side arresters, but the arresters would have to
absorb large quantities of energy.

- 569 -
• The blocking or loss of firing pulses of an inverter combined with flow of direct current
will lead to a similar result. When commutation is no longer occurring, the flow of
current is maintained through the most recently triggered valves, and a line-frequency
ac voltage is impressed on the dc circuit. Especially when the dc circuit resonates with
the line frequency, a high overvoltage can develop which decays only very slowly after
a zero crossing and thus valve blockage becomes effective.

With a modern, fully redundant and failure-tolerant control system, it should be possible to
rule out malfunction of this nature so that temporary overvoltages on the dc side can be
ignored in insulation coordination and arrester dimensioning.

- 570 -
7.6 Transient Overvoltages

The term "transient overvoltages" refers to overvoltages of short duration which occur as a
result of transient events on the ac or the dc side of an HVDC station and which can be
limited to harmless levels by surge arresters. In addition to the classic forms

• Switching surge
• Lightning surge
• Steep-fronted surge

which occur as individual impulses of defined wave front steepness and duration,
oscillating overvoltages are included here. By their nature they are really temporary
overvoltages, but their duration is limited to such a degree by converter control or
protective disconnections that MO arresters are capable of limiting their initial amplitudes
without thermal overload occurring.

7.6.1 Switching Surge Overvoltages on the AC Side

Switching surge overvoltages which originate from the ac side of an HVDC station or in
the associated ac lines are especially significant, not only because of their frequency of
occurrence but also because they determine the level of insulation of the converter and the
dc side of the HVDC station. In course of their transit they are cumulated according to the
number of series-connected converter bridges to a switching overvoltage of multifold
amplitude. However, only line-to-line voltages are transmitted in this manner to the dc
side. Line-to-ground voltages, on the other hand, are not. Switching surge voltages which
occur at the ac busbar of an HVDC station can be caused by one of the events described in
the following sections.

7.6.1.1 Energization of Lines

Energization as well as reclosing of high voltage lines after a fault is cleared, can cause
very high overvoltages at the open line end. On the source side, switching overvoltages are
much more moderate, usually < 1.8 p.u.

7.6.1.2 Energization of AC Filter Circuits

When ac filter circuits or a bank of capacitors is switched on, a voltage drop occurs at first
due to the charging current. The voltage drop is usually followed by an oscillating

- 571 -
overvoltage which usually decays within a few cycles. If the capacitors are initially
discharged (control measures should ensure this), the crest value of the overvoltage will
usually stay < 1.8 p.u. Since the switching of filter circuits and capacitor banks must be
considered a frequent event in reactive power or ac voltage regulation, equipping the
associated ac circuit breakers with closing resistors should be considered. By using this
measure, the switching overvoltage can be limited to < 1.4 p.u. The intervention of the
busbar arresters will then be limited to rare cases of fault clearing close to the station.

7.6.1.3 Single-Phase Line Fault

At the instant of single-phase line-to-ground fault initiation, switching overvoltages occur


in the unaffected phases. In solidly grounded systems, which can always be assumed in
HVDC stations, overvoltages of this type usually remain < 1.4 p.u. and exist
predominately in the zero-sequence system.

7.6.1.4 Clearing of Line Faults

When line faults, in particular three-phase line faults, are cleared, oscillating overvoltages
occur. The initial amplitude of which usually remains in the range < 1.6 p.u. Only in
systems with little damping, higher values have to be considered which in combination
with transformer saturation and resonance between system reactance and filter capacitance
at low harmonics can cause considerable overvoltages.

Fig. 7.12 shows an oscillogram of a three-phase busbar short circuit at the rectifier station.
This oscillogram was taken using the parity simulator of the FGH. The crest value of the
overvoltage at the moment of fault clearing is approximately 1.6 p.u. This overvoltage,
however, is immediately suppressed by the immediate commencement of commutation of
the direct current at zero dc voltage, resulting in a greatly increased absorption of reactive
power. In this case it is not a problem for the MO arresters to limit the first overvoltage
peak.

The overvoltages described above occur initially between line and ground, i.e., in the zero-
sequence system and they may be limited by arresters at this point. Line-to-line
overvoltages are generally considerably lower. Theoretically, the line-to-line voltage,
however, can reach twice the value of the limiting voltage when the arresters of two
phases conduct at the same time.

- 572 -
Figure 7.12: Clearing of a three-phase short circuit at the ac busbar of the rectifier station
UA = AC busbar voltage
Pd = Active power
QA = Reactive power demand
Id = Direct current

- 573 -
7.6.2 Switching Overvoltages on the DC side

There are no dc circuit breakers in existing or planned HVDC systems. Even in the future,
use of HVDC circuit breakers will be limited to multiterminal systems or two-point
connections with parallel-connected (redundant) dc lines. Nevertheless, there are
switching-surge overvoltages even in present-day HVDC systems caused by events on the
dc side (converters being included here with the dc side).

7.6.2.1 Line-to-Ground Fault of the HVDC Overhead Line

At a line-to-ground fault of one pole of a bipolar HVDC overhead line, which may be
caused by a lightning strike or insulator contamination flashover, an overvoltage is
induced in the unaffected pole. The amplitude of this overvoltage in relation to operating
voltage is dependent upon the ratio of line-to-line and ground-to-line capacitances. This
can be clearly seen in Fig. 7.13. Independent of line length and location of the fault, the
induced overvoltage, which theoretically can attain around 1.8 p.u., is damped to values
below 1.5 p.u. before it enters the HVDC station.

CK

CE

Figure 7.13: Line-ground fault of a bipolar HVDC line


CK Coupling capacitance
CE Line-to-ground capacitance

- 574 -
7.6.2.2 Ground Fault of One Station Pole
The likelihood of a ground fault of a station pole is small. Nevertheless, this event must be
incorporated into the insulation coordination. Overvoltages in the form of a switching-
surge voltage occur both in the affected pole - particularly at smoothing reactor and dc
filter-circuit reactors, resistors and the overvoltage arresters which protect them - and in
the unaffected pole, though the overvoltage factors are low. In addition, components
which are connected to the station neutral bus and the associated arresters are stressed by
switching-surge voltages.

7.6.2.3 Switching Operations at the DC Side


Switching operations on the dc side of an HVDC station can be a source of switching
overvoltages even when there are no HVDC circuit breakers. Connecting a dc filter circuit
to the pole voltage triggers switching-in oscillations which can result in overvoltages. Still
more critical is the connecting of an entire station pole to the other pole while in operation.
This may happen in preparation for parallel operation of the station poles upon loss of one
line pole. As discovered in the Cahora Bassa project, this can result in dangerously high
overvoltages if the switching device is not equipped with a closing resistor.

7.6.2.4 Malfunction of Converters


Malfunctions of converters can also result in overvoltages on the dc side which are similar
to a switching surge voltage. Such malfunctions include the temporary loss of a control
pulse or the loss of all control pulses of a converter bridge or group, particularly in the
inverter station. It is possible to essentially eliminate malfunctions of this kind with
redundant converter control systems in use today. In contrast, the commutation failure
must be considered a frequent event and its consequences must be precisely defined. All of
these events result in the superimposing of line-frequency ac voltage components on
system dc voltage. In the event of resonance of the dc circuit to line frequency or to a
second or third harmonic, overvoltages can develop which lead to a high level of stress on
the dc-side overvoltage arresters. This phenomenon may be reinforced by interactions with
the ac side. The line-frequency component superimposed on the direct current creates a dc
component and second harmonic in the alternating currents on the ac side of the converter
bridges. Interaction of transformer saturation and resonance between ac system and filter
circuits at the second harmonic can result in amplification of the alternating currents
superimposed on the direct current to the point of harmonic instability as described earlier
(see Section 6.4.5).

- 575 -
7.6.2.5 Insulation Fault in the Converter Zone

In addition to internal disruptive punctures and external flashovers - for example at the
valve-side bushings - insulation faults include malfunctions of dc switchgear associated
with a converter group. Along with steep-surge voltages, which will discussed later,
overvoltages occur which are similar to switching surge voltages. One example is an
unintended closing of a bypass switch in an HVDC station which has more than one
converter group per station pole. As depicted schematically in Fig. 7.14, at the instant of
switch closing, the system direct voltage - supported by line capacitances and dc filter
circuits - is switched to the lower converter group, with the smoothing reactor limiting the
discharge current. The discharge current is superimposed on the system direct current in
the opposite direction. As a result, at least in low load operation, the current can be
extinguished. In such a case the C arrester, which is connected in parallel to the lower
group, must assume the overvoltage protection function and absorb the discharge energy.
Current and voltage regulation insure that quasi-steady-state transmission operation is
restored in a few milliseconds (in this example, at one-half nominal dc voltage) and the
overvoltage situation is eliminated.

Figure 7.14: Unintended closing of a bypass switch

- 576 -
The insulation fault in a valve-side transformer bushing in the upper bridge, as illustrated
in Fig. 7.15, results in an extremely great stress on a V arrester.

The "upper" V arrester associated with this phase must absorb the energy of the line and
filter circuit capacitances charged to the transmission voltage. This represents an extreme
degree of stress in relation to the relatively low nominal voltage of the V arresters,
particularly in long-distance cable transmissions. For this reason, it is often necessary to
employ multiple-column MO arresters for "upper" V arresters or even use several arresters
connected in parallel.

Figure 7.15: Insulation fault of the valve-side transformer bushing on the upper bridge

- 577 -
7.6.3 Lightning-Surge Voltages

It can be assumed that lightning surge overvoltages will not occur in HVDC stations. The
travelling waves triggered by lightning strikes on the ac or dc overhead lines extending
from the station encounter the series inductances of the converter transformers or of the
smoothing reactors and the shunt capacitances of the ac or dc filter circuits connected
there. These series inductances and shunt capacitances strongly attenuate the lightning
surge overvoltages. The front of the travelling waves is thus flattened and the crest value is
reduced. The travelling waves assume the shape of switching surge overvoltages as
discussed in Sections 7.6.1 and 7.6.2.

This statement assumes, of course, that the HVDC station is reliably protected against
direct lightning strikes in the zone between converter transformers and smoothing reactors,
as far as not housed in the valve hall. This can be accomplished by suspending lightning
shield wires above the system. With wires spaced at 10 meter intervals, the critical current
with which a lightning discharge is able to penetrate the shield is limited to approximately
2 kA.

The overvoltage arresters provided at the line terminals limit the entering lightning-strike
travelling wave to the protective level which is:

US = IC × RA

where IC = coordinating current


} with 8/20 ms surge current
RA = arrester impedance

The lightning discharge is here considered to be a current source. The impedance of the
path for the current (i.e., the arrester) determines the voltage.

In general, 10 kA is specified as the crest value of the coordination current, or in networks


with nominal voltage > 400 kV, 15 or 20 kA.

- 578 -
7.6.4 Steep-fronted Surge Voltages

If an HVDC station is equipped with metal-clad SF6 insulated ac switchgear, steep-fronted


surge overvoltages with a front time of several 10 ns can occur during current
interruptions. These overvoltages must be taken into consideration in insulation
coordination of the ac side, but they are irrelevant for voltage and insulation dimensioning
of the converters.

Ignoring future application of an SF6 dc switchgear, no steep-surge overvoltages should be


expected on the dc side of the HVDC station. In contrast, flashovers to ground in the
converter zone, which represent a voltage collapse, can cause over-stress with steep-
fronted surge voltage for some components, particularly valves. The fault event depicted
in Fig. 7.15 causes an extremely steep voltage rise in the blocking direction at the valve of
the affected phase in the upper commutation group when it is not conducting at the
moment. The same applies for the associated V arrester. The steepness of the voltage rise
can easily be several 1000 kV/ms. In order to provide effective protection in the event of
such steep overvoltage rises, an arrester must have very low inductance and must be
connected directly to the object which is to be protected, in this case to the valve. The
design principle of the V arrester being directly mounted on the valve tower as used by
Siemens represents an optimal solution.

- 579 -
7.7 Protection Level and Test Voltages

When all relevant overvoltages triggered by events on the ac and the dc sides and in the
converter station itself and furthermore the characteristics of MO arresters are known, the
protection level for each type of overvoltage can be determined. These protection levels
are the basis for the insulation coordination of the HVDC station and for establishing the
test voltages.

7.7.1 Arrester Protection Level

The protection levels of an MO arrester are dependent upon the steepness of the front of
the arrester surge current. The typical wave shapes of coordinating currents are

• Steep surge (1/20 ms)


• Lightning surge (8/20 ms)
• Switching surge (30/60 ms)
• Slow front surge (1/2 ms)

Since the residual voltage of the arrester increases with the steepness of the front of the
surge current, a standardized coordinating current should always be used for testing the
arrester and establishing the protection level whose front steepnes is equal to or greater
than that which would occur in the event of the overvoltage which is to be limited.

Therefore the procedure is as follows:

• The protection level for steep surge stress (coordinating current shape 1/20 ms) applies
for anticipated arrester surge currents with front times < 8 ms.

• The protection level for lightning surge stress (8/20 ms) applies for anticipated arrester
currents with front times between 8 and 30 ms.

• The protection level for switching surge stress (30/60 ms) applies for anticipated
arrester surge currents with front times between 30 and 1000 ms.

• The protection level for slow front surge stress (1/2 ms) applies for anticipated arrester
surge currents with front times > 1 ms.

- 580 -
7.7.2 Safety Factor

The next step in insulation coordination is the establishment of the safety factor, i.e., the
relationship of test voltage for a device to the protection level of the overvoltage arrester
which protects the device. Imprecision in determining the anticipated overvoltage stress
and the expected deterioration of the insulation during the life of the device are accounted
for by means of the safety factor. In this way the most important parameters for
dimensioning of insulation of high voltage devices and for HVDC systems are determined.
Main parameters in specifying equipment insulation strength are the following:

• Lightning surge withstand voltage


Per IEC: Lightning impulse withstand voltage (LIWV)
Per ANSI: Basic lightning impulse insulation level (BIL)

• Switching surge withstand voltage


Per IEC: Switching impulse withstand voltage (SIWV)
Per ANSI: Basic switching impulse insulation level (BSL)

Based on many years of experience in the insulation dimensioning of HVDC devices and
systems, safety factors are usually specified as follows:

• For switching surge stress


In thyristor valves: 1.10...1.15
In other components: 1.15...1.20

• For lightning surge stress


In thyristor valves: 1.15...1.20
In other components: 1.20...1.25

• For steep surge stress


In thyristor valves and other
components: 1.20...1.25

- 581 -
The reason for the smaller safety factor for thyristor valves is a fundamentally different
process of deterioration of insulation over time. It is believed today that the off-state
blocking voltage of a thyristor does not change over its life. From time to time there will
be thyristor failures. For this reason a few redundant thyristors are provided in the
dimensioning of a valve. But when the failed thyristors are changed-out during a
maintenance shut-down, the valve will again achieve its original insulation condition. This
assumes that the other insulation structures of the valve, which of course are subject to
deterioration, are designed for a higher test voltage.

The philosophy of insulation coordination presented here presumes that the protection
level of the MO arresters do not change in the course of time. Aging experiments of the
FGH have shown, however, that an oxygen depletion of the atmosphere in the arrester
housing, caused by partial discharge, can cause a structural change in arrester disks, thus
causing a change in the current-voltage characteristic. This effect, however, will lead at
most to a lowering, never to a raising of the protection level.

An aging effect of MO arresters has to date not been observed in HVDC systems. It could
result in the endangering of the arresters themselves, but not of the devices protected.

7.7.3 Test Voltages

The lightning surge and switching surge withstand voltages derived from the arrester
protection levels and the specified safety factors are minimum values for the test voltages
and the dimensioning of insulation. Usually, however, standardized preferred values in
international regulations are specified. This applies, at the least, to all devices and system
components at the ac side of the HVDC station, and usually also for the line side of the
converter transformer. This procedure, however, would be extremely expensive for the
thyristor valves. Manufacturing costs and losses of thyristor valves increase approximately
proportionally with voltage rating and thus with the test voltages to be used. For this
reason, BIL and BSL values are generally adapted precisely to system-specific
requirements. The maximum anticipated voltages of the valves and the required
coordination currents of the MO arresters obtained from studies make it possible to
establish the arrester data. Thus the protection level is defined, from which - using the
relevant safety factors - the required surge withstand values are obtained, and at the same
time the test voltages.

- 582 -
For thyristor valves and for the other dc-side devices and components of an HVDC station,
including the valve-side windings and bushings of the converter transformers, tests with
standard lightning surge and switching surge voltages are not at all adequate to cover the
voltage stresses occurring in actual operation. Therefore, test methods have been
developed in the course of time which to date have been only partially reflected in
international regulations or recommendations. For this reason, various test methods and
voltages are found in specifications for HVDC systems. A few generalizations, however,
can be made.

7.7.3.1 Converter Transformers and Oil-Insulated Smoothing Reactors

The voltage tests should document that the complexly structured oil paper insulation will
withstand the continuous operating voltage composed of a d.c. component and
superimposed a.c. components of a wide frequency range and furthermore withstand the
occasionally appearing overvoltages which are limited by the relevant surge arresters. The
tests shall further prove that the frequently occuring polarity reversals of the d.c.
component will not cause any insulation deterioration by partial discharges.

Because of the impossibility of achieving a test voltage exactly like the actual voltage
shape, substitution tests such as the following must be performed:

• DC voltage test with 1.5 times the crest value of continuous operating voltage
(representing dc voltage components plus 50% of the crest value of superimposed line-
frequency ac voltage) for a period of one hour.

• Polarity reversal with a dc voltage of 1.1 to 1.25 times the crest value of continuous
operating voltage, with the dc voltage being applied for a minimum of 30 minutes
before and after the polarity reversal.

• Alternatively, if a polarity reversal cannot be performed: DC voltage test with twice the
crest value of continuous operating voltage. Duration: two minutes.

• Switching surge voltage test:


This test can be performed on the valve-side winding of the converter transformer only
with short-circuited windings as shown in Fig. 7.16. Otherwise, the voltage induced in
the line winding would exceed its switching surge withstand voltage. Therefore only
the transformer main insulation and the insulation against iron core and transformer
tank are tested by this procedure.

- 583 -
Figure 7.16: Test of valve-side transformer windings with switching surge voltage

• Lightning surge voltage test:


This test is performed with the open winding grounded on one side. Thus the winding
insulation is also tested. For the valve-side winding and the smoothing reactors, both
ends must of course be stressed.

7.7.3.2 Thyristor Valves

The testing of thyristor valves for HVDC systems is described in IEC Publication 700.
This is a very extensive list of tests to be performed which are covering all valve stresses,
not just the voltage stresses. Some of the tests are performed on valve modules or
equivalent valve sections, some on the entire valve, and some on the complete multiple
valve.

With respect to withstand voltage, the IEC publication prescribes very specific tests such
as the firing of a valve when the parallel V arrester is stressed with the coordination
current as a result of a switching overvoltage. Thus functionality is tested under maximum
overvoltage stress. In this process, high di/dt is critical during the commutation of the
arrester current onto the valve.

- 584 -
7.7.3.3 DC Switchgear

To date, there are no international test regulations or IEC publications for voltage testing
of dc switchgear of an HVDC station. The CIGRE Working Group 13/14.08, however,
has established several principles for the specification and testing of dc switchgear for
HVDC systems in a series of Electra publications. In the case of standard surge voltage
tests (BIL, SIL), the preferred values applicable for high-voltage ac systems should be
used, which at a minimum are above the protection level determined by the arrester
residual voltage by at least the safety factor.

7.7.4 Testing of the MO Arresters

There are extensive tests to be performed to assure that the MO arresters used in an HVDC
system can perform their demanding functions as described above in the framework of
insulation coordination and overvoltage protection. These include the following:

• Measurement of the current-voltage characteristic


• Determination of protection levels
• Documentation of energy withstand capability
• Documentation of thermal stability
• Pressure-relief test
• Insulation test of the housing
• Function test with contaminated insulators

These tests should not only determine or document the relevant data for insulation
coordination. They should also demonstrate that all stresses which could result in an
arrester fault will be withstood. The condition for acceptance is that the specified data
must be within the approved tolerance, that no external or internal flashover occur during
the tests, and that the housing does not rupture during the tests except, of course, during
the pressure relief test. Thermal stability must be documented in a long-term test. At the
conclusion of the type test, it must be demonstrated that the current-voltage characteristic
has not changed.

- 585 -
7.7.4.1 Current-Voltage Characteristic and Protection Level

The statically measured current-voltage characteristic of an MO arrester provides


information on any deterioration due to aging. Residual voltages of the arrester,
determined with the various standard surge stresses with the appropriately defined
coordination currents, however, are incorporated directly into insulation coordination as
protection levels. In order to assure the required overvoltage protection, the specified
upper tolerance limit must not be exceeded. The lower tolerance limit must not be
undercut because this can result in an elevated absorption of energy and thus provoke the
thermal instability of the arrester.

7.7.4.2 Energy Withstand Capability

The test of the energy withstand capability of an MO arrester should ensure that the
arrester can not only withstand a discharge with the defined maximum energy content one
time, but that it can withstand many such discharges throughout its service life without
sustaining damage or perceptibly ageing. The tests are usually performed with individual
sections, rather than with the entire arrester. Unfavorable ambient and bias conditions
which can be expected in operation are simulated. The test may comprise six discharge
events with the specified maximum energy content, or it may consist of 6 groups of three
impulses in rapid sequence which together have the maximum energy content. Time for
cooling is allowed between discharges or groups of discharge impulses.

7.7.4.3 Thermal Stability

A compressed-time ageing test can be performed to document the thermal stability of an


MO arrester. The arrester element is exposed at a block temperature of 115° C for
1000 hours to the equivalent continuous operating voltage (ECOV). This is considered the
equivalent of the stress of 100 years of operation at 40° C operating temperature. During
the long-term test, absorbed power loss is recorded. The criterion for thermal stability is
that following a discharge process with a specified maximum energy absorption, the
arrester must cool down, while subjected to the ECOV, to its temperature before the
discharge process. If it is expected or is specified that an arrester will be exposed to a
temporary overvoltage following a discharge event with maximum energy absorption, this
must also be simulated. In this case also, the operating temperature before the discharge
process must be achieved again following the return of the voltage to the ECOV.

- 586 -
A test to assure thermal stability is not required for filter circuit arresters and for E
arresters connected to the station neutral bus. These arresters do have to have a high
energy withstand capability but are exposed to practically no voltage during continuous
operation.

7.7.4.4 Pressure Relief Test

In the event a test of this type is specified, it can be performed according to the IEC or
ANSI standards for ac arresters. It should be documented that the arrester upon energy
overload relieves from overpressure in a controlled manner. Breakage of the housing is
accepted, but not an explosion.

7.7.4.5 Insulation Test

The only testing of insulation of an MO arrester needed is the checking of flashover


behaviour when the housing insulator is contaminated and wet. During dry conditions, the
arrester protects its housing against every kind of overvoltage.

According to CIGRE Working Group 33/14.05, the purpose of the contamination tests is
to take into consideration the additional warming through leakage currents on the surface
of the porcelain housing. This should be simulated by artificially warming the arrester.

No mention is made of checking the withstand behaviour in the event of defined ESDD
(equivalent salt deposit density) contamination and of complete wetting of the surface
contamination in the presence of ECOV or of temporary overvoltage, though these tests
are undoubtedly more significant.

It is also important that there will be no partial discharge in the event of uneven
distribution of voltage along the surface of the housing caused by contamination. This is to
exclude the possibility of the ageing effect discovered by the FGH.

- 587 -
7.8 Insulation Dimensioning

7.8.1 Introductory Remarks

The definition of arrester protection levels and test voltages provides the basis for
insulation dimensioning. While approved rules and standards for the insulation
dimensioning of ac systems and devices can be used as guide line, the special stresses on
insulation from dc voltage or mixed voltages of complex composition require special
consideration. This attention can be limited to three areas:

• Minimum clearances in air


• Creepage distance of outdoor insulators
• Oil paper insulation in transformers and smoothing reactors

It has been possible through extensive research studies during the last forty years - some of
which are still continuing today - to create reliable foundations for insulation dimensioning
in HVDC systems. As experience with existing HVDC systems demonstrates, the
knowledge which has been acquired is sufficient to adequately guarantee reliable operation
with respect to insulation behaviour. Nevertheless, surprises occured repeatedly in the
form of unexplained insulation failures.

7.8.2 Minimum Clearances in Air

Fundamental tests of the behaviour of rod-rod and rod-plate arrangements subjected to


stress of dc voltage, impulse-shaped surge voltages, and various combinations of the two
have been carried out in numerous high voltage laboratories with great thoroughness. The
results of these tests can be summarized as follows:

• For stresses with pure dc voltage, the flashover voltage is approximately a linear
function of clearance.

• For dc voltage stress in dry air, the rod-plate arrangement demonstrates a 50% lower
flashover voltage with positive polarity than with negative polarity.

• In the presence of rain, flashover voltage is reduced by 45% with negative polarity,
while it does not change with positive polarity.

- 588 -
• With the rod-rod arrangement, critical flashover voltage is not dependent on polarity or
ambient conditions - dry or wet.

• Even with switching surge voltage, the rod-plate arrangement demonstrates a 50%
lower flashover voltage with positive polarity than with negative polarity.

• In the case of switching surge voltage stress of a rod-rod arrangement, the critical
flashover voltage is lower by 25 % with positive polarity than with negative polarity.
This relationship does not change in the presence of rain.

• The stress of lightning surge voltage does not play a role in the dimensioning of
minimum clearance in air, since the critical flashover voltage is always at least 30%
higher than with switching surge voltage.

These results - some of which are rather strange - suggest that with increasingly
inhomogenous field conditions, as occurs in rod-rod arrangements, the extreme
dependencies on polarity and ambient conditions will disappear. This is certainly also true
in realistic configurations. The CIGRE Working Group 33-05 conclusions are paraphrased
as follows:

The breakdown voltage in the air gap is dependent on electrode shape and on the
relationship of the voltage components and their polarity when stressed with composite
voltage shapes and also from the corona inception level. Critical flashover voltages at
positive switching surge voltage as well as at composite voltage composed of positive
switching surge voltage and positive dc voltage of the same crest value result in
dimensioning on the safe side.

To this extent, it is possible to use the simple dimensioning standard in common use in ac
technology. But where the choice of minimum clearances has a large effect on costs, such
as within the valve hall, flashover voltage tests on realistic electrode - configurations with
true voltage shapes may allow to deviate from the conservative dimensioning based on the
critical flashover voltage of a rod-rod configuration with positive switching surge voltage.

- 589 -
7.8.3 Creepage Distance of Outdoor Insulators

The flashover behaviour of outdoor insulators is determined by contamination of the


insulator surface combined with the wetting of the contamination layer through fog, dew,
or drizzle. The insulators perform far more unfavourably with dc voltage stress than with
ac voltage stress for two reasons:

• Contamination particles become charged in a dc electrical field and are accelerated to


the opposite electrode where they are deposited (electrostatic filter effect). In an ac
field, the particles constantly reverse polarity and thus have a good chance of escaping
the field through wind or gravity. For this reason, the degree of insulator contamination,
expressed as ESDD in mg/cm2, is much higher with dc voltage than with ac voltage of
the same field strength.

• A conductive coating is formed on the surface of the insulator when a contamination


layer containing salt becomes soaked with water. This leads to leakage currents and, as
a result, to localized drying-out. There the contamination layer becomes highly
resistive, local elevations of field strength occur, and finally there are discharges in the
form of partial arcs. It is easy to understand that such partial arcs can develop more
easily into a total flashover with dc voltage than with ac voltage in which the partial
arcs periodically extinguish and have to be ignited again every half cycle.

The interrelationships between degree and type of contamination, distribution of the


contamination along the insulator string and between the upper and lower side of insulator
shields and the critical flashover voltage were presented in great detail in Section 3.1.6
using the pin and cap line insulator as an example. It was pointed out that results of tests
always apply only for a specific type of insulator and that the specific creepage distance by
itself is no measure of the flashover behaviour which can be expected.

It should be added at this point that, while the same phenomena occur with station
porcelains such as post insulators, transformer and valve hall bushings and the like, their
influence on critical flashover voltage increases with increasing trunk diameter of the
insulator. Thus, numerous flashovers of station porcelains exposed to dc voltage have been
the cause of criticism in some HVDC systems. In some cases, the problem was solved by
periodic cleaning while in others greasing of the insulators was necessary.

Valve hall bushings with dc voltages ³ 400 kV have shown special difficulties. Even in
the practically contamination-free Dorsey HVDC Station of the Nelson River

- 590 -
Transmission System, flashovers have occurred at the valve hall bushings. These
flashovers have always occurred during light rain and moderately strong wind from a
certain direction. It can only be conjectured that electrically charged rain droplets carried
by the wind under the principle of the Van-de-Graaf generator led to the accumulation of
charge and as a result, to a flashover. In any case, "booster sheds," which protect parts of
the bushing from rain, have helped.

Unexpected insulation flashovers such as these have even been observed with specific
creepage path distances of ³ 4 cm/kV. At the least in the case of bushings, overall lengths
are thus required that are at the limit of manufacturing technology. Periodic cleaning and
greasing is not a satisfactory solution due to the costs in time and personnel. The future
will tell whether coating of the station porcelains with hydrophobic substances such as
silicon rubber can guarantee a long-term solution. Initial results are encouraging.

In any case, it is strongly recommended that long-term tests be performed on site to


determine the insulator contamination which can be expected, before establishing the
specific creepage path distances. For this purpose, original insulators or at the least,
insulator segments with the same trunk diameter and the same skirt shape should be
exposed to the same equivalent dc voltage. Insulators contaminated in this way should
then be subjected to flashover tests in a fog chamber ("clean fog method").

Many other risks in an HVDC station can be avoided through suitable placement of
devices. Wall bushings are made unnecessary by placing converter transformers directly at
the wall of the valve hall with the valve-side transformer bushings penetrating into the
controlled atmosphere of the valve hall; thus they can be configured with indoor
insulation.

On the other hand, it is clear that there are many problems associated with outdoor valve
technology if insulator contamination and wetting are to be expected.

- 591 -
7.8.4 Oil Paper Insulation of Transformers and Smoothing Reactors

The insulation system of valve-side windings of converter transformers and the insulation
of oil-insulated smoothing reactors, essentially existing of paper and oil, are subject to
voltage stress which is composed of dc voltage and superimposed ac voltages of various
frequencies, as described in Section 7.4. This leads to very specific problems which are
not present in ac technology. When a complex structure of dielectrics is stressed by ac
voltage, the distribution of voltage corresponds to the dielectric constants of the insulating
materials. When they are stressed by dc voltage, however, the specific resistances
determine the voltage distribution, whereas a stationary voltage distribution is achieved
only after a relatively long period of time - many minutes to an hour.

The dielectric constants of paper and oil have a ratio of around 1:2. This ratio hardly
changes during operation, i.e. when they get warm. The specific resistances of the
materials, however, have a ratio of around 300:1. The specific resistance of transformer
paper is strongly dependent on temperature, on the field strength applied and on the
moisture content. Thus the resistance ratio in a new "dry" transformer may vary between
300 and 20 and between 100 and 1 in an old, "wet" transformer.

Many of the problems considered in Section 3.2.3 with respect to HVDC cables reappear.
Here again, the unavoidable reversal of polarity in HVDC operation results in particularly
critical stresses on the insulation system.

Fig. 7.17 shows schematically the voltage distribution curve in an oil-paper dielectric.
Curves a) and b) give the voltage curves under pure ac voltage and pure dc voltage stress,
respectively, in steady-state condition. When a dc voltage stress is first switched on, curve
a) applies initially and then gradually transitions into the voltage distribution of curve b),
which is determined by the specific resistances. If dc voltage is abruptly reversed, voltage
curve c) is obtained. This is a result of a superimposed ac voltage distribution caused by
applying a voltage of + 2U (broken-line curve) onto the steady-state voltage distribution at
- U which is still retained due to trapped charges. Curve c) clearly demonstrates that the oil
region, which is almost completely released in steady-state operation from voltage stress,
suddenly encounters a voltage upon the reversal of polarity which can be even higher than
the externally applied voltage. Only gradually, with a strongly temperature-dependent time
constant (minutes to hours), the new steady-state voltage curve d) is reached.

- 592 -
+2U

+U
c)

d)
0

b)
a)

-U

Oil Paper

Figure 7.17: Voltage curves in an oil-paper, two-layer dielectric


a) With ac voltage
b) With dc voltage
c) Immediately after polarity reversal
d) Final condition after polarity reversal

The process of steady-state voltage reversal described here can be classified in many
HVDC systems as a rare event occurring only when there is a change in energy direction.
The transient voltage reversal, however, must be considered as a frequent event. In every
line-to-ground fault of the HVDC overhead lines and in every commutation failure of the
inverter, the rectifier reverses the d.c. voltage under the influence of its current regulation
or of the forced retard function.

Each time this occurs, the voltage distribution in the dielectric of the transformer valve
side and of the smoothing reactor transiently assumes the configuration of curve c),
corresponding to a stress with 1.4 to 1.6 times the dc voltage. This is the reason why the
insulation of HVDC transformers and smoothing reactors is rather expensive and bulky.

- 593 -
7.9 System Studies

7.9.1 Introductory Remarks

In the preceding sections, it has become obvious that proper insulation coordination of an
HVDC station is critical with respect to both, reliability and costs. The basis for insulation
coordination is reliable determination of the following information for every device
exposed to a voltage stress and for every component of the HVDC station:

• Continuous operating voltage


• All types of expected overvoltages
• An adequate arrester protection scheme
• Arrester protection level

These parameters must be known for all operating modes and ranges and under the
specified ambient conditions.

To accomplish this, it is necessary to treat the HVDC as a system, considering all of its
components and the interactions with the ac networks on both sides. This means that
system studies must be conducted. Usually it will be necessary to perform a general study
in the tendering stage of the project to determine the basis for dimensioning components
such as thyristor valves. A detailed insulation coordination study is indispensable in the
final design stage of the project. This study shall provide reliable documentation for
voltage stresses, insulation dimensioning of devices and components, and for the
specification of MO arresters.

7.9.2 Basis of the Studies

The following information is necessary in order to perform a system study on insulation


coordination:

• The configuration of the HVDC stations in all operating modes according to the
specification.

• Detailed data and characteristics of station components, including saturation behaviour


of converter transformers, the smoothing reactor inductance as function of direct
current, filter circuit impedances as function of frequency, etc.

- 594 -
• Control, regulation and protection functions of the HVDC control and protection
system with its steady-state characteristics, transient response, triggering criteria, etc.

• Detailed data on the ac and HVDC overhead lines including the distributed inductances
and capacitances in the operating and in the zero-sequence system, corresponding cable
data, data on transition resistance of the earth electrodes, data for the electrode lines, the
most precise data possible on specific resistances of layers of soil near the surface along
the right of way, and to the extent obtainable, data on the conductivity of deeper layers
in the vicinity of earth electrodes.

• Configuration and data of the ac networks on both sides with maximum and minimum
short-circuit power and associated impedance angles as well as the changes in these
data to be expected later. The network impedance versus frequency curve above the
fundamental frequency up to several kHz.

• Current-voltage characteristics of the MO arresters and their typical protection level for
standard surge-voltage stresses. Maximum energy absorption capacity of the arresters.

In performing the preliminary insulation coordination study in the tendering stage, much
of this information will not yet be known. At this stage it is important to apply the
knowledge and experience acquired in earlier projects in order to achieve results which do
not differ from the final ones by more than a few percentage points. In this way,
miscalculations in one direction or the other are avoided. Simulation of the ac network
using suitable and easy-to-handle substitute circuits is very important in this process.
In order to keep the costs of this preliminary work in bounds, it is important to limit the
study to a small number of events which are determinative for voltage and insulation
dimensioning. The selection of these events and the determination of the most
unfavourable parameters, however, require much experience.

The detailed insulation coordination study in the "final design" usually requires the
approval of the client or his consultant. Therefore it is helpful to invite representatives of
the client to participate in the studies at least part of the time. This also makes sense
because the results provide important data for the specification of ac devices and system
components which may not be included in the bid package. The question of what residual
risks in the case of unlikely fault events or coincidences can be accepted is more easily to
be answered in cooperation with the client.

- 595 -
7.9.3 Events to be Studied

For insulation coordination, only those events which result in the greatest insulation stress
for a component of the HVDC system and those which cause the maximum stress of the
MO arresters utilized for overvoltage protection are important. The maximum arrester
residual voltage must be considered as insulation stress for the components which are to
be protected. At least, the following events should be included in the study:

7.9.3.1 HVDC System Load Rejection

Fig. 7.18 shows the oscillogram of a 100% load rejection in an HVDC system caused by a
three-phase line fault close to the inverter station. Busbar voltage of the rectifier station
UDS, transmission power P, reactive power Q drawn form the rectifier, and direct current Id
are shown. This is a station with an extremely weak ac network connection. The effective
short-circuit ratio (ESCR) is only 1.86. For this reason, line voltage collapses to about
50% at the moment of the disturbance. The high temporary overvoltage which is to be
expected at fault clearing is suppressed by the converter control system. The flow of direct
current is maintained at approximately 40% of the corresponding value before the
disturbance by the VDCOL function. This maintains the reactive power at a somewhat
constant level, preventing a temporary overvoltage. Only in the moment the dc voltage is
ramped up by the inverter station after the clearing of the fault the direct current briefly
goes to zero. A high line overvoltage occurs for about one cycle. This can be easily
limited, however, by an MO arrester.

- 596 -
Figure 7.18: 100% load rejection of an HVDC system (quantities of the rectifier station)
See text for explanation. ESCR = 1.86

Fig. 7.19 shows the oscillogram of a 50% load rejection caused by a line-to-ground short
circuit of the bipolar HVDC overhead lines. A direct current zero for approximately 100
ms can be seen. During this time, the unaffected pole of the HVDC is overloaded by
approximately 50% through intervention of the converter control system. As a result, the
reactive power absorbed by the rectifier remains practically unchanged and a temporary
overvoltage is largely avoided. Upon return of the disturbed pole, the overload of the other
pole is brought back to normal. Even during this process, the reactive power remains
approximately constant.

- 597 -
Figure 7.19: 50% load rejection of an HVDC system (same quantities as in 7.18)

The oscillograms prove that even with extremely unfavourable ac system conditions, a
temporary overvoltage which would be determinative for voltage dimensioning and
insulation coordination can be largely avoided by optimization of the HVDC control
system. Naturally, such measures require "back-up" protection for cases when the system
may fail. Rapid disconnection of ac filter circuits and capacitor banks can be a suitable
back-up solution. A procedure of this kind has been shown in Fig. 7.11.

7.9.3.2 Switching Processes on the AC Side

The energization of overhead lines and transformers from the ac busbar of an HVDC
station can cause switching overvoltages which must be considered in insulation
coordination. The inrush current caused by energization can furthermore result in
considerable voltage distortions. If a resonance exists near the second harmonic (due to the
system inductance and the filter-circuit capacitance), persistent temporary overvoltages
may occur. A possible countermeasure against this is the use of closing resistors in the
circuit breakers. The clearing of ac line faults, for which the closing resistors are
ineffective, must then be considered.

- 598 -
Disconnecting of high-voltage lines from the ac bus of an HVDC station will result in
significant temporary overvoltages only in exceptional cases e.g. if there is a disruption of
significant reactive power consumption.

Critical overvoltages need not be feared during the energization and de-energization of ac
filter circuits. Nevertheless, the permitted degree of change of voltage may be exceeded.
This can compel the use of special control measures discussed in Chapter 5.

7.9.3.3 AC Line Faults and Fault Clearing

Single-phase and three-phase line faults on the rectifier and the inverter sides should be
studied. Occasionally it is also demanded that two-phase line faults with and without
ground contact must be studied. In addition to busbar faults, distant system faults should
be included in the study, especially on the inverter side. They provide information on
whether the inverter will withstand the voltage disturbance associated with the line fault
without a commutation failure. This type of study will also show whether normal
commutation will be resumed, i.e., energy transmission will be restarted if a commutation
failure does occur. These studies are related to insulation coordination only to the extent
that the commutation failure itself can result in overvoltages.

Fig. 7.20 shows an oscillogram of a three-phase busbar fault at the inverter station. The
effective short-circuit ratio (ESCR) in this case is 3.15. The direct current only goes to
zero at the end of the fault and the overvoltage following the fault clearing disappears very
rapidly because the inverter resumes commutation immediately. Thus reactive power is
again drawn from the ac system. Transformer saturation caused by the switching inrush
upon the return of voltage is unlikely to result in temporary overvoltages even though it
causes noticeable voltage distortion.

- 599 -
Figure 7.20: Three-phase busbar short circuit at the inverter station (ESCR = 3.15)

- 600 -
7.9.3.4 DC-Side Line Faults

The overvoltage induced in the unaffected pole during a line-to-ground fault of a bipolar
HVDC overhead line usually remains below 1.5 p.u. Thus it is not relevant for insulation
coordination of the HVDC station. The processes which take place upon the return of the
faulted HVDC pole are more critical. This is not only because of overvoltages which may
be caused, but more importantly because of the risk of inverter commutation failures being
triggered. Such failures not only delay the resumption of the full transmission operation
but also can trigger overvoltages. This problem is particularly important if the HVDC line
consist in part of cables.

7.9.3.5 DC-Side Station Faults

Short circuits and ground faults on the dc side of the HVDC station do not cause any
overvoltages evident from the outside, but they may involve extreme stresses for
individual components of the HVDC station and for the overvoltage arresters which
protect them. The study of dc-side faults therefore is an important part of insulation
coordination.

A ground fault of the dc busbar of one station pole produces a stress with steep surge
voltage on the components of the dc filter circuits, reactors, and resistors which exceeds
the continuous operating voltage by several times. This voltage is limited by MO arresters
parallel to reactors and resistors. The arresters must absorb the energy stored in the main
capacitor of the dc filter circuit.

An event particularly critical for the thyristor valves is the short to ground on the ac-side of
a converter bridge. This can be a flashover of a valve-side transformer bushing or of a wall
bushing of the valve hall. When this fault occurs, a steep surge overvoltage appears on at
least one of the valves which must be protected by the parallel-connected arresters,
requiring them in some cases to absorb very large quantities of energy.

- 601 -
7.9.3.6 Lightning Strikes

Unless the specifications expressly say otherwise, the possibility of lightning strikes to the
electrical systems of an HVDC station caused by failures of the shielding should be
disregarded. On the other hand, overvoltages caused by lightning strikes on the ac- and dc-
side overhead lines extending from the HVDC station are an essential part of insulation
coordination studies. Here, the proper representation with respect to high frequency
response of converter transformers, smoothing reactors, and filter circuits is very
important. In addition, the characteristics of the MO arresters used and their physical
distances from the site of origin of the lightning-surge travelling wave and to the objects
which are to be protected must be known and incorporated into the study in order to
properly depict the voltage to which the components of the HVDC station will ultimately
be subjected.

7.9.3.7 Malfunction of the Converter Control System

Inverter commutation failures due to loss of firing pulses must be included among the
consequences of a malfunction of the converter control system. This also refers to the
unintended closing of a bypass switch in a station pole consisting of at least two converter
groups connected in series. Another event is the unintentional blocking of inverter firing
pulses during normal power transfer and the resulting extinction of an individual
commutation group or bridge during light-load or dynamic processes. Unintended opening
of the MRTB (metallic return transfer breaker) in weak-load or dynamic processes in
monopolar operating mode also belongs in this category.

In most cases, the events to be studied and considered in insulation coordination are
established in the specifications. Otherwise consideration must be given to what kinds of
malfunctions, given the redundant and largely fault-tolerant control system, are possible at
all and can be expected with at least a slight degree of probability. Such events must then
be examined to determine whether they have any relevance for voltage and/or insulation
dimensioning. Only then they should be included in the insulation coordination studies.

- 602 -
7.9.4 Tools for Studies

The various events which must be investigated in a system study for insulation
coordination include voltage stresses with quite different time responses. They extend
from a few microseconds (ms) in the case of steep surge voltages to many minutes in the
case of temporary overvoltages. Often the various forms of overvoltage stress occur in
combination. For this reason, widely differing tools must be employed in system studies.
These tools - differentiated with respect to the number of components which can be
simulated, the flexibility and precision of the modelling, and especially the time resolution
which can be achieved - will be presented briefly below.

7.9.4.1 Transient Network Analyzer (TNA)

This simulator, also referred to in German as "Netzmodel" (network model) or


"Schwingungsmodell" (oscillation model) was developed for the study of transient events
in high-voltage ac networks. It has proven to be an effective means for the study of
unknown problems relating to overvoltage.

A TNA consists of scaled physical models of system components such as transformers,


lines, switches, loads of different kinds, capacitor banks, shunt reactors, overvoltage
arresters, etc. Power plants are usually simulated by voltage sources behind reactances.
Some TNAs, however, have detailed machine models with simulation of the rotor system
with field windings and damping windings and of the stator with leakage reactances and
magnetization characteristic. Often simulation of turbines including the governer is
included. Simulation of the dynamic behaviour related to the inertias is important. Great
value is placed on good simulation of non-linear aspects such as transformer saturation,
negative sequence systems of network voltage, etc. The quality of the models used is very
important for the faithful reproduction of oscillations and their damping.

TNAs are customarily operated at network frequency, i.e. they provide real-time
simulation. The power level usually is between 10 and 100 VA.

In order to utilize the TNA for overvoltage studies of systems which include HVDC
systems, models of converter stations were added, initially very simple ones,but
subsequently quite detailed models.

- 603 -
Figure 7.21 shows a single-line diagram of the test set-up for a system study involving an
HVDC back-to-back link on the Siemens TNA.

Static capacitor with


regulation, protection,
and tyhristor actuator
Transformers
with saturation

Turbines
with power
regulation G
Overvoltage
Generator with Netz A arrester
voltage regulation
Filter circuits

Smoothing reactor
and possibly
Network fault dc line
(monopolar and
multipolar) Netz B
HVDC station with
AC source 3~ regulation, protection,
(fixed or with and thyristor bridges
programmed
dynamic)

Motors or synchronous
condensers
Converter
Filter circuit M
Arc furnaces

Figure 7.21: Single-line diagram of the test set-up for a system study on the TNA

Actual control and regulation devices can be employed in the HVDC portion so that both,
the reactions of the HVDC system to transient events in the ac network which may cause
commutation failures and actions of the HVDC to support the ac network, such as
suppression of temporary overvoltages, can be studied with adequate precision.

The main advantage of the TNA - as with all physical models - is its authenticity. This is
especially useful for employee and client personnel training where the use of original
devices of converter control is particularly important. Another advantage of the TNA is the
ability to investigate phenomena for which the theoretical basis is unclear and which
therefore cannot yet be represented in a system of differential equations.

In comparison with the HVDC simulator, the TNA has the advantage that extensive,
complex ac systems can be represented. This allows even to investigate the load-flow
controlled by the HVDC and various stability criteria of the network with supporting
actions on the part of the HVDC.

- 604 -
7.9.4.2 HVDC Simulator

The first HVDC simulators were built in the 1930s while the construction of the first
HVDC systems was being considered. At that time, the only models for converter valves
were mercury-arc valves such as thyratrons or ignitrons. In order to obtain somewhat
realistic relations, considering the unavoidable arc voltage drop of approximately 20 volts,
a nominal voltage in the kV range was mostly selected for the models.

Fig. 7.22 shows a converter bridge of the Rheinau simulator with nominal data of
1000V/20A. The converter transformers were overdimensioned and their leakage
reactances were simulated by means of additional air-core chokes. In this way, a quality
factor of 10 was achieved. A total of eight such bridges allowed the simulation of a bipolar
HVDC system with two series-connected bridges, i.e., one twelve pulse group per station
pole. The nominal current of 20 A was high enough that even with the means available at
that time, measurement of all essential quantities was possible without significant error.

Figure 7.22: Converter bridge of the Rheinau HVDC simulator

The nominal power of the simulator of ±2 kV, 20 A, or 80 kW, which was found
necessary for high accuracy and parameter flexibility, resulted in capital investment in the
millions and a large requirement for space.

- 605 -
A previous synchronous condenser building belonging to the RWE in Rheinau, which was
no longer in use, was available to the companies of the Arbeitsgemeinschaft HGÜ for
installation of the HVDC simulator which was constructed in the mid-1960s as part of the
German HVDC research program.

Fig. 7.23 shows the dc line model in the Rheinau HVDC simulator which consisted of
twenty double p elements, allowing the simulation of a bipolar HVDC overhead line 500
to 1500 kilometers long.

The weak point of a mammoth system of this kind is of course the simulation of the ac
network. In Rheinau, a converter-fed Leonard motor-generator-group was available for
this purpose. The second network was simulated by simple reactors, connected in series
with the power supply. Despite all of these limitations, this model system was good
enough for proper project planning of the two highest power HVDC systems of the time.
No serious problems were encountered when both, Cahora Bassa and Nelson River II were
commissioned and during their subsequent operation.

Figure 7.23: Transmission line model of the Rheinau ±2000V/20A HVDC simulator

- 606 -
With the advent of thyristors, HVDC simulators could be built with lower power levels
and with much better modelling features. Most importantly, it was possible to compensate
the on-state voltage drop of the thyristors electronically. Today, nominal voltages between
20 and 100 V and nominal currents between 0.2 and 1 A are customary for modern HVDC
simulators. At this level of power, all components of an HVDC station and overhead lines
and cables can be represented with satisfactory accuracy and frequency response in the
kHz range. Nowadays even temporary overvoltages and switching-surge voltages can be
investigated.

However, the modelling of complex ac networks by simple network representations are


somewhat a limitation so that many system problems can not be investigated.

In this respect, the Parity Simulator developed by the FGH in Rheinau represents a big
step forward. The individual components are simulated by electronic analog computers so
that all parameters, including quality factor and time behaviour, can be scaled to represent
the actual component. The basic principle of the parity simulator is the conversion of
current and voltage values obtained by the analog computer into properly scaled terminal
quantities. This allows individual components to be connected to each other by wires, as in
the case of a physical model. The feature of concreteness is retained. The interaction of all
components is not mathematically formulated; instead, one can devote his attention to the
much simpler parameter representation of the individual components, the correctness of
which can easily be validated in field tests.

- 607 -
Figure 7.24: Parity Simulator of the FGH in Rheinau with measured value processing

The space requirement of a parity simulator suitable for investigating very extended
systems is by comparison very small, as can be seen in Fig. 7.24. The present
configuration of the FGH parity simulator in Mannheim-Rheinau comprises the following
components:

8 converter units (twelve-pulse) with regulators for applications in HVDC and in


FACTS
430 Line elements (single phase) for single-phase and multiphase lines such as for
implementing a bipolar HVDC line network, length: 4300 km with a limit
frequency of around 4 kHz or 20 ac single lines, each 200 km long (limit
frequency around 2 kHz)
130 RLC filter circuits (single phase)
3 Detailed synchronous machine models with regulators
6 Three-phase voltage sources with inertia and damping behaviour
120 Linear reactors (single phase)
20 Resistive loads (three phase)
15 Capacitive loads (three phase)
30 ZnO arresters
Computer-controlled measuring system for documentation of the results

- 608 -
With this equipment, even HVDC multiterminal systems with their ac networks simulated
in detail can be represented and studied. The use of actual control systems and components
is possible, although there is also a very flexible control system included in the simulator.
In the area of insulation coordination, switching surges and temporary overvoltages can be
studied and the protection functions be optimized. Interaction of the ac system with the
converters and their control, regulation and protection equipment can be evaluated.

A disadvantage of every simulator is the long preparation time for setting up a test system
and putting it into operation. An other technical limit is set by the frequency response
which is around a few kHz. Lightning strike and steep surge voltage stresses cannot be
investigated.

7.9.4.3 Advanced Fully Digital and Hybrid AC/DC Real-Time Simulator

The benefits of new technology applications achievable on site depend mainly upon the
quality of its preceding planning process. System planning is the basic prerequisite for a
system upgrade by HVDC.

Siemens uses powerful computer simulation tools such as NETOMAC and EMTDC for
System Planning. For system development and for project type-testing, an Advanced
AC/DC Real-Time Simulator is used to verify the performance of the real control and
protection equipment.

Computer simulation is the initial step for both, new developments and large project
applications: to define the optimum control algorithm that matches the given specific
system requirements. Especially the Siemens program NETOMAC provides a tremendous
variety of simulation options for these feasibility studies: there is practically no limitation
to the number of generators, lines, nodes and HVDC components to be simulated either in
stability mode or commutation mode or in combination of both of them.

As the next step of a project, the control and protection system will be built up and then be
brought into the real-time simulator for pre-commissioning and type testing. For these
tests, a reduced system-benchmark needs to be defined by the preceding computer study.
This benchmark must cover the real system worst-case conditions: minimum and
maximum short-circuit level including system resonances and possible system interactions
of the new and the already existing HVDC or FACTS installations with special regard to
the system protection.

- 609 -
Fig. 7.25 gives an overview of the bandwidth of system and converter interactions, that
have to be regarded. The simulations must cover the complete band of frequencies
beginning close to 0 Hz with the power oscillations (at the mechanical side of the
machines), followed by subsynchronous resonances, which can cause major problems
especially in series compensated systems. At nominal frequencies, unbalances can degrade
the quality of power supply. Harmonics are mainly produced by the converters especially
when the filter design is not sufficient or during system disturbances. Of main interest for
any control and protection equipment is the nominal frequency, which is filtered out by
proper signal processing technologies. Usually, the AC system frequencies are limited to
some kHz, whereas the converter spectrum reaches 10 kHz and more. Higher frequencies
are difficult to be simulated by real-time facilities, so for special effects from 40 kHz up to
the very fast transients (VF), computer simulation and field measurements need to be
done.

The facilities of the Advanced Real-Time Simulator are shown in Fig. 7.26. Six test
stations allow parallel independent work on different projects. Four of these test stations
are specially designed for HVDC investigations as well as FACTS and GTO studies. One
test station covers the special requirements for protection tests including high performance
voltage and current amplifiers for the relay inputs.

A unique feature of this SIEMENS simulator is the possibility of combined converter


control and protection testing with original equipment, which is often used for
feasibility studies as a part of system upgrade projects.

For the converter simulation, mainly analogue models are in use with regard to the high
frequency requirements (ref. to Fig. 7.25), whereas the system models (generators, lines,
transformers, loads etc.) are step by step now replaced or complemented by new digital
models.

- 610 -
FACTS
40 MHz

Computer Simulation Field Measurements 10 MHz VF Transients

40 kHz
AC/DC Real-Time Simulation 10 kHz

Converter 5 kHz

1 kHz Harmonics
Control
60 Hz System and 60 Hz Unbalances
Protection Power oscillations
SSTI

10 Hz 10 Hz SSTI
Power
oscillations 0 Hz 0 Hz

Figure 7.25: Bandwidth of System Interactions for HVDC, FACTS and


Protection

Digital
Power Sequence
...
Generation Controllers

AC/DC Positive and


Systems ... Zero Sequence
Components

HVDC/FACTS Protection Custom Power


1 ... 4 5 6 Measuring
6 Test
Stations ... Protection
Control

Signal
Simulator
=1 =1 Data Acquisition System Generation and
Interfaces
Recording
Real-Time Playback
Computer RTDS N E T O M A C, E M T D C, E M T P Computer
Simulation Simulation

Figure 7.26: Advanced AC/DC Real-Time Simulator Facilities

- 611 -
The simulator development was done in close co-operation with the well-known RTDS
Technologies Inc., Canada. RTDS models can be used both, for fully digital real-time
simulation and in combination with the analogue simulator as a hybrid simulator
extension. These different ways of simulation have been verified by tests and then
applied for large HVDC and SVC system studies. The hybrid simulator (see Fig. 7.27)
uses a voltage-source model with new developed hardware and software interfacing
technologies.

Analog > 5kHz

RTDS #

< 1kHz ~ 1kHz (70kHz)

< 3msec

Figure 7.27: Transient Requirements for Hybrid Simulator Extension

In addition to the RTDS digital real-time closed-loop simulation, the method of digital
data injection (playback of EMTP/EMTDC or NETOMAC runs) can be applied at each
test station.
At the protection test station (see Fig. 7.28), all simulator models from digital data
injection over digital on-line real-time simulation to the conventional analogue
simulation are available. The protection relays are fed through high quality power
amplifiers for voltage and current inputs. Facilities for signal transmission for the relay
communication, eg. in case of line protection, are available.
Recently, a new sixth test station has been added to the simulator for Custom Power
applications, ref. to Fig. 7.26. This special station uses fast power transistors like IGBTs
and it actually operates to improve the 400 V supply of the main infeed.
All these features offer a wide range of simulation options, with which the technical and
economical optimum for each application can be chosen. The philosophy and the
benefits of this combined design verification by computer and real-time simulation are
explained in Fig. 7.29.

- 612 -
Digital Simulator RTDS Analog Simulator

WS Digital/Analog Converter
Fault
Sequencer L
ZL(A)
R
PC ETHERNET
Selector
Switch
# Measuring
Transducers
EMTDC
NETOMAC F1 F2 F6 F5 F3 F4
ZS(L) ZS(R)
CVT
Simulation ZL(B)
I V
Amplifiers for
Voltage and Currents
7SA513 7SA513
Current/Voltage Current/Voltage
Inputs Inputs

16 Analog Inputs

32 Binary Inputs 32 Binary Inputs


Fault/Data
Recording

SI SI
G G
DI DI

Figure 7.28: Simulation Facilities for Real-Time Protection Testing

AC/DC System Performance Verification


Control, Measuring and Protection

Development:
EMTDC/ Optimized Advanced
NETOMAC Real-Time
Computer Studies Commissioning:
Simulation
Short + Effective

Detailed and Complete Future Upgrading: On-Line Tests of


System Representation Easy + Economic Original Equipment

Benefits
of Simulation

Figure 7.29: Benefits of Combined Computer and Advanced AC/DC


Real-Time Simulation

- 613 -
7.9.4.4 Digital Computer

Digital computer studies are indispensable with system studies for insulation coordination.
The ability to model actual quality factors and frequency responses, and thus the ability to
represent transient events with any frequency content and their damping, is a necessity for
insulation coordination studies. There is a danger, however, that stray capacitances and
inductances will be taken into consideration while damping influences inherent in the
system, which are difficult to determine, are ignored. Hence overvoltage peaks can easily
be computed which do not occur in the actual system. Basically all types of transient
events can be studied. However, the investigation of temporary overvoltages requires a
large amount of computer time and is therefore expensive. High frequency events such as
lightning surges, their dissemination as travelling waves, the reflection of the travelling
waves, and residual voltage stress on the components through the intervention of MO
arresters are the true domain of the digital computer simulations.

Programs suitable for the digital modelling of HVDC systems include the widely-used
EMTP, the EMTDC program which is specifically intended for HVDC studies, or the
highly adaptable NETOMAC program developed by Siemens. They solve the differential
equations which describe the system on an incremental basis. The length of the time
increments is adapted to the required time resolution. In addition to linear components ,
non-linear elements such as transformer saturation and altered topologies which occur
during switching, the commutation processes, or during flashovers in the system must be
representable. In contrast, detailed simulation of control and regulation functions is not
significant in the study of fast transient events.

- 614 -
The time required for a computer run depends not only upon the time interval to be studied
but also to a large degree upon the size of the time increments. Rapidly changing events or
discontinuous events require very small time increments, while much larger time
increments are adequate for events which occur at network frequency.

The development of programs to simulate a specific system configuration can require a


large amount of time. Modern programs provide help in the form of comfortable Graphical
User Interfaces, GUIs, such as shown Fig. 7.30 depicting the GUI for a NETOMAC
simulation.

7.5
EXCITER VOLT.
GENERATOR 1
(GAS)
0.0

-7.5
75
POLANGLE
GENERATOR 1
(GAS)
0
POLANGLE
GENERATOR 2
(STEAM)
-75
1 1
REACTIVE POW. ACTIVE POW.
GENERATOR 1 GENERATOR 1
(GAS) (GAS)
0 0

-1 -1
1
MECH. TORQUE
GENERATOR 1
(GAS)
0
ELEK. TORQUE
GENERATOR 1
(GAS)
-1
1 1
+0.7 pu VOLTAGE
AT BETR.SIE
IN PU
0 0
-0.7 pu

-1 -1
5
CURRENT PH.R
GENERATOR 1
(GAS)
0
GENERATORVALUES
PAGE 1 OF 6
-5
0.00 0.38 0.75 1.13 1.50
3 Phase Short Circuit (tsc=250 ms)
SIEMENS AG, EV NP Instantaneous Value Mode (0.00 to 0.675 s) and Stability Mode (0.675 to 1.60 s)
Calculated with NETOMAC for Windows
NEUGRAF SIEMENS AG EV_NP2-dn0040/Ru 1
(R)
25.5.1997 14:29 Produced with NETOMAC NETOMAC is a registered trade-mark of Siemens AG

Figure 7.30 Example of Userfriendly GUI (NETOMAC)

- 615 -
7.9.4.5 Analog High-Frequency Simulator

In studying particularly rapid transient events in the range of 0.1 to 100 ms, which do play
an important role in the insulation coordination of an HVDC system, specially prepared
analog high-frequency simulators can be used. They may be justified as a supplement to
results obtained with digital computers e.g. as confirmation.

7.9.4.6 Summary

The following table published by CIGRE Working Group 33.5 gives a quick overview of
the tools for system studies for insulation coordination. A column has been added here for
the real-time computer.

Events studied TNA HVDC Digital Real-time Analog


Simulator Computer Computer HF Model
Temporary overvoltages
a) with converters ignored X
b) with converters included (X) X X
c) precise evaluation of details X X
AC-side switching surges
a) Normal switching X X X X
b) Faults and fault clearing (X) X X X
DC-Side overvoltages
a) Caused on the ac side (X) X X (X)
b) Caused on the dc side X X X
c) Caused by converter faults X X X
Rapid transient events X X
MO arrester dimensioning and X X X X
effectiveness

- 616 -
8. HVDC Bibliography

8.1 Comments

There is a vast quantity of publications on HVDC systems and technology. New articles
are being continuously added. For this reason, the author intentionally chose not to give
references to literature in the body of the text. Where statements in the text are based
specifically on studies and publications of HVDC official entities, this fact was
specifically noted. These sources are cited again below. Furthermore a few standard books
on HVDC are listed below which the reader may refer to, when more theory is needed for
the basis of project planning of new types of HVDC configurations or applications.

8.2 HVDC Books

a) Edward W. Kimbark:
Direct Current Transmission
Wiley Interscience, a Division of John Wiley & Sons Inc.
New York, London, Sydney, Toronto; 1971

b) Erich Uhlmann:
Power Transmission by Direct Current
Springer Verlag Berlin, Heidelberg, New York; 1975

c) Robert Jötten:
Leistungselektronik, Vol. 1: Stromrichter- Schaltungstechnik
(Power Electronics, Vol. 1: Converter Circuit Technology)
Vieweg Verlagsgesellschaft mbH Braunschweig; 1977

d) Jos Arrillaga
High Voltage Direct Current Transmission
Peter Peregrinus Ltd, London; 1983

- 617 -
8.3 EPRI Publications

Transmission Line Reference Book HVDC +/- 600 kV


Electric Power Research Institute, Palo Alto, CA, USA; 1975

8.4 IEC Publications

a) Terminology For High Voltage Direct Current Transmission


IEC Publication 633; 1978

b) Performance of High Voltage D.C. (HVDC) Systems


Part 1: Steady State Conditions
IEC Publication 919-1; 1988

c) Performance of High Voltage D.C. (HVDC) Systems


Part 2: Faults and Switching
IEC Publication 919-2; 1991

c) Performance of High Voltage D.C. (HVDC) Systems


Part 3: Dynamic Conditions
IEC Publication 919-3; 1993

Bureau Central de la Commission Electrotechnique International, 3 rue de Varembe,


Geneva, Switzerland

8.5 CIGRE Publications

a) CIGRE Working Group 13/14.08:


Switching Devices Other Than Circuit Breakers for HVDC Systems
Part 1: Current Commutation Switches
Electra 125: 1989

b) CIGRE Working Group 13/14.08:


Switching Devices Other Than Circuit Breakers for HVDC Systems
Part 2: Disconnectors and Earthing Switches
Electra 134; 1991

c) CIGRE Study Committee 14, DC Links:

- 618 -
Protocol for Reporting the Operational Performance of HVDC Transmission; 1989

d) CIGRE Working Group 33.05:


Application Guide for Insulation Coordination and Arrester Protection of HVDC
Converter Stations
Electra 96; 1984

e) CIGRE Working Group 33/14.05


Application Guide for Metal Oxide Arresters Without Gaps for HVDC Converter
Stations

- 619 -

Вам также может понравиться