577A

Electrical Transient Interaction Between

Transformers and the Power System

Part 1- Expertise

Joint Working Group

A2/C4.39

April 2014
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

ELECTRICAL TRANSIENT INTERACTION

BETWEEN TRANSFORMERS AND THE
POWER SYSTEM – PART 1: EXPERTISE
JWG A2/C4.39

Members
A. da C. O. Rocha, Convenor (BR), A. Holdyk (DK), B. Gustavsen (NO), B. J. Jaarsveld
(ZA), A. Portillo (UY), B. Badrzadeh (AU), C. Roy (ES), E. Rahimpour (DE), G. H. da C.
Oliveira (BR), H. Motoyama (JP), M. Heindl (DE), M-O. Roux (CA), M. Popov (NL), M.
Rioual (FR), P. D. Mundim (BR), R. Degeneff (US), R. M. de Azevedo (BR), R. Saers (SE),
R. Wimmer (DE), S. Mitchell (AU), S. Okabe (JP), T. Abdulahovic (SE), T. Ngnegueu
(FR), X. M. Lopez-Fernandez (ES)

Corresponding members
A. Troeger (CH), C. Alvarez-Mariño (ES), D. Peelo (CA), D. Matveev (RU), G. A.
Cordero (ES), J. C. Mendes (BR), J. Leiva (AR), J. M. Torres (PT), J. Veens (NL), M.
Reza (SE), R. Asano (ES), R. Malewski (CA), S. Yamada (JP), U. Savadamuthu (IN), Z. J.
Wang (CN)
Copyright © 2014

“Ownership of a CIGRE publication, whether in paper form or on electronic support only infers right
of use for personal purposes. Unless explicitly agreed by CIGRE in writing, total or partial
reproduction of the publication and/or transfer to a third party is prohibited other than for personal
use by CIGRE Individual Members or for use within CIGRE Collective Member organisations.
Circulation on any intranet or other company network is forbidden for all persons. As an exception,
CIGRE Collective Members only are allowed to reproduce the publication.

Disclaimer notice

“CIGRE gives no warranty or assurance about the contents of this publication, nor does it accept any
responsibility, as to the accuracy or exhaustiveness of the information. All implied warranties and
conditions are excluded to the maximum extent permitted by law”.

ISBN : 978-2-85873272-2

Page 1
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Electrical Transient Interaction between

Transformers and the Power System –
Part 1: Expertise
Table of Contents
EXECUTIVE SUMMARY............................................................................................................... 6
1 INTRODUCTION ...................................................................................................................... 8
1.1 References ................................................................................................................................................................ 9

2 STANDARDS AND SERVICE EXPERIENCE ............................................................................. 10

2.1 Introduction ............................................................................................................................................................ 10
2.2 Standards............................................................................................................................................................... 10
2.3 Service Experience............................................................................................................................................... 13
2.3.1 Generator Step-Up Transformer 990MVA 21.45/500kV – Canada ................................................... 13
2.3.2 Auto-Transformer 200MVA, 500/345/13.2kV – United States ............................................................. 14
2.3.3 Generator Step-Up Transformer 75MVA, 230/16kV – Mexico ............................................................. 14
2.3.4 Autotransformer 240MVA 400/132kV – United Kingdom ...................................................................... 14
2.3.5 Autotransformer 150MVA 230/161/13.8kV – Brazil .............................................................................. 14
2.3.6 Autotransformer 400MVA 500/345/13.8kV – Brazil .............................................................................. 14
2.3.7 Regulator Transformer 33MVA, 230/66.9-44/13.2kV – Brazil ............................................................. 14
2.3.8 Regulator Transformer 300MVA, 500/460/13.8kV – Brazil .................................................................. 15
2.3.9 Distribution Transformer 10MVA, 77kV – Japan ........................................................................................ 15
2.3.10 Rectifier Transformer 26.4kV – United States ........................................................................................... 15
2.3.11 Autotransformers 315MVA, 400/220/22kV and 330/220/22kV – South Africa ......................... 15
2.3.12 Generator Step-Up Transformer 185MVA, 500kV – Brazil .................................................................. 15
2.3.13 Auto Transformer bank 1100MVA, 750/410/18kV – Hungary .......................................................... 15
2.3.14 Autotransformer, 500/275kV – Japan ....................................................................................................... 16
2.3.15 Generator Step-Up Transformer 500MVA, 765/25kV – Auxiliary Transformer
80MVA, 765/138/34.5kV – United States ........................................................................................... 16
2.3.16 HVDC Convertor Transformer 234 MVA 400/93 kV Yy-Yd – India ................................................... 16
2.4 Work of Previous Groups................................................................................................................................... 16
2.5 References ............................................................................................................................................................. 17

3 ELECTRICAL NETWORK TRANSIENT MODELLING ................................................................. 19

3.1 Introduction ............................................................................................................................................................ 19
3.2 Substations ............................................................................................................................................................. 19
3.3 Upstream network ................................................................................................................................................ 20
3.3.1 Low frequencies studies ..................................................................................................................................... 20
3.3.2 Switching studies.................................................................................................................................................. 20
3.3.3 Lightning studies .................................................................................................................................................. 21
3.4 Overhead lines and underground cables ....................................................................................................... 22
3.4.1 Parameter determination .................................................................................................................................. 22
3.4.2 Travelling wave-type models ........................................................................................................................... 22
3.4.3 Lumped-parameter type models ..................................................................................................................... 23
3.5 Surge arresters ..................................................................................................................................................... 23
3.6 Circuit breakers and disconnectors................................................................................................................... 27

Page 2
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

3.6.1 Circuit Breakers .............................................................................................................................................. 27

3.6.2 Disconnectors................................................................................................................................................... 29
3.7 References ................................................................................................................................................................... 29

4 TRANSFORMER MODELLING ................................................................................................ 33

4.1 Simplified procedures ......................................................................................................................................... 34
4.1.1 Power frequency standard model with external capacitance added .................................................... 34
4.1.2 Concentrated Capacitance Model for Fast Transients ............................................................................... 35
4.1.3 Frequency dependent transformer model for fast transients .................................................................... 36
4.2 White box approach........................................................................................................................................... 40
4.2.1 Frequency Dependence ..................................................................................................................................... 41
4.2.2 Computation Methods for Transients .............................................................................................................. 41
4.2.3 Mathematical Model .......................................................................................................................................... 41
4.2.4 Lossy Lumped Parameters ................................................................................................................................. 42
4.2.5 Illustration. Practical Application ..................................................................................................................... 43
4.2.6 Validation. Measurements and Simulations ................................................................................................... 43
4.3 Black box approach ............................................................................................................................................ 46
4.3.1 Characterization of the transformer behavior ............................................................................................. 46
4.3.2 Model extraction ................................................................................................................................................. 48
4.3.3 Model interface with EMTP-type simulation software ................................................................................ 49
4.3.4 Simplifications to the black-box approach ................................................................................................... 50
4.3.5 Validation of measurement setup and modelling procedure .................................................................... 51
4.4 Grey box approach ............................................................................................................................................ 51
4.4.1 The grey box transformer model .................................................................................................................... 51
4.4.2 Frequency Response Analysis for parameter estimation............................................................................ 52
4.4.3 Layered Model .................................................................................................................................................... 52
4.4.4 Determination of terminal transfer functions ................................................................................................. 53
4.4.5 Model parameter estimation ............................................................................................................................ 53
4.5 Comparison Table ................................................................................................................................................ 55
4.6 References ............................................................................................................................................................. 56

5 NETWORK INTERACTION WITH TRANSFORMER .................................................................. 60

5.1 General .................................................................................................................................................................. 60
5.2 Introduction to overvoltages in transformers and resonant conditions...................................................... 60
5.2.1 Resonance........................................................................................................................................................ 60
5.2.2 Resonant interaction between transformer and external system ........................................................ 61
5.3 Transformer external overvoltages .................................................................................................................. 62
5.3.1 Impinging overvoltage and voltage transfer between windings ............................................................. 62
5.3.2 External resonance behavior explained by simplified transformer model............................................ 63
5.3.3 Resonant interaction between the transformer and a feeder cable ....................................................... 65
5.4 Transformer internal overvoltages ................................................................................................................... 67
5.5 Resonant overvoltages: Topologies, switching operations, faults .............................................................. 70
5.5.1 Cable-transformer networks ............................................................................................................................. 70
5.5.2 Phenomena caused by closing a circuit-breaker.......................................................................................... 71
5.5.3 Phenomena caused by fault initiation............................................................................................................. 73
5.5.4 Phenomena caused by circuit-breaker opening ........................................................................................... 74
5.6 Restrikes and circuit breaker technologies...................................................................................................... 75
5.7 Disconnector switching ......................................................................................................................................... 75
5.8 Lightning overvoltages ........................................................................................................................................ 77
5.9 Frequency of occurrence..................................................................................................................................... 78
5.10 Conclusion............................................................................................................................................................... 78
5.11 References ............................................................................................................................................................. 79

Page 3
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

6 ASSESSMENT OF TRANSFORMER VOLTAGE STRESSES ........................................................ 81

6.1 Introduction ............................................................................................................................................................ 81
6.2 Time Domain Waveform Conversion – Conventional approach used by manufacturers ..................... 81
6.3 Time Domain Severity Factor ............................................................................................................................. 84
6.3.1 Introduction ........................................................................................................................................................... 84
6.3.2 Time Domain Severity Factor Computation ................................................................................................... 84
6.4 Frequency Domain Severity Factor .................................................................................................................. 86
6.4.1 Introduction ........................................................................................................................................................... 86
6.4.2 Methodology for analysis in frequency domain .......................................................................................... 86
6.4.3 Severity factor ..................................................................................................................................................... 87
6.4.4 Consideration about the FDSF application .................................................................................................... 88
6.4.5 Trends .................................................................................................................................................................... 88
6.4.6 Example ................................................................................................................................................................ 88
6.5 Conclusion............................................................................................................................................................... 89
6.6 References ............................................................................................................................................................. 90

7 IMPACT ON TRANSFORMER INSULATION ........................................................................... 92

7.1 Introduction ............................................................................................................................................................ 92
7.1.1 Insulation strength................................................................................................................................................ 92
7.1.2 Main insulation ..................................................................................................................................................... 93
7.1.3 Internal insulation ................................................................................................................................................ 95
7.2 Insulation aging ..................................................................................................................................................... 96
7.2.1 Introduction ........................................................................................................................................................... 96
7.2.2 Examples of measurement for shell-type transformers .............................................................................. 96
7.2.3 Examples of measurement for core-type transformers .............................................................................. 97
7.2.4 Summary ............................................................................................................................................................... 99
7.3 Effect of repetitive impulses............................................................................................................................... 99
7.3.1 Introduction ...................................................................................................................................................... 99
7.3.2 V-N characteristics below and at 500kV class ..................................................................................... 100
7.3.3 V-N characteristics at UHV class .............................................................................................................. 102
7.3.4 Summary ........................................................................................................................................................ 104
7.4 References ........................................................................................................................................................... 104

8 TRANSIENT SIMULATION SOFTWARE BENCHMARKING – FICTITIOUS

TRANSFORMER...................................................................................................................... 106
8.1 Introduction .......................................................................................................................................................... 106
8.2 The Fictitious Transformer ................................................................................................................................. 106
8.2.1 Electrical Characteristics .................................................................................................................................. 106
8.2.2 Constructive Details .......................................................................................................................................... 107
8.2.2.1 Core – “Three-Legs Core” ........................................................................................................................... 107
8.2.2.2 LV Winding – “Continuous Disk Type” ...................................................................................................... 108
8.2.2.3 HV Winding – “Interleaved Disk Type” ................................................................................................... 109
8.2.2.4 Core & Windings Layout ............................................................................................................................. 110
8.2.2.5 Tank – “Rectangular with Radiators” ........................................................................................................ 111
8.2.2.6 Losses and Reactance Calculation ............................................................................................................. 112
8.2.3 Modelling of Windings and Nodes and Branches Numbering ............................................................... 112
8.3 List of Participants .............................................................................................................................................. 116
8.4 Questionnaire ...................................................................................................................................................... 116
8.5 Variants to Calculate ......................................................................................................................................... 117
8.5.1 Lightning Impulse ............................................................................................................................................... 117
8.5.1.1 Nodes and Branches Maximum Voltage Values .................................................................................... 118
8.5.1.2 Node and Branches Voltage Wave Shapes ........................................................................................... 120
8.5.2 Switching Impulse .............................................................................................................................................. 121
8.5.2.1 Definition of the Damped Oscillatory Wave Shape ............................................................................. 121

Page 4
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

8.5.2.2 Relation between Time Constant and Damping Factor ......................................................................... 122
8.5.2.3 Damped Oscillatory Wave Shapes - Numerical Expressions.............................................................. 122
8.5.2.4 Fictitious Transformer Responses ................................................................................................................ 123
8.6 Time Domain and Frequency Domain Severity Factors.............................................................................. 124
8.7 Conclusions ........................................................................................................................................................... 125
8.8 References ........................................................................................................................................................... 126

9 RECOMMENDATIONS ......................................................................................................... 127

9.1 General considerations on System Aspects .................................................................................................. 127
9.1.1 Shunt capacitor banks energization .............................................................................................................. 127
9.1.2 Shunt capacitor banks interruption ................................................................................................................ 127
9.1.3 Transmission lines energization ...................................................................................................................... 127
9.1.4 Transformers energization .............................................................................................................................. 128
9.1.5 Switching in GIS (Gas Insulated Substations) .............................................................................................. 128
9.1.6 Lightning overvoltages ..................................................................................................................................... 128
9.1.7 Protection against multiple restrikes ............................................................................................................. 128
9.2 General considerations on Transformer Design Practices ......................................................................... 129
9.3 Transformer Specification................................................................................................................................. 131
9.3.1 Dielectric tests .................................................................................................................................................... 131
9.3.2 System studies .................................................................................................................................................... 133
9.3.3 Transformer high frequency modelling ........................................................................................................ 134
9.3.4 Terminal Model (Black Box model) ............................................................................................................... 136
9.3.5 Grey Box Model ............................................................................................................................................... 136
9.3.6 Low frequency model ....................................................................................................................................... 137
9.3.7 Insulation stress assessment ............................................................................................................................. 137
9.3.8 Transformer Failure Analysis .......................................................................................................................... 139
9.3.9 Interaction between manufacturer and user ............................................................................................... 141
9.4 Transient Measurements .................................................................................................................................... 142
9.4.1 Permanent Setups ............................................................................................................................................ 142
9.5 References ........................................................................................................................................................... 143

APPENDIX A - FICTITIOUS TRANSFORMER: TRANSIENT SIMULATION RESULTS .................. 148

A.1 Lightning Impulse – Nodes Maximum Voltage Values ............................................................................... 148
A.2 Lightning Impulse – Branches Maximum Voltage Values ........................................................................... 150
A.3 Variant FT1 - Lightning Impulse – Temporal Wave Shapes ..................................................................... 152
A.4 Variant FT2 - Lightning Impulse – Temporal Wave Shapes ..................................................................... 158
A.5 Variant FT3 - Lightning Impulse – Temporal Wave Shapes ..................................................................... 164
A.6 Variant FT4 - Lightning Impulse – Temporal Wave Shapes ..................................................................... 168
A.7 Variant FT5 - Switching Impulse – Temporal Wave Shapes .................................................................... 172
A.8 Variant FT6 - Switching Impulse – Temporal Wave Shapes .................................................................... 173
A.9 Variant FT7 - Switching Impulse – Temporal Wave Shapes .................................................................... 174
A.10 Variant FT8 - Switching Impulse – Temporal Wave Shapes .................................................................... 175

Page 5
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

EXECUTIVE SUMMARY
A number of transformer dielectric failures have been attributed to transient overvoltages, even when good
practices for insulation design and insulation coordination have been followed. CIGRE WG A2/C4.39 was formed
with the objective to clarify possible reasons for such failures and to recommend remedial actions, in the context of
high-frequency transients and insulation design practices.

The principal conclusions of this work are:

The current factory proof tests contained in the standards do not completely address all types of transient
events that occur in the field. The use of the standard lightning impulse wave shape is not appropriate in
the case of the fast-front or oscillatory waveforms occurring in actual service conditions with reactor
switching, HVDC converters, capacitor banks switching, GIS switching and transformer energization via
feeder cable. In addition, these tests are performed with the non-excited terminals grounded which do not
adequately take into consideration the voltage transfer between terminals.

The manufacturing industry and transformer purchasers have assumed that the problems of transient
voltage have been adequately addressed by current impulse standards. This is not the case. There are
still failures recently reported due transients, and many unknown failures are of dielectric origin and may
be related to transient phenomena.

Other working groups have addressed this problem but it still requires attention.

For certain network configuration, there is a high probability that system-initiated transients may contain
oscillatory voltage wave at the transformer's terminals which coincide with the transformer's natural
frequencies. These internal voltages can exceed the insulation withstand capability of the transformer by
resonant voltage buildup. Failures may occur even if their amplitude of the impinging overvoltage is much
lower than the arrester protection level. As far as the transformer design is concerned, this type of
vulnerability cannot be avoided.

The transformer affects the wave shape of the transient overvoltage at its terminals due to its frequency-
dependent impedance. An appropriate model of the transformer should therefore be applied in transient
simulations. There are several different approaches and levels of sophistication for obtaining such
models. Manufacturers typically create detailed models for studying internal winding stresses based on
information about geometry and material properties. Others, due to a lack of this detailed winding
information, create terminal equivalents based on measurements at the terminals. Most models are
compatible with common circuit simulators.

For the representation of the adjacent power system in transformer overvoltage studies, the standard
simulation tools provide sufficiently accurate models for most situations.

The standard approach to assess the internal transformer voltages stress is to use analysis tools and
design information normally only available to manufacturers. The utility on its side can make an initial
evaluation using the so called frequency domain severity factor (FDSF) which is obtained via a time
domain simulation with a terminal equivalent transformer model. The FDSF approach can thus be used
both for design review upon incoming transients and in analysis of failures. When combined with online
monitoring, it can also be used as indicator of increased transient risks for a unit.

Repetitive transient overvoltages and ageing reduces the insulation withstand capability and must be
recognized in the design of the transformer insulation system. The breakdown characteristics of solid
materials due to high frequency transients are still not well known and deserve future work.

Thirteen case studies are presented in “Part 2: Case Studies”, which demonstrate situations where
system transients lead to excessive overvoltages in transformers. These studies clearly show the
importance of considering not only the peak of these overvoltages but also the frequencies involved.
Some of these case studies are related to failures with overvoltage as probable cause.

Page 6
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

A “Fictitious Transformer” was defined to evaluate the performance of the white box models when
calculating the internal voltage distribution due to different types of transients applied to its terminals. The
simulation results obtained by 11 independent parties (manufacturers, universities, consultants) were in
good agreement in the case of the internal voltages maximum values, but some differences were found in
the wave form shapes.

The resonance frequencies are strongly dependent on the values of the inductances (self and mutual)
and capacitances that were used to represent the transformer. Some members performed examples
using the same values of inductances and capacitances and in that case the internal voltages obtained
with the different softwares were identical.

Manufacturers must improve their models in order to achieve more accurate values for the maximum
internal voltages throughout the winding and consequently better responses in the time domain. An
improvement in these models requires better methods for calculating the inductances and capacitances
that represent the transformer windings.

Good system operation and design practices may help to prevent transformer failures due to transients,
but it is very important that the transformer insulation structure addresses the presence of these
transients. This can be achieved by writing a specification that appropriately reflects the unique
requirements of a utility system (for example, special test voltage). This requires a close cooperation
between the manufacturer and the purchaser. In such cooperation, it is desirable that the manufacturer
provides the utility with an appropriate terminal equivalent of the transformer so that transient studies can
be performed. An outcome of such studies can also be that the excitation of transformer internal
resonances can be detuned by small modifications to the power system.

Page 7
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

1 INTRODUCTION

Transformers are constantly exposed to different types of transient events during their daily operation which
imposes high stresses on their insulation structure. Field experience has shown that even when good insulation
coordination studies and well-accepted insulation design practices are applied, a significant number of transformers
suffer dielectric failure as reported in the literature. Such failures may occur due to transient events which are not
necessarily related to any system event at the time of its occurrence. The analysis of the failures and their future
prevention requires an in depth knowledge of the transient interaction between transformer and the power system.

In this context, another important aspect to consider is the fact that, under the new power system deregulation
scenario, the necessity to integrate different agents, such as the transmission system operators, generators and
distributors, requires the development of new operation procedures, when compared to the operation procedures
previously used. These new system operation conditions in combination with a more extensive usage of transient
generating technologies and the trend of keeping the equipment longer in operation create a new electrical
environment for transformers with an expected increase of the dielectric stress on their insulation.

Although previous IEEE and CIGRE working groups [1.1, 1.2] have reported important findings on this subject,
additional evaluations with a wider scope was found necessary to improve transformer reliability regarding
transients [1.3]. The extended scope should include transformer design and testing with consideration to its
insulation system high frequency behavior and its modelling for system studies. With this focus, CIGRE JWG
A2/C4-39 “Electrical Transient Interaction between Transformers and the Power System” was formed as an
additional contribution to this task. This Joint Working Group began its operation in 2008, comprising members
representing generation, transmission and distribution utilities, transformer manufacturers, universities and
research centers. A significant number of technical contributions were received throughout the work from experts of
20 countries.

This technical brochure presents a summary of the investigation carried out by the group and has been divided into
two parts, “Part 1: Expertise” and “Part 2: Case Studies”. The “Part 1: Expertise” has nine chapters, dealing with:

Chapter 2 presents an overview of the work of previous group on this subject, some examples of
transformer failures due to transients and the status of the current standards regarding these phenomena.

Chapter 3 discusses some aspects regarding the network modelling for transient studies.

Chapter 4 deals with the state of art of transformer modelling covering different approaches such as
black box, grey box and white box modelling.

Chapter 5 covers some theoretical aspects concerning high frequency transformer resonant overvoltages.

Chapter 6 describes new concepts of analysing the proper electrical stress imposed on the power
transformer due to non-standardized impulse.

Chapter 7 discusses different aspects regarding the impact of transients on transformer insulation.

Chapter 8 compares different computational tools for transient voltage calculation along the winding of a
“Fictitious Transformer”.

Chapter 9 presents general recommendations regarding transformer specification, transient

measurement and dielectric tests.

The second part of this brochure, presents case studies carried out covering transformer failure analysis, examples
of interaction with circuit-breakers and different modelling application.

The main goal of this document is to provide an update in the study of this broad and complex topic with focus on
some relevant aspects, including resonant overvoltages. It should be borne in mind that new approaches and
challenges are expected to arise as new technologies are introduced together with different power system
scenarios.

Page 8
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

It is clear that a good knowledge of the possible transient interactions between the transformer and the power
system cannot be reached without a close contact between manufacturer and clients with their respective
expertise. Good communication, not only during the transformer procurement process but through its life in
operation, is essential in this pursuit.

1.1 References
[1.1] Study Committee A2/B3/A3, JWG 21 “Electrical Environment of transformers”, Electra No.219, Feb 2005.

[1.2] IEEE Guide to Describe the Occurrence and Mitigation of Switching Transients Induced by Transformer,
Switching Device, and System Interaction, C57.142, 2010.

[1.3] Cigre A2/C4 committee. JWG A2/C4.39 Term of Reference, Electrical Transient Interaction between
Transformers and the Power System.

Page 9
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

2 STANDARDS AND SERVICE EXPERIENCE

2.1 Introduction
From its early development, it has been a constant effort of engineers to maintain and improve the power grids
reliability. This has meant not only the reliability of operation but more specifically the reliability of components. A
power transformer is one of the most strategic and costly piece of equipment in the power system, requiring a high
level of reliability and availability. As such, transient over-voltages have long been recognized as one of the
important causes of equipment failure and thus, unavailability.

The present chapter addresses how high frequency transient over voltages in equipment, such as transformers,
have been dealt with by users, standards, and manufacturers. Additionally, several examples of experience in
service are provided.

Appreciating that improving the standards and the common practices is a continuous process, the chapter ends
with a short summary of the contributions of previous Cigre and IEEE Working Groups, relating to transient
interactions between power transformers and the power Grid.

2.2 Standards
Power transformers in the electric system may be subjected to transient over voltages of varying magnitudes,
shapes and frequency content, as a result of atmospheric discharges or interaction with the power system as fault
conditions, switching, etc.During the design process, different approaches or adapted tools are usually applied by
the manufacturers to predetermine the electrical stresses on the windings, and thus, to size the insulating structure
necessary to withstand those electrical stresses resulting from the transient over voltages.

As current general practice, the design and the overall manufacturing process of power transformers are validated
during the factory acceptance tests (FAT); the applicable tests, recommendations or guides for these being given
by the standards.

Most countries have their own standards and some of them are applied regionally. Nevertheless, two main families
of standards are internationally used. On one hand, there is the IEC (International Electrotechnical Commission)
family of standards, which have been historically applied mostly in countries around the world where the power
system frequency is 50 Hz. On the other hand, there are the ANSI/IEEE American standards, which have been
historically applied mostly not only in USA but also in most of the countries where the power system frequency is
60 Hz. It is important to observe that, based on these standards, some large companies and utilities have built their
own specifications including special requirements.

IEC has grouped the main power transformers standards under the 60076 series number, excluding from the
scope, some specific applications as rectifier transformer, railways applications, etc, which are dealt with in different
other IEC standard series. Regarding high voltage testing, the IEC related standards are grouped under the series
number 60060.

On their side, the IEEE standards relating to transformers are grouped under the C57 series. The IEEE C57
transformers standard series have a wider scope as they also cover special transformers, instrument transformers,
as well as different ranges of transformers, which are not included in the IEC 60076 series scope.

The transformer rated power definition is also different in these two standards. Whereas the IEC standards define
the rated power in terms of the input to the primary side, the IEEE standards define it as the power delivered at the
secondary terminals. For large power transformers, however, this is not relevant since internal losses of the
transformers, which account for the difference in the definitions, are only a small percentage of the throughput
power.

The verification of power transformer withstand to transient over voltages are part of the dielectric tests. The
different types of dielectric standard impulses were defined to represent most of the conditions that can occur in the
network. The test voltage levels are chosen in accordance with the voltage level tables included in the national or

Page 10
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

the international standards for each transformer winding. These voltages levels must be in accordance with system
insulation coordination studies.

Insulation coordination studies are necessary/mandatory in order to provide information on all essential actions that
must be considered in the design to avoid incidents with the equipment, caused by over voltages. Insulation
coordination is a topic more specifically dealt with by system specialists and reflects the needs of local network.The
choice of insulation level depends on as many factors as: The highest system voltage,exposure to over voltages
and level of protection of protective devices, the network grounding method, transformers three phase coupling,
etc.

In order to verify the transformers internal insulation structure capability with respect to transient over voltages, the
standards have defined varying wave shapes. At this point it is interesting to make a historical review to understand
the origin of each of these tests that are commonly used today.

In the early twentieth century, the start in operation of the first long distance high voltage lines interconnecting
consumers and generation centers made necessary the evaluation of the effects of lightning impinging on them.
This led to the application of grounding wire on the lines and the use of arcing horns and gap-type surge arresters
to protect the transformers.

In addition to that, it was necessary to develop tests to reproduce the field conditions in order to evaluate the
transformer’s ability to withstand the resulting stresses. This was the origin of the Full Wave Lightning Impulse Test
which tries to simulate the lightning over voltages impinging on the transformer and also of the Chopped Wave
Lightning Impulse Test to simulate the action of arcing-horn and the gap-type surge arresters commonly used at
the time to protect the transformer or to simulate insulator flash over.

With the development of the gap-less ZNO surge arrester, the arcing horns and the Chopped Wave Lightning
Impulse Test were considered unnecessary and therefore this test was classified as special test in the standard
IEC 60076-3: 2000. Recently, some transformers faults were attributed to internal resonance in the transformer
windings due to switching system operations, especially in gas-insulated substations (GIS). This has initiated
discussion about the possibility of new tests to cover these operational conditions and, in particular, has put into
general consideration again the Chopped Wave Lightning Impulse Test as a valid alternative. The IEC 60076-3
was reviewed and now the Chopped Wave Lightning Impulse Test is a routine test for Um>170kV.

Another type of transient overvoltage that needed to be considered with the introduction of high voltage long
transmission lines were the switching impulse overvoltage due to sudden changes in the state of power systems,
e.g. switching operations or faults. It has become generally recognized that switching impulse voltages are usually
the dominant factor affecting the design of insulation in high voltage power systems for rated voltages of about 300
kV and above.

The necessity to evaluate the switching impulses and their effects became especially evident in the sixties, when
high voltage and extra-high voltage up to 750 kV started to be extensively applied. Thus, similar to the lightning
impulses, there was a need to create tests that covered these new operating conditions.

With this goal, the switching impulse test was devised and the various international standards began to recommend
that equipment designed for voltages above 300 kV be tested according to it.

This brief historical review shows how the tests are dynamic in time, the system evolution in extension and voltages
creates new demands on electrical equipment connected to the system, and this in turn requires the development
of new tests to evaluate the behavior of the equipments against these new contingencies. Obviously, all possible
wave shapes arising in the power system cannot be covered by the standards. The standards rather attempt to
provide envelop of the practical existing conditions, at the time they are developed.

The current standards define a series of impulse tests, according to the system voltage which are described below:

The Full Wave Lightning Impulses which aim to cover the lightning events and resulting transient over
voltages waves that travel the electric power system and impinge on the transformers. Typical shape is

Page 11
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

seen in Figure 2.2-1. It is referred to standards (IEC 60076-3, IEC 60076-4, IEC 60060-1 and/or IEEE
C.57-12 90, IEEE C57.12.00, IEEE C57-98) for more details on sequences and conditions.

Figure 2.2-1: Lightning Impulse Typical Waveform requirements [2.1], [2.2].

The Chopped wave lightning impulses which aim to address the over voltage shapes, which occur
when the lightning voltage waves are modified and reduced after a flash over occurs across insulators.
Typical wave shape is seen in Figure 2.2-2. It is referred to standards (IEC 60076-3, IEC 60076-4, IEC
60060-1 and/or IEEE C.57-12 90, IEEE C57.12.00, IEEE C57-98) for more details on sequences and
conditions.

Figure 2.2-2: Chopped Lightning Impulse Waveform requirements [2.1], [2.2].

• The Switching Surges which attempts to address those electrical stresses and over voltages that can
occur as results of switching operations in the electric network. Typical shape is seen in Figure 2.2-3. It is
referred to standards (IEC 60076-3, IEC 60076-4, IEC 60060-1 and/or IEEE C.57-12 90, IEEE C57.12.00,
IEEE C57-98) for more details on sequences and conditions. It is well known that this test does not
reproduce well the oscillatory waves that appear in the system due to switching and faults as it does not

Page 12
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

cover the broad range of frequencies involved. Although, this test is necessary to verify the external air
insulation distances.

Figure 2.2-3: Switching Surge typical waveform requirement [2.1], [2.2].

Power grid components, structure and operation modes have been under continuous development/evolution.
These system changes subject HV equipment, such as power transformers, to newer electrical stresses. These
transient overvoltages contain wave shapes and frequency content not directly addressed by the existing
standards. Additionally, the effects on electrical insulations ageing is not yet fully understood or not considered.
This situation is illustrated by the fact that some transformers may fail in the field whereas they had passed the
FAT.

With these in mind and, also taking into account evolution of HV equipment procurement processes (globalization),
sometimes, along with standardized dielectric tests, additional special wave shapes or terminal conditions may be
agreed between transformer manufacturers and users; references [2.3], [2.4] and [2.5] give some examples of this
type of agreement.

2.3 Service Experience

In several cases, evidences of failures related to Transient Phenomena have no obvious immediate cause-effect
correlation. As a matter of fact, the transformers are internally sized to face varying types of stresses as, the
operating steady state electric conditions, thermal aspects related to admissible hot spot temperatures, also
vibrations aspects; this implies the following up of a transformer unit through monitoring techniques based on
chemical contents in the oil, partial discharge activity measurements, …etc. Those aspects also have to be
considered when failures occur, the failures resulting generally from combination of causes, cumulated over the
time (operating conditions, repetitive energizations, aging of the insulations, aging of the oil, high steady-state
voltages/new grid-codes, vibrations, etc).

In this context, this section presents few case studies from the literature related to transformer failures, where from
analysis, transient over voltages interactions with the power system were finally pointed out as the possible failure
root cause. It is to be observed that, this type of failures involving transient interaction between transformer and
power system happened in different type of electrical network systems, in various countries all over the world.

2.3.1 Generator Step-Up Transformer 990MVA 21.45/500kV – Canada [2.6]

One unit of 990MVA EHV generator step-up transformer bank failed in one of the major generating stations of the
utility. The unit was operating in a “back-feeding” mode supplying power to the station via the Unit Service
Transformer, with the generator not connected. Physical examination revealed that arcing had taken place in
several areas in the bottom half of the HV winding on one core leg, including disks at the line end. Investigations

Page 13
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

concluded that while there was no direct cause-effect situation that damaged the transformer, there were
unacceptably high disk-to-disk voltage stresses near the HV line terminals in the case of bus faults in the SF6
substation.

2.3.2 Auto-Transformer 200MVA, 500/345/13.2kV – United States [2.7]

During a single-phase to ground fault, caused by lightning on phase 1 of a 345kV transmission line, a three-phase
bank of auto-transformers installed approximately 145 miles from the fault location was tripped by differential
relays. A post-mortem analysis concluded that a flashover had occurred between the taps of the no-load tap
changer, which caused a distortion of the coils connected between them.

One month later, in another lightning storm, a series of eight single-phase to ground fault occurred on phase 3 of
the 345kV line at approximating the same location as the previous one. The auto-transformer bank was tripped
again by differential relays, this time due to the failure of phase 3 winding during the line faults. A flashover was
detected between the same taps, as on phase 1 unit, which caused the failure of the winding between these taps.

2.3.3 Generator Step-Up Transformer 75MVA, 230/16kV – Mexico [2.8]

Seven single phase shell-form 230kV generator step-up transformers failed in service. Analysis of these failures
has revealed insufficient protection by conventional silicone-carbide lightning arresters and excitation of an internal
winding resonance by steep-front lightning impulses. Short duration impulse coming from transmission line is
reflected at transformer bushing and at the station bus-bar end resulting in an oscillation which coincides with one
of transformer high frequency natural resonance.

2.3.4 Autotransformer 240MVA 400/132kV – United Kingdom [2.9]

United Kingdom has experienced several unexplained dielectric faults and failures in a group of its power
transformers. Evidence of problems arisen from interactions with the system. All the problems could be attributed to
a common failure mode: internal over-voltages arising from part-winding resonance initiated by remote
energization, either manually or by the action of delayed automatic re-closure schemes as can be shown in one of
the dielectric failure mentioned below.

A new protection system was being commissioned on a circuit comprising a 92km overhead line route incorporating
a 2km long cable 9km from the substation. At the substation a 34 year old 400/132kV 240MVA autotransformer
was connected directly (without a circuit breaker) to the line. As part of the protection commissioning the
transformer was energized 6 times from the remote end of the line. On the sixth energization there was an
explosion within the tank that operated the Buchholz oil surge trip and caused a minor rupture in a tank weld.
Subsequent inspection revealed that one phase of the centre-entry series winding had flashed over from about half
way down the lower half of the winding, along an insulating wrap to the stress ring at the bottom 132kV end of the
winding.

2.3.5 Autotransformer 150MVA 230/161/13.8kV – Brazil [2.10]

Failure of one phase of a 230/161/13.8kV transfer bus. Analysis carried out by CEMIG GT together with
manufacturer indicated that high frequency voltages originated by repetitive strikes of the electrical arc generated
by the disconnect opening operation were amplified within the 13.8kV winding causing the failure.

2.3.6 Autotransformer 400MVA 500/345/13.8kV – Brazil [2.10]

Unexplained dielectric failures of two 400MVA 500/345/13.8kV autotransformers, within a week in February 1995,
led the company to revise their traditional view about reliability of its transformers. After exhaustive analysis, it was
concluded that the most likely cause of the failures, was the occurrence of internal resonance excited by abnormal
disconnector switching near the equipment.

2.3.7 Regulator Transformer 33MVA, 230/66.9-44/13.2kV – Brazil [2.10]

In the 90's a three-phase transformer 33 MVA, 230/66.9-44/13.2 kV failed after the interruption of the magnetizing
current by a disconnector switching. The substation layout, at the time, was such that for every two transformers,
there was only one circuit breaker at the high-voltage side and de-energization was achieved by opening the low

Page 14
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

voltage circuit breaker followed by the opening of the high voltage side disconnector switching. After this
occurrence, all of the no-load current opening operations through disconnector switching were prohibited in
EletrobrasEletrosul system for all voltage levels. Details of internal inspection of the transformer are unknown.

2.3.8 Regulator Transformer 300MVA, 500/460/13.8kV – Brazil [2.10]

In 1988, a few minutes after a phase-to-ground short-circuit in a 460 kV transmission system followed by automatic
re-closure, a dielectric failure occurred at one phase of a 300 MVA, 550/460-13.8 kV transformer bank. Internal
inspection showed that there was an electrical discharge between load tap changer contacts. The regulation
winding frequency response measurement showed significant resonance in the range 4-6 kHz, which is typical of
switching surges.

2.3.9 Distribution Transformer 10MVA, 77kV – Japan [2.11]

Two distribution transformers in a 77kV power system were damaged three times by winter lightning from 1997 to
2001. The normal operating conditions of surge arresters indicated that the incident surges were limited to below
the lightning impulse withstand voltage of the transformers. Study results showed that internal resonance
overvoltage of transformer caused by incident lightning surge was inferred as the cause of the failures, even
though there was no evidence indicating a relationship between the frequency components of incident surges and
the internal resonance overvoltages of transformers.

2.3.10 Rectifier Transformer 26.4kV – United States [2.12]

A 26.4kV rectifier transformer with 150kV-BIL primary winding failed because of a temporary overvoltage (TOV)
following a switching. The primary winding was damaged in the middle region. Analysis determined a rectifier with
dc link capacitors and the transformer inductance formed an internal resonance that was excited by the switching.
Fails in the middle of the transformer primary winding are typical from LC series resonance.

2.3.11 Autotransformers 315MVA, 400/220/22kV and 330/220/22kV – South Africa [2.13]

Autotransformers of 400/220kV and 330/220kV both with a 22kV tertiary reactor failed several times due to
overvoltages on the regulating winding leads. All failures occurred when the tertiary reactors were in circuit. An
investigation using a series of low-voltage pulse-injections, current-interruptions and impulses indicated the
occurrence of potentially severe oscillations in the regulating windings during surges on any one of the HV
terminals. As the failures occurred when tertiary reactors were in circuit, it gave an indication that the worst part-
winding overvoltages were generated during de-energization of the reactor-loaded transformer. A series of HV
measurements in the field confirmed the occurrence of re-ignition during reactive switching.

2.3.12 Generator Step-Up Transformer 185MVA, 500kV – Brazil [2.14]

In the period of two years, three generator step-up units produced by the same manufacturer failed. All
occurrences happened in the same bank. Investigations concluded that circuit breakers switching (especially the
ones without pre-insertion resistors) with amplitude value below surge-arresters operation level reach transformers
with components in the same frequency range of faulty transformers family natural oscillation. Surges that reach
HV windings and which frequency components coincided with natural oscillation frequencies of windings sections
will be amplified leading, in unfavourable conditions, to exceeded supportability level.

2.3.13 Auto Transformer bank 1100MVA, 750/410/18kV – Hungary [2.15]

Overvoltages of considerable magnitude had been generated in the tapped winding of the regulating transformers
connected into the neutrals of the single-phase auto transformers. These overvoltages occurred when the open
ended 750kV line was connected to the transformer already energized from the 400kV side. The single-phase units
consist of a 750/√3/410/√3/18kV ratio auto-transformer and of a regulating transformer connected into the neutral of
the former unit. The regulating transformer is energized by the 18kV winding of the main transformer. According to
measurements performed, overvoltages inside the tapped winding in some cases may be much higher than the
calculated values. The measurements also showed that the high-frequency oscillations due to the rather long
earthing cable may increase the internal voltages of the tapped winding. Thus the frequent flashover of the
protecting spark gaps of the on-load tap-changer can presumably be explained partly by the resonance-like
phenomenon and partly by the voltage increasing effect of the oscillations caused by the long earthing cable.

Page 15
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

2.3.14 Autotransformer, 500/275kV – Japan [2.16]

A 500/275kV autotransformer failed just after closing of a 500kV cable line by GCB (Generator Circuit Breaker) at
an adjacent substation connected by two-circuit cable line. The probable cause was a step wave transmitted via
cable because of closing GCB at a remote substation. Oscillatory rectangular wave produced by reflection in cable
was applied to primary side of transformer with little attenuation in the surge propagation because of the cable
system. Frequency of the oscillatory surge applied to the transformer coincided with LC resonating frequency of
transformer secondary side causing internal overvoltage.

2.3.15 Generator Step-Up Transformer 500MVA, 765/25kV – Auxiliary Transformer 80MVA,

765/138/34.5kV – United States [2.17]
Seven single phase transformer failures at 765kV installation were experienced in the early years of 1980’s. One
common scenario suggested that some transient event or events caused incipient dielectric damage which later
developed into a complete failure under normal system voltage. It meant that the failures were associated to
gradual weakening of their insulation mostly due to switching surges (in kHz range).

2.3.16 HVDC Convertor Transformer 234 MVA 400/93 kV Yy-Yd – India [2.23]
Failures occurred in a bi-pole back to back HVDC station utilizing 1 convertor per pole and requiring a total of 12
large single phase transformer units for the whole station operation. A single phase transformer unit is described as
wound on 2 legs, with the HV sections from the 2 legs connected in parallel to constitute the HV (AC) winding.
Externally, the HV line windings from 3 single phase units are connected YN; the valve windings from one leg are
connected y (star), and the valve windings from the other leg are connected d (delta). The transformers had been
in service for several years when, a series of dielectric failures occurred between HV tapping leads, only
associated with the tapping leads of the star legs. Dielectric design had a significant margin and there was a long
service experience. Based on the failure characteristics, it was suspected that internal resonance excited by an
external transient could have caused the flashover at tapping leads.

The lumped parameters model, validated by the Recurrent Surge Oscillation (RSO) and by the Frequency
Response Analysis (FRA) tests, was used to demonstrate possibility of having internal resonance between the
tapping leads, resulting in excessive overvoltage amplitudes. The effect of possible transient voltage incoming from
the AC neutral was identified as potentially the most detrimental. Monitoring system installed on site, recorded
transients of the predominant frequencies corresponding to the potential resonance frequency pointed by the
calculations, thus supporting the failure hypothesis. Mitigation measures were then taken accordingly; in particular,
the external earthing circuit was modified in order to detune from the natural frequency of the star leg tapping
leads.

2.4 Work of Previous Groups

In the last thirty years the subject of fast front transients and transformers under transient conditions has been dealt
with by several international working groups. In the following section a short summary of their work is briefly shown.

In 1979, Cigre working group 12.07 [2.18] stated that the resonance phenomenon is not a matter of a passive
structure (transformer) alone, but also requires an active structure providing various sources of oscillating voltages.
The group stated that transformer resonance is very difficult to occur and that following simultaneous conditions
need to be fulfilled: transformer’s damping is low; its winding’s natural frequency and excitation voltage frequency
coincide; amplitude of excitation voltage is sufficiently large and of appropriate duration. The group also proposed
measures for avoiding resonances. For transformers these are design modifications or use of additional equipment,
for instance de-tuning capacitances or non-linear resistors to avoid natural frequencies in range of 5 kHz to 100
kHz as much as it is possible.

In 1988, Cigre JWG 33/13.9 [2.19] investigated one of the possible sources of excitation of transformers windings,
namely the very fast transients associated with gas insulated substations. The group indicated that extremely high-
peak resonances might be developed in transformers windings due to oscillations within GIS. For directly
connected transformers, frequencies up to several MHz can be transmitted while for indirectly connected
transformers about 1 MHz; however, no recommendations regarding resonances in transformers were given.

Page 16
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

The interaction between a transformer and the network was assessed and reviewed by Cigre JWG A2-A3-B3.21 in
2005 [2.20]. The group indicated that standards for testing transformers defined waveforms but lacked description
of extremely steep wave fronts (very fast transients) and oscillating voltage waves matching the natural frequency
of transformer’s winding. It was also noted that the assessment of a severity of the interplay of transformer and its
environment is impossible without detailed study of each particular case and it cannot be done in a general way.
However, the group proposed a risk management methodology which then could be used as one of the factors in
decision making.

A number of transformer dielectric failures in Brazilian electric power system was the main incentive to form a
Brazilian-Cigré JWG A2/C4.03 [2.21]. The group’s primary task was to increase the understanding of oscillatory
phenomenon which results from the interaction between the transformer and its electrical environment after a
switching event. For this purpose a number of switching transients’ simulations at different voltage levels and
substation arrangements in the Brazilian network has been performed. A Frequency Domain Severity Factor has
been introduced to ensure that the electric stresses arising from a system event are covered by the transformer’s
dielectric tests. The group recommended considering maximum values together with frequency spectrum of
transient voltages in the insulation coordination studies.

The recent IEEE guide [2.22] emphasizes the importance of detailed modelling of components and enumerates
systems and components of concern: rather simple systems as transformer connected by one line or cable;
transformers that are unloaded, lightly- or inductively loaded, grounded or ungrounded; and often non-linear loads.
The guide concludes that as the occurrence of an event that will produce excessive voltages cannot be predicted
with certainty, frequently switching transformers that are lightly loaded or unloaded should be considered more
likely to produce conditions that will overstress the transformer. For transformers of modest size the guide
suggests the application of a capacitive resistive snubber will mitigate the potential problem of resonant
overvoltage.

2.5 References
nd
[2.1] Alstom Grid, ‘’Power Transformers: Vol 1 Fundamentals”, 2 Edition ©Alstom Grid, 2012.

nd
[2.2] Alstom Grid, “Power Transformers: Vol 2 Expertise”, 2 Edition © Alstom Grid, 2012.

[2.3] L.B Wagenaar, JM Schneider, JH Provanzana, WN Kennedy, DA Yannuci, “Rationale and

Implementation of new 765 kV generator step-up transformer specification”, Cigre paper 12-202-1990.

[2.4] L.B Wagenaar, JM Schneider, JA Fleeman, “EHV transformer dielectric specification improvements”,
IEEE Transactions on Power Delivery, Volume:9, Issue: 1 , PP : 265 – 284, Jan 1994.

[2.5] Jeff Fleeman & Loren Wagenaar, “Transformer Specifications That Work-Exceeding Standards”,
IEEE/PES Transformers Committee, Fall 2002, Oklahoma City, Meeting Tutorials/Prensentations,
October 21, 2002.

[2.6] Morched, A.S., Martí, L., Brierley, R.H., Lackey, J.G., “Analysis of Internal Winding Stresses in EHV
Generator Step-Up Transformer Failures”. IEEE Transactions on Power Delivery, Vol. 11, No. 2, April
1996.

[2.7] Margolis, H.B., Phelps, J.D.M., Carlomagno, McElroy, A.J., “Experience with part-winding resonance in
EHV auto-transformers: Diagnosis and corrective measures”, IEEE Transactions on Power Apparaturs
and Systems, vol. PAS-94, no. 4, July/August 1975.

[2.8] Cancino, A., Ocón, A., Enríquez, G., Malewski, R., “In Service Failure of 230kV Transformers due to
Steep-Front Lightning Over Voltages at Mexican West Coast”, Cigré 2006, Paper A2-201.

[2.9] Lapworth. J.A., Jarman, P.N., Breckenridge, T., “Transformer Internal Over-Voltages caused by Remote
Energization”, Cigré 2006, Paper A2-305.

Page 17
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

[2.10] Cigré-Brazil JWG-A2/C4-03, “Electrical Transient Interaction between Transformers and Power Systems -
Brazilian Experience”, Paper 257, IPST - International Conference on Power Systems Transients,
Kyoto,Japan. June 2009.

[2.11] Hori, M., Mitani, M., Sakurano, H., Motoyama, H., “Observation and Analysis of Incident Surge Voltage
Waveforms in Substations by Winter Lightning”, IEEE Transactions on Power Delivery, Vol. 22, NO. 1,
January, 2007.

[2.12] Shipp, D.D., Dionise, T.J., Lorch, V., MacFarlane, B.G., “Transformer Failure Due to Circuit-Breaker-
Induced Switching Transients”, IEEE Transactions on Industry Applications, Vol. 47, NO. 2, March/April,
2011.

[2.13] Pretorius, R.E., Goosen, P.V., “Practical Investigation into Repeated Failures of 400/220kV Auto
Transformers in the ESCOM Network – Results and Solutions”, International Conference on Large High
Voltage Electric Systems, 12-10. August/September, 1984.

[2.14] Alexandre Neves, Bianca Leyen, Helvio J. A. Martins, Jose A. Duque, Ricardo Cerbino, Roberto C.
Menezes, Walter Cerqueira, Miguel Carlos Medina Pena, Roberto de Barros de Lima, Jose Carlos
Mendes, “Investigation About Failures in 500kV/185MVA Generator Transformers at Luiz Gonzaga
Hydroelectric Plant”, 12th International Symposium on High Voltage Engineering, August 2001.
Bangalore, India.

[2.15] Hámos, I., Kerényi, D., “Experience with the 750kV Transformer of the Hungarian Electric System,
Acquired during on site Testing”, Cigré 1986, Paper No 12-02.

[2.16] Ebisawa Y., Kobayashi T., “Experience of transformer resonance caused by remote closing of cable line”,
CigréBienal 2006 Paris, Group Meeting SC A2 PS3 Q3.4.

[2.17] Schneider J.M., Fromholtz E.N., Nichols D.K., Ware B.J. “The Rockport Transient Voltage Monitoring
System”, Cigré 1988, Paper No 23-04

[2.18] Study Committee 12 (Transformers) – WG 12-07, “Resonance Behavior of High-Voltage Transformers”,

in Proc. 1984 CIGRÉ Large High Voltage Electric Systems Int. Conf.

[2.19] CIGRE Working Group 33/13-09, “Very fast transient phenomena associated with gas insulated
substations“, Cigré 1988, Paper No. 33-13

[2.20] CIGRE Joint Working Group A2-A3-B3.21, “Electrical environment of transformers – Impact of fast
transients“, ELECTRA No 218, Feb 2005.

[2.21] CIGRE-Brasil Joint Working Group A2/C4-03, “Interaction Between Transformers and the Electrical
System with Focus on High Frequency Electromagnetic Transients“, Brochure 012 (Original version in
Portuguese), May 2011.

[2.22] IEEE PC57.142/B8.2, “IEEE Guide to Describe the Occurrence and Mitigation of Switching Transients
Induced by Transformer, Switching Device, and System Interaction“, April 2011.

[2.23] T. Ngnegueu, M. Saravolac, S. Lelaidier, N. Macleod, R. Mukhedkar, “Design of HVDC Converter

Transformers with due consideration to Transient Phenomena”,Cigre SCA2&D1 Joint Colloquium
Transformer materials and Emerging techniques, Kyoto, Japan, September 11-16 2011.

Page 18
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

3 ELECTRICAL NETWORK TRANSIENT MODELLING

3.1 Introduction
When modelling an electrical network it is important to consider the phenomena involved. Different models may be
required for phenomena such as low frequency overvoltages, switching overvoltages, slow front overvoltages or
fast front overvoltages.

Those aspects have already been considered in the past, for insulation coordination purpose [3.1], [3.2], [3.3], [3.4]
where the goal was the integration of the transformers into the networks with acceptable overvoltages at their
terminals. More recently it has been proposed that additional aspects should be considered, which will demand
more complex network modelling to be associated with the transformers involved. An example is asset
management issues in aged networks. In this case, the aim is to reduce the equipment stress in order to maximize
the transformer’s operating lifetime within the network. Another example is the connection of renewables [3.5],
[3.6], [3.7], [3.8]. This will take into account new types of networks and technologies such as power electronics and
UHV transmission.

When a transformer model is to be included in an electrical system study involving transient overvoltages, EMTP-
type modelling and simulations [3.9] are usually the most practical approach (EMTP-RV, PSCAD, ATP, Digsilent,
etc). With this approach, each relevant power system component (overhead lines, cables, surge arrester etc) is
modelled in sufficient detail in order to consider the phenomena involved.

EMTP-type programs have prescribed support routines, which assist the program user to create models from
available data [3.10]. The following sections provide advice regarding the modelling of the power system for the
purpose of simulating transformer overvoltages.

For certain applications, the link between those tools and electromagnetic 3D programs may be applied.

3.2 Substations
Substations may be modelled in different ways depending on the frequency range of the phenomena involved. For
low frequencies, which occur during transformer energizations, substations may be considered as sub-networks
that take into account the circuit-breaker, protective devices like surge arresters, and associated cables or lines
connected to the observed transformer. Propagation aspects and the equipment capacitances may in fact be
neglected at those frequencies.

For higher frequencies, more detailed modelling has to be taken into account.

Basically, substation models are derived from their layout drawings showing the exact position of substation
equipment.

Busbars and conductors inside the substation are represented by line sections. These line sections are modelled
by un-transposed distributed parameters with modal surge impedances calculated from the structural geometry,
and with the propagation velocity set to be equal to the speed of light. Note that the minimum section length with
this representation dictates the simulation time step and it depends on the maximum frequency of the phenomena.
For smaller sections, a lumped parameter inductance is used.

The substation equipment, such as circuit breakers, disconnector switches, and CVTs, are represented by their
stray capacitances to ground. Such data is based on supplier information or extracted from literature [3.11], [3.12],
[3.13]. With regards to circuit breakers and disconnectors, appropriate capacitances should be added to the model,
taking into account their open/closed status. The capacitance to ground of all insulators should also be represented
since the substation capacitance is one of the parameters that can modify surge wave shapes.

Modelling of Gas-Insulated Substations (GIS) components also makes use of electrical equivalent circuits. These
circuits are comprised of lumped elements and distributed parameter lines. GIS sections are considered as
concentric cylinder distributed parameter models. On the assumption that the external enclosure is perfectly

Page 19
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

grounded, only the internal mode (conductor-enclosure) is taken into account. If ground potential rises have to be
considered, it is necessary to add one more mode (enclosure-ground). The capacitive effect of spacers, flanges,
elbows, corona shields and other connections, should be included by assuming the propagation velocity is 0.9-0.95
the velocity of light. Other equipment is represented by their stray capacitance.

3.3 Upstream network

3.3.1 Low frequencies studies
For low frequency studies, a point to remember is that the upstream-network is generally represented by its
Thevenin equivalent [3.13], [3.14] located at the substation where the line or cable is connected to the observed
transformer. The source impedance, which is typically inductive, may create a resonant circuit with the capacitance
of the cable or line that is connected to the transformer. When this resonant frequency is low, it may lead to high
overvoltages at transformer terminals.

For certain applications where the non-linear characteristics of equipment dominate, the reduction of the network to
a linear Thevenin equivalent may not be sufficient. In such cases, a more detailed network modelling approach will
be required in order to take into account the non-linear aspects and their impact on the total network.

In fact, when higher frequencies are involved, propagation aspects have to be considered along the lines. Pi
circuits, or pi in cascade, are not sufficient to represent the line. Phenomena at these higher frequencies will be
considered in the following section.

3.3.2 Switching studies

For higher frequency studies, for example those involving switching surges, the distance between the fault location
in the network and the transformer is a major issue. This distance defines the frequency of the travelling wave that
propagates towards the transformer. For example, the occurrence of a fault or its elimination can lead to
phenomena in the frequency range of 0-20kHz, whereas a lightning stroke has a frequency range of 250 kHz-
1 MHz.

Different cases have to be taken into account, especially the short-circuit impedance of the up-stream network. In a
case where the short-circuit reactance of the upstream network is negligible with respect to the line reactance (high
short circuit power), a travelling wave may propagate when the fault occurs. In such a case, the internal
overvoltage will reach its maximum when the frequency of the travelling wave is equal to the resonant frequency of
the upstream network. As a consequence this frequency is equal to:

fo = v/4X, v being the velocity and X the distance between the fault location and the transformer.

A typical travelling wave may be described by figure 3.1 below:

Figure 3.1: Travelling wave/overvoltage at the terminal of the transformer.

Page 20
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

In a case where the short-circuit impedance is not infinite, the critical distance from the transformer is dependent
upon the capacitance of the line, the short-circuit reactance and the resonance frequency of the transformer. Note
that each internal resonant frequency mode of the transformer needs to be considered.

Different surge arresters, one located in the main substation and possibly one at the neutral of the transformer,
have to be considered in the studies in order to calculate the stresses correctly as they may reduce the overall
impact.

When higher frequency oscillations occur, such as lightning strokes, more detailed network modelling of the
substation is required.

3.3.3 Lightning studies

3.3.3.1 Phenomena involved
Overvoltages on overhead lines can be caused by lightning flashes that strike the line. When the lightning leader
connects with a grounded object, a return stroke current starts propagating up along the lightning channel, which
effectively gives a current injection at the stroke location.

The associated current injection has typically a peak value of tens of kA. The current is characterized by an upward
concave-shaped front with duration of a few microseconds followed by a tail with duration of tens to hundreds of
microseconds. The flash may consist of several subsequent strokes in addition to the first return stroke, all taking
place in the same lightning channel. Extensive lightning statistics are available which characterize the shape and
the peak value of the lightning stroke current for the first and subsequent strokes as well as for negative and
positive polarity flashes. An extensive summary is found in CIGRE Brochure 63. [3.15]

The development of the lightning discharge is in the form of a stepped leader which propagates from the cloud
down towards the earth. The theory of fractals has also provided new approaches for the determination of the
stroke current, including non-linear effects [3.16]; this may give better results when comparing results obtained from
simulations and on site measurements.

The actual point of terminating to ground objects is related to the length of the last step ("striking distance"), which
is dependent on the prospective peak value of the return stroke current. This relation has led to the development of
the so-called electro geometrical model (EGM). The EGM takes as input the lightning characteristics (peak value
and stroke position relative to the overhead line) and the overhead line geometry (position of phase wires, ground
wires, and tower) and determines the termination point (ground, phase wire, ground wire, or tower top) based on
the computed lightning striking distance. Details about the EGM are found in CIGRE Brochure 63.

3.3.3.2 Modelling aspects

Once the termination point has been determined, the stroke current can be modelled by an ideal current source,
which connects the remote ground.

The overhead line can be modelled in detail by segmenting into individual spans. Each span is modelled by a
transmission line with mutual coupling between conductors; towers are modelled as transmission lines; tower
footing groundings are modelled in as much detail as possible. Flashovers on the line are most likely to take place
across insulator strings. Accordingly, each insulator string is modelled as a controlled switch that is governed by a
voltage-time curve, or more accurately by an integration method based on an area calculation, or the Leader
Development Method. Again, CIGRE Brochure 63 is an excellent source of information, see also [3.17].

In practice, only the spans near the stroke location needs to be modelled in detail. The flashover of insulator strings
is dependent on the instantaneous 50/60Hz operating voltage on the phase conductor and this voltage must
therefore be included in the simulation. Moreover, the effectiveness of substation surge arresters against lightning
overvoltages is dependent on the instantaneous 50/60 Hz voltage. It is therefore convenient to use fast transient
simulation software that can automatically initialize from the 50/60 Hz initial conditions.

Given the statistical variations in the lightning flash parameters and stroke position, it is often recommended to
perform a Monte Carlo simulation where the lightning current parameters are derived from their statistical
distribution and the point of termination is computed by the EGM. When combined with an EMTP-type simulation

Page 21
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

2
tool and knowledge about the average lightning ground flash density (flashes per year and km ), one obtains a
statistical distribution of the overvoltage peak value at nodes in the network (e.g. transformer terminals), i.e. the
probability of exceeding a given peak value per year.

The distribution of the peak overvoltage can then be applied in insulation co-ordination studies where the
overvoltages are assumed to be representative with the respect to the test voltages. However, since failures in
transformers are more related to the oscillating nature of the overvoltages rather than their peak value, one may
conclude that the typical Monte Carlo simulation approach, which only focuses on peak voltages, is insufficient. It
should be mentioned that the use of the surge arresters, generally located at the substation entry, may also have
an impact on the shape of the overvoltage propagating towards the transformer.

Other practices include the calculation of the overvoltage based on the assumption of the worst case conditions,
e.g. the high lightning current peak value in combination with a short front time.

Furthermore, lightning overvoltage will result when the lightning strikes in the vicinity of the line. Such overvoltage
conditions are however, much smaller in magnitude than those of a direct stroke to the overhead line, and are
called indirect strokes. It is unlikely that indirect strokes will cause failures in HV power transformers.

Phenomena which involve the propagation of a wave towards a transformer with frequencies above a few hundred
of kHz, require a high frequency transformer model. Such a model will need to take into account capacitive
elements (see Chapter 4 on the HF modelling of transformers).

A better approach would be to perform a Monte Carlo simulation based on the FD severity factor which will be
introduced in Chapter 6.

3.4 Overhead lines and underground cables

3.4.1 Parameter determination
Overhead lines and underground cables should be modelled by taking into account their distributed-parameter
nature and frequency-dependent effects when high frequencies are involved. In EMTP-type programs, linear
phenomena like the skin effect are taken into account. The effect of corona, which might occur at very high
voltages, is not taken into account because of its complicated representation within these programs. Fortunately,
this effect is not of fundamental importance for the applications and the phenomena considered in this brochure.

In EMTP-type programs, such models can be created starting from a geometrical description of the overhead line.
Such a description will include the position (x,y) of each phase conductor and ground wire, their diameter, DC
resistance and the ground resistivity. A similar approach is used for underground cables requiring the specification
of the relative position of the cable in the (x,y) system, its diameter and DC resistance of each phase conductor, its
thickness and resistance of each metallic screen, burial depth in ground and ground resistivity.

From this information, the per-unit-length parameters of series impedance Z(ω)=R(ω)+jωL(ω) and shunt admittance
Y(ω)=G(ω)+jωC(ω) are calculated as a function of frequency assuming plane wave propagation. These quantities
are matrices of dimension n, where n is the number of phase conductors. The considered frequency-dependent
effects are those of skin effect in conductors and earth [3.18]-[3.19], with the assumption of cylindrical-symmetrical
current distribution in conductors. Transient analysis and wave propagation in underground cables has been
studied in [3.20]. During fast transients, the effect of the ground to the wave propagation may play an important role
as well [3.21], [3.22].

3.4.2 Travelling wave-type models

The most suitable type of transmission line model is the travelling wave type where the voltages and currents are
decomposed into forward and backward waves. This class of model achieves high computational efficiency as well
as high accuracy. The line is characterized by its characteristic admittance Yc (or characteristic impedance Zc) and
propagation function, H. These quantities are calculated from Z, Y and the line length. A number of transmission
line models are available which properly account for the frequency dependency in Yc and H [3.23], [3.24], and
[3.25]. They are based on fitting the elements of Yc and H with rational functions and time delays. In addition, a

Page 22
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

sub-category of line models [3.26] exist which are based on a modal decomposition assuming a constant, real
eigenvector matrix. The latter category has some accuracy limitations, in particular for underground cables.

Yc (ω) = Z−1 ZY (3.1)

H(ω) = e− YZl
(3.2)

3.4.3 Lumped-parameter type models

In the case of very short lines or cables, the travelling wave formulation suffers from the requirement that the
simulation time step must be shorter than the line travel time (delay). This dictates a very small simulation time step
and thus slow simulations. In such situations, it is often better to represent the line/cable with a lumped pi section.
Most EMTP-type programs have options for such model type generation.

In the case of transformer energization, pi-sections may be associated in cascade, in order to represent the
impedance of the upstream network correctly for up to a few hundred hertz. This representation leads to a proper
determination of the overvoltages as well as inrush currents, however, in order to represent the switching
surges/travelling waves when the circuit-breaker is located quite far from the transformer, a propagation model may
be required. This last model may not be adequate to represent the homopolar mode correctly; the pi sections are
more effective for steady-state frequencies up to a few hundred hertz.

Other line models

In addition to the frequency-dependent travelling wave models, EMTP-type programs can also represent lines and
cables with constant-parameter travelling wave models with losses lumped in a few places (e.g. the line ends and
in the middle), however, the frequency-dependent effects are excluded.

3.5 Surge arresters

The effect of multiple reignitions and voltage escalation can be limited and eliminated by protective devices:

- a R-C suppression branch set as close as possible to the terminals of an inductive load.
- a series connected capacitor and a resistor (the resistor connected in parallel to a ZnO varistor) and
surge arresters [3.27].

The simplest way to model the surge arrester is to use a piece-wise linear resistor as considered by all EMTP-
based software tools. However, representing a surge arrester only by a non-linear inductance is not enough when
fast and very fast transients should be studied because of the frequency dependent behavior of the arrester.
Several models have been introduced for the modelling of the surge arresters. A general model was proposed by
Durbak [3.28] who represented the arrester by a number of non-linear resistors connected by parallel R-L filters.

The IEEE surge arrester committee [3.29] approved a simplified form of this model consisting of only two non-linear
resistors describing a U-I hysteresis curve. It is not easy to simulate this characteristic for a specific arrester
because the parameters of the R-L filters depend strongly on the geometrical data of the arrester. Pinceti and
Giannettoni [3.30] have adapted this model.

Page 23
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Figure 3.2: Measured U-I curve of an arrester.

Figure 3.3: Simplified surge arrester model by Schmidt [3.31].

A surge arrester model can also be developed by making use of only a measured U-I arrester curve as shown in
Fig. 3.2. With this curve one can use the simplified version of the model described by Schmidt et al. [3.31] (Fig.
3.3). The capacitance is based on the capacitance of the arrester block and is non-linear because it depends on
the rate of rise of the transient voltage wave. In reality, this capacitance is variable and it is in the order of 1.5 nF/kV
for distribution transformers and 5 nF/kV for station transformers. The parameters Ra and La represent the physical
behavior of the ZnO grain, but when the arrester is connected to the transformer, these parameters also depend on
the loop inductance made by the connection between the arrester and the transformer. The inductance is
estimated from the distance of the arrester to the transformer terminals, and it is in the order of 5-20 µH/m. This is a
practical value of the inductance for distribution transformers.

Page 24
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Figure 3.4: Response of the surge arrester when excited with current impulse 10 kA, 0.5 µs.
a) Calculated residual voltage.
b) Current impulse 10 kA, 0.5 µs.

Figure 3.5: Response of the surge arrester when excited with current impulse 10 kA, 8/20 µs.
a) Calculated residual voltage.
b) Current impulse 10 kA, 8/20 µs.

Figures 3.4 to 3.6 show the results of a 12 kV surge arrester model. According to ANSI/IEEE Std C62.11-1993, the
arrester model should give sufficient information for the surge impulse (45 – 60 µs time to crest), lightning impulse
8/20 µs and front-of-wave 0.5 µs.

For the arrester, instead of a surge impulse of 45-60µs, a current impulse of 500 A 30/60µs is applied. The
computation of the arrester residual voltages is performed by making use of a modified current amplitude and
shape. The parameters used for the calculation of the arrester are: Ra=0.6Ω , La=0.5 µH and Ca=0.1 nF. The
current sources can be produced by means of the Heidler function [3.32] as given in Fig. 3.4 to Fig. 3.6.

Page 25
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Rated voltage (in kV) 12

MCOV (in kV rms) 10
V10 kA – current impulse 8/20 µs (in kV) 44
V10 kA – current impulse 0.5 µs, front-of-wave (in kV) 51.6
V500 A - current impulse 30/60 µs (in kV) 32.3
Energy absorption (in kJ/kV) 4.3
Tab. 3.1: Arrester data.

In Tab. 3.1 one can see that the measured residual voltages V10 kA for 8/20µs and front-of-wave 0.5 µs current
impulse are 44 kV and 51.6 kV respectively. The calculations result in 43.2 kV and 51.8 kV, which is in good
agreement with the measured data. The measured value of the current impulse, V500 A for 30/60 µs is 32.3 kV, and
calculated 31.7 kV.

It should be pointed out that for current impulses with higher amplitude and very short front-of-wave impulses,
computed residual voltage is slightly higher than those given in the catalogue data. This implies that the U-I
hysteresis must be taken into account if this model is used for very fast transients such as lightning impulses.

Figure 3.6: Response of the surge arrester when excited with current impulse 500A, 30/60 µs.
a) Calculated residual voltage.
b) Current impulse 10 kA, 30/60 µs.

From the perspective of overvoltage reduction, surge arresters are an important component in substations. They
are one of the key components in insulation coordination studies [3.33], [3.34] and may decrease the impact of high
frequency waves, such as lightning surges.

They can dissipate energy when overvoltages have amplitudes of 2 p.u., when switching phenomena are involved,
and reduce the amplitude of the overvoltage when fast front phenomena like lightning surges are involved. They
should be regularly included in the network when the overvoltage amplitude is high.

Page 26
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

3.6 Circuit breakers and disconnectors

3.6.1 Circuit Breakers
This section deals with the efficient modelling of circuit breaker (CB) reignitions. The model requires chopping
current, the withstand voltage characteristic of the CB and the arc quenching capabilities of the CB.

The modelling procedure that will be presented in this section applies to a vacuum circuit breaker (VCB) which is
represented by a simplified circuit.

When dealing with overvoltage estimation and small inductive current switching, the model of the VCB may have to
include HF reignition components (depending on the properties of the VCB and the surrounding network). The VCB
is modelled by means of the:
• cold withstand voltage characteristic of the VCB,
• HF quenching capability,
• chopping current.

The cold withstand voltage characteristic of the VCB is a function of the contact distance. One of the parameters
that is of influence is the speed of contact separation, with many researchers having investigated the withstand
capability experimentally [3.35, 3.36]. It is known that the data varies with a statistical distribution. Smeets [3.37]
represented the withstand voltage characteristic with an exponential expression, while Glinkowski et al. [3.38]
showed that the reignition can take place at short gaps (<1 mm), so it is sufficient to use a straight line.
The HF quenching capability is defined by the slope of the HF reignition current (when the HF current is zero).
Authors of earlier research assumed the slope to be constant [3.39, 3.40], but later it has become clear that the
slope also depends on the reignition voltage [3.41] and shows time dependent behavior.

The chopping current depends primarily on the contact material, however the surge impedance of the load side
does influence the chopping current level. We are able to find different expressions for the calculation of the mean
chopping level [3.42, 3.43]. In this analysis however, the chopping current is constant and equal to 3 A. The
characteristics describing whether or not reignition occurs are [3.44]:

( )
U b = AA t − t open + BB (3.3)

( )
di / dt = CC t − t open + DD (3.4)

where topen is the instant of contact opening. The quantities Ub and di/dt represent the dielectric and arc quenching
capability of the VCB respectively. An example will be shown where the constant values in (3.3) and (3.4) are:
7 10 2 6
AA=1.7x10 V/s, BB=3400 V, CC=-3.40x10 A/s , DD=255x10 A/s. These characteristics are randomly varied with
10% standard deviation for a normal statistical distribution. In the following example reignitions are modelled by
means of the MODELS controlled routine in ATP [3.9, 5.10]. After the contacts have opened, the current continues
to flow until its absolute value is equal to the chopping current. Thereafter a switch opens, and its further operation
is controlled by the expression (3.3) and (3.4). The VCB modelling method is explained in [3.38].

This approach is demonstrated by using a simple test circuit as shown in Fig. 3.7. The data for the circuit
parameters are: Ln=5 mH, Cn=100 nF, Rσ=2 Ω, Lσ=40 µΗ, Rs=100 Ω, Ls=50 nH, Cs=100 pF, RL=1E5 Ω, LL=120 mH,
CL=10 nF.

Figure 3.7: Circuit for reignition modelling.

Page 27
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

The simulated results are depicted in Fig. 3.8 and Fig. 3.9. Fig. 3.8 shows the current and voltage of the VCB as
well as the load side voltage, whereas the enlarged trace of the first reignition is shown in Fig. 3.9. After the current
has been chopped the TRV occurs, resulting in two different oscillations. The oscillations depend on the type of the
circuit. One frequency is in the range of a few MHz (typically 2.5 MHz). It can approximately be calculated with:

(
f 1 ≈ 2π Lσ Cs ) −1
(3.5)

The other frequency is lower and is determined by the load parameters. It is the natural frequency of the load with
values in the order of kHz (typically around 4.6 kHz) and can be estimated with:

(
f 2 ≈ 2π LL CL ) −1
(3.6)

When the TRV has surpassed the withstand voltage level of the vacuum gap, the VCB reignites and a HF current
flows through the network. The reignited current has two HF components superimposed on the power frequency
current. One has a value of approximately 250 kHz and is calculated by:

( )
−1
f 3 ≈ 2π Lσ CL (3.7)

Figure 3.8: Simulated reignited current and voltages across the VCB and the load.

Figure 3.9: Simulated VCB current, voltage and load voltage of the first reignition.

Page 28
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

The frequency of the other oscillation is very high and is superposed on the current oscillations with frequency f3.
This HF oscillation finds its origin in the parasitic capacitance and inductance of the gap and can be calculated
with:

( )
−1
f 4 ≈ 2π LsCs (3.8)

The frequency is in the range of a few tens of MHz (typically around 70 MHz). This HF current normally causes
voltage spikes as per those shown in Fig. 3.9. Measurements however, show more UHF oscillations being present
[3.45, 3.46]. These are superimposed on the reignited HF current and a very small calculation time step must be
used if they are to be taken into account.

3.6.2 Disconnectors
The task of the disconnectors in high voltage power systems is to provide electrical and visible isolation of one part
of the system. The isolation generally takes two forms:

o Isolation of the normal day-to-day operation of the power system. For instance, shunt reactors required only
during light load periods are switched out using circuit breakers and then isolated by the disconnectors during
peak load periods.
o Isolation related to repair or maintenance on transmission lines or station equipment such as transformers or
circuit breakers.
To achieve isolation disconnectors are operated under energized conditions and will thereby interrupt current in air,
the type of which being dependent on the circumstances.

The types of current interruption with disconnectors are loop switching for which the range of application is from up
to 1600 A between busbars within a substation, down to hundreds of amperes between transmission lines and
distribution feeders. Furthermore, interruption of magnetizing and capacitive currents can also be accomplished by
a disconnector [3.47].

For each of these switching types, current chopping may occur under conditions similar to those of circuit breakers
[3.48] and reignitions may result. The modelling procedure and analysis of multiple reignitions is the same as that
used for circuit breakers, however when disconnectors are located outside, weather conditions can have significant
influence on the withstand capability of the disconnector and will need to be taken into account.

3.7 References
[3.1] G. Sybille, M.M. Gavrilovic, J. Bélanger, "Transformer Saturation Effects on EHV System Overvoltages", IEEE
Transactions on Power Apparatus and Systems, Vol. PAS-104, No. 3, March 1985.
[3.2] CIGRE WG 33.10, "Temporary over-voltages withstand characteristics of extra high voltage equipment",
Electra, No. 179, August 1998.
[3.3] M. Rioual, P. Guuinic, D. Laval, M. Adelghani, N. Schaefer, M. Schäfer, “Modeling of the energization of a
power transformer in a 110 kV network and its validation by on site tests. Description of ferroresonance
phenomena involved and means to avoid them”, IEEE General Meeting (Cat. n° 086M1335), Pittsburgh, 20-
24 July 2008.
[3.4] N. Nenemenlis, M. Ené, J. Bélanger, G. Sybille, L. Snider, "Stresses in metal-oxide surge-arresters due to
temporary harmonic overvoltages", Electra, No. 130.
[3.5] Entsoe, “Requirements for Grid Connection Applicable to all Generators”, Working Draft, October 2010
[3.6] EU Smart-Grids Plattform, “Vision for networks 2010-2030”.
[3.7] K. Chen, Member, L. W. Montgomery, G.Klempner, J. Yagielski,, J Amos, M. Brimsek, “Comparing IEEE
50.13 and IEC 60034, Standards for Large Cylindrical Rotor, Synchronous Machines”, 978-1-4244-8357-
0/10/©2010 IEEE.

Page 29
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

[3.8] M. Rioual, J.C. Reveret, “Modeling of cables for the connection of windfarms to the Distribution Network:
Application to the energization of transformers and its validation by on-site tests“, (GM-978-1-4244-6/09)
IEEE General Meeting, Calgary, 24-29 July 2009.
[3.9] Electromagnetic Transients Program reference manual (EMTP Theory Book), prepared by H.W. Dommel for
Bonneville Power Administration, P.O. Box 3621, Portland Oregon 97208 U.S.A., August 1986.
[3.10] J.A. Martinez, “Power system transients. Parameter determination”, CRC Press, 2010, ISBN: 978-1-4200-
6529-9.
[3.11] “Guidelines for Representation of Network Elements When Calculating Transients”, Report of CIGRE
Working Group 33.02, 1990.

[3.12] “Modelling Guidelines for Fast Front Transients”, Report of IEEE Fast Front Transient Task Force, IEEE
Transactions on Power Delivery, Vol. 11, No. 1, January 1996.
[3.13] D. Povh, H. Schmitt, O. Volker, R. Witzmann, “Modelling and Analysis Guidelines for Very Fast Transients“,
IEEE Transactions on Power Delivery, Vol. 11, No. 4, October 1996.
[3.14] M. Rioual, C. Sicre, “Energization of a no-load transformer for power restoration purposes: Modeling and
validation by on site tests.” IEEE Volume 3, 2000, Jan. 23-27, Page(s):2239-2244.
[3.15] M. Rioual, B. Bernin, C. Crepy, “Determination of transient phenomena when energizing a 340 MVA
transformer having a highly non linear characteristics: modelling of the equipments and their validation by on
site tests”, IEEE General Meeting, Minneapolis, 25-30 July 2010.
[3.16] CIGRE Brochure 63, "Guide to procedures for estimating the lightning performance of transmission lines",
Working Group 01 report for Study committee 33, October 1991.

[3.17] IEEE PES Special publication, "Modeling and analysis of system transients using digital programs", 1998.
Prepared by IEEE WG 15.08.09.

[3.18] Jeremy A. Riousset, Victor P. Pasko, Paul R. Krehbiel, Ronald J. Thomas, William Rison, “Three dimensional
fractal modeling of intracloud lightning discharge in a New Mexico thunderstorm and comparison with
lightning mapping observations”, Journal of Geophysical Research, Vol. 112, D15203,
doi:10.1029/2006JD007621, 2007.
[3.19] S.A. Schelkunoff, “The electromagnetic theory of coaxial transmission lines and cylindrical shields”, Bell
System Technical Journal, vol. 13, pp. 532-579, 1934.
[3.20] L.M. Wedepohl and D.J. Wilcox, “Transient analysis of underground power transmission system; system-
model and wave propagation characteristics”, Proceedings of the IEE, 120(2), pp. 252-259, February 1973.
[3.21] J.R. Carson, “Wave propagation in overhead lines with ground return”, Bell System Technical Journal, vol.
5, pp. 539-554, 1925.
[3.22] A. Deri, G. Tevan, A. Semlyen, and A. Castanheira, ”The complex ground return plane. A simplified model
for homogenous and multi-layer earth return”, IEEE Trans. Power Apparatus and Systems, vol. 100, no. 8,
pp. 3686-3693, August 1981.
[3.23] T. Noda, N. Nagaoka, and A. Ametani, “Phase domain modelling of frequency-dependent transmission lines
by means of an ARMA model”, IEEE Trans. Power Delivery, vol. 11, no. 1, pp. 401-411, January 1995.
[3.24] A. Morched, B. Gustavsen, and M. Tartibi, “A universal model for accurate calculation of electromagnetic
transients on overhead lines and underground cables”, IEEE Trans. Power Delivery, vol. 14, no. 3, pp. 1032-
1038, July 1999.
[3.25] A. Ramirez, J.L. Naredo, and P. Moreno, “Full frequency-dependent line model for electromagnetic transient
simulation including lumped and distributed sources”, IEEE Trans. Power Delivery, vol.20, no.1, pp. 292-
299, Jan. 2005.
[3.26] J.R. Marti, “Accurate modelling of frequency-dependent transmission lines in electromagnetic transient
simulations”, IEEE Trans. Power Apparatus and Systems, vol. 101, no. 1, pp. 147-157, January 1982.

Page 30
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

[3.27] P.G. Slade: “Vacuum Interrupters: The New Technology for Switching and Protecting Distribution Circuits”,
IEEE Transactions on IA, Vol. 33, No.6, November/December 1997, pp.1501-1511.

[3.28] D.W. Durbak: “Zinc-Oxide Arrester Model for Fast Surges”, EMTP Newsletter, Vol. 5, No. 1, January 1985.

[3.29] IEEE WG 3.4.11: “Modelling of Metal Oxide Surge Arresters”, IEEE 28Transactions on Power Delivery, Vol.
7 No. 1, January 1992, pp. 302-309.

[3.30] P. Pinceti, M. Giannettoni: “A Simplified Model for Zinc Oxide Surge Arresters”, IEEE Transactions on Power
Delivery, Vol. 14, No. 2, pp. 393- 397, April 1999.

[3.31] W. Schmidt, et al.: “Behaviour of MO-Surge-Arrester Blocks to Fast Transients”, IEEE Transactions on
Power Delivery, Vol. 4, No. 1, pp. 292-300, January 1989.

[3.32] O. Hevia: “Sources of Type Input in the ATP”, EEUG News, November 1998.

[3.33] IEC Standard 71.1 “Insulation coordination” / Definitions.

[3.34] IEC Standard 71.2 “Insulation coordination” / Application Guide.

[3.35] A. Greenwood, M. Glinkowski: “Voltage Escalation in Vacuum Switching Operations”, IEEE Transactions on
Power Delivery, 1988, pp.1698-1705.
[3.36] A. T. Roguski: “Experimental Investigation of the Dielectric Recovery Strength Between the Separating
Contacts of Vacuum Circuit Breaker”, IEEE Transactions on Power Delivery, Vol. 4, No. 2, pp. 1063-1069,
April 1989.
[3.37] R.P.P. Smeets, et al: “Types of Reignition Following High-Frequency Current Zero in Vacuum Interrupters
with Two Types of Contact Material”, IEEE Transactions on Plasma Science, Vol. 21, No. 5, pp. 478-483,
April 1993.
[3.38] M. Glinkowski M., et al.: “Voltage Escalation and Reignition Behaviour of Vacuum Generator Circuit Breakers
During Load Shedding”, IEEE PES Summer Meeting, July 28-August 1 1996, 96 SM 402-8 PWRD.
[3.39] L. Czarnecki, M. Lindmayer: “Measurement and Statistical Simulation of Virtual Current Chopping in Vacuum
Switches, XI-th International Symposium on Discharges and Electrical Insulation in Vacuum”, Berlin, GDR,
September 1984, pp. 1-8.
[3.40] J. Kosmac, P. Zunko “A Statistical Vacuum Circuit Breaker Model for Simulation of Transient overvoltages”,
IEEE Transactions on Power Delivery, Vol. 10, No. 1, January 1995, pp. 294-300.
[3.41] J. Helmer, M. Lindmayer: “Mathematical Modelling of the High Frequency Behaviour of Vacuum Interrupters
and Comparison with Measured Transients Power Systems”, XVII-th Symposium on Discharges and
Electrical Insulation in Vacuum, Berkeley, California, USA, July 21 - 26, 1996, pp. 1-9.
[3.42] R.P.P Smeets: Low Current Behaviour and Current Chopping of Vacuum Arcs, Ph.D. thesis, TU Eindhoven,
1987.
[3.43] L. Dube, I. Bonfanti: “MODELS: A New Simulation Tool in the EMTP”, European Transactions of Electrical
Power, Vol. 2, No. 1, January/February 1992, pp. 45-50.
[3.44] M. Popov, L. van der Sluis, G. C. Paap: “Investigation of the Circuit Breaker ReignitionOvervoltages Caused
by No-load Transformer Switching Surges”, European Transactions of Electrical Power, Vol. 11, No. 6,
November / December, 2001, pp. 413-421.
[3.45] R.P.P. Smeets, et al: “Essential Parameters of Vacuum Interrupters and Circuit Related to Occurrence of
Virtual Current Chopping in Motor Circuits”, International Symposium on Power and Energy, Sapporo,
Japan, 1993.
[3.46] R.P.P. Smeets, et al: “Types of Reignition Following High-Frequency Current Zero in Vacuum Interrupters
with Two Types of Contact Material”, IEEE Transactions on Plasma Science, Vol. 21, No. 5, pp. 478-483,
April 1993.

Page 31
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

[3.47] D. F. Peelo: “Current Interruption Using High Voltage Air-break Disconnectors”, Ph.D. thesis, TU Eindhoven,
The Netherlands, ISBN 90-386-1533-7.
[3.48] V.S. Rashkes, L.D. Ziles: “Very High Frequency Overvoltages at Open Air EHV Substations During
Disconnect Switch Operation”, IEEE Transactions on Power Delivery, Vol. 11, No. 3, July 1996, pp. 1618-
1623.

Page 32
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

4 TRANSFORMER MODELLING

Mathematical modelling of a dynamic system, such as the transient voltage response of a power transformer, can
generally be divided into two basic approaches in terms of procedures for selecting the model structure and
calculating the model parameters [4.1]-[4.3]: White-box (or physical) modelling and Black-box modelling.

The White-box modelling approach is associated with methods that use a complete knowledge of the internal
behavior of the dynamic system being modelled for selecting the model structure and computing the model
parameters. Such models are derived from first principles, that is, physical and electrical laws. White-box models
do not depend on measurement data and their parameters have a direct relationship to the transformers physical
structure.

An alternative approach is Black-box modelling. This approach is associated with methods where the model
structure and parameters are computed based only on external input and output data. The model structure and
realization is chosen from a well known set of models. The parameters have no physical meaning or relationship
with the system’s electrical principles (in this case a transformer), since the model structure is just a mathematical
equation that matches the model output with the observed data. In this case, no prior knowledge about the system
physical arrangement or characteristics is used.

A methodology that is a compromise between these two approaches is the Grey-box model. This terminology is
associated with methods and models that can be put on a scale ranging from a pure White-box physical model to a
pure Black-box parameterized model. Together with the knowledge about the first principles and other system
physical information, measurement data is used in the Grey-box approach to derive the model structure and to
compute the model parameters. Usually the model structure is based on prior system knowledge and the
associated model parameters are estimated using measurement data.

As far as transformer models are concerned, the White-box approach uses network equivalent parameters such as
resistances, self and mutual inductances, capacitances and conductances to build the model structure. There is
usually a focus on a certain frequency range of interest. This type of model is useful for analysing events that occur
within the transformer and can be constructed to represent either all parts of the transformer in great detail.
Alternately, the model can be constructed to represent gross physical components such as winding layers. One of
the basic purposes of such models is the calculation of internal voltages and dielectric stresses to ensure that the
insulation system can withstand all occurring dielectric stresses during field test procedures. It is common to
evaluate these stresses as well as these voltages for assuring design safety margins are met.

Black-box models for power transformers are the terminal equivalent that reproduce the transformer behavior as
seen from its terminals, over a wide frequency band. The available techniques are only capable of representing
linear effects. This modelling approach is particularly suitable for studying the high-frequency interaction between a
transformer and the network, analysing transferred overvoltages between windings, insulation coordination of a
power system. These models can be employed to evaluate the current and voltage wave shapes at the terminals
of the transformer.

On the other hand, there are many examples in the literature and in the engineering practice describing simplified
methods and models for representing power transformers. Generally such practice uses models valid only in
certain a range of frequencies. This range is chosen depending of the desired application. It can be viewed as
grey-box models since the structure is built primarily on the equipment physic and the parameters obtained by
measurement tests. These very easy-to-use methods and models will be referred to here as simplified models.

Modelling power transformers for general electromagnetic studies is a difficult task since it should consider both the
non-linear core behavior and the wide-band frequency dependent behavior associated with the winding resonances
This chapter will not discuss low-frequency transients associated with core non-linear effects such as inrush current
and ferroresonance since the main focus of this brochure is the impact of high-frequency overvoltages transients.
Reference can be made to [4.4] for the modelling the transformer low frequency non-linear behavior.

In this chapter, an overview of the main procedures for wide-band modelling of transformers for high-frequency
transient studies is given. Section 4.1 presents simplified procedures for representing power transformers. Section
4.2 White-Box modelling procedures for obtaining a detailed model of the windings based on the transformer
internal geometry. Section 4.3 reviews procedures for Black-Box modelling where a terminal equivalent is extracted
via frequency sweep measurements and rational fitting techniques. Finally, the Grey-Box approach is discussed in

Page 33
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Section 4.4, where the parameters of the transformer's ladder network representation are determined from
frequency sweep measurements and other methodologies.

4.1 Simplified procedures

Exact modelling of transformer is a difficult and time consuming task. Fortunately, however, in many practical
situations, simple models and modelling procedures are suitable and sufficient for transient analysis. Such
simplified models are usually based on small RLC networks (the transformers are represented on digital computer
programs by a circuit using resistances, inductances, ideal transformers, saturable inductors and capacitances)
related with the equipment basic structure. The necessary details included in the model consider the context of the
transient phenomena being analysed. So, usually the selection of a suitable representation for a given
electromagnetic phenomena, which can include or not the surge transfer from one winding to another, depends on
the frequency range, but also on the equipment design and on the available data.

This section presents three types of simplified transformer modelling approaches. A common idea of such cases is
that the model parameters should be easily obtained compared to Black-box or White-box representations. So, the
presented cases are based on parameters provided by manufactures, test reports, typical surge capacitance to
earth or frequency response analysis. The first case discussed is a power frequency standard model with external
capacitance added. The second case represents the transformer by a few concentrated capacitance. This model
type is commonly applied for very fast transient studies. The third and last case is a procedure to obtain a simple
RLC network based model using the asymptotes and resonances of the measured frequency response.

4.1.1 Power frequency standard model with external capacitance added

The power frequency standard model with external capacitance added is a simply approach to transformer
modelling. It starts with a conventional power frequency model and adds lumped capacitors to the transformer
terminals. It is usually applied to slow transients in the range of less than 100 kHz. The basic idea is illustrated in
Figure 4.1, which contains the model of a two-winding single-phase transformer, where: C1s and C2s are the
series capacitances of the windings, C12 is parallel capacitances between windings, C10 and C20 are
capacitances between windings and earth. R1, L1 R2 and L2 are ohmic resistances and leakage inductance of the
windings and Ro and Lo represents the core of the transforms. TI represents an ideal transformer.

Figure 4.1: Combining power frequency transformer model with external capacitive network.
These capacitances generally can be provided by manufactures. The other parameters can be determined from
usual transformer test reports.

However, sometimes these capacitances are not available. In that case one can consider the capacitances
measurements from transformer test report data. Such measurements represent the total capacitance for the

Page 34
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

winding with the three terminals bonded, i.e. 3C10, 3C20 and 3C12, see also Figure 4.2. However, the series
capacitances are not included in such measurements.

Figure 4.2: Transformer capacitances from measurements.

The same approach can be applied to three-phase transformers. In that case, capacitances should be added
between phases on the low-voltage side and the high-voltage side.

4.1.2 Concentrated Capacitance Model for Fast Transients

Traditionally, in fast-front transients, such as lightning surge studies, only the capacitive behavior of the power
transformer is relevant. So, the mutual capacitive coupling between phases is ignored and often the real and
reactive parts as well, leaving only a capacitive per-phase equivalent as shown in Figure 4.3 and Figure 4.4. Such
a model could be used for representing the impact of a transformer on an incoming wave on a cable or overhead
line, but is way too simple for capturing the detailed high-frequency interaction between the transformer and the
network, or the transfer of overvoltages between windings.

Figure 4.3: Simplified capacitive representation (one terminal).

Figure 4.4: Simplified capacitive representation (two terminals).

The parameter C1, C2 and C3 are the surge capacitances and will depend upon the range of frequencies
considered. The parameter C1, of Figure 4.4, can be provided by the manufacture or calculated from impedance
frequency response measurements. If these values aren’t available from measurements or from manufactures,
typical values are suggested in the literature. Some values are proposed, as example, in Table 4.1 [4.5]. Typical
values for auto transformers capacitance are presented in Table 4.2 [4.6]. This subject is also discussed in [4.7]
and in [4.8] where a list of typical values are also provided.

Page 35
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Nominal Voltage [kV] Capacitance [pF]

500 3000

187-220-275 2500

110-154 1500

66-77 1000
Table 4.1: Transformer capacitance to earth [4.5].

Voltage [kV] 115 400 765

Autotransformer [pF] 3500 2700 5000

Table 4.2: Transformer capacitance to earth [4.6].

Finally, another possibility is the calculation of the capacitance to earth of autotransformers using the following
expression for Y connection transformer, where C is the capacitance in nF, and S is the rated power in MVA. For a
delta connection, C has to be divided by 2 [4.7].

C = 0.52 × S 0.4

4.1.3 Frequency dependent transformer model for fast transients

Frequency characteristics of transformer windings may have a great influence on transient calculations. In the case
of multiple resonance frequencies representation, an asymptotic approach can be used to obtain a low order
representation of winding behavior in the frequency domain [4.8], [4.9]. The model described here belongs to the
class of models based on the frequency dependent response at the terminals of the transformer (black box) which
is reproduced by means of equivalent circuits. This procedure is more suitable for cases where surge transfer from
one winding to another is not of interest (e.g. unloaded transformer, so that the mutual coupling between terminals
can be ignored). Nevertheless, it is possible to approximately represent all transformer windings by mean of
terminal impedances and a standard EMTP transformer model.

If the impedance versus frequency Z(f) has been measured at the terminals of an actual transformer, it can be
modelled by a combination of several damped series resonance circuits connected in parallel to its main
inductance and surge capacitance [4.8]. In the following a simple procedure for constructing such network directly
from the given frequency response is illustrated. Alternatively, curve-fitting techniques can be used (see Section
4.3).

An example of such a procedure is illustrated using a 600 MVA single phase transformer with three windings, the
voltage ratios are 765/√3-230/√3-13.8 kV. Figure 4.5 shows the field measurements of a terminal impedance of this
transformer. These measurements were performed at the terminals of the secondary winding, with the other
terminals opened.

Page 36
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

(a)

(b)
Figure 4.5: Terminal impedance field measurements – (a) amplitude and (b) angle.
Modelling of main inductance and surge capacitance
The main inductance is obtained from the low frequency range of the impedance. As shown in Figure 4.5, the
terminal impedance is inductive below 300 Hz. For instance, Z is equal to 4600 Ω at 60 Hz, which gives L = 12.2 H.

In this example, the surge capacitance is easily identified in the frequency range from 10 kHz to 200 kHz. For
example, Z is equal to 182 Ω at 100 kHz, which gives CS = 8.7nF. This is the surge capacitance mentioned in
Section 4.1.2. At the first resonance f1, the terminal impedance has a resistive behavior and it is equal to 290,000
Ω. Thus, the first circuit of the model is obtained. Figure 4.6 compares the terminal impedance measurement
(green curve) and the frequency response of the first circuit (red curve).

Page 37
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Figure 4.6: Frequency response of the first circuit.

Modelling of multiple resonances
The resonant RLC series circuit is related to the anti resonance, like frequency f2 shown in Figure 4.5 and the
parameters are calculated based on two definitions of quality factor. The first one is related to the relation between
resistance and reactance:

Q = 2 π fo Lo / Ro (4.1)

Q can also be characterized by a bandwidth relative to its center frequency, as shown in Figure 4.7:

Q = fo / df (4.2)

The bandwidth df [Hz] is defined as the range of frequencies in which the signal's spectral density is above (below)
a certain threshold relative to its maximum (minimum). Most commonly, the threshold refers to 3 dB:

ZdB = 20 log10 (Rdf / R0) → Rdf = 1,414 R0 (4.3)

Combining definitions (4.1) and (4.2) of Q gives:

Lo = Ro / (2 π df) (4.4)

Figure 4.7: A graph illustrating the concept of Q.

The value of the capacitor can be determined by the following expression:

Co = 1 / (Lo (2 π fo) )
2
(4.5)

The RLC values of anti resonance frequency f2 are calculated as follow:

• From Figure 4.5, we have R0 = 242 Ω and f0 = 5890 Hz;

• Equation (4.3) gives Rdf = 342 Ω, which defines df = 900 Hz in Figure 4.5;

Page 38
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

• L0 and Co are calculated from equations (4.4) and (4.5).

Frequency R [Ω] L [mH] C [µF]

f2 242 42.79 0.0170
Table 4.3: Parameters of series RLC circuit.
Figure 4.8 shows the frequency characteristics of synthesized circuit and field measurement.

Figure 4.8: Frequency characteristic of Z: synthesized circuit and field measurement.

The last task is the modelling of the resonance associated with surge capacitance and inductance of connections.
This series RLC circuit is calculated by equation (4.5), where Co = CS = 8.7 nF and f0 = f3 = 405.6 kHz, and the
resistor is equal to the terminal impedance Z at f = f3. Figure 4.9 shows the final result.

Figure 4.9: Frequency characteristics of the synthesized model.

Page 39
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

4.2 White box approach

Each type of model has its own advantage/disadvantage depending on how it is to be used [4.10]. Thus, if the
principal interest is on the behavior of the transformer as a component into a power system, a terminal model could
be adequate. But if the concern involves events within the transformer, such as the transient stresses on the
winding insulation, a detailed and more complex model has to be used, also known as White Box model.

If switching operations or surge excitation closely matches one of the natural frequencies of a transformer’s
windings, there is potential for exciting an internal resonance which can produce high internal voltages [4.11]. This
resonant response may eventually result in the catastrophic failure of the transformer. To study this phenomenon,
manufacturers convert physical geometry and material characteristics into a lumped RLCG Detailed Model. The
parameters of this model are determined based on numerical field analysis methods, analytical methods or charge
simulation methods, etc [4.7] [4.13].

When the wavelength corresponding to the frequency components present in the voltage surges is comparable
with dimensions of the winding, the transformer winding has to be treated as if it were transmission lines. When the
model of a transformer has to be used for the MHz frequency range, it requires that every turn has to be
represented as well as all mutual, inductive and capacitive coupling with every other turn [4.14].

The high frequency Detailed Models of transformer windings can be classified into three basic types[4.10] [4.14]:

1) Lumped parameter model

2) Transmission line model (Distributed parameters)

3) Hybrid model

Classically, in the Lumped Parameter Model (LPM) the length of windings is far less than the wavelength. However,
there is no limit to modelling each coil employing lumped parameters whilst the winding length is shorter than the
electromagnetic wavelength. Therefore in order to extend the application range of RLCG Lumped model to a few
MHz, it is necessary to use a turn-to-turn modelling procedure instead of disk-to-disk modelling[4.19][4.30].

In the transmission line model, the Multi-Transmission Line (MTL) Model is the more widely used. Each coil is taken
as a distributed line, with the transformer windings represented by a group of interconnected and coupled
transmission lines [4.14]. In the case where windings are constructed from uniform disks, they can be modelled by
a Single Transmission Line (STL) where it cannot include the mutual inductance between turns or winding coils.
The STL model describes each elementary unit such as disks, while the MTL model is used to model the
interaction between each elementary unit [4.10].

The Transmission Line model has been proposed to take into account electromagnetic wave propagation.
Transformer windings are modelled as a group of interconnected and coupled Transmission lines (the turns). In the
transmission line theory it is usual to use distributed parameter. The model is based on a zigzag connection of
different conductors, where each transmission line representing a complete turn of the winding. The literature
indicates that lumped-parameter model can give adequate results for fast transients (up to 1 MHz) and for very fast
transients (above 1 MHz) a distributed-parameter model provides better results [4.10].

The hybrid model combines the STL procedure of the whole winding together with the MTL. First, each coil is
represented by an STL model where voltages at the coils ends are obtained. Then, each coil is represented by an
MTL model to compute the distribution of the inter-turn over-voltages independently from one coil to another.
Another alternative is to combine STL or MTL together with LPM or even finite element methods, for example.

The ability of the Detailed Model to faithfully reproduce the transient characteristics of the transformer depends on
accurate values of its RLCG parameters in the range of the intended frequencies.

The transient response can be represented by partial differential equations such as the case of transmission line
model, but their solution is very difficult. If the windings are subdivided into sections in which inductances and
capacitances are lumped, the calculation becomes easier since the partial differential equations can be now
replaced with close approximation by ordinary simultaneous differential equations. These ordinary differential
equations can be solved by analytical analysis for homogeneous winding or by numerical analysis for complex
configurations of windings. The accuracy of results obtained from the circuit representation depends predominantly
on the degree of sophistication used in the representation of windings.

Page 40
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Nowadays, commercial software packages are available on the market. However, manufactures prefer to use their
own tools integrated into their design environment. Into that scheme, a detailed method for modelling and
simulating VFETs in PT is presented [4.14]. Special attention is paid in the construction of the lossy LPM (skin and
proximity effects, and dielectric losses) [4.15]. Numerical solutions are carried out applying modal analysis and
numerical Laplace transform (NLT) [4.16].

4.2.1 Frequency Dependence

The inclusion of losses in a Detailed Model is indispensable, particularly when internal stresses are evaluated in
the design-stage of a transformer [4.19]. Without the implementation of losses, it is believed that the stresses will
be higher in the model than in reality and the design will become more expensive than necessary to manufacture
and hence less cost competitive.

When a surge voltage arrives at the terminal of a transformer, different frequencies penetrate the winding at
different velocities. From knowledge of the surge voltage in the frequency domain, it is possible to see how
windings respond to this particular frequency spectrum [4.29].

Although both the capacitance and inductance are of distributed nature, for practical computation purposes these
can be lumped in varying degrees according to desired accuracy. Thus, the transformer can be approximated as a
collection of lumped circuit elements. The parameters R, L, C and G of the transformer model can be independent
of frequency. However for high frequency performance the dependence of parameters with the frequency must be
taken into account if a good simulation is to be achieved by the model. This frequency dependence is a
consequence of the penetration of the magnetic flux into a conductor, and the electric field into the insulation.

Frequency dependent losses in the conductors occur due to eddy currents caused by time-varying magnetic fields,
and dielectric losses within the insulation structure due to time-varying electric field.

The assessing of the surge behavior in the transformer by an equivalent network, the values of the frequencies and
their related amplitudes, depend on the parameters of the network and the boundary conditions. The total number
of natural frequencies of the model is limited by the number of free nodes considered [4.19]. The amplitude is also
a function of the location of node because of the presence of the internal damping associated with the dissipation.

The magnitude of the natural frequencies are primarily dependent upon inductances and capacitances but with
negligible dependence upon resistance and conductance.

4.2.2 Computation Methods for Transients

Transient response is a result of the flow of energy between the distributed electrostatic and electromagnetic
characteristics of a device, in this case the transformer. The very fast electromagnetic transient (VFET)
encompasses rise times in the range of 50 to 100 µs and frequencies from 0.5 to 30 MHz. For all practical electrical
(power transformer) PT winding configurations, the VFET simulation is quite complex and can only be realistically
computed by constructing a detailed lossy model of the winding structure [4.15]. This is generally accomplished
using lumped parameter model (LPM) and then carrying out a numerical solution for the internal transient voltage
response. Once the voltage distribution is known, its effect on the PT insulation structure can be assessed with the
calculation of resultant electric field using 2D or 3D finite element method (FEM) [4.17] [4.18] [4.29].

Nowadays, commercial software packages are available on the market. However, often manufactures prefer to use
their own computation methods for transients as tools integrated into their design environment. Into that scheme, a
detailed method for modelling and simulating VFETs in PT is presented [4.19]. In that proposal, a special attention
is paid in the construction of the lossy LPM (skin and proximity effects, and dielectric losses) [4.19] [4.20].
Numerical solutions are carried out applying modal analysis and numerical Laplace transform (NLT) [4.21].

4.2.3 Mathematical Model

For the simulation of VFETs in transformers with W windings Maxwell’s equations for transverse electromagnetic
(TEM) waves for lossy transmission lines can be applied [4.19]. Such W windings can be discretized into blocks,
which are represented by means of equivalent π-circuits with lumped parameters. As a result, Maxwell’s equations
can be expressed as in [4.19],

Page 41
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

W Kj Ki
Vi (ki ) = Vi ( K i ) + ∑∑ ∑ Z
j =1 m =1 r = k i +1
ij ( r , m) I j ( m) (4.6)

W K j Ki
I i ( ki ) = I i ( K i ) + ∑ ∑ ∑Y
j =1 m =1 r = k i + 1
ij ( r , m ) V j (m) (4.7)

where ki = 0, 1, …, Ki, i = 1, 2, …, W, Zij and Yij are the impedance and admittance submatrices of the windings in
which the ith-winding is discretized in Ki blocks, and jth-winding is discretized in Kj blocks. Vi(ki) is the voltage at
kith-node of the ith-winding, and Ii(ki) is the current flowing from node ki-1 to node ki. From (4.7) and (4.8), the
resultant equation system (4.9) is expressed by a set of matrix equations [4.21],
 I B   A B  VB 
 V'  = C D   V'  (4.8)

where A, B,C,D are matrices which are functions of impedance Z and nodal admittance Y matrices, of the
equivalent network of transformer, formed by Zij and Yij. VB represents voltage at the winding terminals by equation
(4.10), IB is the current entering into each winding and V’ represents the voltage at the internal nodes [4.19] [4.22].
Manipulating (4.9), the modal form of the solution is obtained in frequency domain. The time domain solution is
computed applying the NLT [4.5] [4.9], by equation (4.16),

2N
VB (c + j ( 2k - 1) ∆ω ) = ∑V ((2n - 1)∆t ) e−
T ( c+ j ( 2k - 1) ∆ω ( 2n - 1) ∆t )
B (4.9)
2N n =1

Δ
t
V' ((2n - 1)∆t ) = ∑ V' (c + j (2k - 1)∆ω ) ((2k - 1)∆ω ) e(c+ j ( 2k -1) ∆ω ( 2n - 1)
1 N
)
(4.10)
T k =1

where ∆t is the time step by N samples, ∆ω the integration step in the frequency spectrum, T is the total time
of simulation, σ((2k-1)∆ω) is the Hanning data window for attenuating the Gibb errors produced by the truncation of
the frequency, and c reducing the aliasing error produced by the discretization of the frequency. Wilcox proposed
as c = 2∆ω in [4.23].

Figure 4.10: Representation of transformer windings. (a) Discretization of windings into

elementary lumped blocks. (b) Equivalent lumped parameter model for each pair of blocks.

4.2.4 Lossy Lumped Parameters

The proposed model requires dividing the W windings of transformer into elementary lumped blocks (see Figure
4.10.a). These lumped circuits incorporate mutual coupling between RLCG parameters (Figure 4.10.b). Each block
is formed by one coil-turn or combining several turns. Thus the simulation of transient voltage distribution in PT
winding depends on the calculation of RLCG lossy parameters [4.18], which has to be done from the geometric
dimensions and physical properties of the winding materials [4.19] [4.20] [4.24].

Page 42
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

According to Figure 4.10.b, the series capacitances inter-turns CS have been taken into consideration [4.20]. Also,
earth capacitances Cg of the turn facing the core, tank and yoke have to be included [4.20]. Shunt capacitances Cp
between adjacent turn face windings have been introduced to model the electrostatic coupling between the
windings [4.24]. These capacitances are grouped between blocks (Figure 4.10.b) [4.25] [4.26].

The self-inductance of each turn is calculated assuming the turn is a circular wire with a rectangular cross-section.
The mutual inductance is computed using Lyle’s method [4.27]. The mutual inductance between turns of different
windings has also been considered. These inductances are grouped and the resultant inductance included in
Figure 4.10.b [4.27]. It is assumed that the flux does not penetrate into the core at high frequency [4.20] [4.24].

Losses play an important role in an accurate simulation of the distributed voltages. Therefore, frequency dependent
copper and dielectric losses are considered. Thus Zij and Yij matrices are obtained from the resistance Rij,
inductance Lij, capacitance Cij and conductance Gij matrices with lossy parameters by (4.12) and (4.15) in each r
and m blocks according to (4.7) and (4.8) expressions [4.19],

Z ij (r , m ) = Rij (r , m ) + jω Lij (r , m ) = R DC , ij (r , m )+ Rskin , ij (r , m )+ R prox , ij (r , m )+ jω Lij (r , m ) (4.11)

π rh ωµo
Rskinhh =
(ah + bh )
(4.12)
2σ c

 2ω 
R proxhq =   Lhq , h ≠ q (4.13)
 µ oσ c d hq2 
 

Yij (r , m ) = (ω tan δ + jω ) Cij (r , m ) = Gij (r , m ) jω Cij (r , m ) (4.14)

where µo is the permeability of vacuum, ω is the angular pulsation of the sinusoidal input voltage, tanδ is the loss
tangent of the insulation, σc is the electrical conductivity, rh, ah and bh are radius, axial and radial dimensions of
each hth-turn respectively, and dhq is the distance between the hth and qth-turns. The real term in (4.12) represents
the Joule losses due to direct current, skin and proximity effects, where the two last are calculated by (4.13) and
(4.19). In (4.15), the real part corresponds to the dielectric losses.

Once all RCLG parameters are calculated, each block is represented by an equivalent π-transmission line. Finally,
the LPM can be built, where the minimum network pattern is between two blocks as shown in Figure 4.10.b. This
model is flexible enough to consider different number of blocks in each winding. Further more information is
available in [4.18].

4.2.5 Illustration. Practical Application

A real three-phase EFACEC 40 MVA, 60/10.5 kV, core type transformer was modelled according to the previous
methodology in order to verify its accuracy. Each primary phase presents three windings: main high voltage
winding (HV), coarse tapping winding (PS) and fine tapping winding (REG). The HV winding consists of 50
continuous disks and 6 turns per disk. PS winding is a two layer winding with 65 turns. REG winding is a
interleaved helical winding with 13 taps in parallel and 5 turns per tap. Each secondary phase consists in the low
voltage winding (LV) with two layers and 54 turns per layer. The configuration of one phase is shown in Figure
4.11.a, where the electric representation with windings grouped in different blocks can be seen in Figure 4.11.b.
Each block, a rectangular box within a broken line, is represented by an electric circuit as described in Figure
4.10.b. Figure 4.11.c is a symbolic representation of all one-phase windings.

4.2.6 Validation. Measurements and Simulations

A pulse generator was connected to the TS(HV) terminal (Figure 4.11). Different transient voltages tests were
applied to the transformer without tank. Only simulation results and measurements in response to standard
lightning impulse of 1.2/50 µs are presented in this section for validation.

Page 43
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Figure 4.11: Outline of one phase of the tested transformer. (a) Geometric configuration. (b)
Electrical representation. (c) Symbolic.representation.

 ) voltage transients. a) At
Figure 4.12: Comparison between simulation (····) and measured (
the end node (no. 51) of HV winding TR(HV) with maximum tapping. b) At the end node (no. 14)
of PS winding TR(PS) with minimum tapping.

Page 44
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Figure 4.13: Simulated results of internal voltage distribution along all nodes of the HV winding
for different times. a) With maximum tapping. b) With minimum tapping.

Figure 4.14: The simulated internal voltage along the HV winding (nodes) vs. time after an input
standard lightning .impulse. a) With maximum tapping. b) With minimum tapping.
The simulations were computed applying a SLI voltage waveform mathematically represented as [4.28]

(
Ve(t ) = Vmax A e −αt − e βt ) (4.15)

Where Vmax is the maximum amplitude, t is the time expressed in µs, A=1.0167, α=0.01423, β=6.0691 are adjusted
so that Ve rises to its maximum value in 1.2 µs and decays to half its value in 50 µs (Figure 4.13).

The results of two different configurations, maximum and minimum tapping, are included in this work and shown in
Figure 4.13 when standard lightning impulse is applied at TS(HV). In both configurations, the tap changer is placed
at the maximum position connecting the 13-th tap by means of the switching (SW) at TR(REG) terminal.

Following the connections of Figure 4.11, the maximum tapping is that in which the TR(HV) terminal is connected
to the TS(PS) terminal, the TR(PS) to the TS(REG) terminal and the TR(REG) terminal connected to ground.
Minimum tapping is that in which the TR(HV), TS(PS) and TS(REG) terminals are connected to ground, meanwhile

Page 45
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

the TR(PS) and TR(REG) terminals are floating points. In Figure 4.13 the computed distribution voltage in all the
nodes of HV winding along the time is shown, which permits to identify the coils with higher dielectric stresses.
Finally, Figure 4.14 represents the computed internal distribution voltage for both configurations in all nodes of HV
winding vs. time. For further information, refer to [4.19] [4.29] [4.30].

4.3 Black box approach

With the black-box approach [4.31]-[4.39], the transformer is modelled as a terminal equivalent that attempts to
reproduce the behavior of the transformer as seen from its terminals, over a wide frequency band. The available
techniques are only capable of representing linear effects, i.e. any core non-linear effects cannot be faithfully
represented. This modelling approach is particular suited for studying high-frequency interaction between the
transformer and the network as well as transferred overvoltages between windings. Currently, the black-box
approach can only be performed by experts, due to difficulties in characterizing the transformer behavior as well as
the model extraction step. Most experience with this approach is for distribution transformers but the principles for
modelling and measurements are the same for large transformers. Case study 1 in “Part 2: Case Studies” shows
an example of black-box modelling of a 410 MVA step-up transformer.

4.3.1 Characterization of the transformer behavior

The characterization of the terminal behavior is most conveniently shown in the frequency domain via the terminal
admittance matrix Y which defines the relation between voltage v and currents i at the transformer terminals.

i(ω ) = Y(ω ) v(ω ) (4.16)

The admittance matrix Y can be obtained in several ways, via measurements or by computations from detailed
geometrical data, e.g. starting from a white-box model. In practice, usage of measurements is likely to produce a
more accurate model.

A. Measurements
Using a vector network analyser, the voltage/current relations can be measured at a set of discrete frequencies
(frequency sweep). For an n-terminal transformer, the admittance matrix can be computed from a set of n linearly
independent voltage applications V and corresponding current responses I,

Y (ω ) = I (ω )V −1 (ω ) (4.17)

The admittance matrix can also be measured directly by choosing V equal to the identity matrix. One experimental
setup for achieving the direct measurement is shown in Figure 4.15, left panel. With this setup, the elements of Y
can be measured one-by-one by rearranging connections on the box. Some measurement setups are commercially
available for characterization and modelling of transformers. With the ABB setup (SoFT) [4.35] (right panel in
Figure 4.15), the measurements are done by choosing the voltage application equal to the eigenvectors of Y. This
approach leads to higher accuracy compared to a direct measurement of Y by giving a more accurate
representation of its small eigenmodes. Increased accuracy of small modes can also be achieved by choosing V
and I as responses coming from a mix of short-circuit and open-circuit responses [4.39] (hybrid method).

With all setups, the zero sequence system associated with ungrounded windings (e.g. a delta-connected winding)
poses a difficulty when extending the measurements to lower frequencies (below 10 kHz). The problem is related
to the very small current associated with a zero-sequence voltage application.

Page 46
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Source Ref. Input

Jumpers
to ground Attenuator

6
5
4
3
2 Current
sensor
1 A

Figure 4.15: Measurement setups for characterizing transformer admittance matrix.

Left panel: Direct measurement of elements [4.33];
Right panel: ABB “SoFT” modal measurements.
The accuracy of the measurements often deteriorates at low frequencies due to the low impedance of the
transformer, compared to the internal 50 Ω impedance of the network analyser. As a result, it can be difficult to
apply the model in simulations which require correct 50/60 Hz initial conditions, e.g. some switching transient
studies. The initialization problem can be solved using superposition techniques [4.55]. Transformers with one or
more ungrounded windings are in particular difficult to model at low frequencies as some eigenmodes are much
smaller than the others [4.34]. The loss of accuracy can lead to large error magnifications in a simulation,
depending on the terminal conditions in the simulation. Reference [4.34] addresses this latter problem by a special
measurement and modelling procedure.

The cables between the measurement unit and the transformer are in practice several meters long and can at high
frequencies contribute substantially to the measured admittance matrix. The measurement procedure should
therefore be able to reduce the effect of the cables. One procedure which removes the cable capacitive effects into
account is shown in [4.33]. Still, the length of the cables will in practice place an upper limit fmax of the useable
frequency in the measurement due to the occurrence of standing waves in the cable. In practice, the upper
frequency limit fmax must be chosen well below the quarter wave resonance frequency fλ/4 ,

v
f max < f λ / 4 = (4.18)
4l
with v and l being the cable propagation velocity and length, respectively. For instance, a propagation velocity of
198 m/µs (RG 214) and a cable length of 10 meters gives a value fλ/4 = 5 MHz.

Standard FRA measurement systems are not directly applicable for measuring Y. One reason is that they are
based on only a single input terminal in addition to the source and reference terminals. In principle, one could
measure the matrix of voltage transfer from each terminal to the other terminals. In practice, however, this is not
feasible because the usage of 50 Ω input impedances will change the voltage transfer functions when moving the
input from one transformer terminal to the next.

B. Calculations from white box model

The admittance matrix can also be computed from the white box model. The CIGRE WG (B2/C4.39) suggests that
the transformer manufacturer could provide a computed Y(ω) when possible, for the purpose of modelling and

Page 47
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

simulation. It is remarked that providing this information would not reveal any proprietary information about the
transformer design.

4.3.2 Model extraction

From the tabulated Y(ω), a model needs to be constructed. For that purpose, a model based on rational functions is
preferable as its leads to computationally efficient models that can be easily made compatible with standard circuit
solvers, i.e. EMTP-type simulation programs. Most approaches attempt at fitting a rational model on state space
form to the data (4.20). In order to obtain a symmetrical model, it is convenient to use a pole-residue representation
with common poles (4.21). The model (4.21) can be directly converted into (4.20).

Y(ω ) ≅ C( jωI − A)B + D (4.19)

N
Rm
Y(ω) ≅ ∑ + R0 (4.20)
m=1 s − am

A number of approaches are available for fitting the models (4.20) and (4.21) to the data [4.40], [4.45]. Vector
Fitting-based methods [4.42], [4.45] have proved to be quite suitable for this task. Vector Fitting (VF) is based on
relocating a set of initial poles to better positions by iteratively solving a set of linear equations. Application of VF to
(4.20) leads to a rational model which satisfies the physical requirements of being symmetrical, causal, and be
formulated using stable poles only. Due to the large dynamic variation of the elements in Y, relative error control is
required to obtain a satisfactory accuracy for all matrix elements over the full band. Such error control is achieved
by inverse magnitude weighting of the (linear) least squares problems used by VF.

Although the model obtained by VF is guaranteed to have only stable poles, the resulting time domain simulation
may still be unstable due to passivity violations. Small passivity violations can be removed by perturbing the
model’s residue matrices [4.46]-[ 4.50]. The passivity enforcement step is a non-trivial task as the model can easily
become corrupted by the perturbation. Passivity enforcement is required also for alternatives to VF.

Figure 4.16 shows an example where the measured admittance matrix Y has been fitted using a passive pole-
residue model (4.20). It is noted that Y has many resonance peaks and that the elements differ very much in
magnitude.

Figure 4.16: Example of measured and fitted admittance matrix of 300 kVA distribution
transformer [4.33]. Subscript “H” and “L” refers to high and low voltage windings.

Page 48
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

4.3.3 Model interface with EMTP-type simulation software

A. Lumped Circuit Equivalent
The model (4.20) can be directly synthesized with a lumped circuit equivalent consisting of resistors, inductors and
capacitors [4.31] [4.51]. The network has branches between all terminals, and between and each terminal and
ground. This is illustrated in Figure 17 for a six terminal transformer (not all branches between nodes are shown).
The network is entirely fictitious and topologically incorrect. Therefore, no physical meaning can be associated with
each branch and many network elements will be negative. Nevertheless, the total behavior as observed from the
external terminals reproduces exactly that of the model (4.20). Some loss of accuracy can however result with
some circuit simulators because of the finite precision in the input data format [4.54].

Figure 4.17: Structure of circuit equivalent.

B. Convolution
The interface can also be achieved via a companion model and recursive convolution, see Figure 4.18 [4.52],
[4.53]. The model is represented by a conductance matrix and a set of controlled current sources between ground
and the external transformer terminals. The current sources are updated in each time step based on past history
values of the terminal voltages. This is shown in Figure 4.18 for a one-terminal network.

G Ihistory(t)

Figure 4.18: Interface by companion model.

Figure 4.19 shows one example of measured vs. simulated overvoltages on a distribution transformer [4.29]. In this
case, the transformer model was obtained via frequency sweep measurements and modelling via VF and passivity
enforcement by perturbation. The final model validation is done in the time domain by applying a step voltage to
one of the terminals. In the test, lumped circuit elements have been connected to the terminals. By including the
transformer model, the lumped circuit elements, and the applied voltage (represented by an ideal voltage source) in
an EMTP-type simulation, a close agreement is obtained between the measured and simulated overvoltage on
terminals 2, 3, and 6.

Page 49
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Figure 4.19: Example of measured vs. simulated overvoltages on transformer terminals. 30 kVA
distribution transformer [4.29].

4.3.4 Simplifications to the black-box approach

The application of black-box approaches can require a large number of measurements, even when taking into
account the symmetry of Y. For instance, to measure Y of a two-winding transformer requires 21 separate element
measurements, which can be very time consuming. In the case of three-winding transformers, 45 measurements
are needed.

The required number of measurements can be greatly reduced by taking into account the actual terminal conditions
for the transformer. To explain this, consider a situation where one wishes to study the effect of a transient event
occurring on the HV side of the transformer with the secondary connected to an overhead line, see Figure 4.20. In
this situation, one may in the frequency sweep measurement connect a resistive network to the secondary that
represents the characteristic impedance matrix Zc of the overhead line. As a first approximation, Zc can at high
frequencies be assumed to be a real and balanced matrix, allowing the overhead line to be represented by a
resistive network defined by the positive and zero sequence component of Zc. (With Zcs and Zcm denoting the self
and mutual terms of the (balanced) characteristic impedance matrix, we have Zc+ = Zcs − Zcm, Zc0 = Zcs + 2Zcm). This
representation of the overhead line is valid until the transmitted wave from the transformer becomes reflected back
to the transformer, i.e. a time t=2τ=2L/v, where L is the line length and v its propagation velocity. In the frequency
sweep measurements, terminals 4, 5 and 6 are simply ignored. Thus, the obtained model by this procedure is a
model of the transformer plus that of a connected overhead line, being valid for a time 2τ after the initiation of the
transient event.

If required, the overvoltage on the LV terminals can be obtained by measuring the voltage transfer matrix from the
HV terminals to the LV terminals. These measurements can be synthesized using rational functions and included in
the circuit simulation [4.52].

In addition to reducing the required number of measurements, the suggested procedure can simplify the modelling
procedure as well. In particular, the difficulties in enforcing passivity for the model can be greatly simplified.

Ra = Re{Z c + } (4.21)

Re{Z c 0 } − Re{Z c + }
Rb = (4.22)
3

Page 50
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

1 4
Electrical 2 5 Overhead Electrical
network 3 6 line network

Ra
1 Rb
2
3

Figure 4.20: Representing six-terminal transformer as a three-terminal device during

measurements by treating overhead line as a resistive network.

4.3.5 Validation of measurement setup and modelling procedure

Due to the many challenges in the black-box modelling approach, several stages of validation are recommended.

•
T
Verify that the measured Y is symmetrical, Y=Y
•
H
Verify the the measured Y is passive, i.e. the matrix K has positive eigenvalues: K=Y+Y , with superscript
H denoting transpose and conjugate.
• Compute the voltage ratio of the transformer from Y and check against a direct measurement of the
voltage ratio.
• Following the model extraction step, check again the voltage ratio of the transformer.
• Perform time domain measurements with a low-voltage transient excitation (e.g. step voltage) and
compare with a simulation result obtained by the model. See e.g. Figure 4.19.

4.4 Grey box approach

The grey box transformer model is a compromise between the black and white box modelling approaches.
However unlike the white box approach, the grey box model is derived without prior knowledge of the internal
specifications of the transformer [4.56]-[4.60]. This latter point is pertinent since the design specifications for a
transformer are generally not available due to the intellectual property restrictions of the manufacturer [4.38].

Whilst a black box model does not take into account the physical attributes of the transformer under test, both the
white and grey box models are based on a transformer’s geometry. The difference between these two approaches
lies primarily in the determination of the model parameters. Since it is assumed for the white box approach that all
internal specifications are known, model parameters can be readily determined through finite element analysis
and/or analytical means. In the case of a grey box model, parameters must be estimated, for example, directly from
the transformer’s frequency response measurements. Other useful support information which will make the
estimation process more robust includes terminal resistance measurements, nameplate details and external tank
dimensions. Model construction assumptions are generally based on typical manufacturing practices [4.56].

Following, an approach for grey-box modelling is discussed.

4.4.1 The grey box transformer model

As in the case of the white box model, the grey box model is a mathematical representation of the complex
electromagnetic relationships that exist within a transformer. A common approach is the use of the ladder network
model which can originally be traced back to Blume and Boyajian’s work of 1919 [4.61]. Their work has
subsequently been improved by many other researchers over the following decades including [4.62]-[4.67]. A
generic phase ladder network model from [4.56] is given in Figure 4.21 where L, R and C represent the frequency
dependent inductance, resistance and capacitance elements respectively. In this case generic phase references

Page 51
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

(X, Y and Z) are used due to their modelling flexibility when considering 3 phase transformers. Given the grey box
transformer model structure, the next step is to estimate the model parameters.

Figure 4.21: Generic phase transformer model.

4.4.2 Frequency Response Analysis for parameter estimation

Frequency Response Analysis (FRA) as initially proposed by Dick and Erven [4.68], is a diagnostic tool which can
be used to look for changes in the frequency response of a transformer. An FRA test yields a signature which is
unique to the transformer’s mechanical geometry from the perspective of the two measurement terminals [4.69].
One approach to estimating the transformer model parameters is to fit the transformer model transfer function to
the corresponding FRA terminal “signature”. This approach has been adopted by several researchers including
[4.70]-[4.72]. This method can be further improved by incorporating the FRA signatures of several different types of
FRA test into the model parameter estimation process [4.56].

4.4.3 Layered Model

A convenient way to visualize the modelling of a three phase transformer undergoing a variety of FRA tests is to
treat the system as three distinct layers (Figure 4.22). The first layer represents three instances of the generic
phase model of Figure 4.21. The next layer is the vector group topology of the transformer (Dyn connection in this
example). The last layer is the FRA test terminal connection currently under consideration. The Cigre WG A2.26
has detailed a comprehensive overview of FRA testing [4.69].

Page 52
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Figure 4.22: Layered Modelling Approach.

4.4.4 Determination of terminal transfer functions

Analysis of the Kirchhoff current and voltage relationships that exist within the transformer model structure will
facilitate the construction of a state space representation of the model. The state space model will be of the form,

(4.23)

where x is the state vector, y the output vector, , B and C are time-dependency matrices, u is the input argument
and t is time. Note that time-dependent terms introduced into (4.23) are due to a range of phenomena. These
include permeability attenuation as a result of magnetic skin effect within the transformer core, winding skin and
proximity effects, and the complex permittivity of the dielectric materials [4.73]. Such nonlinear properties make
difficult to incorporate this model in EMTP-type software.

A transfer function of the model, H , can then be calculated by taking the Laplace Transform of (4.23),

(4.24)

The transfer function (4.24) can then be used to determine the model’s frequency response for a given set of
parameters.

4.4.5 Model parameter estimation

The next step is to determine the best fit between each transfer function and its corresponding FRA data set by
finding the model parameters that will minimize an appropriate cost function. One approach is to consider the 2-
norm between each data set. For example, for a High Voltage Winding End to End Open Circuit FRA test [4.68] on
a three phase transformer, the cost function could be defined as,

 2 2 2
 (4.25)
 Hˆ AC ( jω )   Hˆ BA ( jω )   Hˆ CB ( jω ) 
J = min  log10   + log10   + log10    ,
  G AC ( jω )   GBA ( jω )   GCB ( jω )  
 

Page 53
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

where J is the cumulative total of each test’s 2-norm, Hˆ AC is the estimated transfer function between terminals A
and C, GAC is the FRA between terminals A and C, and the Laplace Transform complex number s is assumed
purely imaginary such that s = jω . Likewise for the other terminal permutations (BA and CB). In order to avoid
local minima, model parameters should also be constrained, wherever possible, using transformer details that can
be obtained through external measurements and knowledge of common manufacturing practices. Failure to
appropriately constrain the parameter set could result in a solution set which is not physically representative of the
transformer.

A suitable algorithm is then applied to the cost function in order to determine the global minima. There are many
algorithms which could be used including non-linear least squares, simulated annealing, genetic programming, and
bacterial swarming [4.38] [4.71] [4.72]. These algorithms are generally readily available using an applicable Matlab
toolbox. It should be noted that a model parameter estimation algorithm could be quite computationally expensive
and may take many hours to process.

After estimating the cost function’s global minima, the model parameters which give the best overall FRA fitting
have now been determined and the transformer’s grey box model is complete. This approach will be discussed in
the Itaipu GIS case study in “Part 2: Case Studies”.

Page 54
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

4.5 Comparison Table

Black Box White Box Grey Box

Typical System interaction and

System interactions System interactions
applications internal overvoltages

Approximately 20th
up to 2MHz (depends on natural frequencies,
Typical Model that is,
the measurements up to 500kHz
Bandwidth
quality) up to 500k to 800kHz
for large transformers

Possible to be
obtained, it depends on
how each model
Very Fast
Difficult to obtain reliable element will be
Transients,
measurements represented. The initial
above 2MHz
part of the winding
should be well
represented.

Measurements and Design

Data Basis Measurements Design Geometry
Data

numerical field analysis

Model Optimization
methods, analytical Optimization
Extraction
methods

Model
Medium / High High Small / Medium
Complexity

Simulation
Small Small / High Small/High
Time

Integration Yes
with EMTP (RLC model or Yes Yes, for linear representations
type software Convolution)

Page 55
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

4.6 References
[4.1] Ljung, L. System Identification, Theory for the user. 2nd ed. Prentice Hall. 1999

[4.2] Bosh, P. P. J. van den; A. C. van der Klauw. Modeling, Identification and Simulation of Dynamic Systems,
CRC Press, 1994

[4.3] Nelles, O. Nonlinear System Identification. Springer. 2001

[4.4] Martinez, J. A. and B. Mork, Transformer modeling for low- and mid-frequency transients: A review. IEEE
Transactions on Power Delivery, 20(2), 1625–1632, April 2005

[4.5] CRIEPI, “Lightning protection design for power stations, substations and underground transmission lines”,
CRIEPI Report T40, 1995 (in Japanese).

[4.6] I EEE TF Report, “Modeling guidelines for fast front transients”, IEEE Trans. on Power Delivery, Vol. 11, No.
1, pp. 493-506, 1996.

[4.7] Greenwood A., “Electrical transients in power systems, Second edition”, Jon Willy & Sons Inc., ISBN 0-471-
62058-0, 1991.

[4.8] CIGRE WG 33.02, “Guidelines for representation of network elements when calculating transients”, CIGRE
Technical Brochure 39, 1990.

[4.9] Morched A., L. Marti, J. Ottevangers, “A high frequency transformer model for the EMTP,” IEEE Trans.
Power Delivery, vol. 8, no. 3, Jul. 1993, pp. 1615-1626

[4.10] J.A. Martinez-Velasco, Power System Transients Parameter Determination, CRC Press, 2010.

[4.11] Degeneff, R.C. A general method for determining resonances in transformer windings, IEEE Transactions
on Power Apparatus and Systems, Vol. PAS-96, No. 2, March/April 1977, pp. 423–430.

[4.12] E. Bjerkan and H.K. Hoidalen, High frequency FEM-based power transformer modeling: Investigation of
internal stresses due to network-initiated overvoltages, IPST 2005, Montreal, Canada, June 19-23, 2005.

[4.13] S.M.H. Hosseini, M. Vakilian, and G.B. Gharehpetian, Comparison of transformer detailed models for fast
and very fast transient studies, IEEE Transactions on Power Delivery, 23(2), 733-741, April 2008.

[4.14] M. Popov, L.V. Sluis, and G.C. Paap, Computation of very fast transient overvoltages in transformer
windings, IEEE Transactions on Power Delivery, 18(4), 1268-1274, October 2003.

[4.15] Electrical Power Transformer Engineering, James H. Harlow, CRC Press LLC, (2004), ISBN 0-8493-1704-
5

[4.16] Rocha A. C. O., Electrical Transient Interaction Between Transformers and Power Systems, Cigre Session
(2008), 1-10

[4.17] Yamashita H., Cingoski V., Nakamae E., Namera A., Kitamura H., Design Improvements on Graded
Insulation of power Transformers Using Transient Electric Field Analysis and Visualization Technique,
IEEE Trans. on Energy Conversion, 14 (1999), No. 4, 1379-1384

[4.18] Khaligh A.,Vakilian M., Naderi M. S., A Method for Power Transformer Insulation Design Improvements
Through Electric Field Determination, ScientiaIranica, 10 (2003), No. 4, 410-418

[4.19] Lopez-Fernandez X. M., Alvarez-Mariño C., Computation Method for Transients in Power Transformers
with Lossy Windings, IEEE Trans. Magn., 45 (2009), No. 3, 1863-1866

[4.20] Shibuya Y., Matsumoto T., Teranishi T., Modelling and Analysis of Transformer Winding at High
Frequencies, Proc. Int. Conf. Power Syst. Trans., (2005), No. IPST05-025, 1-6

Page 56
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

[4.21] Wilcox D. J., Hurley W. G., Mchale T. P., Conlon M., Application of modified modal theory in the modeling
of practical transformers, Inst. Electr. Eng. Proc.-C, 139 (1992), No. 6, 513-520

[4.22] Wilcox D. J., McHale T.P., Modified Theory of Modal Analysis for the Modelling of Multiwinding
Transformers, IEE Proc. C., 139 (1992), No. 6, 505-511

[4.23] Wilcox D. J., Numerical Laplace Transformation and Inversion, Int. J. Elect. Enging. Educ., 15 (1978), 247-
265

[4.24] Popov M., Van Der Sluis L., Smeets R.P.P., Roldan J.L., Analysis of Very Fast Transients in Layer-type
Transformer Windings, IEEE Trans. On Power Delivery, 22 (2007), No.1, 238-247

[4.25] Chatterjee A. De, N., Part Winding Resonance: Demerit of Interleaved High-Voltage Transformer Winding,
IEE Proc. Electr. Power Appl., 147 (2000), No. 3, 167-174

[4.26] Honorati O., Santini E., New Approach to the Analysis of Impulse Voltage Distribution in Transformer
Windings, IEE Proc., 137 (1990), Pt. C, No. 4, 283-290

[4.27] Wirgau K. A., Inductance Calculation of an Air-Core Disk Winding, IEEETrans. Power Apparatus and
System, PAS-95, (1976), No. 1, 394-400

[4.28] Munshi S., Roy C.K., Biswas J.R., Computer studies of the performance of transformer windings against
chopped impulse voltages, IEE Proc.-C, 139 (1992), No.3, 286-294.

[4.29] Xose M. López-Fernandez, Casimiro Alvarez-Mariño, R. Lopes, D. Couto and J. Ramos, Modeling and
Insulation Design Methodology in Power Transformer under Fast Transients, IEEE XPlorer-International
Conference on Electrical Machines (ICEM2010), pp. 1-6, Roma, 2010.

[4.30] Xose. M. Lopez-Fernandez, C. Alvarez-Mariño, D. Couto, R. Lopes, A. Jácomo,"Modelling, Simulation and

Measurements of Very Fast Transients in Lossy Transformer Windings with Tap Changer",
Przegladelektrotechniczny, ISSN 0033-2097, R. 86 NR 5/2010 pp. 141-144.

[4.31] A. Morched, L. Marti, and J. Ottevangers, “A high frequency transformer model for the EMTP,” IEEE Trans.
Power Delivery, vol. 8, no. 3, pp. 1615-1626, July 1993.

[4.32] M.J. Manyahi and R. Thottappillil, “Transfer of lightning transients through distribution transformers,” Proc.
Int. Conf. Lightning Protection, September 2-6, 2002, Cracow, pp. 435-440.

[4.33] B. Gustavsen, “Wide band modelling of power transformers,” IEEE Trans. Power Delivery, vol. 19, no. 1,
pp. 414-422, Jan. 2004.

[4.34] B. Gustavsen, “Frequency-dependent modelling of power transformers with ungrounded windings,” IEEE
Trans. Power Delivery, vol. 19, no. 3, pp. 1328-1334, July 2004.

[4.35] M. Tiberg, D. Bormann, B. Gustavsen, and C. Heitz, “Generic and automated simulation modelling based
on measurements,” Proc. Int. Conf. Power Systems Transients, Lyon, France, June 4-7, 2007, 6 p.

[4.36] M. Popov, L. van der Sluis, and R.P.P. Smeets, “Evaluation of surge transferred overvoltages”, Electric
Power Systems Research, vol. 78, no. 3, pp. 441-449, 2008.

[4.37] A. Borghetti, A. Morched, F. Napolitano, C.A. Nucci, and M. Paolone, “Lightning-induced overvoltages
transferred through distribution power transformers”, IEEE Trans. Power Delivery, vol. 24, no. 1, pp. 360-
372, Jan 2009.

[4.38] B. Gustavsen, “Study of transformer resonant overvoltages caused by cable-transformer high-frequency

interaction”, IEEE Trans. Power Delivery, vol. 25, no. 2, pp. 770-779, April 2010.

[4.39] B. Gustavsen, “A hybrid measurement approach for wide-band characterization and modelling of power
transformers”, IEEE Trans. Power Delivery, vol. 25, no. 3, pp. 1932-1939, July 2010.

Page 57
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

[4.40] E.C. Levy, “Complex curve fitting”, IRE Trans. Automatic Control, vol. 4, pp. 37-44, May 1959.

[4.41] C.K. Sanathanan and J. Koerner, “Transfer function synthesis as a ratio of two complex polynomials”, IEEE
Trans. Automatic Control, vol. 8, pp. 56-58, 1963.

[4.42] B. Gustavsen and A. Semlyen, “Rational approximation of frequency domain responses by vector fitting”,
IEEE Trans. Power Delivery, vol. 14, no. 3, pp. 1052-1061, July 1999.

[4.43] B. Gustavsen, “Improving the pole relocating properties of vector fitting”, IEEE Trans. Power Delivery, vol.
21, no. 3, pp. 1587-1592, July 2006.

[4.44] D. Deschrijver, B. Haegeman, and T. Dhaene, “Orthonormal vector fitting: a robust macromodelling tool for
rational approximation of frequency domain responses”, IEEE Trans. Advanced Packaging, vol. 30, no. 2,
pp. 216-225, May 2007.

[4.45] D. Deschrijver, M. Mrozowski, T. Dhaene, and D. De Zutter, “Macromodelling of multiport systems using a
fast implementation of the vector fitting method”, IEEE Microwave and Wireless Components Letters, vol.
18, no. 6, pp. 383-385, June 2008.

[4.46] B. Gustavsen and A. Semlyen, “Enforcing passivity for admittance matrices approximated by rational
functions”, IEEE Trans. Power Systems, vol. 16, pp. 97-104, Feb. 2001.

[4.47] D. Saraswat, R. Achar, and M. Nakhla, “A fast algorithm and practical considerations for passive
macromodelling of measured/simulated data,” IEEE Trans. Compon. Packag. Manufact. Technol., vol. 27,
no. 1, pp. 57–70, Feb. 2004.

[4.48] S. Grivet-Talocia, “Passivity enforcement via perturbation of Hamiltonian matrices,” IEEE Trans. Circuits
Syst. I, Reg. Papers, vol. 51, no. 9, pp. 1755–1769, Sep. 2004.

[4.49] B. Gustavsen, “Fast passivity enforcement for pole-residue models by perturbation of residue matrix
eigenvalues”, IEEE Trans. Power Delivery, vol. 23, no. 4, pp. 2278-2285, Oct 2008.

[4.50] Song Gao; Yu-Shan Li; Mu-Shui Zhang; , "An efficient algebraic method for the passivity enforcement of
macromodels," IEEE Trans. Microwave Theory and Techniques, vol.58, no.7, pp.1830-1839, July 2010.

[4.51] B. Gustavsen, “Computer code for rational approximation of frequency dependent admittance matrices”,
IEEE Trans. Power Delivery, vol. 17, no. 4, pp. 1093-1098, October 2002.

[4.52] A. Semlyen and A. Dabuleanu, “Fast and accurate switching transient calculations on transmission lines
with ground return using recursive convolutions”, IEEE Trans. Power Apparatus and Systems, vol. 94, pp.
561-575, March/April 1975.

[4.53] B. Gustavsen, and O. Mo, “Interfacing convolution based linear models to an electromagnetic transients
program”, Proc. Int. Conf. Power Systems Transients, Lyon, France, June 4-7, 2007, 6 p.

[4.54] A.C.S. Lima, B. Gustavsen, and A.B. Fernandes, “Inaccuracies in fitted frequency dependent networks due
to finite precision of RLC branches”, Proc. Int. Conf. Power Systems Transients, Lyon, France, June 4-7,
2007, 5p.

[4.55] B. Gustavsen, A.P. Brede and J.O. Tande, “Multivariate analysis of transformer resonant overvoltages in
power stations”, IEEE Trans. Power Delivery, TPWRD-00912-2010, in press.

[4.56] S.D.Mitchell, J.S.Welsh, “Modelling Power Transformers to Support the Interpretation of Frequency
Response Analysis”, IEEE Trans. Power Delivery, vol.26, no.4, pp.2705-2717, October 2011

[4.57] C.C Brozio, H.J. Vermeulen, "Wideband equivalent circuit modelling and parameter estimation
methodology for two-winding transformers," Generation, Transmission and Distribution, IEE Proceedings,
vol.150, no.4, pp. 487- 492, 14 July 2003

Page 58
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

[4.58] G.B. Gharehpetian, H. Mohseni, K. Moller, "Hybrid modelling of inhomogeneous transformer winding for
very fast transient overvoltage studies," IEEE Trans. Power Delivery, vol.13, no.1, pp.157-163, Jan 1998

[4.59] K. Ragavan and L. Satish, “Construction of physically realizable driving point function from measured
frequency response data on a model winding,” IEEE Trans. Power Delivery, vol. 23, no. 2, pp.760 –767,
april 2008.

[4.60] S. Xu, S. D. Mitchell, and R. H. Middleton, "Partial Discharge Localization for a Transformer Based on
Frequency Spectrum Analysis," in AUPEC Christchurch New Zealand, 2003.

[4.61] L. F. Blume and A. Boyajian. “Abnormal voltages within transformers”. American Institute of Electrical
Engineers, Transactions of the, XXXVIII(1):577 –620, jan.1919.

[4.62] P. A. Abetti, “Transformer models for the determination of transient voltages,” Power Apparatus and
Systems, Part III. Transactions of the American Institute of Electrical Engineers, vol. 72, no. 2, pp. 468–
480, Jan. 1953.

[4.63] J. H. McWhirter, C. D. Fahrnkopf, and J. H. Steele, “Determination of impulse stresses within transformer
windings by computers,” Power Apparatus and Systems, Part III. Transactions of the American Institute of
Electrical Engineers, vol. 75, no. 3, pp. 1267 –1274, jan. 1956.

[4.64] P. Fergestad and T. Henriksen, “Transient oscillations in multiwinding transformers,” Power Apparatus and
Systems, IEEE Transactions on, vol. PAS-93, no. 2, pp. 500 –509, march 1974.

[4.65] W. Mcnutt, T. Blalock, and R. Hinton, “Response of transformer windings to system transient voltages,”
Power Apparatus and Systems, IEEE Transactions on, vol. PAS-93, no. 2, pp. 457–467, March 1974.

[4.66] D. Wilcox, W. Hurley, T. McHale, and M. Conlon, “Application of modified modal theory in the modelling of
practical transformers,” Generation, Transmission and Distribution, IEE Proceedings C, vol. 139, no. 6, pp.
513 –520, nov 1992.

[4.67] F. de Leon and A. Semlyen, “Complete transformer model for electromagnetic transients,” Power Delivery,
IEEE Transactions on, vol. 9, no. 1, pp. 231–239, Jan 1994.

[4.68] E. Dick and C. Erven, “Transformer diagnostic testing by frequency response analysis,” Power Apparatus
and Systems, IEEE Transactions on, vol. PAS-97, no. 6, pp. 2144–2153, Nov. 1978.

[4.69] “Mechanical condition assessment of transformer windings using frequency response analysis (FRA),”
ELECTRA-CIGRE WG A2.26 Report 342, vol. 237, April 2008.

[4.70] A. Shintemirov, W. Tang, and Q. Wu, “Transformer core parameter identification using frequency response
analysis,” Magnetics, IEEE Transactions on, vol. 46, no. 1, pp. 141 –149, jan. 2010.

[4.71] A. Shintemirov, W. Tang, W. Tang, and Q. Wu, “Improved modelling of power transformer winding using
bacterial swarming algorithm and frequency response analysis,” Electric Power Systems Research, vol. 80,
no. 9, pp. 1111 – 1120, 2010.

[4.72] G. Zambrano, A. Ferreira, and L. Caloba, “Power transformer equivalent circuit identification by artificial
neural network using frequency response analysis,” in Power Engineering Society General Meeting, 2006.
IEEE, 0-0 2006, p. 6 pp.

[4.73] S.D.Mitchell, J.S.Welsh, “The Influence of Complex Permeability on the Broadband Frequency Response
of a Power Transformer”, IEEE Trans. Power Delivery, vol.25, no.2, pp.803-813, April 2010.

Page 59
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

5 NETWORK INTERACTION WITH TRANSFORMER

5.1 General
The withstand capability of the transformer insulation systems against overvoltages in HV and EHV networks is
only tested with a standard lightning impulse voltage or a standard switching impulse voltage. Such standard test
voltages do however not ensure sufficient withstand against oscillating overvoltages. It is an accepted view that
resonant overvoltages can lead to transformer dielectric failure by exciting internal resonances in the transformer,
even when the overvoltage magnitude is much lower than the peak value of the standard test voltage [5.1]. In fact,
the use surge arresters at the transformer's terminals are usually not an effective measure to protect against
oscillating overvoltages.

The following sections explain the phenomena which may cause overvoltages to build up inside a transformer by
resonance, with some reference to actual failures.

• Sections 5.2−5.4 gives an introduction to the topic by explaining the nature of resonance and how it applies
to a transformer and its interaction with the network. The text covers transformer external and internal
overvoltages as well as transferred voltages between windings.

• Section 5.5−5.8 describe phenomena that may potentially lead to resonant overvoltages in transformers:
operation of circuit breakers and disconnectors, fault initiation and lightning discharges. The significance of
the network topology is emphasized.

• Section 5.9 discusses the importance of the phenomena in terms of frequency of occurrence.

5.2 Introduction to overvoltages in transformers and resonant conditions

5.2.1 Resonance
In general terms, ‘resonance’ describes the condition where the frequency of an applied stationary voltage or
current source coincides with one of the natural frequencies of a complex, passive, frequency dependent network
function, for instance a driving point impedance or admittance function. In resonance, the network function attains a
real value, i.e. the imaginary part vanishes. This means that the corresponding RLC network consumes only real
power. As a result, the value of current or voltage at the network terminals reaches a (local) maximum. As seen
from the outside, this state can be regarded as a ‘resonance’ condition. On the other hand, RLC networks may
reach ‘resonance’ states at certain frequencies at which the voltage (or current) over certain branches or at certain
nodes inside of the network reaches a maximum value. Due to damping effects and superposition of resonances,
the exact frequencies of ‘inside’ and ‘outside’ resonances do in general not coincide. Regarding the interaction
between a transformer, as an oscillatory network, and the power grid as the source of excitation, the aspect of
inner resonances is of great importance. Any built transformer naturally features one or more inner resonance
frequencies, at which

• The voltage distribution across the windings is non-uniform,

• Currents or voltages inside the windings reach values greater than the steady state proportion of a harmonic
excitation signal.
In practice, the transient voltages to which the transformer is exposed, contains a spectrum of superimposed
oscillations. If the transient contains a strong oscillation whose frequency is matching one of the ‘inner’ resonance
frequencies of the winding, very high internal overvoltages can occur inside the transformer. Clearly, also the
adjacent network is of importance to the transient interaction, e.g. the length of transformer feeder lines which
directly influences the dominating frequency of oscillatory transients.

Page 60
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

5.2.2 Resonant interaction between transformer and external system

One important aspect of resonance is the interaction between the transformer and the network where energy is
interchanged over the transformer's terminals. The phenomenon itself can be explained by the circuit in Figure 5.1
(left panel) where two circuits, circuit #1 (L1-C1) and circuit #2 (L2-C2) are connected in series. In terms of
transformer resonant overvoltages, circuit #1 would represent the external network (e.g. a feeder cable) and circuit
#2 is the transformer. A unit step voltage is applied to circuit #1. Very high overvoltages can now result at point C if
the following conditions are met:

1. The two circuits have approximately the same natural frequency,

1 1
≈ (5.1)
L1C1 L2C2
2. Circuit #2 has a much higher surge impedance than circuit #1,

L2 L
>> 1 (5.2)
C2 C1
For the given example we use the circuit parameters and associated parameters listed in Table 5.1 which meet the
two conditions (5.1) and (5.2). The right panel of Figure 5.1 shows the simulation result when applying a unit step
voltage to point A. Energy is now exchanged back and forth between the two circuits, causing the observed beat
phenomenon. The voltage reaches in this is case a maximum of 8 V. Modifying the circuit parameters to increase
the surge impedance of circuit #2 (or lower that of circuit #1) leads to even higher overvoltages.

L C 1 L
LC C
Circuit #1 1 mH 1 µF 31623 rad/s 31.6 Ω
Circuit #2 50 mH 0.02 µF 31623 rad/s 1581 Ω
Table 5.1: Circuit parameters for Figure 5.1.

A L1 B L2 C

C1 C2

Figure 5.1: Resonance between two circuits conforming with the conditions of (5.3.2) and
(5.3.3), circuit #1 (L1-C1) and circuit #2 (L2-C2). Left panel: exciting circuit #1 with a unit step
voltage; Right panel: Voltage response at points A, B, and C [5.2]. (© 2010 IEEE).

In reality, the situation is a lot more complex as the transformer is essentially a highly complex RLC network but the
same principles apply and excessive resonant overvoltages on a transformer (internally, at taps, or at other winding

Page 61
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

terminals) can only result when

• An oscillating overvoltage appears on one or more terminals with a frequency f0 that coincides with a peak in
the voltage transfer function from the external point to some internal point, or other external points.

• The transformer input impedance at f0 is much higher than the output impedance of the feeding network.

5.3 Transformer external overvoltages

5.3.1 Impinging overvoltage and voltage transfer between windings
Overvoltages travel as waves along transmission lines, i.e. overhead lines and underground cables. When an
overvoltage meets a transformer, part of the voltage wave becomes reflected back while another part appears on
the other winding(s) as a transmitted wave. This is illustrated in Figure 5.2 where an incoming voltage wave enters
from the left towards winding #1.

Figure 5.2: Incoming, reflected and transferred voltage wave.

The reflection and transmitting of waves is strongly dependent on the transformer impedances as seen from its
terminals, relative to the characteristic impedance of the connected lines. In general, the finite impedance of a
transformer will contribute to reducing the front steepness of an incoming wave.

The impact of the transformer impedance on overvoltage front steepness and transferred overvoltage is
investigated by the example in Figure 5.3 where a traveling unit step voltage meets terminal 1a on the HV side of a
distribution transformer. The transformer is at all terminals connected to a transmission line of semi-infinite length,
of either 30 Ω (cable) or 400 Ω (overhead line). The resulting overvoltage on terminal 1a (HV side) and 2a (LV side)
are shown in Figure 5.4. Clearly, the front steepness of the impinging overvoltage at 1a (left panel) decreases with
higher transmission line characteristic impedance. Also, the transferred voltage (right panel) is affected by the line
characteristic impedance. The oscillations observed in Figure 5.4 are related to the characteristic resonances of the
transformer. In the simulations, the transmission lines are assumed uncoupled, lossless and distortionless. The
transformer is represented by a wide-band black-box model of a 300 kVA distribution transformer that was
extracted from frequency sweep measurements [5.2].

1V 300 kVA
1a 2a
1b 2b
1c 2c
11 kV 0.23 kV

Figure 5.3: Unit step voltage arriving on phase "a" at transformer HV side. Transformer
connected to semi-infinite transmission lines.

Page 62
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Figure 5.4: Resulting voltage at transformer terminals. Parameter: transmission line

characteristic impedance. Left panel: voltage on terminal 1a (HV side); Right panel: voltage on
terminal 2a (LV side).

5.3.2 External resonance behavior explained by simplified transformer model

The following gives an explanation of the resonant behavior of transformers by means of a simplified transformer
model, see Figure 5.5. The model parameters have been derived from a real transformer (220 MVA, 400 kV/115
kV, relative short-circuit voltage 12%).

Figure 5.5: Simplified transformer model with capacitances lumped at terminals.

Figure 5.6 shows the differential input admittance at the HV side with alternative loads connected (differentially) at
the LV side. Clearly, the observed input admittance is strongly dependent on the LV side loading, and therefore on
the terminal conditions in general. Therefore, a single frequency sweep measurement does not fully characterize
the transformer with respect to resonances.

Page 63
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

-1
10
RL = 50 Ω

-2
RL = 100 Ω
10
RL = 500 Ω
RL = 1k Ω
|Y(f)| in 1/Ω

-3
10 RL = 10k Ω
RL = open
-4
10

-5
10

-6
10 2 3 4 5 6
10 10 10 10 10
Frequency f in Hz

Figure 5.6: Input admittance at transformer HV side with alternative loads on the LV side.

The real part of the input admittance reveals the frequency ranges where the active power consumed by the
transformer reaches a maximum, i.e.

max{P ( jω )} = U 2 ⋅ max{Re{Y ( jω )}} (5.3)

Dangerous resonances may occur at frequencies where the real part is maximum, since there will be large currents
oscillating internally in the transformer.

Figure 5.7 shows the input admittance of the transformer by its magnitude and real part, for the case with open LV
side. If the transformer is stimulated by transient voltages containing significant signal energy in frequency ranges
matching resonance areas, voltages greater than the input voltage may occur, leading to excessive electrical field
stress of the winding insulation.
-1
10
|Y(f)| (RL = open)

10
-2 Re{Y(f)} (RL = open)
|Y(f)| in 1/Ω

-3
10

-4
10

-5
10

-6
10 2 3 4 5 6
10 10 10 10 10
Frequency f in Hz

Figure 5.7: Input admittance of HV winding (magnitude and real part).

Page 64
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Evaluation of the transformer impedance and/or voltage transfer functions alone may give an ‘idea’ about potential
transformer overvoltage related to resonances. The actual spatial voltage distribution within the winding however
can only be calculated with a detailed white box model requiring precise knowledge about the geometry of the
winding.

5.3.3 Resonant interaction between the transformer and a feeder cable

It is important to recognize that the transformer by its impedance affects the impinging overvoltage at its terminals
and therefore the occurrence of resonant overvoltages. To see this, we revert to the wide-band transformer model
of a distribution transformer in Section 5.3.1. This transformer is seen to have a peak in its voltage transfer function
at about 2 MHz, see Figure 5.8.

Figure 5.8: Voltage transfer from HV side to LV side, with open LV side. Diagonal element:
Voltage transfer from a phase on HV side to same phase on the LV side, with all other HV
phases grounded. Off-diagonal element: Voltage transfer from a phase on HV side to a different
phase on the LV side, with all other HV phases grounded.

One important source of resonant overvoltage is due to the switching of a connected cable. In Figure 5.9, a unit
step voltage is applied to the end of a cable which is connected to the HV side of the transformer with the LV side
open. The cable length is chosen such that its quarter wave resonance frequency matches the peak of the voltage
transfer function from high to low at about 2 MHz by (5.1). This implies a cable length of about 20 m by (5.4) where
τ is the cable travel time. The cable is lossless and has a characteristic impedance of 13.3 Ω.

1 v
fλ / 4 = = (5.4)
4τ 4l

In general the dominant oscillation frequency of system “feeder cable – transformer” is less than quarter wave
resonance frequency because of influence of transformer surge impedance (capacitance). The lesser the cable
length and bigger the surge capacitance of transformer, the bigger the difference of dominant frequency of this
system from quarter wave resonance frequency.

Figure 5.10 shows a simulation of the voltage wave form on the energized phase, at the HV side (1a) and LV side
(2a). The left panel shows the result when the loading effect of the transformer is ignored. The voltage at the cable
end (1a) is a square wave which alternates between zero and 2 V. The voltage at the LV side of the transformer
(2a) builds up to about 1.9 V by resonance, which is consistent with the highest peak in the frequency domain plot
of Figure 5.8. The right panel shows the same result when the transformer input impedance is taken into account.
Clearly, the input impedance reduces the voltage at the cable end and thereby limits the voltage build-up to a much
lower peak value, from 1.9 V down to below 1.0 V.

Page 65
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

1V
300 kVA
Cable, 20 m 1a 2a
1b 2b
1c 2c
11 kV 0.23 kV

Figure 5.9: Unit step voltage excitation of lossless cable that is connected to a transformer.

Figure 5.10: Unit step excitation of lossless cable connected to LC circuit. Left panel: ignoring the
transformer impedance; right panel: taking the transformer impedance into account.

The voltage buildup shown in Figure 5.10 can only be observed if the input impedance of the transformer is much
higher than the cable characteristic impedance. Otherwise, the cable end voltage becomes damped as energy is
transferred from the cable to the transformer. This impact is quantified using the circuit in Figure 5.11 where the LC
circuit represents the transformer. L and C is chosen such that the resonant frequency is f0 =400 kHz while the
surge impedance is either 3 kΩ or 300 Ω. The cable quarter wave resonance frequency is chosen equal to that of
the transformer (400 kHz) and its characteristic impedance is 30 Ω. Figure 5.12 shows the step voltage response at
point C. It is observed that decreasing the transformer characteristic impedance from 3 kΩ to 300 Ω greatly
reduces the voltage buildup. These cases correspond to the transformer input impedance being respectively 100 or
10 times greater than the cable characteristic impedance.

Figure 5.11: Unit step excitation of lossless cable connected to LC circuit.

Page 66
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Figure 5.12: Voltage on capacitor (point c) with alternative values for the LC circuit characteristic
impedance.
Left panel: 3000 Ω impedance; right panel: 300 Ω impedance.

5.4 Transformer internal overvoltages

This section presents the basic parameters considered in the representation of a transformer when facing transient
overvoltages generated by the system and a concise description of how these overvoltages are distributed inside
the transformer. The term “transformer” is used generically in this chapter but all the concepts may also apply to
auto transformers and reactors.

An equivalent circuit of the transformer is created by subdividing the winding into a cascade of equivalent RLC
circuits. The RLC circuit parameters are calculated considering the ohmic value of resistance between the
terminals of each winding, the value of self-inductance of each coil, the values of mutual inductances between the
coils, the capacitance values found along each coil (series capacitances - Cs) and the values of capacitances
between adjacent coils and between coils and grounded structural parts of the transformer (capacitances to earth -
Cg).

An example of such winding model is illustrated in Figure 5.13, excluding mutual inductances. In this example each
of the windings is divided into six branches or inductive elements. By increasing the number of circuits by finer
subdivisions, the accuracy can be extended to higher frequency.

Page 67
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Figure 5.13: Example of a simplified equivalent circuit for a transformer submitted to a voltage
transient (where: Núcleo = Core, Tanque = Tank, Bobina BT = LV Winding, Bobina AT = HV
Winding).

In the case of a very steep-fronted voltage impinging the transformer terminal, the current flows initially through the
capacitive elements of the coil, which results in an initial stress distribution determined primarily by the
capacitances of the circuit. Figure 5.14 shows a simplified representation of a coil of the transformer for the initial
instants of the transient.

Figure 5.14: Capacitive circuit (coil in the initial instant of the transient).

Page 68
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

The initial voltage distribution along the length x of a coil having total length l, for a voltage step function applied to
the coil, is well represented by the curves of Figure 5.15 [5.3] where:

CG TOTAL
α= (5.5)
CS TOTAL

Figure 5.15: Initial voltage distribution in the capacitive model of a coil with a grounded end.

When the voltage distribution in the windings is uniform, the voltage gradients near the terminal reached by the
transient will be less concentrated. Therefore it is desirable that the value of the constant α is minimized. As Cg is
practically determined by a compromise between minimum acceptable electrical distances and maximum
mechanical distances consistent with a technically and economically feasible design, Cs becomes the main
variable to control the distribution of initial voltage. The optimization of Cs may require the adoption of more
complex types of windings such as windings with interleaved turns or containing internal shields.

After the initial instants of the transient, for a voltage step applied to the coil, the current through the inductive
elements increases and the voltage wave propagates in the winding, together with the typical oscillations of an
RLC circuit in time.

In other words, oscillating with different frequencies, the voltage along the winding has different amplitude at every
moment (Figure 5.16) [5.3]. These voltages oscillate around the value corresponding to the final (or inductive)
distribution (t → ∞), and the closer the initial distribution is to the final distribution the smaller these voltages will be.
Here again, with the oscillatory overvoltages, it is desirable that the value of the constant α is minimized.

To include the core effects in the model in the low frequency range, it is possible to add a common flux to all
inductive elements of the circuit. This common flux is needed for the model to represent the power frequency
behavior.

Page 69
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Figure 5.16: Voltage oscillations in a coil with a grounded end (α ≈ 10).

This oscillatory response can be obtained analytically, in theory, through traveling waves and stationary waves, for
a uniform winding. In practice, however, there are complex windings presenting discontinuities. For example,
regions of voltage taps or even where the constructive type of the winding is modified in order to minimize the
impact of transient overvoltages. Such discontinuities are easily handled with the adopted numerical solution
approaches based on an equivalent circuit with concentrated parameters (see Figure 5.14) that represent exactly
such discontinuities. It is observed, however, that this concentrated parameters model is limited to certain
frequency range (maximum up to 500 kHz) due to the difficulties to calculate parameters to build and to solve a
turn to turn model.

5.5 Resonant overvoltages: Topologies, switching operations, faults

All transformers have resonances in the range of tenths of kHz to hundreds of kHz, so any event in the system
which generates frequencies in this range can potentially provoke internal overvoltages in the transformer. This
section discusses situations that can lead to oscillating overvoltages on the transformer terminals, thereby
representing a risk to the transformer insulation system via resonant voltage magnification.

5.5.1 Cable-transformer networks

A significant proportion of transformer dielectric failures caused by overvoltages can be related to switching
operations by a circuit-breaker that is separated from the transformer by a cable of short length. To explain how
such configurations can lead to failure we use variants of the topology in Figure 5.17. A load transformer is
connected to the substation busbar via a feeding cable and a circuit breaker. Several other cables are possibly
connected to the same busbar. The substation is fed from a transmission supply.

Page 70
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Figure 5.17: Description of the electrical network considered [5.1] (© IEEE). IEEE Std C57.142 -
2010. Reprinted with permission from IEEE, Copyright 2010 IEEE. All rights reserved.

5.5.2 Phenomena caused by closing a circuit-breaker

• When energizing the transformer via a cable from a bus where n other cables are connected, with the circuit-
breaker involved located at a certain distance of the transformer at the upstream network side, the bus
appears initially as an ideal step voltage source behind an impedance Zc/n, where Zc is the cable
characteristic impedance. In the case with many installed cables, a travelling wave will be observed on the
primary side of the load transformer provided that the cable between the transformer and bus is substantially
shorter than the other cables. The frequency f0 of the phenomena involved is closely related to the cable
traveling time by (5.4). Transformer failure can result if the frequency of the travelling wave, which is
generated when closing the circuit-breaker poles, matches an internal resonance frequency f0 in the
transformer and is of sufficient magnitude and duration [5.1]-[5.8].

Figure 5.18 shows an example of measurements made on a 200 MVA single-phase autotransformer [5.4].
The voltage transfer from the high-voltage terminal to the (differential) voltage across the tap range was
measured with a 400 Ω loading which represents the characteristic impedance of a long overhead line.

It can be seen that the voltage is about 13 times higher at resonance (29 kHz) than at 60 Hz. The lower plot
shows that the input impedance at 29 kHz is around 3 kΩ, which is 100 times higher than that of the
connected cable (30 Ω). The 3 kΩ input impedance is too high to damp out a 29 kHz oscillation on the
cable, see also simple analysis in Section 5.2.2 and simulation result in Figure 5.12.

Page 71
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Figure 5.18: Measurement of voltage transfer from transformer high-voltage terminal to winding
mid-point. Upper right plot: Voltage transfer function; Lower right plot: Input impedance. [5.4].
(© 1974 IEEE)
• Switching a capacitor bank directly to the bus gives effectively a temporary short circuit of the involved
phases. The resulting step-like voltage application to the bus leads to oscillating overvoltages on connected
cables, not much different from the situation of cable energization. Figure 5.19 shows simulation results
using a wide-band distribution transformer which has a peak in the voltage transfer function at about 2 MHz,
see Figure 5.8. The transformer is connected to the busbar by a short cable whose length is chosen such
that the cable quarter-wave resonance frequency (5.4) is about 2 MHz, matching the peak in the voltage
transfer. Energizing the capacitor bank effectively gives a short circuit at the transformer's high-voltage side,
causing a semi-stationary resonant oscillating voltage on the transformer HV side (V3) which gives an
excessive overvoltage on the LV side by resonance. Since the nominal voltage ratio is 48:1, a 1 p.u. voltage
on the LV side implies a voltage magnification by a factor 48. It is noted that the cable voltage (V3) is
substantially attenuated due to the loading effect of the transformer. Without this loading effect (infinite
transformer impedance), the transformer voltage on the LV side would reach a much higher value as
demonstrated in Section 5.3.3.

Page 72
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Figure 5.19: Resonant overvoltage build-up in transformer LV side (V4, V5, V6) caused by
energizing capacitor bank by closing one of the breaker poles. Circuit breaker represented by
ideal switch [5.2]. (© 2010 IEEE).
• Internal switching in GIS can give rise to very high frequency transients which are oscillatory in nature.
Figure 5.20 shows an example of a measured transient overvoltage whose dominant resonance frequency is
about 1 MHz. We refer to [5.9] for an in-depth discussion of very high frequency phenomena associated with
GIS and their impact on apparatuses including transformers.

Figure 5.20: Measurement transient voltage at load end in GIS [5.9]. (© 1988 CIGRE).

5.5.3 Phenomena caused by fault initiation

• Oscillating overvoltages can also result from line faults on a connected overhead line or cable as the voltage
changes from its steady state value to zero while the faulty conditions occur in the network. In that case, the
fault causes a voltage wave to propagate towards the transformer which is reflected back and forth between
the transformer and the fault location. This phenomenon is shown in Figure 5.21, in the case of a fault
occurring on the EHV network and propagating towards a 1650 MVA transformer.

Page 73
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Figure 5.21: Phase to ground overvoltages appearing at the entry of the transformer.
• Depending on the distance to the location that a fault occurred, the resonant voltage on the transformer
terminals can coincide with an internal resonance frequency of the transformer. In the case of high short
circuit powers, the fault current may continue flowing through the arc during the transient event without
disruption so that the fault effectively represents a solid grounding. In this situation, the overvoltages become
maximum when the frequency of the traveling waves matches the internal resonant frequency of the
transformer [5.10]. From (5.4), this leads to the critical distance X between the transformer and the fault:
v
X= (5.6)
4 f0

where fo is a resonant frequency of the transformer, v is the velocity of the traveling wave, and X is the
distance from the transformer to the fault point. Example: when the transformer has two resonant
frequencies of 6.5 kHz and 13.3 kHz, the location of the fault from the transformer leading to the most severe
overvoltages is 11.5 km and 5.6 km, respectively, for an overhead line with propagation velocity of 300 m/µs.

• AEP experienced several autotransformer transformer failures in the late 1960s and early 1970s due to
distant line faults [5.11]-[5.12]. Resonance of a few kHz was identified as a likely cause for the failures.

• Internal faults in GIS can give rise to very high frequency oscillating overvoltages not much different from the
switching event. The amplitude of the oscillating component in the internal faults in GIS is likely to be higher
than that in the case of a switching event since a 1 p.u. voltage excitation results at the fault location as the
voltage instantly drops from 1 p.u. to zero. Also, the dominant frequency component depends on the fault
location.

5.5.4 Phenomena caused by circuit-breaker opening

• When the breaker contacts start separating, the arc will burn between the contacts until the current passes
through the current natural zero point. Normally, the arc will become unstable and current chopping takes
place shortly before the current zero point. With modern breakers the current chopping level can be kept
below a few Amperes. Following the current chopping, a voltage buildup takes place between the contacts
due to voltage oscillations on both sides of the breaker. When breaking the inductive currents, the voltage
oscillations can reach levels where they can damage the transformer via resonance. Examples of inductive
loads include

o A transformer loaded with a reactor

o A transformer loaded with stalled motors
The overvoltages across the breaker following current chopping may reach such levels that a restrike takes
place across the breaker poles, giving rise to further voltages. Such phenomena are described in detail in
Section 5.6.

Page 74
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

5.6 Restrikes and circuit breaker technologies

The current chopping level is below 5 A with vacuum circuit breakers (VCB) and below 1-2 A with SF6 breakers
[5.1]. The actual chopping level depends on the contact material for the VCB technologies, and on the breaker
parallel capacitance as well as the forced cooling process for the SF6 technologies. The chopping level also
depends on the contact distance when the decreasing current approaches the prospective current zero.

If the transient recovery voltage across the breaker exceeds the breaker dielectric withstand capability, a restrike
takes place and another transient event occurs. The interruption of the current may take place at the next natural
current zero. Unlike SF6 breakers, vacuum circuit breakers are able to interrupt fast oscillating currents, sometimes
causing current interruption during the transient passage following a restrike.

Repeated restrikes/chopping during an opening event can lead to an escalating voltage on the load side of the
circuit-breaker as the contact distance and hence the dielectric strength increase with time (Figure 5.22). Such
escalating voltages tend to be sawtooth-like in appearance and their peak value can exceed the system operating
voltage. As the results, very high internal transformer overvoltages may be triggered by resonance. An
experimental study of the phenomenon by using a distribution transformer is presented in [5.5].

During closing operations, the breaker will show flashover at some points before reaching to the metallic contact.
This is called the prestrike. In the case of VCB technologies, the resulting high-frequency current may become
interrupted and several prestrikes/current interruptions may occur in rapid succession. Since the contacts are
closing, the voltage on the load side cannot escalate but it could still cause transformer failure by triggering a
resonance.

Figure 5.22: Voltage escalation due to multiple re-ignitions in a circuit-breaker [5.22] (© 2010
CRC Press).

5.7 Disconnector switching

The operation of a disconnector leads to a large number of sparkovers across the switch contacts [5.13]-[5.15].
Disconnector switching inside GIS has been reported to cause transformer failures (for instance, see the
mentioning to Tucurui HPP Substation in [5.9]).

A dielectric testing method, developed in a high-voltage laboratory, [5.17] was employed in order to check the
reliability of disconnector operation in the worst conditions, as shown in Figure 5.23. In the figure, B is a bushing, D
is a disconnector, B.S. is a GIS bus, C.B. is a circuit breaker, and S1 and S2 are voltage sensors.

Page 75
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Figure 5.23: Description of the test setup.

In this configuration, the circuit-breaker is closed. The transient voltages generated by the switching of the
disconnector were recorded both in the power frequency and in the high frequency range (tens of MHz). Figure
5.24 shows an example of overvoltages recorded during the opening sequence of the circuit:

Figure 5.24: Voltages appearing during disconnector opening with VT present (C.B.
closed).Upper track: source voltage; Lower track: load side voltage
Because of the phase and frequency differences between the load and the source voltages, in the extreme case
(∆U = 2 p.u.), restrikes may occur for full phase opposition. This aspect is observed near the end of the operation
(see figure above), where the final restrike takes place for ∆U equal to 1.9 p.u.

The modelling of the GIS components in the high frequency range is an important issue for the determination of the
overvoltages generated by switching at the location of the disconnector. The equipment is modelled by its
capacitance and the busbar sections are modelled by their surge impedances and time delays [5.18]. Figure 5.25
shows the compared results between the measured voltage in S2 and the calculated one at the first strike of a
disconnector closing.

Page 76
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Figure 5.25: Comparison between the measured voltage in S 2 (solid line) and the calculated one
(dotted line) at the first strike of a disconnector closing.
In real GIS, the amplitude of the overvoltages may reach 2.5 p.u. Those overvoltages, having a rise-time of the
order of 5 ns since frequencies of a few tens of MHz are generated, may be injected into the transformers
connected to the GIS substation.

Failures have also been reported in air-insulated substations due to disconnector operation. An experience at 525
kV and 787 kV air-insulated substations conducted by Russian research team has also shown that the operation of
disconnectors may have contributed to transformer failure by VFTO [5.19]. One particularly vulnerable configuration
is that of Figure 5.26 where a bus is being disconnected using a disconnector [5.19]. Here, the occurrence of
multiple restrikes causes high-frequency oscillating voltages on both sides of the disconnector, which pose a risk to
the transformer insulation through resonance. One of the case studies also describes the case of the air-insulated
“Tres Marias substation” in Brazil, with the analysis of transformer overvoltages resulting from the operation of a
disconnector.

To line

Live
network

To line

Figure 5.26: Disconnecting a bus using disconnector switch [5.19]. The part of the circuit taking
part in the transient event is indicated by solid lines.

5.8 Lightning overvoltages

Another source of overvoltages is the lightning discharge. The lightning flash consists of one or more return stroke
currents that are injected at the flash location. Overhead lines are normally fitted with overhead ground wires that
act as a lightning shield, at least for the spans close to a substation. As explained in Section 3.3.3, the actual flash
location (ground, ground wires, tower, and phase conductors) can be determined based on the line geometry in
combination with the lightning "striking distance" which is a function of the first stroke peak current. From this
Electro Geometrical Model it follows that only flashes with a small first stroke peak current can terminate directly to
a phase conductor, so-called shielding failure. For well-designed overhead lines, the maximum current due to

Page 77
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

shielding failure is quite small, e.g. below a few kA, but subsequent stroke currents may have a higher peak value.
A shielding failure causes an initially unipolar voltage wave to start propagating in both directions along the
overhead line with a peak voltage of about 400 kV, for a 2 kA stroke current and 400 Ω line characteristic
impedance. The voltage wave is subject to reflections as it propagates towards the substation, often given rise to a
waveshape at the transformer terminals with oscillatory components.

Return strokes with higher peak current values may terminate on the tower top or the ground wires. In the case of a
large peak current in combination with a small front time, the voltage on the tower top may reach such a high level
that a flashover takes place across the insulator strings to phase conductor(s), so-called backflashover. The actual
voltage wave shapes that result in substations are in general irregular, being aperiodic with some oscillatory
component. The phenomena by which the overvoltage results are highly complex, including multiple wave
reflections between the stroke locations and the tower footing groundings along the line. It is generally believed that
it is backflashovers close to the line entrance that are most likely to produce very high overvoltages at the
transformer terminals, but this is not necessarily the worst case scenario considering the transformer's vulnerability
to oscillatory overvoltages and their specific frequency content. In this report, transformer overvoltages due to
lightning backflashovers are analysed in two of the case studies in the Appendix.

5.9 Frequency of occurrence

The actual switching overvoltages which will occur for a given transformer, depend on a number of parameters
which can vary from one switching operation to the next. Examples are breaker statuses in the substation (in/out),
the transformer load current, the point-of-wave where the breaker contacts start separating, the stochastic nature of
the current interruption and breaker dielectric withstand, etc. In order for dangerous overvoltages to reach
dangerous levels in the transformer, the impinging overvoltage must in addition have a dominant frequency
component which matches a peak in the voltage transfer from the given terminal to some internal points along the
winding. It follows that excessive overvoltages in a given transformer is likely to occur in only a fraction of all
switching operations [5.1]. Therefore, frequently switched transformers are in general more likely to fail than seldom
switched transformers. The insulation withstand capability may possibly be reduced in the case of many repetitive
overvoltages. This topic is discussed in detail in Chapter 7. Oscillatory overvoltages on transformers also result
from nearby faults and lightning discharges. Such overvoltages are however rare events compared to switching
overvoltages.

5.10 Conclusion
This chapter has described resonant phenomena which may potentially lead to excessive dielectric stresses to the
transformer's insulation system.

The resonance phenomenon

• A transformer can be viewed as a complex circuit of connected RLC branches. As such the transformer has
a large number of resonances which are observed as peaks in the frequency response of node voltages and
branch voltages/currents. Internal resonances are not necessarily observable from the external terminals.
• The observed resonance frequencies at external terminals are strongly dependent on the terminal
conditions, i.e. their loading impedances.
• Transient events that result in an oscillating overvoltage at the transformer terminals may result in a voltage
buildup at points due to resonance, provided that the dominant frequency component of the impinging
voltage coincides with a peak in the voltage transfer to the considered point. The overvoltage can appear
o Internally (along windings)
o Externally (e.g. transferred voltage to other winding)
o Tap points
• The transformer may impact the impinging overvoltage by its input impedance, in some cases mitigating the
overvoltage. It follows that the transformer is at higher risk in the case that the resonance occurs at a
frequency where the transformer input impedance is much higher the output impedance of the connected

Page 78
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

system.
• Dielectric failures by resonance can result also when the transformer is protected by surge arresters.

Situations that may lead to high overvoltages by resonance

• Energizing a transformer from a bus by closing a circuit breaker may lead to a resonant overvoltage on the
switched side of the transformer when there is a cable between the breaker and the transformer. In general,
the oscillating component of the overvoltage becomes higher and of longer duration if there are many other
cables connected to the same bus.
• Similar oscillating overvoltages may result from ground fault initiation in or close to the substation, or
capacitor bank energization.
• In the past, transformers have failed due to ground fault initiation of connected overhead lines. In that case,
the oscillating overvoltage at the transformer terminals triggered a resonance associated with the regulating
winding.
• Usage of vacuum circuit switchgear may lead to a sequence of high-frequency repetitive oscillations due to
prestrikes or restrikes.
• Separating the transformer using a disconnector may lead to large number of restrikes. Such practice has
caused transformer failure both for GIS and AIS.
• Lightning discharges may lead to overvoltages at transformer HV terminals with an oscillatory component.

5.11 References
[5.1] IEEE Std C57.142-2010, IEEE Guide to describe the occurrence and mitigation of switching transients
induced by transformers, switching device, and system interaction.

[5.2] B. Gustavsen, “Study of transformer resonant overvoltages caused by cable-transformer high-frequency

interaction”, IEEE Trans. Power Delivery, vol. 25, no. 2, pp. 770-779, April 2010.

[5.3] “Interaction between transformers and the electric system with focus on high frequency electromagnetic
transients”, issued by Cigre Brasil JWG A2/C4-03 (original version in Portuguese), May 2011.

[5.4] W. J. McNutt, T. J. Bialock, and R. A. Hinton, "Response of transformer windings to system transient
voltages", IEEE Trans. Power Apparatus and Systems, vol. 93, no. 2, pp. 457-467, March 1974.

[5.5] M. Popov, L. van der Sluis, G.C. Paap, “Investigation of the Circuit Breaker Reignition Overvoltages caused
by no-load transformer switching surges”, European Transaction of Electric Power, ETEP Vol. 11, No. 6,
November/December 2001, pp. 413-422.

[5.6] G. C. Paap, A.A. Alkema, and L. Van der Sluis, L., “Overvoltages in power transformers caused by no-load
switching,” IEEE Trans. Power Delivery, vol.10, no.1, pp. 301-307, Jan 1995.

[5.7] R. Degeneff, W.J. McNutt, W. Neugebauer, J. Panek, "Transformer response to system switching
overvoltages", IEEE Trans. Power Apparatus and Systems, vol. 101, no. 6, pp. 1457-1470, June 1982.

[5.8] B. Gustavsen, A. P. Brede, and J.O. Tande, “Multivariate analysis of transformer resonant overvoltages in
power stations”, IEEE Trans. Power Delivery, vol. 26, no. 4, pp. 2563-2572, October 2011.

[5.9] CIGRE Working Group 33/13-09, "Very fast transient phenomena associated with gas insulated
substations", 1998 Session, August 28- September 3, 1988.

Page 79
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

[5.10] T. Adielson, A. Carlson, H. Margolis, J. Halladay, “Resonant overvoltages in EHV transformers -Modeling
and Application”, IEEE Transactions on Power Apparatus and Systems, vol. PAS-100, no. 7, pp. 3563 -
3572, July 1981.

[5.11] M. B. Margolis, J. M. Phelps. A. A. Carlomagno and A. J. McElroy, "Experience with part-winding

resonance in EHV auto-transformers: Diagnosis and corrective measures", IEEE Trans. Power Apparatus
and Systems, vol. 94, no. 4, pp. 1294-1300, July/August 1975.

[5.12] A. J. McElroy, "On the significance of recent EHV transformer failures involving winding resonance", IEEE
Trans. Power Apparatus and Systems, vol. 94, no. 4, pp. 1301-1316, July/August 1975.

[5.13] Juan Martinez-Velasco, Power System Transients. Parameter Determination, CRC Press, 2010.

[5.14] H. Fujinami, T. Takuma, Y. Aoshima, “Impulse breakdown characteristics in SF6 Gas in the presence of a
local spark”. IEEE Transactions on Electrical Insulation, vol. EI-18, no 4, pp 429-433, August 1983.

[5.15] S. A. Boggs, F. Y. Chu, N. Fujitomo, A. Krenicky, A. Plessis, D. Schlicht, “Disconnect switch induced
transients and trapped charge in gas insulated substations”. IEEE Transactions on Power Apparatus and
Systems, vol. PAS 101 no.10, pp. 3593-3602, October 1982.

[5.16] O. Bosotti, "Phenomenon associated with switching capacitive currents by disconnectors in metal enclosed
SF insulated switchgear", CIGRE 1982 - Report 13.06.

[5.17] J. Lalot, A. Sabot, X. Kieffer, S.W. Rowe, “Preventing earth faulting during switching of disconnectors in
GIS including voltage transformer” IEEE Trans. on PWRD, vol. PWRD-1, no. 1, pp. 203-211, 1986.

[5.18] IEEE PES Special publication, "Modeling and analysis of system transients using digital programs",1998.
Prepared by IEEE WG 15.08.09.

[5.19] V. S. Rashkes and L. D. Ziles, "Very high frequency overvoltages at open air EHV substations during
disconnect switch operations", IEEE Trans. Power Delivery, vol. 11, no. 3, pp. 1618-1623, July 1996.

Page 80
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

6 ASSESSMENT OF TRANSFORMER VOLTAGE STRESSES

6.1 Introduction
This chapter describes both conventional and new proposals for transient stress assessments. It starts by outlining
the conventional approach used by the manufacturer to assess the transformer voltage stresses. It is followed by
sections on new concepts of analysing the proper electrical stress imposed on the power transformer due to non-
standardized impulse. One of the methodologies proposed is in Time Domain and the other in Frequency Domain.
Two important factors, Time Domain Severity Factor and Frequency Domain Severity Factor, are presented and
discussed in details in the following.

6.2 Time Domain Waveform Conversion – Conventional approach used by

manufacturers
The purpose of the voltage stress assessment is to establish the proper design of the transformer insulation
system. For that, the manufactures have to assess the electric field distribution into the geometry of the transformer
in order to find the appropriated design performance of the insulation system [6.1]. Thus, the obtained electric field
values of each material inside the transformer should be less or equal to the maximum permissible (dielectric
strength) value of its dielectric breakdown [6.2]. Therefore, in order to establish an acceptable insulation design
necessary to understand the voltage stress (kV) but also the electric field analysis in relation to material dielectric
strength values (kV/mm) [6.3]. Each manufacturer will have their own safety-ratio between dielectric strength and
field strength performance, later identified here as safety factor (SF).

In service, the transformer insulation system is subjected continuously to operating voltages and occasionally to
overvoltages. The former established the working voltage stress of the insulation. The overvoltages can be caused
by oscillatory surges and/or temporary overvoltages. Due to these transient voltages transformers must be
submitted in factory to impulse tests to prove the Basic Insulation Level (BIL) according to specific standards.
Impulse tests are dielectric tests that when applied produce to the transformer terminals test the transformers
insulation structure to withstand a high frequency steep wave voltage between windings, and between windings
and ground. The BIL of a transformer is a method of expressing the magnitude of the voltage surge that a
transformer will tolerate without dielectric breakdown.

The transformer insulation system can be categorized as major insulation and minor insulation. The major
insulation is the insulation between windings, between windings and limb/yoke, and between high voltage leads
and ground. The minor insulation is the internal insulation within the windings, viz. inter-turn and inter-disk
insulation.

There are basically four different types of standard tests, viz. lightning impulse test to check the Basic Impulse
Level (BIL), switching impulse test to check the Switching Impulse Level (SIL), short duration power frequency test
(AC Applied Voltage 50/60 Hz or higher, 1 minute or less but not less than 15 seconds) and long duration power
frequency test (AC Induced Voltage 50-240 Hz, 1 hour) with partial discharge measurement [6.4]. While the first
three tests check the insulation’s overvoltage withstand capability, the long duration test is mainly intended for
verify the insulation behavior under the working voltage stress. So, the inter-disk spacing is decided by impulse
stresses, whereas the end insulation (between windings and yoke) may get predominantly decided by the power
frequency test voltages [6.5]. The conductor insulation is decided by either the impulse stress or working voltage
stress.

Therefore, because the distributions of electric stresses are different under different electric excitation conditions, it
is important to understand that the voltage stress distribution in the transformer insulation system geometry should
be calculated for all those types of voltage (kV) tests. Once this is accomplished the resultant electrical stresses
(kV/mm) should to be analysed for each particular path and critical location [6.6]. The calculated electrical stresses
for any type of voltage must be acceptable for insulation used inside transformer according to the manufacturer SF.
Because of this point, transient overvoltage calculation has been investigated deeply by different methods [6.7-
6.10].

Page 81
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

In practice, there is a consensus in the market that assumes the performance of insulating materials for oil-filled
transformers is quite accurately characterized by means of breakdown strength curves. These curves are usually
obtained under short duration power frequency test (AC Applied Voltage 50/60 Hz, 1 minute). Since the breakdown
voltage depends not only on its amplitude but on its time duration, this characteristic must also be taking into
account for design purposes. Therefore, it is quite meaningful to convert voltages (e.g. BIL, SIL, AC/1hour) to their
AC equivalent voltage level, which is usually the short duration power frequency (AC/1min). This equivalent voltage
level establishes a Design Insulation Level (DIL), and the resultant voltage VDIL-Test is expressed in kVrms. Thus, a
conversion factor (DIL-factor)Test is obtained from experimental data of each designer to adapt the breakdown
strength curves for different time duration to 1min. The (DIL-factor)Test used by different manufacturers for
converting the calculated voltages (under various tests) to AC one-minute r.m.s. value may not be identical but they
fall in a narrow range, as shown in Table 6.1 [6.5,6.5, 6.11].

Standard voltage test (DIL-Factor) Test

(Multiplication factor)
Lightning Impulse Level (BIL) 1.2/50 Full Wave Impulse (DIL-Factor)BIL ≈ 2.30 (0.44)
Switching Impulse Level (SIL) 250/2500 (DIL-Factor)SIL ≈ 1.80 (0.55)
Short duration power frequency voltage (AC Applied voltage 50/60 Hz, 1 min) (DIL-Factor)AC_1m ≈ 1.00 (1.00)
Long duration power frequency voltage (AC Induced voltage 50/60 Hz, 1 hour) (DIL-Factor)AC_1h ≈ 0.80 (1.25)
Table 6.1: Design Insulation Level Factors (DIL-Factor)Test

Therefore, at any point inside the transformer, there is only one DIL, which is the maximum of the equivalent one-
minute power frequency voltage levels during the four different standard tests VDIL-Test, which take into account that
the breakdown voltage depends not only on the amplitude of applied voltage but also on the time duration of
voltage [6.5]. This approach is widely followed by transformer designers simplifying the insulation design process.

Although voltages and electric fields can be calculated to almost any desired accuracy, assuming the material
properties and transformer geometry are well known, the same cannot be said for breakdown fields in solids or
liquids. The theory of breakdown in solid or liquid is complex and still ongoing investigation, where the current
design rules have evolved based on experience [6.12]. Nevertheless, with suitable margins, the rules generally
produce successful designs. For those reasons, each manufacture has his house rules supported by safety factors
(SFs) for applying during the transformer insulation system design. One definition of a SF is proposed in the
following formula (6.1) [6.3]:

_ ,!,"
, , (6.1)
,!."

Where Ei(x,y,z) is the electric field stress along path i(x,y,z) and EP_i(x,y,z) is the maximum permissible electric field
(electric field strength) on the same path to avoid dielectric breakdown of the insulation material included along that
path i(x,y,z) as recreated in Figure 6.1. The final value of SFcrit will be the minimum value of SFi(x,y,z) along path
i(x,y,z). An insulation design is considered acceptable when all SFcrit along all critical paths are higher than a
certain value fixed by the manufacturer.

E
SF

SFi(x,y,z)
SFcrit

EP_i(x,y,z)

Ei(x,y,z)
Margin

Along a path i(x,y,z)

Figure 6.1: Performance of dielectric safety factor (SF) related with the electric field stress

Page 82
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

v.s. maximum permissible electric field strength along a path i(x,y,z).

As indicated previously, the breakdown voltages or permissible stresses are mostly given or defined for standard
lightning impulse, i.e. 1.2/50 us. However, the applied non-standard lightning impulse and the distributed voltage
along the windings are far from standard wave shape so the total time of applied voltage should be considered
carefully. Because of this fact and to have correct electrical stress analysis, the voltages should be converted to the
standard lightning wave shape. Regarding electrical stress analysis, the conversion must be in that way that the
effect of equivalent standard impulse is the same as non-standard original wave shape. To apply this, the
amplitude of standard impulse should be determined, see Figure 6.2. With increasing the time duration of original
wave form, the amplitude of equivalent standard impulse becomes larger. After such suitable conversions, the
electrical stress rules for the standard lightning impulse can be applied.

Figure 6.2: Equivalent standard lightning impulse 1.2/50 us.

This is the methodology currently applied by the manufacturer to make the correspondence of the different non-
standard transients that occur or might occur from power system with the transformer in operation to the standard
lightning impulse. These actual surge waveforms are complex due to the equipment composition (such as the
location of lightning arresters and the layout of main bus and branches) and other causes, and often more steep
and damp more quickly than the standard lightning impulse waveform of 1.2/50 µs. Therefore, conversion of non-
standard field overvoltage shape to standard impulse waveform is required for assessment of voltage stresses.

IEC60071-1 “Insulation co-ordination Part1” [6.13] defines “representative overvoltages Urp as standard shape
overvoltage assumed to produce the same dielectric effect on the insulation”, and there is a note that “overvoltage
shapes may have to be considered”. In the step of determining representative voltages “Urp” based on system
analysis results in the “flow chart for the determination of rated or standard insulation level” in IEC60071-1, it is
probable that real surge waveforms have different effects on equipment dielectrics from the standard impulse
waveform (1.2/50 µs).

As for this subject, it is described in IEC60071-2 “Insulation co-ordination Part2” [6.14] that “As a general
recommendation, the dependence on the overvoltage shape should also be considered in the determination of the
representative amplitudes, in particular, for external insulation and oil-paper insulation”. Further, Annex H illustrates
with an example and states that “if the assumed actual overvoltage has a shape different from test shape, the
representative overvoltage may have to be modified accordingly so that tests truly verify the insulation strength”.

Recent researchers analysed field surges, clarified insulating characteristics for actual surges in comparison with
for the standard impulse waveform in the lightning surge time region [6.15], and proposed the waveform evaluation
method for an oil-filled transformer (hereafter just as a transformer) as well as GIS [6.16]. According to them, the
decay of the field overvoltage is generally large, making the insulation requirements less severe than those of the
standard lightning impulse waveform. Consequently, it should be possible to use less severe insulation
specifications [6.17]. The proposed conversion method by S. Okabe et al. ([6.16] and [6.18]) for non-standard
lightning impulse waveform could be applicable up to 1100kV UHV (ultra high voltage) system. This procedure has
a draw back as it does not taken into account possible part-winding resonance amplification and the decrease in
insulation withstand due to multiple impulses which will be discussed in sections 6.3 and 6.4.

Page 83
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

6.3 Time Domain Severity Factor

6.3.1 Introduction
Reliability of power transmission depends on reliable operation of transformers insulation systems. Therefore, the
impulse test is required to verify the correctness of the winding insulation design and manufacturing process.

It is possible to determine the transfer function either by using time or frequency domain measurements.
Classically, from time domain analysis the voltage values along each winding are possibly evaluated for a particular
input transient voltage. Thus, the voltage drop between facing turns and turns and ground are calculated at each
step time in order to verify that the maximum dielectric strength value is lower than the permissible dielectric
strength value of each material for its dielectric breakdown.

Therefore, the visualization in time domain of the electric field intensity distribution within the transformer insulation
structure makes design decisions easier.

Previous work already consider the evaluation of the overvoltages in transformer associated to switching
transients by coefficients, such as the frequency domain severity factor (FDSF). But this factor, as a global
coefficient, could not assess the severity along windings to localize dielectrically weak points. To overcome this
limitation an alternative coefficient was proposed and was identified as time domain severity factor (TDSF) in the
Alvarez-Mariño´s paper [6.19]. The aim TDSF is to assess the severity supported along transformer windings when
the transformer is subjected to a transient voltage waveform from the power system.

Since each transient waveform depends on the electrical interaction between transformer and the power system, it
implies that each of those combinations is characterized by a TDSF. To obtain the TDSF implies the use of two
different models of the transformer under consideration. First, a terminal model (black box model) of the
transformer is built to compute the transient voltage waveform at the transformer terminals during the transient
event that occurred in the power system where the transformer is connected [6.20]. Then, a detailed model (white
box model) of the transformer is used to compute the internal transient voltage distribution along transformer
windings.

6.3.2 Time Domain Severity Factor Computation

The TDSF coefficient assesses the severity in terms of overvoltage due to the internal transient response along
transformer windings induced by the transients coming from the power system, compared to the internal transient
response due to standard dielectric tests in the time domain. The expression of this coefficient is [6.19]:

$%
∆'()* +, -
∆'./0.123. -
(6.2)

where ∆Vmax_sw is the maximum voltage drop between disks along of the windings, or turn to turn, due to the
transient event occurred in the power system (switching operation of VCB, for example) and ∆Venvelope is the
maximum voltage drop between disks along windings for all standard dielectric tests (envelope waveform of
standard tests).

Page 84
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Figure 6.3: Flowchart to compute the TDSFs according to Alvarez-Mariño´s paper [6.19].
In Figure 6.3, the flowchart for TDSF computation is shown. First, the building of a transformer detailed model from
the transformer geometry and material data is carried out [6.20].

The detailed model allows the internal transient voltage distribution along transformer windings for each standard
dielectric test to be obtained. Once the transient voltage distribution is available, the maximum voltage differential
distribution between facing disks of each winding and the envelope of the waveforms of the maximum voltage drop
of all standard dielectric tests are available [6.3,6.15,6.20].

With the detailed model available: the terminal model can be constructed, which can be used with a power system
model in EMTP to obtain the transient waveforms at transformer terminals. With the waveform at the transformer
terminal available the internal transient voltage distribution within the transformer winding can be computed. Once
the transient voltage distribution is available, the maximum voltage drop distribution between facing disks of each
winding is evaluated [6.20].

Finally, the maximum voltage drop during the transient operation is compared to the maximum voltage drop during
all standard dielectric tests along each winding, checking the TDSF [6.19]. If the computed values of TDSF along
windings are lower than the unit, the transformer insulation system is well designed for that particular transient
event occurred in the power system and supported by the transformer. Otherwise, the transformer insulation
system might not be suitable to be used into the power system and must be modified.

An application example illustrating how TDSF can be applied is shown in Chapter 8.

Page 85
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

6.4 Frequency Domain Severity Factor

6.4.1 Introduction
In an attempt to systematically classify the severity of voltage transients on power transformers, Malewski et al.,
[6.21] suggested that the analysis be performed in the frequency domain. When the paper was published in 1988,
the digital recording technique was still in its youth, and a limited number of transients could therefore be recorded
and analysed using this method. With modern acquisition hardware, and much larger digital storage space, a
statistical based study in frequency domain is today possible.

6.4.2 Methodology for analysis in frequency domain

Transformer transient protection system, using surge arresters, is designed to protect against overvoltage based
on the amplitude. Protection against steep front transients is considered in the transformer design, and the
transient withstand is tested with various acceptance tests before delivery. The chopped wave test is the most
critical acceptance test for steep front transients.

As a complement, for transients recorded or simulated, several techniques are available for frequency analysis,
e.g. Fourier analysis, Wavelet analysis, etc. Analysing transient records in frequency domain can highlight
additional information compared to analysis in time domain. In Malewski et al., [6.21] the frequency content of a
transient is determined using fast Fourier transform, FFT. The analyses performed yielded the amplitude of
different frequency component and the corresponding phase vectors were ignored. Systematic analysis of
transients was the main argument in [6.21], as time domain analysis requires high skill and experience (see section
6.3 for one possible method). In frequency domain, a steep front is, as an example, identified as it contains higher
frequency components than slow front transients.

When analysing in frequency domain, care has to be taken into using FFT correctly. In comparison to the analytic
Fourier transform where the time window is infinite, FFT analysises recorded traces of a length set by the recording
system. Reference is made to the help of the software used for the FFT, as FFT algorithms are not standardized.
Care has to be taken into windowing techniques (if required), trace length, scaling factors etc. As an example there
is a potential risk of misinterpretation of the result, depending only on the amplitude information of the spectra, and
ignoring the phase information. Several different pulses shapes can have the same frequency content if only
spectral amplitudes are considered. This is illustrated in figure 6.4, using the same amplitude spectrum. Even
though the spectral amplitude contents of these pulses are equal, the resulting stress on a transformer winding will
differ for the three cases. In the right of figure 6.4, three reverse FFTs of the frequency spectrum shown in figure
6.6 are presented. To the left, the corresponding phase vectors are plotted for the red and blue cases. It can be
observed that depending on the choice of phase vector, the time domain data can have almost any shape.

Figure 6.4: Phase information (left), and amplitude in time domain (right) for different pulses
having the same spectral amplitude information (figure 6.6 lower-right).

Page 86
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Correctly analysed, frequency domain analysis is a valuable tool. Internal resonances in the transformer can be
measured with Frequency response analysis or computed by the design engineer. Knowing the resonance
frequencies, and measuring the harmonic content of typical transient events in the substation, potentially
hazardous situations can be avoided. Analysing large quantities of transient data, trends in frequency domain,
could indicate potential risk due to changes in the transformer environment or possibly even internal faults.
Nevertheless, no conclusion should be drawn without a corresponding analysis in time domain, as proposed in
section 6.3, including the manufacturers design team.

6.4.3 Severity factor

In the paper by Malewski et al., [6.21] they presented a spectral comparison, between a transient voltage
interacting with a transformer, and the acceptance test of the same transformer. The authors discussed how to
record transient voltages in time domain, and with help of FFT study the absolute values of the spectra. The
transient spectra were then compared to the envelope spectrum of the acceptance tests of the transformer. Figure
6.5 ilustrates comparison of a certain transient with the spectrum of each standart test wave. Malewski et al., [6.16]
used similar figures to evaluate the impact of transients for 735kV transformers. A conclusion, was that all
transients with a spectrum lower than any acceptance test spectra would not be exposed to a risk, whereas
transient spectra, with amplitudes above the combined envelope of the acceptance test spectra would potentially
be dangerous to the transformer [6.21].

F( f )

Figure 6.5: Amplitude of the spectral content of lightning impulse (LI red), switching impulse (SI
blue), chopped wave (CW green), and an example of a transient (black).

Asano et al. [6.22] did further analysis using the method, and in [6.23] Rocha introduced a new parameter,
Frequency Domain Severity Factor (FDSF),
|7 89 :;<=> ?=@ |A
% 45 |7B 89 |A
, (6.3)

where CD 45 is the envelope of the acceptance test frequency responses. The definition of FDSF varies in the
literature. The definition in equation 6.3 follows the method of [6.24] and is the square of the definition methods of
[6.21] and [6.23]. The main difference between FDSF and the equation 6.3 approach is that this factor is calculated
based on the frequency responses spectral energy densities [6.24], whereas reference [6.21] considers the only
frequency responses of the signals. This means that in FDSF we are comparing energy involved in the transient
with the energy involved in the standard test waves.
A computational example about how to calculate this factor can be found in
http://www.eletrica.ufpr.br/~gustavo/severityfactor/ . Application examples illustrating how FDSF can be applied are
shown in some of the examples of “Part 2: Case Studies” and also in Chapter 8.

Page 87
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

6.4.4 Consideration about the FDSF application

a) Acceptance test
In equation 6.3, the FFT of the acceptance test is used as a reference for the severity of an incoming transient.
Malewski et al., [6.21] pointed out the difficulty of compensating for reduced withstand of the transformer insulation,
as the transformer is ageing. In the paper it is argued for the use of 70-80% of the acceptance test level, to
compensate for all ageing. Using the acceptance test without this correction, the user can overestimate withstand
of old transformers which do not have the same insulation properties as a newly produced unit.
b) Internal resonances
It must be pointed out that the FDSF does not take into account for the phase angle of different frequency
components of the transient. However, if the FDSF is below one in all frequencies and the transients considered
has a single dominant frequency, then it is highly expected that the transformer is protected against failures due to
internal ressonances. On the other hand, if the FDSF is above one, the transient being analysed is not well
represented by the standart wave tests.

Only an analysis in the time domain considering the grey or white-box approach will be able to address the
transformer internal stress due to this transient. The model should include damping to avoid over-estimating the
over-voltages produced at resonance. Sweep frequency resonance analysis (FRA) is a complementary analysis
tool to evaluate existing resonance frequencies.

6.4.5 Trends
A transformer installed in the field is continuously exposed to transients due to breakers, lightning strikes, etc.
Recording the transient activities over time, and presenting it in frequency domain gives a fingerprint of what the
transformer is exposed to. The transformer has been exposed for all the recorded frequency components, at the
recorded level, without failing. With the same argumentation as for the acceptance test being a record of
experimentally verified withstand, the spectral fingerprint is here considered as a withstand reference. Therefore,
complementing the FDSF analysis with a trending of transients could give the transformer owner an alarm of
irregular activities. Changes of transient event fingerprint could indicate a risky situation, and further analysis of the
substation and transformer is advisable.

6.4.6 Example
As an example for the use of transient trends, figure 6.6 presents the result of a study at a generator, where the
unit failed after one year of operation.

In the figure there are two histograms of the transient activities shown.The left is the first six months of operation,
and the right is the second (and last) six month of operation. Both plots are histograms of transient activities, with
frequency on the x-axis, and FDSF level (approximated). The color is the number of events (truncated to 20) at a
certain frequency and FDSF level.
Using the first six month as a fingerprint of the typical transient activities measured at the unit, one can see an
increase of transient activity in the second (and last) six month period of operation. In the studied case, the change
in transient activity was followed by a failure of the unit. It is however not evaluated if the transients caused the
failure, or if the failure caused the higher transient activity.

The creation of a finger print, and the trending of such, requires transient measurements over a significant duration
of time to validate the result.

Page 88
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Figure 6.6: Trending of FDSF for a generator unit that failed.

6.5 Conclusion
In this chapter, several tools for transient withstand have been presented. The most convenient tool to be used, will
in most cases depend on the available data, such as transient recordings, acceptance test files, and design data.
All the presented tools contain approximations, and should therefore, if possible, be used in combination. Relating
incoming transient stress to well understood acceptance tests facilitates procurement, design and test of new
transformers. Computation of the voltage stress at each position in the transformer, and comparing it to the stress
at the acceptance test let transformer owners evaluate the risk of their assets. Knowing the typical transient
spectra will help the system designer to avoid hazardous situations, and spectral trend shifts should be further
analysed in detail. The methods presented can be considered as tool in substation planning studies,
complementary to standard insulation coordination analyses already performed and as indirect information to the
manufacturer of the transient electrical environment of the substation. With the same argument, the methods
presented in this chapter can be considered as additional tools in failure investigations, as will be shown in chapter
9. The TDSF and FDSF methods are compared in an example of chapter 9. The FDSF is also included in several
examples of “Part 2: Case Studies”.

Page 89
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

6.6 References
[6.1] Xose M. Lopez-Fernandez, H.B. Ertan, J. Turowski, “Transformers, Design and Measurement”, Ed. CRC
Press, 2012.
[6.2] Dahinden, V., Schultz, K., and Kuchler, A., “Function of solid insulation in transformers”, Transform 98,
April 1998, Germany, pp. 41–54.
[6.3] Lopez-Fernandez, X.M., Alvarez-Marinño, C., Couto, D., Lopes, R. and Jacomo-Ramos, A., “Modeling and
insulation design methodology in power transformer under fast transients”, The XIX International
Conference on Electrical Machines (ICEM 2010), Rome, 6-8 September, pp. 1-6.
[6.4] IEC 60076-3, Power Transformers, part 3: “Insulation levels, dielectric tests and external clearances in air”,
2000.
[6.5] S.V. Kulkarni and S.A. Khaparde, “Transformer Engineering”, CRC Press, Taylor & Francis Group, New
York, 2004.
[6.6] W. Ziomek, K. Vijayan, D. Boyd, K. Kuby and M. Francheck,”High Voltage Power Transformer Insulation
Design”, Electrical Insulation Conference, pp. 211-215, Annapolis, Maryland, 5-8 June 2011.
[6.7] A. Miki, T. Hosoya and K. Okuyama, “A Calculation Method for Impulse Voltage Distribution and
Transferred Voltage in Transformer Windings”, IEEE Trans. Power Apparatus and Syst., Vol. 97, pp. 930-
939, May/June 1078.
[6.8] M. Nothaft, “Untersuchung der Resonanzvorgänge in Wicklungen von Hochspannungsleistungs
transformatoren Mittels Eines Detaillierten Modells”, Ph.D. Thesis, TH Karlsruhe, Karlsruhe, Germany,
1995.
[6.9] E. Rahimpour, J. Christian, K. Feser and H. Mohseni, “Modellierung der Transformatorwicklung zur
Berechnung der Übertragungsfunktion für die Diagnose von Transformatoren”, Elektrie, Vol. 54, No. 1-2,
pp. 18-30, 2000.
[6.10] R. Malewski, M.A. Franchek and J.H. McWhriter, “Experimental Validation of A Computer Model Simulating
An Impulse Voltage Distribution in HV Transformer Windings”, IEEE Trans. Power Delivery, Vol. 9, No. 4,
pp. 1789-1798, October 1994.
[6.11] D.J. Tschudi, “AC Insulation Design. Paper-Oil Insulation Systems”, WICOR Insulation Conference, pp. 1-
9, Rapperswil, Switzerland, Sept. 1996.
[6.12] R. Del Vechio, et. al, “Transformer Design Principles”, CRC Press, 2002.
[6.13] IEC60071-1, Ed. 8.1, “Insulation co-ordination Part1 : Definitions, principles and rules”, 2010.
[6.14] IEC60071-2 “Insulation co-ordination Part2 : Application guide”, 1996.
[6.15] S.Okabe, M.Koto, T.Kawashima, T.Inoue, T.Teranishi, S.Nagaoka, “Dielectric Characteristics of Oil-Filled
Transformer Insulation Models under Non-standard Lightning Impulse Voltages”, 11-ISH Vol.3, 3.345.P4,
1999.
[6.16] S.Okabe, S.Yuasa, M.Koto. and E.Zaima, “Evaluation of lightning surge waveform for LIWV reduction of
substation equipment”, 13-ISH P.05.66, 2003.
[6.17] W.Schmidt, R.Malewski, Special Report for Group33 (Power System Insulation Coordination), CIGRE
Session-2000, No.33-00, 2000.
[6.18] S.Okabe, M.Koutou, T.Teranishi, A.Takeda, T.Saida, “High Frequency Model of Oil-immersed Transformer
and Lightning Surge Analysis”, T. IEE Japan, Vol. 119-B, No.8/9, 1999.
[6.19] Casimiro Alvarez-Mariño and Xose M. Lopez-Fernandez, Antonio J.M. Jacomo Ramos, Ricardo A.F.
Castro Lopes, and Jose Miguel Duarte Couto, “Time domain severity factor (TDSF) Induced transient
voltage between transformer and vacuum circuit breakers”, COMPEL: The International Journal for
Computation and Mathematics in Electrical and Electronic Engineering, Vol. 31 No. 2, pp. 670-681, 2012.

Page 90
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

[6.20] Alvarez-Marinño, C. and Lopez-Fernandez, X.M. (2011), “Computation of fast transient voltage distribution
in transformer windings caused by vacuum circuit breaker switching”, Proceedings of the XV International
Symposium on Electromagnetic Fields in Mechatronics, Electrical and Electronic Engineering (ISEF2011),
Funchal, Madeira, 1-3 September, pp. 1-8.
[6.21] R. Malewski, J. Douville, L. Lavallée, “Measurement of switching transients in 735-kV substations and
assessment of their severity for transformer insulation”, IEEE Trans. Power Delivery, vol. 3, no. 4, October
1988.
[6.22] R. Asano, A.C.O. Rocha, G. M. Bastos, “Electrical transient interaction between transformers and the
power system” Cigré-33 Brugge 2007, Cigré Brazil JWG A2/C4-03.
[6.23] A.C.O. Rocha, “Eletrical transient interaction between transformers and the power system”, Cigré C4-104,
Paris 2008, Cigré Brazil JWG A2/C4-03.
[6.24] A. V. Oppenheim, A. S. Willsky and S. Hamid, “Signals and systems”, Prentice Hall, 2nd edition 1996.

Page 91
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

7 IMPACT ON TRANSFORMER INSULATION

7.1 Introduction
Insulation design is one of the most important parts of transformer design. Particularly, in large power transformers
the insulation design has a great effect on the final cost of the transformer. Transformer insulation must not only be
designed to withstand the normal service voltages but also the effects of transient overvoltages like lightning
impulse. Consequently, to design the insulation of power transformer properly, transient voltage distribution inside
transformer should be calculated under different types of applied voltages.

Whole insulation inside transformer can be divided to two different groups; one is the insulation inside windings,
which is called internal insulation. The other is the insulation between windings and between windings and
grounded parts such as core and tank, i.e. main insulation. Internal insulation and main insulation are studied
separately in the following sections.

7.1.1 Insulation strength

The breakdown voltage of a dielectric material is a function of its physical and chemical properties, as well,
impurities present in it. Failures in an insulation material may be initiated by higher electrical stresses, interrelated
thermal, chemical phenomenon, mechanical factors, and so on. Hence, evaluation of insulation strength is a
difficult issue. With the aging and deterioration of electrical and mechanical properties of insulation materials, it
becomes even more difficult to predict their performance [7.1].

In transformers, insulation system is a combination of oil and solid insulation (paper, pressboard) [7.2]. The
breakdown in oil has an erratic behavior, especially when it is used alone. Hence, larger oil ducts are always
subdivided by solid insulation into smaller ducts in order to have more stable insulation system, as shown in Figure
7.1. Compared with breakdown processes in gases, little is known about the processes which initiate and lead to
breakdowns in the oil.

Oil
Winding A

Winding B

Pressboard

Figure 7.1: Insulation structure between two windings.

Page 92
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Insulation strength is indeed a statistically value [7.1], [7.3]. However, most of the factors affecting this strength are
known. Following parameters should be considered in analysing the insulation strength:

- Moisture [7.4] [7.5]

- Impurities [7.2] [7.3]
- Insulation thickness [7.1] [7.2]
- Oil velocity [7.1]
- Temperature [7.4]
- Duration of voltage application [7.1]
- Frequency of voltage application [7.1]
- Pressure [7.2] [7.3]
- Aging [7.6] [7.7]

7.1.2 Main insulation

Between cylindrical windings mounted on a common limb but having different voltage ratings, the layer insulation
generally consists of oil and some kind of solid insulation. Similar layer insulation is considered between winding
and tank and or between winding and core. This insulation construction is called main insulation. Figure 7.2
demonstrates an example of main insulation consisting of multiple barriers (not all the key spacers or sticks are
shown) between high voltage and low voltage windings [7.2].

Up to about 35 kV rating, the oil layer is divided into ducts of 3 to 15 mm widths by pressboard or paper-based
phenol tubes, whereas above 35 kV the tubes are made only of pressboard. Except for the winding ends, the
electric field between cylindrical windings may be regarded, with good approximation, as homogeneous, so that the
voltage appearing across the layers can easily be determined by calculation [7.3].

End collars
HV Winding
LV Winding

Pressboard barriers

(a)

Page 93
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Key spacers

Sticks

Pressboard barriers

(b)

Figure 7.2: Main insulation structure between windings: a) Side view, b) Top view.
Namely, in such a case, the electric field imposed on the various layers will be inversely proportional to their
permittivity. If the permittivity of the layers of thicknesses d1, d2, ..., dn are ε1, ε 2, ..., ε n respectively, then on
applying voltage U to the entire width of insulation, the field strength in the part of thickness d1 will be:

U
E1 =
d d d 
ε1  1 + 2 + ... + n 
 ε1 ε 2 εn 

In the other layers:

HG HG HG
EF EG EI EG … .. EK EG
HF HI HK

Since the permittivity of the oil (εo=2.2) is lower than of paper-based phenol (εpph=5 to 6) or pressboard (εpb=4 to 5),
the field strength in the oil will always be higher than that in the solid insulation. In addition, the dielectric strength of
the oil is lower than that of the solid insulation, so that the stress imposed on the oil will by all means be the higher.
Under the effect of an excessively high electric stress, first, sporadic low-intensity discharges of short duration
occur in the oil, the frequency of which gradually increases, and then the process goes over into intense partial
discharges, leading to total breakdown of one of the oil ducts. Thereby, the stress imposed on the remaining parts
of the insulation grows, causing further breakdowns in the oil, whereby an increasingly higher stress is imposed on
the solid insulation. When all oil ducts having suffered a breakdown, the solid insulation will have to withstand the
full voltage. Consequently, the two basic principles of correct designing of the main insulation are as follows [7.3]:

1) The field strength permitted to occur in the oil ducts, assuming a new oil filling free from moisture and gas,
should be sufficiently low to prevent the phenomenon described above from starting even at the test voltage
level.
2) After a shorter or longer time of service, a gradual drop in the dielectric strength of oil ducts, due to
contamination of the oil, should be reckoned with.

Page 94
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

At higher voltage several pressboard cylinders are applied. Dividing the oil into several ducts of small widths is
favourable, since the dielectric strength of the oil in a thinner layer is higher than in a thick layer. Moreover, the
solid insulating cylinders prevent impurities and contaminations from forming long chains leading to total
breakdown through the oil.

In designing the insulation between winding closest to the core and the core itself the role of mechanical strength is
decisive. This winding is usually the one of lower voltage rating, therefore the electrical stress imposed on it is not
excessively high, whereas the short-circuit forces which tend to compress the inner winding may reach substantial
levels.

7.1.3 Internal insulation

All the insulation inside a winding, consists of paper, enamel insulation, pressboard spacers, pressboard sticks and
oil are considered as internal insulation. The strength of internal insulation should be enough to avoid any failure
inside the windings. To design internal insulation in a correct way, voltage distribution in transient and steady states
should be calculated inside the winding. One 400 kV disc winding is shown in Figure 7.3, which all internal
insulation except to oil is shown in.

In the overwhelming majority of cases, copper or aluminum conductors of transformers are insulated with Kraft
paper, though the use of wires with enamel insulation is gaining ground. The paper strip is applied overlapped over
the conductor. The layer thickness of interturn insulation is between 0.2 and 1.5 mm. In small series-capacitance
windings, i.e. in the case of uneven lightning impulse voltage distribution, the designing is generally governed by
the expected lightning impulse stresses. In winding with high series capacitance the distribution of lightning impulse
voltage is uniform but, due to interleaving, a multiple of the power frequency interturn voltage may arise between
geometrically adjacent turns, so that in many cases the power frequency stress will be decisive in the designing of
insulation.

Figure 7.3: One 400 kV disc winding.

A 4 to 12 mm oil duct is generally provided between sections of the disc-type windings of transformers to ensure
the required dissipation of heat. Due to the creepage distance formed by the pressboard spacers placed between
the sections, the voltage stress permitted to arise in the duct is much lower than that tolerable for an oil layer of
equal thickness [7.3]. From knowledge of voltage imposed on the duct, the required width of the duct can be
determined.

Page 95
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

7.2 Insulation aging

7.2.1 Introduction
This section reviews the degradation status of insulating oil and paper of field aged transformers, covering from old
to relatively recent investigative results. Initially, the influence of aging on various characteristics is investigated,
followed by a study on characteristics that may be effective indices to find the degree of degradation of
transformers.

7.2.2 Examples of measurement for shell-type transformers

This section reports the investigation results of the degradation of field aged shell-type transformers [7.8] [7.9]. Of
the insulating materials used for oil-immersed transformers, those related to the life of insulation performance are
mainly insulating oil and paper. Therefore, the various characteristics of insulating oil and paper are measured and
compared with the values of new ones to evaluate the degradation trend.

Ten transformers with voltage classes of 77 kV to 250 kV and capacities of 10 MVA to 100 MVA are examined. The
manufacturing dates are old, ranging from 1940 to 1952. However, since the examination was conducted in around
1977 to 1978, those transformers were about 30 years old.

7.2.2.1 Results of investigation for insulating oil

Although the degree of degradation of insulating oil depends on the type of transformer, the characteristics such as
volume resistivity, dielectric loss tangent, and acidity are deteriorated on the whole. On the other hand, the
breakdown voltages that represent the key characteristics are 66 to 82 kV/2.5 mm, showing high values.

7.2.2.2 Results of investigation for insulating paper and pressboard

Insulating paper and pressboard are materials that determine the life of a whole transformer because they cannot
be easily replaced even if the characteristics are deteriorated due to aging. Figure 7.4 exhibits the measurement
results of various characteristics of these materials. The materials used are Manila paper and pressboard
containing cotton waste. Figure 7.4(a) shows the relationship between the average degree of polymerization (DP)
and the ac short-term breakdown voltage for pressboard, and Figure 7.4(b) shows the relationship between the DP
and the tensile strength for insulating paper and pressboard.

50
120 Insulating paper
40 Pressboard
Tensile strength (MPa)

100
Breakdown voltage

30 80
(kV/mm)

Examples of
new articles
60
20 Examples of
new articles
40
10
20

0 0
0 200 400 600 800 1000 0 200 400 600 800 1000
Average degree of polymerization Average degree of polymerizatiom

(a) (b)
Figure 7.4: Characteristics of insulating paper and pressboard: (a) Relationship between DP and
the AC breakdown voltage; (b) Relationship between DP and the tensile strength.
The DP has decreased from about 1000 for new material to about 200 to 600 due to aging. Along with this
decrease in the DP, the tensile strength also tends to decrease. Regarding the ac breakdown voltages, they vary
among 60% to 90% (80% on average) of that of new material, regardless of the DP. Although the decreasing rate

Page 96
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

is small compared with the DP, attention must be paid to the fact that the breakdown voltages also decrease due to
aging.

7.2.3 Examples of measurement for core-type transformers

This section reports the investigation results of the degradation of field aged core-type transformers [7.10] [7.11]. In
the investigation, similarly to the case of shell-type transformers, various characteristics of insulating oil and paper
are measured to evaluate the degradation trend by aging.

7.2.3.1 Results of investigation for insulating oil

A total of 97 transformers with voltage classes of 11 kV to 525 kV and capacities of 12.5 MVA to 1100 MVA are
examined. With manufacturing years ranging from 1951 to 1977, the transformers were selected uniformly from
each era. The actual examination was conducted in 1978 and the number of years in service of the transformers
was about 0 to 27.

Figure 7.5 shows the relationship between the number of years in which transformers have been in service and the
breakdown voltage. A solid line represents an average regression line and a broken line represents a 95%
confidence interval. For almost all non-highly aged transformers, the breakdown voltage exceeds 40 kV/1.5 mm.
Conversely, for highly aged transformers, particularly those of the SNT type, the breakdown voltage for some
insulating oil is relatively low, or 35 kV/1.5 mm or less. Overall, the breakdown voltage tends to gradually decrease
with age. The average regression line can be given by the following equation:

L 51.4 − 0.15 ×

V: Breakdown voltage (kV / 1.5 mm)

t: Number of years in service (year)

70
Breakdown voltage (kV/1.5 mm)

60

50

40

30

20
● SNT type: Floating-tank-type conservator OHD type
10 ○ OHD type: Membrane-type no-pressure sealed conservator
SNT
0
0 5 1 1 2 2
Numbers of years in which transformers have been in service

75 70 65 60 55 50
Service start year of transformers

Figure 7.5: Relationship between the numbers of years in which transformers have been in
service and the dielectric breakdown voltage

According to this equation, the decrease in insulation strength over 30 years is about 9%. It is also possible that the
9% drop in dielectric strength in 30 years may have resulted from the improvement in the initial-characteristic
dielectric strength due to technical advances made in 30 years, not simply from the degradation of the dielectric
strength. In actually, the two factors must have combined to produce the 9% drop in dielectric strength in 30 years.
Therefore, it is sufficient to assume the 9% dielectric strength drop in 30 years due to degradation because, in
actuality, it cannot be more than that. In addition, now that OHD conservators, which have higher oil conservation
characteristics, are mainly used, the analysis techniques of insulating oil and the management techniques for the
impurities it contains are increasingly advanced. Based on these factors, the potential degradation of the withstand
voltage over 30 years is considered substantially smaller than 9%.

Page 97
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Besides the breakdown voltage mentioned above, various characteristics have been measured. For example, a
significant age-related decline in the characteristics of the volume resistivity and acidity is confirmed. The aging
characteristics of insulating oil are similar to those for shell-type transformers and their difference, attributable to
structural differences, is considered minor.

7.2.3.2 Results of investigation for oil impregnated materials

Eight transformers are investigated, with voltage classes of 60 kV to 275 kV and capacities of 10 MVA to 600 MVA.
The actual investigation was conducted in around 1980 and the number of years in service of the transformers
ranged from several to 40 years. From these transformers, the insulating papers were sampled from diverse
locations. Subsequently, the remaining rates of the tensile strength, DP, and dielectric strength were measured as
shown in Figure 7.6. In the figure, the maximum and minimum values of measurements are represented by
envelopes and the average value of them by broken lines.

The degrees of degradation exhibit a considerably wide range of variations among transformers. It is considered
most significantly attributable to the fact that, since the samples were taken from different locations, the
temperature conditions differed for each location. Besides, the differences in the operation history of each
transformer, the treatment conditions in the assembly process, the advancement of material used, and the
difference in oil conservator are considered to have caused the variability. The variability is also considered partly
attributable to the comparison, not with the initial value of the material used for individual transformers but based on
the equivalent material during the period of the examination.

These measurement results indicate that the mechanical strength, such as the tensile strength and DP, decreased
about to the 50% level over 30 to 40 years. On the other hand, the decreasing rate of dielectric strength is more
gradual than that of mechanical strength. According to the mean value, the decrease in insulation strength over 30
years is about 12%. These characteristics are similar to those for shell-type transformers and their difference,
attributable to the structural differences, is considered minor, similarly to the case of insulating oil.
Remaining rate of average

polymerization degree (%)

100 Max. value 100 Max. value

tensile strength (%)
Remaining rate of

80 80
60 60
40 Mean value Mean value
Min. value 40 Min. value
average

20 20
0 0
0 10 20 30 40 0 10 20 30 40
Number of years in service (year) Number of years in service (year)

(a) Aging in tensile strength (b) Aging in average polymerization degree

100
dielectric strength (%)

Max. value
80 Mean value
Remaining rate of

60
Min. value
40
20
0
0 10 20 30 40
Number of years in service (year)

(c) Aging in dielectric strength

Figure 7.6: Tensile strength, average polymerization degree, and dielectric strength
characteristics of insulating paper due to aging

Page 98
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

7.2.4 Summary
The present section reviewed the degradation status of insulating oil and paper of field-aged transformers, covering
investigation results ranging from about 30 years ago to a relatively recent period. The study results are
summarized in Table 7.1. Though the transformers treated were diverse in terms of, for example, their
manufacturing year, manufacturer, structure, and voltage class, changes in various characteristics due to the aging
of insulating oil and paper were similar.

Basically, the service life of the whole transformer is determined by factors such as the degradation of the
mechanical strength of insulating paper due to thermal stress, and insulation degradation is less likely to cause a
problem earlier. Although it is difficult to directly measure this degree of degradation of insulating paper, it can be
indirectly estimated by measuring the CO2 and CO content and furfural content in insulating oil.

On the other hand, even though the decreasing degree is small, there is a fact that the breakdown voltages of
insulating oil and paper decrease due to aging. Therefore, for highly aged transformers or transformers which are
exposed to overvoltages frequently, consideration must be given to age-related degradation of the insulation
performance. In order to identify the degree of degradation of these transformers, it is effective to measure the
volume resistivity and dielectric loss tangent in electrical terms and the interfacial tension trend in physical terms
and it may possibly be effective to measure the acidity in chemical terms. Analysis of insulating oil components is
also considered effective to evaluate degradation by aging.

Insulating oil Insulating paper and pressboard

Investigation about 30 years ago Investigation about 30 years ago Investigation on the degradation degree
Recent investigation [7.6] [7.7]
[7.8] [7.9] [7.10] [7.8] [7.9] [7.11] [7.12] [7.13] [7.14] [7.15]

Use for about 30 years

Use for about 30 to 40 years • The average polymerization degree is
Use for about 30 years
• The ac breakdown voltage is sufficiently decreased from about 1,000 to the range
• The residual ac withstand voltage is • The decreasing rate of the average
high at 60 kV/2.5 mm. of 200 to 600.
Shell-type sufficiently high, ranging from 66 to 82 polymerization degree strongly correlates
• The degradation due to aging can be • The dielectric strength varies ranging
transformer kV/2.5 mm. with CO2 and CO content, furfural content,
observed in the volume resistivity and from 60 to 90% (80% on average) of new
• The acidity and volume resistivity are and acetone content produced.
dielectric loss tangent in electrical terms, one regardless of the average
degraded due to aging. • The average polymerization degree of a
in the interface tension in physical terms, polymerization degree and the decrease
transformer in operation cannot be directly
and in the acidity in chemical terms. in breakdown voltage is small.
measured but can be indirectly estimated
• The increasing trend of alcohols and
through the measurement of the CO2 and
ketones along with aging can be
Use for about 30 years Use for about 30 years CO content and furfural content produced.
observed, and methyl ethyl ketone is
• The decreasing rate of the dielectric • The mechanical strength is decreased to
Core-type particularly an effective indicator to
strength is 9% or less. about 50%.
transformer evaluate the aging degradation.
• The acidity and volume resistivity are • The decreasing rate of the dielectric
degraded due to aging. strength is 12% or less.

Table 7.1: Summary of insulation degradation by aging

7.3 Effect of repetitive impulses

7.3.1 Introduction
Two important electrical conditions should be taken into account in the study of the insulation breakdown
characteristics of in-service oil-immersed transformers:

• One is the effect of repetitive voltage application (V-N), considering the relationship between the number of
repetitions N and the breakdown voltage V.
• The other is the influence of the voltage shape, especially related to the application of different rise times.

In this section, V-N breakdown characteristics are presented when a lightning impulse or a damped oscillation
wave is repeatedly applied to transformers. This work is based on Japanese experience.

Page 99
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Regarding the second topic, a full understanding of how rapidly rising voltages affects insulation system is still a
challenge and deserve future investigation. Some works in the literature have shown an important influence of the
wave shape on the voltage breakdown and a possible decrease of the withstand with shorter risetimes [7.16],
[7.17], [7.18], [7.19]. A better knowledge of this effect will, for sure, provide an improvement of the design of
insulation system exposed to fast transients [7.17], [7.20].

7.3.2 V-N characteristics below and at 500kV class

The breakdown voltage characteristics when the impulse voltage is repeatedly applied, i.e. the V-N characteristics,
are generally expressed as follows.

V50 = K·N-1/n (%)

According to Figure 7.7 [7.21] and Table 7.2 [7.21], if a voltage with the same polarity is repeatedly applied, n = 38
or greater is obtained, and the breakdown voltage decreases only a little, while the alternate application of opposite
polarities decreases the breakdown voltage to a greater extent with n = 20 or so.

However, since the stress on actual transformers is usually designed to be substantially lower than these test
values, an overvoltage surge intrusion at high enough levels to cause a partial discharge rarely occurs.

n-values indicate the respective insulation models and experimental conditions in Table 7.2

Figure 7.7: Summarized V-N characteristics

Page 100
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Reference Insulation model Experimental conditions Experimental results (No.)

1) Turn-to-turn model 1) Using the turn-to-turn model in oil, *The breakdown voltage decreased
a voltage 1.5 to 1.7 times PDI Vimp by 10% from the initial value after
with 30sec intervals was applied 1000 applications.
continuously. (No.1) n = 65.6
2) Lightning impulses (1.2/50 µs) (The decrease was less than that in
[7.22] were applied. the oil-duct model and other
Conductor
insulation models.)
Coatin

“A” – “A” section

2) Barrier insulation 1) The breakdown test was conducted *The breakdown voltage decreased
using pressboards in oil. after 500 applications as follows,
(approximately 6mm) (No.2-a) Either positive or negative
[7.23] 2) Lightning impulses (1/50µs) were polarity: approx. 15% ···· n = 38.2
applied. (No.2-b) Alternate application of
positive and negative polarities:
approximately 25% ···· n = 21.6

3) Section-to-section model 1) Using the section-to-section *There was no increase of partial

(Model between windings) insulation model, lighting impulses discharge occurrence after lightning
with impulse coronas (90% of the impulses were applied repeatedly.
breakdown voltage) were applied *The breakdown voltage was also
[7.24]
200 times to measure the partial unchanged.
discharge.
2) Lightning impulses (1/40µs) were
applied.
4) Barrier insulation 1) Using insulating paper (0.1mm × *The breakdown voltage scarcely
3
5sheets) in oil, the relationship of decreased with up to10 applications
the breakdown voltage to the of switching impulses.
[7.25] frequency of application was
measured.
2) Switching impulse waves were
applied.
5) Barrier insulation 1) JIS Oil No.1 was impregnated in (No.5-a) 1000 applications of
Oil layer the barrier insulator: Room lightning impulses: 100% of the
temperature (20 to 30°C), Oil initial value n>100
2
pressure 1 kgf/cm (No.5-b) Damped oscillation waves
2) a. Impulses were applied 1000 (with 10 min. intervals): 78% of the
times. initial value n = 27.8
[7.26]
b. Damped oscillation waves (with (No.5-c) Damped oscillation waves
Electrode end
10min. intervals) were applied (with 15sec. intervals): 63% of the
reinforcing 1000 times. initial value n = 14.95
paper c. Damped oscillation waves (with
15sec. intervals) were applied
1000 times.
Table 7.2: V-N characteristics with repeated voltage application

Page 101
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

7.3.3 V-N characteristics at UHV class [7.27]

(1) V-N characteristic of a core-type transformer
a. Lightning impulse V-N characteristics of the turn-to-turn model
Figure 7.8 shows the V-N characteristics of BD and PDI of the turn-to-turn model by lightning impulses. As the
applied voltage level is lowered, the N will be plotted to larger values, finally no BD or PD will occur even at a test
count of 1000 times. This can be expressed in the following equations:
-1/41.2
V50 = 258 × N (kV) (for BD)
-1/41.9
V50 = 138 × N (kV) (for PDI)

The n-value is around 40, smaller than that of the section-to-section model described below, i.e., indicating a
repetitiously deteriorating characteristic. The count (time) shape parameter of the Weibull distribution expression, a-
value was generally 0.3 to 0.7 including both BD and PDI under the four conditions described in this section,
indicating an initial breakdown (BD and PDI) pattern.

b. Switching impulse V-N characteristics of the turn-to-turn model

The V-N characteristics of BD and PDI of the turn-to-turn model by switching impulses can be expressed in the
following equations:
-1/71.8
V50 = 235 × N (kV) (for BD)
-1/46.8
V50 = 132 × N (kV) (for PDI)

The n-value at breakdown is around 70, close to that of the section-to-section model described next. The n-value at
PDI is around 40, close to the lightning impulse characteristic of the turn-to-turn model.

c. Lightning impulse V-N characteristics of the section-to-section model

Figure 7.9 shows the V-N characteristics of BD and PDI of the section-to-section model by lightning impulses. This
can be expressed in the following equations:
-1/72.2
V50 = 296 × N (kV) (for BD)
-1/67.0
V50 = 247 × N (kV) (for PDI)

The n-value is around 70, indicating a considerably flat characteristic.

d. Switching impulse V-N characteristics of the section-to-section model

The V-N characteristics of BD and PDI of the section-to-section model by switching impulses can be expressed in
the following equations:
-1/77.6
V50 = 269 × N (kV) (for BD)
-1/61.6
V50 = 251 × N (kV) (for PDI)

The n-value is around 70, indicating a considerably flat characteristic, close to that of the lightning impulses.

Page 102
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

350
300
250 (1)
(6)

PDI and BD voltage [kV]

200 V50=258N
- 1/41.2 (8)

150
(2)
100 V50=138N-1/41.9

● Breakdown
○ Partial discharge inception
　50
● No BD
○ No PDI
102
　30
1 10 (200 10
3
)
Number of times of application [times]

Figure 7.8: Lightning impulse V-N characteristic of the turn-to-turn model of a core type
transformer

400
350
PDI and BD voltage [kV]

(4)
300 (5) (2)
(8) (4)
250 (3)
(5) (2)
200 -1/72.2
V50=296N
V50=247N-1/67.0
150
● Breakdown
○ Partial discharge inception
● No BD
○ No PDI
100
1 10 102 (200) 103
Number of times of application [times]

Figure 7.9: Lightning impulse V-N characteristic of the section-to-section model of a core type
transformer
(2) V-N characteristics of a shell-type transformer
Figure 7.10 shows the V-N characteristics of BD and PDI under switching impulse voltages. In this figure, a white
circle represents a PDI point and a black circle represents a BD point; A value enclosed in a square represents the
number of times that BD or PD occurred when the voltage is applied once and a value enclosed in a circle
represents the number of times that these phenomena did not occur while the experiment is repeated 1,000 times.

The V-N characteristic gradient n-value is 75.6 for BD and 70.5 for PDI by lightning impulses, 80.9 for BD and 65.1
for PDI by switching impulses, indicating a flat characteristic. Even if n=65.1, the deterioration after application of
1,000 times is about 10%. The number of times (time) shape parameter of the Weibull distribution expression, a-
value is 0.37 to 0.78, indicating an initial breakdown (BD and PDI) pattern.

Page 103
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

400
n=80.9
360

PDI and BD voltage [kV]

3
2

320

280 6
3 n=65.1
②

240

1 101 102 103

Number of times of application [times]

Figure 7.10: Switching impulse V-N characteristic of the barrier-oil-duct model of a shell type
transformer

7.3.4 Summary
Some estimates indicate that the total number of typical overvoltage surges with usual level intruding from
transmission lines is normally 200 from lightning surges and 600 from switching surges [7.21] during the entire
lifetime of the transformer, and that as a practical matter the effect of such repeated voltage applications will be
small. However, for transformers installed in a system that operates switches frequently, it has a potentiality to be
exposed to switching surges more than 600 times during the service period. In such a case, since there is concern
at a possible drop in dielectric strength due to repetitive voltage stresses, sufficient consideration must be taken
when specifying transformers.

7.4 References
[7.1] S.V. Kulkarni and S.A. Khaparde, “Transformer Engineering”, CRC Press, Taylor & Francis Group, New
York, 2004.

[7.2] R.M. Del Vecchio, B. Poulin, P.T. Feghali, D.M. Shah and R. Ahuja, “Transformer Design Principles”, CRC
Press, Taylor & Francis Group, New York, 2001.

[7.3] K. Karasai, D. Kerenvi and L. Kiss, “Large Power Transformers”, Akadémiami Kiadó and Elsevier
Science Publishers, 1987.

[7.4] I. Fofana, H. Borsi and E. Gockenbach, “Fundamental Investigations on some Transformers Liquids
under Various Outdoor Conditions”, IEEE Trans. Dielectrics and Electrical Inssulation, Vol. 8, No. 6, pp.
1040-1047, 2001.

[7.5] J. Dai, Z.D. Wang and P. Jarman, “Creepage Discharge on Insulation Barriers in Aged Power
Transformers”, IEEE Trans. Dielectrics and Electrical Inssulation, Vol. 17, No. 4, pp. 1327-1335, August
2010.

[7.6] M. Kohtoh, S. Kaneko, S. Okabe, and T. Amimoto, “Aging Effect on Electrical Characteristics of
Insulating Oil in Field Transformer”, IEEE, Trans. Dielectr. Electr. I Insul., Vol. 16, pp. 1698-1706, 2009.

[7.7] S. Okabe, S. Kaneko, M. Kohtoh, and T. Amimoto, “Analysis Results for Insulating Oil Components in
Field Transformers”, IEEE, Trans. Dielectr. Electr. Insul., Vol. 17, pp. 302-311, 2010.

[7.8] T. Kusumoto, S. Takami, T. Ishii, and E. Tamaki, “Study of Ageing Transformers for Power Systems”,
National Convention of IEEJ, Paper 643, 1981.

[7.9] E. Tamaki, I. Itakura, and Y. Makino, “Life Assessment Technologies for T Transformers”, Tokyo Branch
Convention of IEEJ, Paper S.5-4, 1985.

Page 104
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

[7.10] T. Yanari, M. Ikeda, Y. Taniguchi, and Y. Ishioka, “Oil Deterioration of Transformer for Long Term”,
IEEJ, Trans. Power and Energy, Vol. 101, pp. 719-726, 1981.

[7.11] Y. Ishioka, H. Okubo, T. Suzuki, and M. Honda, “Degradation Phenomena for Oil- immersed Insulations”,
IEEJ, Insulating Material Study Group, EIM-84-11, 1984.

[7.12] R. Tamura, H. Anetai, T. Ishii, and T. Kawamura, “The Diagnosis on the Aging Deterioration of Insulating
Paper in Transformers by Gas Analysis”, IEEJ, Trans. Fundamentals and Materials, Vol. 101, pp. 30-36,
1981.

[7.13] H. Tsukioka, E. Mori, K. Goto, and S. Suzuki, “Study on Diagnosis of Aging Deterioration of Oil Filled
Transformer by CO2 and CO”, IEEJ, Trans. Fundamentals and Materials, Vol. 106, pp. 331-338,
1986.

[7.14] E. Takahashi and H. Tsukioka, “Recent Diagnosis Techique of Aging Deterioration for Oil-immersed
Transformers”, IEEJ, DEI-93-98 and EC-93-21, pp. 27-36, 1993.

[7.15] Electric Technology Research Association of Japan, “Maintenance Management for Oil-immersed
Transformers”, Report Vol. 54, No.5, Part 1, 1999.

[7.16] B. Sharath and S. Usa: "Prediction of Impulse Voltage-time Characteristics of Air and Oil Insulation for
Different Wavefronts", IEEE Transactions on Dielectrics and Electrical Insulation Vol. 16, No. 6, pp 1693-
1697; December 2009

[7.17] D.Breitfelder et al.: “Dieletric strength of transformer oil under impulse and high frequency voltage stress”.
Fifth Internationl Symposium on High Voltage Engineering, 24-28 August 1987, Federal Republic of
Germany.

[7.18] T. Hammarstrom, T.Bengtsson, J.Blennow and S.M.Gubanski: "Evidence for changing PD properties at short
voltage rise times", IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 18, No. 5, October 2011,
p 1686-1692

[7.19] A.J.Vandermaar, M.Wang, J.B.Neilson and K.D.Srivastava: "The electrical breakdown characteristics of
oil-paper insulation under steep front impulses voltages ", IEEE Transactions on Power Delivery, Vol. 9, No.
4, October 1994, p.1926-1935

[7.20] C. Kieny et al : "Distribution of very fast transient overvoltages in transformer windings ’’ , 12-204 - Cigre
1992 session.

[7.21] Electric Technology Research Association Report, “Rationalization of Insulation Design” Vol. 44, No.3;
Electric Technology Research Association, 1988.

[7.22] T. Inoue, M. Honda, H. Okubo, S. Ogura, “Dielectric Strength of Transformer Insulation Strength in the Case
of the Repeated Impulse Voltage Superposition to the AC Voltage”, T. IEE Japan, Vol.B-106, pp.25-30, 1986.

[7.23] W. G.Standring, P. H. Hughes, “Breakdown under Impulse voltages of solid and Liquid dielectrics in
combination”, PIEE, p.583, 1957.

[7.24] Y. Kamata, K. Endo, “Lightning Impulse volt-time characteristics in transformer Insulation”, IEEE Winter
Meeting, C73, 028-8, 1973.

[7.25] IEE Japan Technical Report, part II, No.47, Testing Voltage Standardization Special Committee WG I: “Study
of the AC Insulation Test of 500kV Oil-filled Transformers in Terms of the V-t Characteristics”, 1976.

[7.26] Kojima et al.: “Switching Surge Dielectric Strength Characteristics of OF Cable Insulator”, T. IEE Japan, Vol.
83, p.43, 1963.

[7.27] S. Okabe, T. Ohno, E. Zaima, “V-t and V-N Characteristics of Insulation Elements with Large Scale Oil-
Immersed Transformers”, 14-th ISH, I-13, 2005.

Page 105
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

8 TRANSIENT SIMULATION SOFTWARE BENCHMARKING – FICTITIOUS

TRANSFORMER

8.1 Introduction
The objective of the work presented in this chapter of the brochure is to define a simple transformer able to test
different transformer “white-box” mathematical models designed to compute the distribution of internal transient
voltages.

For this purpose a “Fictitious Transformer” two winding 100 MVA, 230 kV to 69 kV, transformer geometry was
created. In the following paragraphs this “Fictitious Transformer”, will be completely defined.

After that it is presented the working group members predicted transient behavior for the “Fictitious Transformer”
when it is excited with different waves shapes in different terminal points.

The results of the modelling using various tools on the common geometry of the “Fictitious Transformer” will be
reported in the Appendix A. This will form a valuable contribution to the transformer industry about the state-of-the-
art of the computing of voltage transients inside transformers.

The “Fictitious Transformer” and its transient results can be used in the future during design review activities to
evaluate and benchmark manufacturers softwares used for dielectric design of real transformers.

8.2 The Fictitious Transformer

It is worth mentioning that only the data contained in sections 8.2.1 to 8.2.3 were shared with the participants. The
participants were asked to use their own computational methodology to compute the internal transient response.
There was no specification for type of model to use, capacitance and inductance values used in the model, and if
damping were include or not.

Finally some remarks about the “Fictitious Transformer”:

o This design is not an optimal transformer but could be constructed and it work very well in a real network

o All insulation distances are over sized, e.g., very conservative (manufacturers technical know-how is not
involved in this design)

o For simplicity conductors with large dimensions were used (in practice these dimensions would be reduced
to minimize eddy current losses)

8.2.1 Electrical Characteristics

The principal electrical characteristics of the “Fictitious Transformer” are:

o Two Windings Three Phase Transformer

o 60/80/100 MVA, ONAN/ONAF1/ONAF2

o 230/69 kV, YNd11, 60 Hz, Uz = 13%

o BIL HV: AC 460 kVef/ SI 850 kVcr/ LI 1050 kVr (Line and Neutral)

o BIL LV: AC 140 kVef/ LI 325 kVcr

o ±2×2.5% Off-Circuit Tap-Changer in HV Winding

The polarities and connections of one phase are shown in Figure 8.1.

Page 106
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Figure 8.1: “Fictitious Transformer” polarities and connections

8.2.2 Constructive Details

8.2.2.1 Core – “Three-Legs Core”

The type of core is “Three-Legs Core” with circular cross section formed with 12 steps (Figure 8.2).

o Core Material: Silicon Steel M4 0.27 mm with a Stacking Factor 0.97

o Core Section Diameter = 700 mm

o Section: 12 steps (690/670/650/620/590/560/520/480/440/400/360/310 mm)

o Axial Cooling Ducts: 2 × 6 mm between first and second step

o Stacking Height: 626.8 mm

2
o Core Section = 3379.27 cm

o BMAX = 1.7690 T

All dimensions in Figure 8.2 are in mm.

Figure 8.2: C ore details

Page 107
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

8.2.2.2 LV Winding – “Continuous Disk Type”

The low voltage winding is a conventional “Continuous Disk Type” with three conductors in parallel (Figure 8.3).

Figure 8.3: LV winding details

Construction details for the low voltage winding are:

1 Circuit per Leg

Copper Conductors: 3 × (16 mm ×3.15 mm)

Conductor Section: 3 × 49.85 mm = 149.55 mm (3.230 A/mm )

2 2 2

Insulating Paper (double side thickness): 0.72 mm

Paper Compression Factor: 0.90

Number of Turns: 433

Number of Disks: 92

Number of Turns per Disk: approximately 5 (4.71)

Axial Distance between Disks (from Bottom to Top):

45 × 4 mm + 1 × 6 mm + 45 × 4 mm

Axial Distance between Disks (Total Pressboard): 366 mm

Pressboard Compression Factor: 0.95

Pressboard Radial Spacers: 20 × 30 mm

Non-Compressed Height:

92×16 + 92×0.72 + 366 = 1904.2 mm ≈ 1905 mm

Compressed Height:

92×16 + 92×0.72×0.90 + 366×0.95 = 1879.3 mm ≈ 1880 mm

Radial Width: 5×3×(3.15+0.72)×1.034 = 60.0 mm

Diameters (Internal/External): 780 / 900 mm

Page 108
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

8.2.2.3 HV Winding – “Interleaved Disk Type”

The high voltage winding is a conventional “Interleaved Disk Type” (Figure 8.4).

Figure 8.4: HV winding details

Construction details for the high voltage winding are:

HV Line Terminal in the center of HV Winding (this winding is a center entry disk)

2 Circuits in Parallel per Leg: HV1 lower part and HV2 upper part

Copper Conductors: 1×(13.6 mm ×2.9 mm) in each circuit

Conductor Section: 2 × 38.89 mm = 77.78 mm (3.397 A/mm – Tap 5)

2 2 2

Insulating Paper (double side thickness): 1.68 mm

Paper Compression Factor: 0.90

Total Number of Turns (Tap 1) = 875 in each circuit

Number of Turns in each Tap:

2
Tap kV A/mm Turns

1 241.50 3.074 875

2 235.75 3.149 854

3 230.00 3.227 833

4 224.25 3.310 812

5 218.50 3.397 791

Table 8.1: Number of turns in each tap

Total Number of Disks = 84 (42+42)

76 (38+38) PR Disks (Principal) and 8 (4+4) TAPs Disks

18×PR + 4×TAPs + 20×PR + 20×PR + 4×TAPs + 18×PR

Number of Turns per Disk = 21 (20.82) PR / 21 (21.00) TAPs

Page 109
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Axial Distance between Disks (from Bottom to Top):

HV1 (41): 4×8mm + 14×7mm + (1×7mm + 1×32mm + 1×7mm) + 20×7mm

Between HV1 and HV2 (1): 1×11mm

HV2 (41): 20×7mm + (1×7mm + 1×32mm + 1×7mm) + 14×7mm + 4×8mm

Axial Distance between Disks (Total Pressboard):

68×7 mm + 8×8 mm + 1×11 mm + 2×(2×7 mm + 1×32 mm) = 643 mm

Pressboard Compression Factor: 0.9495

Pressboard Radial Spacers = 24 × 35 mm

Non-Compressed Height =

84×13.6 + 84×1.68 + 643 = 1926.5 mm ≈ 1927 mm

Compressed Height =

84×13.6 + 84×1.68×0.90 + 643×0.9495 = 1879.9 mm ≈ 1880 mm

Radial Width = 21×(2.9+1.68)×1.04 = 100.0 mm

Diameters (Internal/External) = 1060 / 1260 mm

The disks distribution in HV1 and HV2 are defined in Figure 8.5.

Details of the TAPs Disks in HV1 and HV2 are shown in Figure 8.6.

8.2.2.4 Core & Windings Layout

In Figure 8.7 the core and windings layout is defined along with the detail of the insulation between LV and HV
winding, the insulation between LV winding and core and the insulation between phases.

Figure 8.5: D isks distribution in HV1 and HV2

Page 110
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Figure 8.6: Details of the TAPs Disks in HV1 and HV2

Figure 8.7: C ore and windings layout

8.2.2.5 Tank – “Rectangular with Radiators”

The tank is rectangular equipped with plate radiators for cooling.

Figure 8.8 defines the insulation between windings and tank.

Page 111
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Figure 8.8: I nsulation between windings and tank

Based on the information in Figure 8.8 the tank dimensions are:

Width = 2270 mm (= 505 + 1260 + 505)

Large = 4540 mm (= 260 + 1260/2 + 2×1380 + 1260/2 + 260)

Height = 3840 mm

8.2.2.6 Losses and Reactance Calculation

To complete the transformer data and for the purpose of verification the following table contains the no load losses
at nominal voltage (Pfe), the short-circuit losses at 100 MVA and 75ºC (Pcu) and the short-circuit reactances at 100
MVA and 75ºC (Ux) for the nominal and extreme taps. The reactances were calculated using the Rabins Method.

Tap kV Pfe(W) Pcu(W) Ux(%)

1 241.50 73444 338520 12.923

3 230.00 73444 345841 13.026

5 218.50 73444 353925 13.250

Table 8.2: Losses and impedance in each tap

8.2.3 Modelling of Windings and Nodes and Branches Numbering

For modelling purposes each winding is divided in disk pairs. There are 46 branches (pair of disks) in low voltage
winding and 42 (21+21) branches (pair of disks) in high voltage winding.

Each model or software uses its own nodes and branches numbering system.

To make comparisons more convenient, for comparison purposes, the node numbering system detailed in Figure
8.9 is used.

Page 112
Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Figure 8.9: N ode numbering

Page 113
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

The definition, polarity and numbering of low voltage and high voltages branches are detailed in the following table:

Low Voltage Branches High Voltages Branches

BRANCH NODE+ NODE- BRANCH NODE+ NODE-

1 2 1 47 49 48

2 3 2 48 50 49

3 4 3 49 51 50

4 5 4 50 52 51

5 6 5 51 53 52

6 7 6 52 54 53

7 8 7 53 55 54

8 9 8 54 56 55

9 10 9 55 57 56

10 11 10 56 58 57

11 12 11 57 60 59

12 13 12 58 61 60

13 14 13 59 62 61

14 15 14 60 63 62

15 16 15 61 64 63

16 17 16 62 65 64

17 18 17 63 66 65

18 19 18 64 67 66

19 20 19 65 68 67

20 21 20 66 69 68

21 22 21 67 70 69

22 23 22 68 71 72

23 24 23 69 72 73

24 25 24 70 73 74

25 26 25 71 74 75

26 27 26 72 75 76

27 28 27 73 76 77

28 29 28 74 77 78

Page 114
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

29 30 29 75 78 79

30 31 30 76 79 80

31 32 31 77 80 81

32 33 32 78 81 82

33 34 33 79 83 84

34 35 34 80 84 85

35 36 35 81 85 86

36 37 36 82 86 87

37 38 37 83 87 88

38 39 38 84 88 89

39 40 39 85 89 90

40 41 40 86 90 91

41 42 41 87 91 92

42 43 42 88 92 93

43 44 43

44 45 44

45 46 45

46 47 46

Table 8.3: Numbering of low voltage and high voltages branches

According to the agreed upon node numbering scheme the winding terminals are:

Tap Nodes

X1 47

X0 1

H1 70 and 71

H0 48 and 93

Table 8.4: Numbering of winding terminals

The connections for the high voltage winding taps are indicated in the following table:

Tap kV Connect Nodes

1 241.50 58 to 59 83 to 82

2 235.75 59 to 57A 82 to 83A

3 230.00 57A to 59A 83A to 81A

Page 115
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

4 224.25 59A to 57 81A to 84

5 218.50 57 to 60 84 to 81

Table 8.5: Connections for the high voltage winding taps

8.3 List of Participants

The following table lists in alphabetical order the members that participate in the “Fictitious Transformer”
calculations. These participants came from universities, manufacturers and independent consultants:

Member Country Representing

Robert Degeneff United States Utility System Technologies Inc.
Poliana Duarte Brasil ABB
Jorge Leiva Argentina Tubos Trans Electric
Xosé López-Fernández Spain Universidad de Vigo
José Miguel Torres Portugal EFACEC Energia
Daniil Matveev Russia All-Russian Electrotechnical Institute
Triomphant Ngnegueu France Alstom
Álvaro Portillo Brasil WEG Equipamentos Elétricos
Barend van Jaarsveld South Africa Powertech Transformers
Jos Veens The Netherlands SMIT Transformatoren
Shin Yamada Japan Toshiba
Zan Ji Wang China Tsinghua University
Table 8.6: List of participants
Not all the members submitted all the calculations and some members made more than one calculation for each
variant (with different softwares or with the same software with different options for example with and without
damping effects).

To preserve the identity of the members the different calculations will be identify with letters (A to N).

8.4 Questionnaire
The following questionnaire was submitted to each of the participants. This information provided the principal
features and details of the softwares used in the “Fictitious Transformer” calculations:

1. Model Type

• Lumped (or Concentrated) Parameters (R, L, C)

System of Ordinary Differential Equations

• Distributed Parameters

Transmission Line Model or Multiconductor Transmission Line (MTL)

System of Partial Differential Equations

2. Damping Effects (included or not)

3. Solution in Time-Domain or in Frequency-Domain

4. Numerical or Analytical Solution of Equations

Page 116
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

5. Modelling of Core Effects

The following table provides a summary of the responses to the questionnaire:

A B C D E F G H I J K L M N

Model Type LP LP LP LP LP LP LP LP LP LP LP LP LP LP

Damping Effects N Y N Y N Y Y Y N Y Y N Y Y

Time-Domain Y Y Y Y Y N N Y Y Y Y Y Y Y

Frequency-Domain N N N N N Y Y N N N N N N Y

Numerical Solution Y N Y N Y Y N Y Y N Y Y Y Y

Analytical Solution N Y N Y N N Y N N Y N N N Y

Core Effects Y Y N Y N Y Y N Y Y Y N Y Y

LP = Lumped Parameters DP = Distributed Parameters Y = Yes N = No

Table 8.7: Summary of responses to the questionnaire

It can be seen that all the member softwares use lumped parameters to model the transformer and most of then
solve the equations in time domain.

In reference to damping, some members don´t take into account damping effects, other members use constant
resistances tuned to fundamental oscillation frequency and other members use damping effects function of
frequency (in general more frequency means more damping).

8.5 Variants to Calculate

To test the simulation tools for transient calculations inside the transformer windings of the “Fictitious Transformer”,
it will be applied a standard lightning impulse wave (often referred to as a full wave) and a damped oscillatory wave
shapes simulating switching overvoltages.

8.5.1 Lightning Impulse

For this portion of the analysis of the “Fictitious Transformer” response to a lightning Impulse wave, it is defined
four variants according to Figure 8.10:

Variant FT1 Variant FT2

1050 kVcr Lightning Impulse applied to H1 1050 kVcr Lightning Impulse applied to H0

H0, X1, X0 connected to ground H1, X1, X0 connected to ground

HV in Tap 5 (Connect: 57 to 60, 81 to 84) HV in Tap 5 (Connect: 57 to 60, 81 to 84)

Variant FT3 Variant FT4

325 kVcr Lightning Impulse applied to X1 325 kVcr Lightning Impulse applied to X0

X0, H1, H0 connected to ground X1, H1, H0 connected to ground

HV in Tap 5 (Connect: 57 to 60, 81 to 84) HV in Tap 5 (Connect: 57 to 60, 81 to 84)

Page 117
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Figure 8.10: Lightning impulse variants

For this calculation all the participants use the same 1.2/50 µs standard impulse wave defined as (Figure 8.11):

(
V (t ) = VMAX ⋅ A ⋅ e −α ⋅ t − e − β ⋅ t )
A = 1.038 α = 0.015 × 106 β = 2.47 × 106

VMAX is the peak voltage in kV and t is the time in seconds

Figure 8.11: Lightning impulse wave

To compare the results of the different softwares for these four cases, the work is divided in two parts:

o Analysis of maximum values of the voltages in all nodes and branches

o Analysis of wave shapes responses in some of the nodes and branches

8.5.1.1 Nodes and Branches Maximum Voltage Values

The comparative results for these calculations are summarized in Appendix A.1 for the maximum voltages in the
nodes and in Appendix A.2 for the maximum voltages in the branches.

From Appendix A.1 and A.2 comparative results, it can be concluded:

o Relative good agreement in maximum node voltage responses (in the node maximum voltage value, not in the
time in which this maximum occur).

o Poor agreement in maximum branch voltage responses.

Page 118
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

The branch voltages are differences between node voltages of similar wave shapes. This fact increases the
errors in the numerical calculations.

It can be seen an example of that in Figure 8.12 for Variant FT1 without damping effects for Nodes 69 and 70
who define Branch 67 (Node 70 - Node 69).

o In all the cases the agreement is good for values that define the project insulating distances (defined by the
maximum values of nodes and branches voltages).

This means that if all participants work with the same distance design rules (for example distances in function
of voltage tables) all arrive to the same transformer dielectric design.

Figure 8.12: Nodes and Branch voltages

o Neglect damping effects are not always conservative in the range of the first microseconds principally for the
branch voltages. It is possible to obtain bigger calculated voltages in branches in the first microseconds with
damping effects included in the model.

Obviously in the long term (10 to 50 microseconds and more) neglect damping effects is always conservative
and the calculated values neglecting damping effects are higher and more oscillatory than values calculated
considering damping effects.

For example in Figure 8.13 for Node Voltage 69 the red curve (with damping effects) is little less than the blue
one (without damping effects).

In Figure 8.14 for Branch Voltage 67 the red curve (with damping effects) is greater than the blue one (without
damping effects). The peak value of the red curve is later in time and 32% greater than the peak of the blue
curve.

This result is logical if Branch 67 Voltage is considered as the difference between Node 70 and Node 69
Voltages. Node 70 Voltage is the applied impulse (constant and not depending of damping effects) and Node
69 is less for the damping wave. Conclusion: the difference (Node 70 - Node 69 = Branch 67) is higher for the
damping wave.

Comparing the wave shapes for Branch Voltage 67 up to 500 microseconds in Figure 8.15, it can be seen
clearly the effect of damping (the high frequency oscillations of the blue curve disappear in the red one).

Page 119
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Figure 8.13: Node 69 voltage with and without damping

Figure 8.14: Branch 67 voltage with and Figure 8.15: Branch 67 voltage with and
without damping without damping

8.5.1.2 Node and Branches Voltage Wave Shapes

For the analysis of wave shapes responses in some of the nodes and branches the wave shapes are calculated
between 0 and 100 µs, for node and branches voltages for the four variants defined in 8.5.1 according to the
following details:

Variant FT1 Variant FT2

Nodes 73, 72, 71=70, 69, 68 Nodes 48, 49, 50, 93, 92, 91

Branches 73-72, 72-71, 70-69, 69-68 Branches 48-49, 49-50, 93-92, 92-91

Branches 58-59, 82-83, 70-24 Branches 58-59, 82-83, 70-24

Comparative results in Appendix A.3 Comparative results in Appendix A.4

Variant FT3 Variant FT4

Nodes 47, 46, 45 Nodes 1, 2, 3

Branches 47-46, 46-45 Branches 1-2, 2-3

Branches 58-59, 82-83, 70-24 Branches 58-59, 82-83, 70-24

Comparative results in Appendix A.5 Comparative results in Appendix A.6

From comparative results in Appendix A.3 to A.6, it can be concluded:

Page 120
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

o Similar tendencies in all softwares temporal responses

The wave shapes are strongly depending of the resonance or natural frequencies of the transformer. It can be
seen clearly in the waves of the different softwares differences in time responses due to differences in
resonance frequencies

o Relative good agreement in node voltages temporal responses

For high voltage nodes the maximum wave values are in the first microseconds and are very similar for all the
softwares. It can be seen differences in the waves for time higher than 50 microseconds but in this time the
values are less than 0.5 pu and are not important.

For low voltage nodes all the softwares show the effect of travelling wave reflections at 25, 50, 75 and 100
microseconds.

o Poor agreement in branch voltages temporal responses

The maximum values are similar for all the softwares but the wave shapes are similar up to 20 microseconds
but for higher times an increase in the differences can be noted.

o Poor agreement in TAPs voltage temporal responses

Same comments as for branch voltages.

o Poor agreement in HV-LV transferred voltage temporal responses

Same comments as for branch voltages.

For branch, TAPs and HV-LV transferred voltages it is recommend to use higher safety margins in the dielectric
design in view of the spread of software results.

8.5.2 Switching Impulse

To conclude this work with the “Fictitious Transformer”, it will be defined an oscillatory wave shape to apply to the
high voltage terminals of the “Fictitious Transformer” simulating a switching overvoltage (switching not VFTO)
created by the power system.

The first resonance frequency of the transformer will be calculated with all softwares and then a sinusoidal damped
wave with this frequency will be applied to the transformer.

8.5.2.1 Definition of the Damped Oscillatory Wave Shape

To define the Damped Oscillatory Wave Shape the following formula is used:

t
−
V (t ) = MAX ⋅ e T ⋅ sen (ω ⋅ t )
V
FMAX

Where:

2 ⋅π
ω = 2 ⋅π ⋅ f T= ω ⋅ tMAX = Arctg (ω ⋅ T )
ω ⋅ Ln (D )

t MAX
−
FMAX = e T
⋅ sen (ω ⋅ tMAX )

VMAX = 850 kVcr Maximum Voltage (first peak)

Page 121
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

f = 7920 Hz Oscillatory Frequency of the wave (approximately equal to the first resonance or natural
frequency of the transformer)

D = Damping Factor of the wave

T (s ) = Time Constant of the wave

t MAX (s ) = Time of the first maximum of the wave

FMAX = Factor to adjust the value of the first maximum of the wave

8.5.2.2 Relation between Time Constant and Damping Factor

The Time Constant T and the Damping Factor D are related by the formula:
2 ⋅π
T =−
ω ⋅ Ln (D )

A2
The meaning of D is easier to understand than T : D =
A1

Where:

A1 = first positive peak of the wave at ω ⋅ t ≈ 0.5 ⋅ π

A2 = second positive peak of the wave at ω ⋅ t ≈ 2.5 ⋅ π

Analyse of the responses of the “Fictitious Transformer” for different Damping Factors will be done.

8.5.2.3 Damped Oscillatory Wave Shapes - Numerical Expressions

In the next four Figures (8.16 to 8.19) it can be seen the wave shapes for different values of the Damping Factor
D (0.6, 0.7, 0.8 and 0.9) corresponding to the following formulas:
t (s )
−
• D = 0.6 → V (t ) = 962.60 ⋅ e 247.17×10 −6
⋅ sen [49762.83 ⋅ t (s )] in kV
t (s )
−
• D = 0.7 → V (t ) = 927.78 ⋅ e 354.00×10 −6
⋅ sen [49762.83 ⋅ t (s )] in kV
t (s )
−
• D = 0.8 → V (t ) = 898.20 ⋅ e 565.84×10 −6
⋅ sen [49762.83 ⋅ t (s )] in kV
t (s )
−
• D = 0.9 → V (t ) = 872.56 ⋅ e 1198.39×10 −6
⋅ sen [49762.83 ⋅ t (s )] in kV

Page 122
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Figure 8.16: Damped Oscillatory Wave Figure 8.17: Damped Oscillatory Wave
with D = 0.6 with D = 0.7

Figure 8.18: Damped Oscillatory Wave Figure 8.19: Damped Oscillatory Wave
with D = 0.8 with D = 0.9

8.5.2.4 Fictitious Transformer Responses

The oscillatory damped wave shapes will be applied to the high voltage winding (Node 70 = 71) and the response
to this excitation in the center of the lower part of the high voltage winding (Node 57 connected to node 60) will be
analysed.

The following variants will be considered:

Variant FT5 Variant FT6

Damped oscillatory wave shape with D = 0.6 Damped oscillatory wave shape with D = 0.7
applied to H1 (Node 70=71) applied to H1 (Node 70=71)

Comparative results in Appendix A.7 Comparative results in Appendix A.8

Variant FT7 Variant FT8

Damped oscillatory wave shape with D = 0.8 Damped oscillatory wave shape with D = 0.9
applied to H1 (Node 70=71) applied to H1 (Node 70=71)

Comparative results in Appendix A.9 Comparative results in Appendix A.10

For all variants it was considered H0 (Node 48 connected to node 93), X1 (Node 47) and X0 (Node 1) all
connected to ground. If possible, it is desired to calculate the responses in each variant with and without taking into
account the damping effects inside the transformer.

Page 123
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

From comparative results in Appendix A.7 to A.10, it can be concluded:

o There are differences in the calculated first resonance frequency of the transformer with the different softwares
(Between 8 kHz to 15 kHz). As consequence it can be seen differences in the temporal wave shapes (similar
wave forms with phase differences as consequence of different frequencies).

o Similar tendencies in all softwares temporal responses relating to maximum voltage values (with and without
taking into account the damping effects inside the transformer).

o The transformer damping effects, when taking into account, give similar amplitude results for all tested
softwares.

o The transformer damping effects are very important in the amplitude values of the resultant waves. Its influence
increase when the damping factor D of the oscillatory wave shape increase. For D = 0.6 the amplitude
reduction due to transformer damping effects is around 38% and for D = 0.9 the amplitude reduction due to
transformer damping effects is around 66%.

8.6 Time Domain and Frequency Domain Severity Factors

For Variants FT5, FT6, FT7 and FT8 the Time Domain Severity Factor (TDSF) and the Frequency Domain Severity
Factor (FDSF) for the “Fictitious Transformer” are calculated when an oscillatory wave shape with a frequency of
14.91 kHz is applied in the high voltage terminal H1 (Node 70).

The wave shapes in the center of the lower part of the high voltage winding (Node 60) when the oscillatory wave
shape is applied to H1 terminal (Node 70) are showed in Appendix A.7, A.8, A.9 and A.10, in the curves labeled
with F (transformer model taking into account internal damping effects).

The maximum voltages to ground and the TDSF of the HV winding nodes for the different damped oscillatory
waves are shown in the Figure 8.20.
Maximum Voltage in HV Winding Time Domain Severity Factor in HV winding
12 2.2
D = 0.6, f = 14.9 khz D = 0.6, f = 14.9 khz
D = 0.7, f = 14.9 khz 2 D = 0.7, f = 14.9 khz
10 D = 0.8, f = 14.9 khz D = 0.8, f = 14.9 khz
D = 0.9, f = 14.9 khz D = 0.9, f = 14.9 khz
1.8
Envelope Envelope
8
1.6
Voltage (p.u.)

TDSF

6 1.4

1.2
4
1

2
0.8

0
70 68 66 64 62 60 58 56 54 52 50 48 70 68 66 64 62 60 58 56 54 52 50 48
Nodes of HV winding Nodes of the HV winding

(a) (b)
Figure 8.20 a) Maximum voltage to ground of the HV winding nodes for different damped
oscillatory waves. b) TDSF of the HV winding nodes for different damped oscillatory waves.
The maximum branches voltages of the HV winding and the corresponding TDSF for the different damped
oscillatory waves are shown in Figure 8.21.

Page 124
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Maximum Voltage Drop between disks in HV Winding Time Domain Severity Factor in HV winding
1.3 2.5
D = 0.6, f = 14.9 khz
D = 0.6, f = 14.9 khz
1.2 D = 0.7, f = 14.9 khz
D = 0.7, f = 14.9 khz
D = 0.8, f = 14.9 khz
1.1 2 D = 0.8, f = 14.9 khz
D = 0.9, f = 14.9 khz
D = 0.9, f = 14.9 khz
Envelope
1 Envelope
Voltage (p.u.)

0.9 1.5

TDSF
0.8

0.7 1

0.6

0.5 0.5

0.4

0
68 66 64 62 60 58 56 54 52 50 48 46 68 66 64 62 60 58 56 54 52 50 48 46
Branch of the HV winding Branch of the HV winding

(a) (b)
Figure 8.21 a) Maximum voltage drop between HV winding discs for different damped
oscillatory waves. b) TDSF between HV winding discs for different damped oscillatory waves.
Figure 8.22 presents the energy spectral density and the FDSF for different the damped oscillatory waves.

8
Energy Spectral Density in HV winding Frequency Domain Severity Factor in HV winding
10 25
D = 0.6, f = 14.9 khz D = 0.6, f = 14.9 khz
6 D = 0.7, f = 14.9 khz D = 0.7, f = 14.9 khz
10
D = 0.8, f = 14.9 khz D = 0.8, f = 14.9 khz
20
D = 0.9, f = 14.9 khz D = 0.9, f = 14.9 khz
Energy Spectral Density (V.s)2

4
10 Envelope Envelope

2 15
10
FDSF

0
10
10

-2
10

5
-4
10

-6
10 0
2 3 4 5 6 2 3 4 5 6
10 10 10 10 10 10 10 10 10 10
Frequency (Hz) Frequency (Hz)

(a) (b)
Figure 8.22 a) Energy spectral density of the different damped oscillatory waves. b) FDSF of the
different damped oscillatory waves.

Only for the Variant FT5 with D = 0.6 the TDSF is less than one for all the nodes and branches. For the other
cases, with lower damping factors, it can be found nodes and branches with TDSF higher than one. In particular for
the Variant FT8 with D = 0.9 most of the nodes and branches present TDSF higher than one.

This confirm that there is a high fault probability when oscillatory wave shapes with frequency equal to one of the
resonance frequencies of the transformer are applied.

In the frequency domain all the waves present values higher than the reference envelope (impulse test). This
shows that the FDSF is more conservative than TDSF.

8.7 Conclusions
From the analysis of calculations results from section 8.5, for lightning impulse wave and damped oscillatory
waves, the principal conclusions are:

o Good agreement in maximum voltage values for nodes and branches

This result validated these calculation tools for safety dielectric transformer design regarding to internal
insulation distances inside and outside the windings.

Page 125
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

o Differences in natural frequencies with the consequence of poor agreement in temporal responses. Different
transient simulation programs will lead to a different internal oscillating behavior of the winding. For that reason
the simulated admittance matrix might not match exactly the measured admittance matrix.

o The resonance frequencies are strongly dependent on the values of the inductances (self and mutual) and
capacitances that were used to represent the transformer.

Some members performed examples using the same values of inductances and capacitances and in that case
the internal voltages obtained with the different softwares were identical.

o All compared softwares use lumped parameters to model the transformer. For the usual modelling practice for
windings using one branch for each two disc the validity frequency range go up to approximately 500 kHz.

Degeneff [8.1] states the rule to know the validity frequency range of a lumped parameters model: "In a valid
model, the highest frequency of interest would have a period at least ten times larger than the travel time of the
largest winding segment in the model".

o To modelling for higher frequencies using lumped parameters models is necessary divide the windings turn by
turn and the behavior of leads, bushings, tank wall, shield, should be considered detailed too.

o For Very Fast Transient Overvoltages the transformer is modelled using transmission lines (distributed
parameters models). Marjan Popov [8.2] uses a hybrid model which is a combination of the multiconductor
transmission line model (MTLM) and the single-transmission line model (STLM).

o These different oscillating behavior of the models leads to the fact that the simulation of the transient
interaction between transformer and power grid might have a reduced accuracy especially for high frequencies
(higher than 1 MHz) if usual calculation models are not improved.

o This study proof that the tools used by manufacturers are good for transformer dielectric design but are not
capable for accurate determination of the natural or resonance frequencies of the transformers. In every case
that an internal resonance problems in the network-transformer interaction should occurs is necessary the
determination of resonance frequencies by measurement.

o Manufacturers must try to improve their models with the objective not only to obtain good approximations for
the maximum values of internal voltages but also obtain better temporal responses. For this, is essential an
improvement in the used methods for calculating inductances and capacitances that represent the transformer.

8.8 References
[8.1] James H. Harlow: "Electric Power Transformer Engineering", 2004, CRC Press Chapter 3.10: Transient-
Voltage Response by Robert C. Degeneff

[8.2] M. Popov, L. van der Sluis, G. C. Paap and H. De Herdt: "Computation of Very Fast Transient Overvoltages
in Transformer Windings", IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 18, NO. 4, OCTOBER
2003, pp 1268-1274

Page 126
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

9 RECOMMENDATIONS

9.1 General considerations on System Aspects

Ideally, transformers would be designed to be immune to system transients from a viewpoint of insulation
coordination. In most aged networks, being today between 20 and 30 years, insulation coordination studies have
already been performed with the aim of the determining the appropriate insulation levels for the equipment within
the range from steady-state to transients conditions. Thus the standard tests are usually performed on the
transformers being purchased before their installation on the electrical network. However, these standard tests can
not cover all transient wave shapes that may be imposed on the transformer, in particular oscillating overvoltages.
The shape of the overvoltage applied to the transformer terminals will depend on network configuration, type of
switching and external phenomena such as lightning and faults.

In practice, it is impossible to manufacture transformers with immunity to all system transients due to the resonant
nature of windings and the cost associated with over-designing the insulation for protecting it from a very specific
phenomenon. However, it is recommended that utilities design their network to protect the transformers from the
system transients including switching and lightning surges as much as possible.

The following subsections show some examples of practices that would improve the reliability of transformers. The
advantages of such practices on power quality and on equipment reliability justify the cost invested in these
mitigation methods (e.g. controlled switching) for utilities [9.1]-[9.3].

9.1.1 Shunt capacitor banks energization

It is known that the static capacitor bank switching may causes transformer failures due to energizations and re-
strikes [9.4]-[9.6]. When energization occurs, the capacitor appears to the system as a sudden short circuit. The
voltage at the transformer terminals drops abruptly and then recovers through with high-frequency oscillation
determined by the RLC nature of the network. When one or more capacitors are in service on the same bus, back-
to-back switching involving the energization of a capacitor bank causes high frequency inrush currents circulating
between the capacitor banks.

Recommendations: Transients caused by the energization of capacitor banks can easily be controlled by using
closing resistors, series reactors or controlled switching [9.1]-[9.3].

9.1.2 Shunt capacitor banks interruption

Interruption of small capacitive currents can cause a re-strike in modern SF6 breakers. International Standards IEC
62271-100 defines different degrees of re-strike probability: C1 (low probability) and C2 (very low probability). Re-
strike in the circuit breaker when interrupting the current of a capacitor bank is highly likely to produce high-
frequency transients. Multiple re-strikes are also possible. Specifying a class C2 breaker minimizes the probability
of transformer failures. The usage of controlled switching during opening is also highly effective by minimizing the
critical short arcing times [9.1]-[9.3].

9.1.3 Transmission lines energization

The energization of transmission lines or cables causes switching overvoltages during both the energization and
fast re-closing. When the circuit-breaker is separated from the transformer by a cable or overhead line, the
operation of the breaker causes a travelling wave to propagate along the cable/line towards the transformer. The
resulting overvoltage at the transformer may be oscillatory in nature due to repetitive wave reflections and can
therefore excite the transformer winding resonances. The buildup of internal overvoltages within the transformer
has the highest probability of occurring in the case of a moderate length of the line or cable. In that situation, the
frequency of the oscillating overvoltage has a significant probability of matching one or more of the natural
frequencies of the transformer and dielectric failure may result due to the buildup of these internal voltages [9.7].

Recommendations: Transients caused by the energization of transmission lines can be controlled or reduced by
using closing resistors or controlled switching [9.1]-[9.3]. In the case of the closing resistors, the resistance is

Page 127
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

located in the circuit-breakers, with a value close to the surge impedance of the lines, e.g. 300 ohms [9.8] for an
overhead line. As a result, the initial front wave is cut to a low value.

Alternatively, remote energization should be avoided by installing an extra circuit breaker on the transformer side of
the cable/line.

9.1.4 Transformers energization

The energization of transformers with circuit breaker very near its terminal produces inrush currents whose level
and shape depend upon the transformer characteristics, closing times of the circuit-breaker poles and transformer
residual flux. The frequency spectrum of the inrush currents is essentially composed of low-order harmonics,
generally up to 500 Hz.

The presence of a short substation line or cable will generate very high frequency overvoltages at the transformer
terminals similar to the phenomena described in the previous section.

Recommendations concerning the switching surges: These high frequencies transients can easily be controlled by
using circuit-breakers with pre-insertion resistances or synchronous switching [9.1]-[9.3].

Recommendations concerning the use of controlled switching: Controlled switching can be applied to reduce inrush
currents. In order to do that, the switching could takes place (it depends of the residual flux) when the voltage is
near the maximum. This will provoke a very high steep voltage at the transformer terminal that may be dangerous
for the insulation, especially for old equipment. [9.9]

9.1.5 Switching in GIS (Gas Insulated Substations)

The switching of circuit breakers and disconnectors may generate extremely steep-fronted, oscillating overvoltages
in the substations. When closing a breaker, a steep-fronted voltage wave starts propagating into the substation.
This voltage wave results in an oscillating overvoltage at the transformer which may cause the transformer to fail
due to the steep front and/or subsequent high frequency oscillations [9.10]-[9.13]. The operation of disconnectors
may result in multiple restrikes which places the transformer in a higher risk of failure.

Recommendations on very high frequency overvoltages: If possible, operating the disconnectors in a GIS should
be avoided or forbidden. The presence of a short cable or a bay between the transformer and GIS will change the
wave shape of the voltage which may contribute to reduce this problem.

9.1.6 Lightning overvoltages

In the insulation coordination of transformers, lightning overvoltages are one of the most significant factors to be
taken into account. For protecting transformers from lightning overvoltages, surge arresters are installed near
transformers. Although this is the recommended insulation coordination practice, transformer internal winding
damage have been reported [9.14]-[9.15] due to the subsequent oscillating voltage that propagates inside the
transformer. In all these cases, the surge arresters have operated correctly. These results indicate that the incident
surge voltages were reduced to below the Lightning Impulse Withstand Voltage (LIWV) level of the transformers by
the usage of the surge arresters. Some failure reports indicate that some frequency components of incident surges
caused by lightning resonated with the natural frequency of the transformers, and the resultant internal resonance
overvoltages reached the breakdown voltage level of the transformers insulation structure [9.16].

Recommendations on lightning overvoltages: Good practices of transmission line design and substation
coordination insulation including proper shielding, surge arrester location, line surge arrester etc.

9.1.7 Protection against multiple restrikes

Under specific circumstances, when the circuit break opens close to the natural current zero, the current can be
chopped. This phenomenon leads to increased TRV that rises faster than the withstand capability of the circuit
breaker and may result in multiple restrikes. Multiple restrikes contain oscillations within broad frequency range that
may lead to severe overvoltages since these oscillations can easily match one of the many resonance frequencies
of the transformer. In this case, the common practice is to apply R-C snubbers for medium voltage transformers or
surge arresters. When applying surge arrester protection, one should note that the surge arresters limit the

Page 128
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

amplitude of the overvoltages to approximately 3 p.u. However, especially for transformer delta windings, the
voltage to which the winding is exposed is the phase to phase voltage. Besides, in different phases the voltages
may have the same amplitude but different polarity that might result in an overvoltage with double amplitude. Thus,
despite the fact that the overvoltage is limited by the surge arrester within 3 p.u., the total voltage to which the
winding is exposed may reach 6 p.u. Therefore, in particular cases the transformer should be supplied not only with
surge arresters connected phase to ground, but also with three additional arresters connected phase to phase
[9.17].

9.2 General considerations on Transformer Design Practices

In transformer winding internal dielectric design most of the effort is expended in the effort to reduce the internal
oscillating voltages within the windings during transient events.

To reduce internal voltage oscillations in windings during transients is necessary to increase the winding series
capacitance CS in relation to winding parallel capacitance (to ground or between windings) CP.

Usually the S factor is defined as:

UV
S
(9.1)
T
UW

This parameter indicates the degree of deviation of the initial voltage distribution decided solely by winding
capacitances from the final linear voltage distribution which is decided solely by turn’s ratio. The higher the value
of S, the higher are the deviation and amplitude of oscillations which occur between the initial and final voltage
distributions.

For a conventional continuous disk winding (Figure 9.1), the value of X may be in the range of 5 to 30. Any change
in the transformer design, which decreases the distribution constant X of the windings, results in a more uniform
voltage distribution and reduces the voltage stresses between different parts of the winding.

Figure 9.1: Conventional continuous disk winding.

The initial voltage distribution of the winding can be made closer to the ideal linear distribution (X 0) by increasing
its series capacitance and/or reducing its capacitance to ground.

Page 129
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

The capacitance to ground and between windings is mainly determined by insulation levels, and thus is difficult to
reduce practically or economically. The series capacitance of the winding can be modified more easily, e.g. by
using a different type of winding or by rearrangement of the winding conductors.

It is normal practice to use disc type windings for high voltage windings and the series capacitance of disc type
windings can be improved by arrangements using interleaving (Figure 9.2a) or intershielding (Figure 9.2b)
techniques, mainly near the phase end of the winding.

(a) (b)
Figure 9.2: (a) Interleaved disk winding. (b) Intershield disk winding.

More details can be found in references [9.18] and [9.19] for interleaving disc windings and in references [9.20] and
[9.21] for intershielding disc windings. An elaborate treatment of this subject is presented in reference [9.22].

Surge arresters installed in the line or outside the transformer tank in general do not mitigate the consequences of
the (winding-internal) resonance, because resonance can be initiated by transient voltages that are lower than the
level of protection of the arrester.

However, one way to avoid the effects of winding internal resonance is to install ZnO varistors in parallel with the
critical parts of the winding in order to limit the overvoltages [9.23] [9.24] [9.25]. The tap windings or the taps in a
main winding are typically critical points considering resonances and therefore are sometimes protected by ZnO
varistors (provided the transformer specification allows this).

The use of ZnO varistors in parallel with certain parts of the transformer windings is one of the solutions to
suppress the oscillations due to the transient waves and the possibility of resonance. The ZnO varistors used today
(when properly dimensioned) are reliable, have a low rate of failure rate, and by limiting the voltage and reducing
the oscillations, increase the reliability of the transformers. One slight drawback is that the lightning impulse test
during factory acceptance testing is more elaborate [9.26], [9.27].

Another important aspect that should be taken into consideration in the insulation design was discussed in chapter
7 and is related to two electrical conditions that may impose additional stress on in-service oil-immersed
transformer insulation. They are the following:

• The effect of repetitive voltage application (V-N), considering the relationship between the number of
repetitions N and the breakdown voltage V.

• The influence of the voltage shape, especially related to the application of different rise times.

A full understanding of how these conditions affect insulation system is still a challenge. Future investigation in this
field is highly recommended due to its importance to transformer operation reliability.

Page 130
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

9.3 Transformer Specification

9.3.1 Dielectric tests
The study of high-frequency resonances in the windings of EHV transformers is a very complex topic for utilities.
First, it is impossible to measure internal stresses of a transformer in service. Also, the expertise and the tools
which are needed to model transformers are rare and often limited to manufacturers. Utilities often need to
considered transformers as black boxes where only voltages at external bushings can be studied. Moreover, even
with the possibility of getting an appropriate model from manufacturers, the studies often can't be done by utilities in
the planning stage of projects because of insufficient data (e.g. new substation conception unknown) and the
unavailability of the transformer model.

In that regard, an acceptance test program is a necessity for utilities to validate the performance of a transformer's
insulation. One of the main criteria in the design of power transformers is in fact the acceptance test program.
Insulation coordination of transformers is typically calculated in a way that the test is passed with little margin to
limit unnecessary cost for the manufacturer [9.28]. A dielectric test with focus on high frequency electromagnetic
transients is yet to be defined in standards. However, some standardized dielectric tests are susceptible to amplify
the voltage at high-frequency resonances in the windings.

As presented in chapter 2, according to IEC 60076-3, the following test voltage shapes are standardized:

• Switching impulse test (SI) - an impulse voltage having a time to peak of over 100 µs and a time to first
zero crossing of at least µs 500 but preferably 1000 µs.
• Lightning impulse test (LI) - an impulse voltage having a front time of 1,2 µs and a time to half value of
50 µs.

The following special voltage shape is also standardized in the 2013 version of IEC 60076-3 (it was optional
previously) for transformers with Um > 170 kV:

• Lightning impulse chopped on the tail (LIC) - an impulse voltage having a front time of 1,2 µs, a time to half
value of 50 µs and a time to chopping between 2µs and 6µs.

Similar dielectric tests are specified in IEEE C57.12.90.

Considering the high cost and strategic importance of transformers, dielectric tests are a necessity for utilities to
control the quality and the conception of transformers. In this situation, it is essential that the acceptance tests and
its test levels specified by utilities be representative of real service conditions. Recommendations to improve
dielectric tests are made in the following sections.

9.3.1.1 Specifying Lighting Impulse Chopped on the Tail (LIC)

To achieve adequate testing of high frequency resonances, dielectric tests must represent real life conditions. The
frequency spectrum of the test waves applied to transformer must be sufficient over a wide range of frequencies to
cover network transients.

Figure 9.3 shows the Fourier Transform (FT) of the LI, SI and LIC with chopping time ranging from 2 µs to 6 µs,
expressed in volt-seconds, according to [9.28] and [9.64].

Page 131
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Figure 9.3: Frequency spectrum of typical test voltages [9.64]

As discussed in section 6.4.4, after calculating the frequency spectrum of each standard waveform, an envelope
consisting of the highest value found for each frequency can be calculated. This envelope presents a frequency
range which is directly related to each standard waveform spectra as shown below:

the standard switching waveform (100/1000 µs) up to a frequency of 2 kHz;

the lightning waveform (1.2/50 µs) from 2 kHz to 30 kHz and;
the chopped lighting waveforms from 30 kHz.

The spectral analysis of network transients can be compared to this envelope. For example, reference [9.28] shows
that the waveforms spectrum at high frequencies of transient dielectric tests is sufficient to cover typical switching
transient measured on his network. However, it is to be expected that the envelope above be exceeded by some
transients as demonstrated in some of the examples in “Part 2: Case Studies”.

From the FT analysis, the following conclusions are made:

the frequency spectrum of the SI impulse is insufficient compared to the other waveforms to excite
resonance in power transformers which are normally at frequencies over 2 kHz;
the frequency spectrum of the LIC waveforms exceed the frequency spectrum of the LI waveform from
30 kHz (t chop = 6 µs) up to 1 MHz;
the LI waveform is highly susceptible to excite a resonance throughout the windings of transformer which
is often found in the 5 to 30 kHz range.

From these results, utilities should specify the lighting impulse chopped on the tail (LIC).

9.3.1.2 Upgrading Test Levels

As seen in the previous section, the LI (Lighting Impulse) and the LIC (Lighting Impulse Chopped) dielectric tests
are susceptible to excite one or many resonances in the windings of power transformers causing possible
breakdowns. The voltage levels for these tests are standardized by IEC and ANSI. However, specific values are
not given in these standards in favor of limit values (min/max). Utilities have the choice to specified voltage levels
above the minimal and maximal limits recommended by these standards. Utilities with low voltage levels should

Page 132
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

consider upgrading the levels for the LI and the LIC dielectric tests [9.29]. Such a practice will increase the
probability that transients applied during a transformer life will be covered by the standard wave tests [9.28].

9.3.1.3 Adding Unloaded LI and LIC dielectric tests to specification

Adequate testing of high frequency resonances in power transformers can only be achieved if dielectric tests
represent real life conditions. Test connections are defined in standard IEC 60076-3 for each dielectric test:

A) Switching impulse test (SI)

During the test the transformer shall be in a no-load condition. Windings not used for the test shall be solidly
earthed at one point but not short-circuited. Performed this way, the SI test doesn't represent all service conditions
where switching transients will be applied on power transformer (i.e. with all line terminals connected to lines).
However, as seen previously, the switching impulse test (SI) isn't determinant for high frequency resonances. The
impact of such practice should be minimal.

B) Lightning impulse test (LI) and Lighting Impulse Chopped (LIC)

Untested line terminals shall be earthed directly or through low impedance, not exceeding the surge impedance of
the connected line. However, real service conditions are not well represented with grounded terminals under
standard LI and LIC tests. Line terminals of a transformer in service are often unloaded and protected by surge
arresters. Using white-box models and lab measurements, those guidelines were proven to be insufficient to
adequately test the insulation of a transformer in regard to high-frequency resonances [9.28], [9.29], [9.32], [9.33],
[9.64]. A special testing procures where terminal connections are unloaded and terminated by surge arrestors is
proposed in the literature [9.29], [9.33], [9.64]. It is necessary for utilities to investigate the impact of such
guidelines on the effectiveness of dielectric tests in regard of high frequency resonance failures. As an example,
reference [9.64] presents a case study on this topic.

LI and LIC dielectric tests perform with all untested line terminals grounded as recommended in IEC 60076-3 is
insufficient to test appropriately all possible high-frequency resonances that could be excited in the service life of a
transformer. It is recommended that transformers, EHV and strategic transformers at least, be tested with LI and
LIC dielectric tests following these guidelines:

with untested line terminals grounded;

with untested line terminals unloaded and protected by arresters;
with untested line terminals connected to a typical line impedance value and protected by arresters.

9.3.1.4 Additional Dielectric Tests

Additional dielectric tests can be specified by a utility to reflect real service conditions. For example, a Fast Front
Switching Impulse (FFSI) was specified [9.29] by a utility to simulate frequent energization. The FFSI test wave
shape in [9.29] is specified with a 1.2 µs front, 4200 µs time to first voltage zero and a crest value of 2.0 times the
tap's rated voltage. Non-tested terminals are left ungrounded. It is a good practice for a utility to require this
dielectric test when a transformer will be frequently switched. Other dielectric tests could also be specified by a
utility based on personal experience or forecast problematic.

9.3.2 System studies

As far as the transient interaction between transformers and the power system is concerned, the ideal situation
would be if the user could provide the manufacturer with as much information as possible relating to these
transients during the planning stage. This procedure would enable the manufacturers to have a better
understanding of the requirements imposed by the system on the transformer terminals and consider them during
the insulation design stage.

Unfortunately, as the amplitude, frequency and repeatability of these transients are directly affected by different
system aspects, it will be very difficult, in most of the cases, to have enough information in advance to address all
the changes that may occur in the system topology and also in operational procedures during the planning stage.

Page 133
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

The accurate evaluation of the transients involving the power transformer and the power system may demand
wideband frequency transformer modelling. As already discussed in the previous chapters, a simple model may not
be sufficient to represent the transformer in the study of different condition of the system.

“Part 2: Case Studies” presents several case studies which clearly show how sensitive the transient voltages are
according to the type of transformer models used during the electromagnetic transient simulations. Thus, to
determine more realistic values of the transient voltages at the transformer terminals more adequate models may
be needed for new transformers, which are not available at the specification stage.

Generally speaking, as the combined overvoltage phenomena depends on system configurations and transformer
frequency response, there is no practical method for avoiding resonances from the viewpoint of system aspects,
ex. application of lightning arrester, at the planning stage.

As presented in section 9.3.1.2, one way to deal with these difficulties is to upgrade the specified nominal voltage
levels to the higher limit presented by the standards. This practice would result in higher insulation margins, not
only in the time domain but also in the frequency domain, and will certainly increase the probability that the
expected transients will be covered by the standard tests. More expensive equipment is expected to result from this
procedure.

Another possibility to be considered in the technical specification of new transformers is to provide the
manufacturers with some representative impulses based on the user’s experiences with similar electrical
transformer environment. Transient measurement, as described in 9.4, can be carried out with that purpose. For
example, in stations with past history of high lightning occurrence or with critical arrangement, a good knowledge of
the generated impulses can be achieved in this way. The goal is to acquire a better understanding of the
requirements imposed by the system on the transformer terminals in order to improve the insulation project,
especially for critical applications.

These representative impulses can also be used as an input for system transient studies to cover different
substation configurations and their effect at the transformers in a case-by-case analysis. As a high frequency
model is not expected to be available, a simplified model, as presented in Chapter 4, could be considered or one
from similar equipment.

In summary, it will be up to the purchaser to define the best way to provide the manufacturer the necessary
information if non-standard impulses are to be specified and considered during the insulation design. The strategic
importance of the transformer and the system’s past operation experience should be taken into account. The
cooperation between the manufacturer and client is also essential during the design process, as system and
transformer characteristics will have a great influence on the long term performance of the transformer.

9.3.3 Transformer high frequency modelling

It is strongly recommended that the transformer’s technical specification requires the manufacturer to provide either
a high frequency transformer model, adequate to a certain range, or the necessary data to allow its calculation by
the user, for instance provide the inductance and capacitance matrix of the transformer.

As it has been discussed in previous chapters, the availability of a sufficiently accurate transformer model is
important. This is not only for the analysis of the effects of high frequency transients on transformer insulation but
also for the analysis of their effect on other system equipment, such as circuit breakers, in cases where the
influence of transformer frequency response is mandatory.

Some proposals for transformer models, to be required in technical specification are presented in the following
sections. They are based on the concepts presented in Chapter 4.

9.3.3.1 High frequency simplified model

Three alternative types of simplified high-frequency models were discussed in section 4.1 of Chapter 4 of this
brochure.

For the three models, the following information could be requested in the transformer specification:

Page 134
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

External 50/60 Hz capacitances to be connected at the terminals of the power frequency standard model,
between windings and to earth. These capacitances could be measured or calculated.
Surge capacitances between the windings and earthed parts and between the different windings of the
transformer. These capacitances can be calculated from the transformer white box models or be
determined via impedance measurements.
Input impedance at the transformer terminals, ignoring mutual coupling effects. This impedance can be
measured or be calculated via the White Box model. By subjecting the impedance to modelling via the
asymptotic fitting method in Section 4.1.3 (or using general rational fitting techniques), a frequency-
dependent model is obtained. The model can be synthesized in the form of an RLC network for inclusion
in EMTP tools.

Section 4.1 of Chapter 4 discusses the drawbacks of these models and in what kind of studies they are most
successfully applied. Comparison of transient simulation results with these models with more elaborated ones are
presented in some case studies.

9.3.3.2 Electrogeometric Model (W hite Box model)

As it was presented in Chapter 4, the electromagnetic transformer model (white box model) of a transformer
consists of a large network of self-inductances, mutual-inductances and capacitances calculated from the
discretization of the continuous winding parameters. This model will have a great number of elements in order to
represent the transformer as accurate as possible, regarding its high frequency response.

If the range of frequency is specified by the user, according to the type of phenomena to be analysed, a reduced
white box model can be derived by the manufacturer. All the circuit elements (resistances, series and mutual
capacitances and self and mutual inductances) are then calculated considering the maximum frequency required.
In other words, the winding discretization must be in accordance with the maximum frequency of interest otherwise
the model will be not accurate.

The advantage of this model is the ability to assess the voltage response of the internal nodes of the winding
system. This will allow the determination of the transferred transient voltages (partial voltage between nodes and
to ground) originated by its interaction with the system. The time domain severity factor, defined in Chapter 6, can
be calculated with this procedure.

One drawback of this model is its difficulties to reproduce accurately the internal damping associated with the
oscillating transients for a large frequency range. In other words, the frequency dependent losses of the winding
resistance may not be fully considered. Usually the resistance at a particular frequency is determined and applied
as a constant during the complete analysis.

An illustration of a transformer electromagnetic model (white box) model which was applied to an EMT type
program is illustrated in Figure 9.6 below [9.34].

Page 135
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Figure 9.4: EMTP Transformer lumped parameter high frequency model [9.34].

References [9.35], [9.36] and [9.37] present some cases studies where system transient analyses were carried out
considering transformer white box models provided by the manufacturers.

9.3.4 Terminal Model (Black Box model)

The concept of transformer terminal (Black Box) modelling has been covered in detail in Chapter 4.

In this approach, the frequency response of the transformer is identified by rational functions which can be
implemented in EMT type simulations.

In order to have the necessary data for the Terminal Model (black box) implementation, frequency response
measurement in the laboratory and/or field should be requested in the technical specification. They could be
obtained from the typical frequency response measurement, which includes the determination of the winding
terminal impedances and the voltage transfer between the windings (magnitude and angle) [9.38]. The simplified
frequency dependent transformer model for fast transient presented in Chapter 4, section 4.1.3, can be derived
from this data.

A complete characterization of the transformer frequency response behavior requires the determination of its
complete admittance matrix, as described in Chapter 4, section 4.3. This complete matrix can be measured with
specialized FRA equipment as shown in section 4.3 or be computed at the design stage via the White Box model.
This matrix can be requested in the technical specification and discussed during the design review and factory
tests. As this may be a new requirement for the manufacturer, a close relationship with the client is advisable
through the whole process.

The admittance matrix is subjected to model extraction using rational function based curve-fitting techniques,
leading to a state-space type model which can be included in EMTP-type simulations.

Examples of the calculation of black box transformer models following these procedures can be found in [9.39],
[9.40] and [9.41].

9.3.5 Grey Box Model

The Grey Box Modelling approach seeks to calculate the parameter values of a predefined ladder type RLC circuit
structure, based on measured terminal responses in combination with basic information about the transformer such
as name plate data. The resulting model can be utilized in the same way as a black-box model, although the Grey
Box model is not necessarily compatible with EMTP-tools.

In order to have the necessary input data for this type of model, standard FRA measurements should be requested.

Page 136
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

9.3.6 Low frequency model

Although the main focus of this group is the study of the high frequency transient interactions between the electrical
system and transformers, when low frequency phenomena are involved, such as transformers energizations or de-
energizations, it is of fundamental importance to have the value of the air-core reactance of the transformer and the
knowledge of the core hysteretic curve considering different technologies (shell type, 5 or 3 limbs, autotransformers
etc). This information should also be provided by request.

In fact, for energizations studies, the value of the air-core reactance, which defines the last slope of the saturation
curve φ (i) of the magnetic circuit, is a major importance, especially when the this curve is flat corresponding to
highly non linear transformers.

Those values, initially given by the manufacturer with an accuracy of 15% by, generally from abacuses, may be
determined with a very good accuracy taking into account the exact technology of the transformers, from analytical
formulas [9.42], [9.43] ; they can be also provided by the manufacturers from a 3D electromagnetic program, both
approaches giving a good accuracy of 2-3%.

There is an actual tendency to build magnetic circuits with laminations having low losses, for economical purposes;
unfortunately, they lead to highly non linear saturation curves; that is why this parameter is important, and its
knowledge with a good accuracy is mandatory, for a correct determination of this inrush currents [9.44].

For de-energizations studies, for the determination of the residual fluxes [9.45] the knowledge of the hysteretic
curve is mandatory, under steady–state conditions, or up to 1.10 p.u., which is the actual tendency for TSOs
(Transmission Systems Operators) to operate the network [9.46]. This curve may be directly measured, but using a
current divider with a very good accuracy, as those curves are measured under non loaded conditions.

They will also be direct input data for transient tools; the construction of this curve from points in the knee of the
saturation curve and from hysteretic losses is quite complex, as assumptions have to be made on the coercitive
current (flux equal to 0 on the X axis) and the residual flux (current equal to zero on the Y axis).

The reference [9.47] shows the application of a low frequency transformer model in a ferroresonance study. It is a
good example of the necessity to have this type of model available to enable the study of different phenomena of
power system/transformer transient interaction.

9.3.7 Insulation stress assessment

The studies carried out by the manufacturers to support the insulation design of transformers are usually based
only on the standard dielectric test voltages; lightning, chopped wave and switching. As it has already been
mentioned in different sections of this brochure, more realistic wave shapes should in some cases be included in
the insulation design stage.

Accordingly, some proposals for an improvement in transformer technical specification regarding non-standard
wave shapes are presented in the following sections.

9.3.7.1 Transient studies

A simulation of the voltage distribution in the internal parts of the transformer for any given voltage impulse can be
performed by the manufacturer as part of the technical specification. These transient studies are expected to be
carried out with specific digital simulation tools considering the complete geometry of the transformer. These
impulses could be the standard dielectric waves and non-standard waves, representative of some system
conditions expected to occur in operation, for example, typical surges due to switching in substations [9.48] and
short-circuits [9.49]. The transformer terminals, except the one where the impulse is applied, can be simulated to
be open, connected to a surge arrester, a representative surge impedance, or even a complete network.

A frequency response analysis simulation can also be carried out with such digital simulation tool by performing a
sequence of simulations where in each simulation an oscillating impulse voltage is applied to the transformer with a
given frequency, damping factor and amplitude. The frequency is varied from one simulation to the next, effectively
covering the frequency range of interest. This damped oscillating impulse is defined in chapter 8, section 8.5.2, and
is considered a good representation of the surges originated due to occurrence in the power system. The amplitude

Page 137
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

could be, for example, the residual voltage of the surge arrester at the transformer terminals, for a defined current
wave, or the nominal operation voltage. The damping factor suggested in the literature based on typical system
switching surges is 0.8 [9.46].

The result of these transient studies will be the voltages at internal nodes (partial node-to-node voltages and node-
to-ground voltages) of the transformer winding system. From the internal winding voltage response, electrical
stresses can be assessed which will then be compared with the corresponding local withstand limits which
depends on insulation arrangements and technology aspects. The ratio between these two voltages (security
factor) must be greater than unity in order to have some security margin. Thus, the most critical nodes and the
main resonance frequencies of internal parts of the windings will be known. Table 9.2 presents, as example, some
illustrative results considering this procedure application [9.51].

Voltage on Generated Allowable Security

Impulse Voltage Tap Position
Line terminal voltage voltage Factor

1-2 1050 114 144.7 1.27

Full wave Min 3-4 1050 105 144.7 1.38

5-6 1050 73 117. 1.60

1-2 1155 127 211.3 1.66

Chopped wave Min 3-4 1155 115 211.3 1.84

5-6 1155 57 185.8 3.26

Damped 19 kHz 1-2 766 93 144.7 1.56

Resonant Min
voltage 34 kHz 3-4 766 97 144.7 1.49

Table 9.2: Transformer internal nodes security factor for standard impulses and resonant voltages.
Another possibility practiced by some manufacturers [9.52] is to present the maximum amplitude of a damping
oscillating impulses at the transformer terminals in order not to surpass a specified security margin based on the
insulation withstand. This evaluation is carried out for all resonance frequencies determined by a simulated
frequency response analysis of the windings. Table 9.3 presents an example of this type of evaluation. The
numerical values are illustrative.

Tertiary winding
Maximum System
Frequency Internal Withstand
Voltage
[kHz] Nodes [kV]
[kV]
185.6 23-24 88.2 366.7

324.1 9-10 88.2 338.3

432.9 10-11 88.2 249.7

483.5 94-95 88.2 266.7

580.3 7-8 88.2 220.0

724.2 55-56 88.2 199.7

Table 9.3: Maximum amplitude of oscillating impulses at resonances frequencies to guarantee security
margin.

Page 138
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

As it was discussed in section 9.2.1, the winding voltage distribution in response to any kind of excitation will be
influenced by the transformers terminal conditions, i.e. their loads. So in order to apply such procedure, the client
should inform the manufacturer what type of connections to be considered as representative of the normal system
operation.

The concepts presented in chapter 6, which define a time domain and frequency domain severity factor, could also
be applied by the client as an additional evaluation of the winding withstand response to specified input transients
representative of system conditions.

The frequency spectrum of transient voltages for typical system condition can be specified and compared with the
envelope defined by the frequency spectra of standard waveforms used in the design of the transformer insulation.
This comparison is an indication of how representative the dielectric tests are regarding the specified system
transients. The frequency Domain Severity Factor (FDSF), defined in chapter 6 as the ratio between the spectral
density of the calculated transient voltage and the spectral density of the envelope defined by the standard
waveforms, can be requested to be smaller than one to guarantee an insulation margin for critical conditions
defined by the user.

However, as discussed in chapter 6, one must be aware that FDSF analysis and corresponding safety factor
cannot alone be used to address the insulation withstand regarding its maximum allowable stress as the withstand
capability of an insulation structure is a time domain voltage function. The possibility here is then to convert back
the frequency spectrum of a transient voltages at some location of interest to a time domain voltage and/or
electrical field stresses and then compare it with the local corresponding insulation withstand [9.52], [9.53], [9.54],
[9.55]. Alternatively, time conversion factors may be also required to properly assess the local insulation withstand,
stresses and corresponding safety factor as discussed in chapter 7. Transient studies of this kind are also
important in transformer transient related failure analyses and could be requested from the manufacturer
responsible for the transformer refurbishment as presented in some of the examples of “Part 2: Case Studies” and
section 9.2.5.

As it is not a usual practice to specify these types of evaluations, cooperation between transformer manufacturers
and users is essential to achieve good results and avoid misunderstandings. The decision to request these studies
in the specification will always depend upon the client’s decision taking into account its experiences with these
phenomena and the strategic importance of the equipment.

9.3.8 Transformer Failure Analysis

The analysis of a transformer failure is a quite complex task as the consequential damages may in many cases
eliminate important evidence that would have helped to achieve an accurate diagnosis. As a consequence, failure
investigations may indicate several events and phenomena as the possible root cause.

Thus, the action of a multidisciplinary team becomes necessary to evaluate many important aspects such as [9.56]
[9.57]:

the type of failure: electrical, mechanical, thermal and contamination;

maintenance history (for example, the results of dissolved in oil gas analysis);
loading conditions history (pre-failure loading and historic loading data and aging evaluation);
short circuit contributions;
over-excitation;
voltage transients (lightning, close and remote internal faults, breakers and switches operations,
frequency of switching overvoltages including trends, etc);
other operational conditions of the system when the failure occurred.

A key aspect in such type of preliminary evaluation is to understand from available data if the origin of the failure is
related to an insulation withstand reduction (aging, cumulative effect, contamination, excessive moisture, bubbling,
etc) or to an overstress condition (internal or external overcurrent or overvoltage) or due to a combined effect of
both insulation withstand reduction and overstress.

Page 139
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

If the investigation points at transient voltages as a probable cause of failure, a series of actions should be taken,
involving the manufacturer responsible for the repair or refurbishment, to assist in the failure analysis. Some
possible steps are proposed below:

a) The manufacturer responsible for the repair/refurbishment should provide a simplified model of the
transformer appropriate to the relevant frequency range or calculate a “black box” model from frequency
response measurements if a healthy or similar unit is available.
b) The owner should carry out a series of transient studies to determine the most critical conditions to
which the transformer was exposed at the time of failure or during normal operations. In particular, the
situations described in chapter 5 should be paid attention to.
c) The owner should compare the frequency spectrum of transient voltages with the standard waves
according to the FSDF (Frequency Domain Severity Factor) methodology presented in chapter 6, to
have an indication if the transient voltages could have stressed the transformer beyond its design level.
As the transformer insulation design is normally based on the standard (transient) waveforms
(according to IEC and/or ANSI norms), unless the withstand to different transient waveforms were
added to the specification, the FSDF evaluation is an additional tool in the analysis of a possible
dielectric failure.
d) The manufacturer responsible for the transformer repair/refurbishment should analyse the impact of
these transients to internal points of the transformer, mainly in the region of the failure.
e) If it is concluded that the transformer has not withstood satisfactory the transient to which it was
subjected, the manufacturer should be requested to review the insulation design. The effectiveness of
these revisions should be discussed during the Design Review. In addition to this, possible changes to
the arrangement of the substation and/or the adoption of operational procedures to minimize these
stresses should be considered.
f) Whenever possible, frequency response measurements or low-voltage impulse time domain
distribution measurements should be carried out along the failed winding (or similar spare winding) to
determine the voltages at internal points of interest within a certain frequency range, for example, up to
a few MHz. This analysis will identify the most dangerous frequencies for that particular winding.

Practical examples of different types of failures due to transient were summarized in chapter 2 and presented in
more details by some examples in “Part 2: Case Studies”. Regarding failure investigation, references [9.58]-[9.63]
present detailed investigations based on what have been discussed so far.

A good example can be found in references [9.59] and [9.60], which describe the analysis of transformer failure
due to internal resonances excited by winter lightning strokes on transmission lines. The evaluation carried out in
these cases was in line with the approach proposed in this section. These were the steps considered:

(1) Determination of the overvoltage frequency characteristics based on the lightning stroke point and on
the power system configurations.
(2) Measurement of the natural frequency components and the response magnification of the weakest part
of the transformer.
(3) Comparison of these frequency components.
(4) As the natural frequency components of the transformer match the frequency components of the power
systems, protection methods (modification of winding structure, application of surge protection device,
improvement of insulation level of winding, etc.) were applied to the transformer.

In all these analysis, it is clear how important it is to have the participation of the manufacturer responsible for the
repair/refurbishment involved in the analysis of the transformer failure. Ideally, this analysis should have the
involvement of the manufacturer of the original transformer, due to their detailed knowledge of its design.

Page 140
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

9.3.9 Interaction between manufacturer and user

The importance of close cooperation between the manufacturer and the purchaser has been pointed out in different
sections of this document. Although, the technical specification is the most important mean of this communication
as it provides the basic requirements regarding the electrical system, some joint analysis could be carried out still in
the design stage and checked in the design review. The main idea is to provide the manufacturer a better
understanding of the requirements imposed by the system to the transformer.

The flowchart of Figure 9.5 is an example of some steps that could be followed by users and manufacturers of
transformers in the design phase to take into account more realistic dielectric stresses. The application of this
procedure presents some challenges as it may lead to some specific dielectric requirement not initially considered
in the bid with impact on the cost of the project and its deadline.

CLIENT MANUFACTURER

Specification of standard insulation levels

(BIL/SIL) and basic requirements Basic transformer design

Determination of high-frequency High frequency transformer

transient voltages (critical cases) transformer model

Evaluation of the impact of the

Additional specification of non-
new transient voltages in
standard waves
transformer project

Substation

Review of system
NO Are the results
procedures and/or
satisfactory?
transformer project

Transformer YES

Final project and fabrication

Figure 9.5: Example of joint analysis between Client and Manufacturer during the design
process.

Page 141
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

The combination of the completion of the Design Review with Test Frequency response techniques and evaluation
of results of analysis using computational tools has provided a greater understanding of the behavior of the
transformer subjected to the Electrical System requirements, and tends to become a very important tool useful to
all users.

9.4 Transient Measurements

The measurement of transients is an important part of the process of analysing interactions between the
transformer and the power grid. Often, such measurements are performed in order to explain the reason for
equipment failures as a post mortem analysis. Measurements can of course also be used for assessing
permanently the transient environment for transformers, as will be discussed below.

9.4.1 Permanent Setups

Due to the fact that transients occur randomly, it is desirable that the measurement system is kept on-line
constantly and that it meets necessary requirements of monitoring systems for high voltage equipment in order to
be able to record any transient events.

Operation of the transient measurement system has to be carried out in a manner such that there is no risk of
interruption of the normal operation of the transformer. All components have to meet high quality standards in order
to achieve low maintenance and long life of the measurement system.

Figure 9.8 shows a typical setup for transformer voltage transient monitoring. The measurement tap of the bushing
is used to measure the transient voltages on the transformers phase line. The capacitances C1 and C2 of the
bushing comprise the capacitive divider. Typically, a capacitive coupling device is connected to the measurement
tap, where CA >> C2. For low frequencies, the voltage ratio of the coupling device then is

U out C1 (9.2)
≈
U in CA

being typically in the range of 1/1000 to 1/4000.

The setup in Figure 9.6 shows an example of voltage transient measurements using the tap of the transformer
bushings but it is possible also to measurement using an external capacitive voltage divider. It has to be mentioned
that Figure 9.6 shows only a principle drawing. Several details like ensuring that the bushing tap will never exceed
a certain voltage, separation between bushing tap potential from the recording instrumentation, linear
characteristic, etc has to be taken into consideration. Due to safety reasons for extra high voltage (EHV), ultra high
voltage (UHV) transformer and high frequency measurements (MHz range) it is recommended to use a current
transformer (CT) instead of the secondary capacitor CA [9.65].

Page 142
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Conductor

Voltage
Equalizers

C1

C2 Bushing

Figure 9.6: Measurement of voltage transients at the measurement tap of a transformer bushing.

For higher frequencies however, the transmission ratio is not only determined by the capacitances but also by the
parasitic inductances within the signal path. A good coupling device shows linear characteristic up to several MHz
without resonances. It is remarked however that most measurements involving transformer resonant interaction
with the system requires only a good bandwidth up to a few MHz.

9.5 References
[9.1] CIGRE WG A3.07, “'Controlled Switching of HVAC Circuit - Breakers Benefits & Economic Aspects”',
CIGRE Technical Brochure No. 262.

[9.2] CIGRE WG A3.07, “'Controlled Switching of HVAC Circuit - Guidance for Further Applications Including
Unloaded Transformer Switching, Load and Fault Interruption and Circuit-breaker Uprating”', CIGRE
Technical Brochure No. 263.

[9.3] CIGRE WG A3.07, “'Controlled Switching of HVAC Circuit Breakers - Planning, Specification and Testing
of Controlled Switching Systems”', CIGRE Technical Brochure No. 264.

[9.4] J. C. Das, “Analysis and Control of Large-Shunt-Capacitor-Bank Switching Transients”, IEEE Trans. on
Industry Applications, Vol. 41, No. 6, pp. 1444-1451, 2005.

[9.5] H. M. Pflanz and G. N. Lester, “Control of Overvoltages on Energizing Capacitor Banks”, IEEE Trans. on
Power App. Syst., Vol. PAS-92, No. 3, pp. 907-917, 1973.

[9.6] R. S. Bayless, J. D. Selmens, D. E. Traux and W. E. Reid, “Capacitor switching transformer transients”,
IEEE Trans. on Power Delivery, Vol. 3, No. 1, pp. 349-357, 1988.

[9.7] J. Lapworth and A. Wilson, “Transformer Internal Over-Voltages Caused by Remote Energization”, IEEE
PES Power Africa 2007 Conference and Exposition., Johannesburg, South Africa, pp. 16-20, 2007.

[9.8] K. Ragaller, “Surges in high voltage networks”, Edition Plenium, 1980.

[9.9] CIGRE WG 13.07, “Controlled switching of HVAC circuit breakers: Guide for application lines, reactors,
capacitors, transformers – 2nd part,” ELECTRA, no. 185, pp. 37–57, August 1999.

Page 143
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

[9.10] J. Lalot, A. Sabot, X. Kieffer, S.W. Rowe, “Preventing earth faulting during switching of disconnectors in
GIS including voltage transformer” IEEE Trans. on Power Delivery, vol. 1, no. 1, pp. 203-211, 1985.

[9.11] H. Fujimi, T. Takuma, Y,Aoshima, “Impulse breakdown characteristics in SF6 Gas in the presence of a
local spark”. IEEE Trans. on Electrical Insulation, vol. EI-18, no. 4, pp 429-433, 1983.

[9.12] S .A. Boggs, F.Y. Chu, N. Fujitomo, A. Krenicky, A. Plessis, D. Schlicht, “Disconnect switch induced
transients and trapped charge in gas insulated substations”. IEEE Trans. on Power Apparatus and
Systems, vol. PAS-101 no. 10, pp. 3593-3602, 1982.

[9.13] O. Bosotti "Phenomenon associated with switching capacitive currents by disconnectors in metal
enclosed SF insulated switchgear", CIGRE 1982 - Report 13 .06.

[9.14] N. Ito, “Local Resonance in a Main Transformer Winding”, The Thermal and Nuclear Power, vol. 47, no.
6, pp. 651-654, 1996 (in Japanese).

[9.15] M. Hori, M. Nishioka, Y. Ikeda, K. Noguchi, K. Kajimura, H. Motoyama, and T. Kawamura, “Internal
Winding Failure Due to Resonance Overvoltage in Distribution Transformer Caused by Winter Lightning”,
IEEE Trans. on PWRD, vol. 21, no. 3, pp. 1600-1606, 2006.

[9.16] M. Hori, M. Mitani, H. Sakurano, H. Motoyama, T. Kawamura, “Observation and Analysis of Incident
Surge Voltage Waveforms in Substations by Winter Lightning”, IEEE Trans. on PWRD, vol. 22, no. 1, pp.
316-322, 2007.

[9.17] M. Popov:Switching three-phase distribution transformers with a vacuum circuit breaker: Analysis of
overvoltages and the protection of the equipment, Ph.D thesis, Delft University of Technology, 2002.

[9.18] A.T. Chadwik, J.M. Ferguson, D.H. Ryder, and G.F. Stearn: “Design of power transformers to withstand
surges due to lightning, with special reference to a new type of winding”, Proceedings IEE, Pt. II, Vol. 97,
1950, pp. 737–750.

[9.19] E.J. Grimmer and W.L. Teague: “Improved core form transformer winding”, AIEE Transactions, Vol. 70,
1951, pp. 962–967.

[9.20] R.M. Del Vecchio, B. Poulin and R. Ahuja: “Calculation and measurement of winding disk capacitances
with wound-in-shields”, IEEE Transactions on Power Delivery, Vol. 13, No. 2, April 1998, pp. 503–509.

[9.21] M. Bagheri, M. Vakilian, A. Hekmati and R. Heidarzadeh: "Influence of Electrostatic Shielding of Disc
Winding on Increasing the Series Capacitance in Transformer", 2007 IEEE Lausanne Powertech, pp.
1780-1784.

[9.22] S.V. Kulkarni and S.A. Khaparde: “Transformer Engineering: Design and Practice”, chapter 7, CRC
Press; 1st edition (May 24, 2004), ISBN-10: 0824756533.

[9.23] R.Baehr, CIGRE Study Committee 12: “Use of ZnO-Varistors in Transformers”, ELECTRA Nº143, August
1992, pp 32-37.

[9.24] Wen Yuanfang and Chengke Zhou: “Experimental Studies on the Use of MOV in Transformer Windings
Inner Protection”, IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 20, NO. 2, APRIL 2005, pp
1441-1446.

[9.25] N.V. Buthelezi, N.M. Ljumba and A.C. Britten: “Suppression of transients cross the tap windings of an
auto-transformer”, Energize, June 2005, pp 22-24.

[9.26] C57.12.90, IEEE Standard Test Code for Liquid-Immersed Distribution, Power, and Regulating
Transformers.

[9.27] IEC 60076-3, Power Transformers, part 3: Insulation levels, dielectric tests and external clearances in air.

Page 144
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

[9.28] R. Malewski, J. Douville et L. Lavallée, Institut de Recherche d’Hydro-Québec (IREQ), “Measurement of

Switching Transients in 735-kV Substation and Assessment of their Severity for Transformer Insulation”,
IEEE Trans. on Power Delivery, vol. 3, no. 4, pp. 1380-1390, 1988.

[9.29] L.B. Wagenaar, J.M. Schneider and J.A. Fleeman, American Electric Power Service Corp., “EHV
Transformer Dielectric Specification Improvements”, IEEE Transaction of Power Delivery, January 1994.

[9.30] “Electrical Environment of Transformers Impact of Fast Transients”, CIGRE JWG A2/A3/B3.21,
ELECTRA, 2005, pages 24-37.

[9.31] R. Malewski, M.A. Franchek and J.H. McWhirter, “Experimental Validation of a Computer Model
Simulating An Impulse Voltage Distribution in HV Transformer Windings”, IEEE Transaction of Power
Delivery, 1994.

[9.32] C. Vaillers, R. Malewski, Xuan Dai-Do and Jacques Aubin , Centre d’Innovation sur le Transport
d’Énergie du Québec (CITEQ), “Measurements of Dielectric Stress in EHV Power Transformer
Insulation”, IEEE Transaction of Power Delivery, October 1995.

[9.33] W.P. Seitlinger, H. Foschum, L.B. Wagenaar and J.A. Fleeman, “Investigations of an EHV
Autotransformer Tested with Open and Arrester Terminated Terminals”, IEEE Transaction of Power
Delivery, January 1996.

[9.34] Abhinandan De, Dilip Debnath et Abhijit Chakrabarti; “A Study on the Impact of Low Amplitude oscillatory
Switching Transients on grid Connected EHV Transformer Windings in a Longitudinal Power Supply
System”, IEEE Transaction of Power Delivery, vol. 24, April 2009.

[9.35] Okabe S., Koto M., Ueta G, Saida T, Yamada S.; “Development of High Frequency Circuit Model for Oil
Immersed power Transformer and its Application for Lightning Surge Analysis”, IEEE transaction on
Dielectrics and Electrical Insulation, vol. 18, No. 2; April 2011.

[9.36] Rohrbach P, Mendes J.C., Lacorte M.; “550kV GIS VFT Simulations a Support for Transformer Design”;
IPST 2001 Rio de Janeiro Brazil, June 24-28, 2001.

[9.37] Rocha A.C.O, Duarte L.C. “CEMIG experience in the analysis of transient recovery voltages associated
with Transformer-fed faults”, Symposium Transient Phenomena in Large Electric Power Systems,
Zagreb, Croatia, April 18-21, 2007.

[9.38] “Mechanical condition assessment of transformer windings using frequency response analysis (FRA),”
ELECTRA-CIGRE WG A2.26 Report 342, vol. 237, April 2008.

[9.39] Morched A.S., Marti l., Brierley R.H., Lackey J.; “Analysis of Internal winding stresses. In: EHV generator
step-up transformer failures” IEEE Transaction on Power Delivery, vol. 11, no2, April 1996.

[9.40] Savadamuthu U, Rocha A.C.O, Moreira S., Pena A., Mendes J.C., Lima A.,Balaji S.; ”Investigation of
Disconnector Switching on Two Generator EHV Step-Up Transformer Failures”. Cigre SCA2&D1 Joint
Colloquium Transformer materials and Emerging techniques, Kyoto, Japan, September 11-16 2011.

[9.41] Mendes J.C, Nakamura J.K., Mundim P.D, Rocha A.C.O, Pena A. M., “Power Transformers: Voltage
transient Analysis-Application of Soft Frequency Domain Characterization”. Cigre SCA2&D1 Joint
Colloquium Transformer materials and Emerging techniques, Kyoto, Japan, September 11-16 2011.

[9.42] M. Rioual, C. Crepy “Modelling of transformers in saturated conditions: determination of the air-core
reactance in the air by analytical formulas and its validation by on site tests”. IEEE PowerTech 2009, 1-4
July 2009, Bucarest, Romania.

[9.43] M. Rioual, C. Crepy, Y. Guillot. “Determination of the air-core reactance of transformers by analytical
formulas for different topological configurations and its comparison with an electromagnetical 3D
approach; discussions”. (GM-978-1-4244-6/09) IEEE General Meeting, Calgary, 24-29 July 2009.

Page 145
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

[9.44] M. Rioual, B. Bernin, C. Crepy “Determination of transient phenomena when energizing a 340 MVA
transformer having a highly non linear characteristics: modeling of the equipments and their validation by
on site tests”, IEEE General Meeting, Minneapolis, 25-30 July 2010, Paper n°2010 GM1381.

[9.45] M. Rioual, Y. Husianicia “Determination of the residual fluxes when de-energizing a power transformer &
Comparison with on site tests”, “IEEE PES General Meeting (IEEE Cat. N° 05CH37686)”, San Francisco,
June 2006.

[9.46] K. Mayor, L Montgomery, K. Hattori, J. Yagielski “Grid code impact on electrical machine design”, IEEE
General Meeting, San Diego, ref 978-1-4673-2729-9, July 2012.

[9.47] M. Rioual, P. Guuinic, D. Laval, M. Adelghani, N. Schaefer, M. Schäfer, “Modeling of the energization of
a power transformer in a 110 kV network and its validation by on site tests. Description of ferroresonance
phenomena involved and means to avoid them”, IEEE General Meeting (Cat. N° 086M1335), Pittsburgh,
20-24 July 2008.

[9.48] Rohrbach P., Mendes J.C., Lacorte M.; “550 kV GIS VFT Simulations a a Support for Transformer
Design”; IPST 2001 Rio de Janeiro Brazil, June 24-28, 2001.

[9.49] Morched A.S., Marti l., Brierley R.H., Lackey J.; “ Analysis of Internal winding stresses in EHV generator
step-up transformer failures “ IEEE transaction on Power delivery ,Vol 11, no2, April 1996.

[9.50] J. Musil , G. Preininger, E.Schopper, S. Wonger ; “Voltages stresses produced by aperiodic and
oscillating system overvoltage in transformer winding “IEEE transaction on Power Apparatus and
system. vol. Pas-100, no. 1. January 1981.

[9.51] TOSHIBA, Análise de Sobretensões Transitórias Autotransformador Trifásico 345 kV SE Lafaiete - EP –

17779 – In Portuguese.

[9.52] SIEMENS. CEMIG Usina Três Marias: Transformador Monofásico Fabricação Siemens - Schukert -
Ocorrência em Operação. Technical Report ETTRTLME, SIEMENS Brasil. Jundiaí, SP, 02/02/2010 - In
Portuguese.

[9.53] Savadamuthu U, Rocha A.C.O, Moreira S., Pena A., Mendes J.C., Lima A.,Balaji S.; “ Investigation of
Disconnector Switching on Two Generator EHV Step-Up transformer Failures”. Cigre SCA2&D1 Joint
Colloquium Transformer materials and Emerging techniques, Kyoto Japan, September 11-16 2011.

[9.54] Mendes, J.C. “Sobretensões Ressonantes: Fundamentos, Falhas e Monitoramento Contínuo de

Transformadores de Alta Tensão”. Dissertação de Mestrado. Escola Politécnica da Universidade de São
Paulo, EPUSP. São Paulo, 514p, Out.1989 ( in Portuguese).

[9.55] Mendes, J.C., “Redução de Falhas em Grandes Transformadores de Alta Tensão”, Tese de Doutorado.
Escola Politécnica da Universidade de São Paulo, EPUSP. São Paulo, 166p, Dez. 1995 ( in Portuguese)

[9.56] IEEE “Guide for Failure Investigation, Documentation, and Analysis for Power transformers and Shunt
Reactors” (C57.125).

[9.57] IEEE C57.117-1986: “IEEE Guide for Reporting Failure Data for Power Transformers and Shunt Reactors
on Electric Utility Power Systems”.

[9.58] A. Cancino*, R. Ocón, G. Enríquez, r. Malewski, “In Service Failure of 230 kV Transformers due to
Steep-front Lightning Over Voltages at Mexican West Coast”, A2-201 Cigre Biennial Session 2006.

[9.59] M. Hori, M. Nishioka, Y. Ikeda, K. Noguchi, K. Kajimura, H. Motoyama, and T. Kawamura, “Internal
Winding Failure Due to Resonance Overvoltage in Distribution Transformer Caused by Winter Lightning”,
IEEE Trans. on PWRD, vol. 21, no. 3, pp. 1600-1606, 2006.

Page 146
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

[9.60] M. Hori, M. Mitani, H. Sakurano, H. Motoyama, T. Kawamura, “Observation and Analysis of Incident
Surge Voltage Waveforms in Substations by Winter Lightning”, IEEE Trans. on PWRD, vol. 22, no. 1, pp.
316-322, 2007.

[9.61] H. B. Margolis, J. D. Phelps, A. A. Carlomagno and A. J. McElroy, “Experience with Part-Winding

Resonance in EHV Auto-Transformers: Diagnosis and Corrective Measures”, IEEE Trans. on Power
Apparatus and Systems, vol. PAS-94, no. 4, pp. 1294-1300,1975.

[9.62] K. Nakanishi, S. Fujita, H. Kurita, A. Kishi, T. Hasegawa, Y. Shibuya, “High Frequency Voltage Oscillation
in Transformer Windings and Electrical Breakdown Properties of Interturn Insulation Immersed in Oil at
VFT Voltage”, Trans. IEE of Japan, vol. 118-B, no.11, pp. 1228-1234, 1998. (in Japanese).

[9.63] A.S. Morched, L. Marti, R.H. Brierley, J.G.Lackey, “Analysis of Internal Winding stress in EHV generator
step-up transformer failures”, IEEE Trans. on Power Delivery, vol. 11, No.2, April 1996.

[9.64] M.-O. Roux, C. Rajotte, P. Riffon, “Improving EHV Hydro-Québec Transformer Test Program to Prevent
Failures due to Internal High-Frequency Resonances”, CIGRÉ SC A2/C4 Joint Colloquium, Zurich,
September 2013.

[9.65] Stepan, L.G., Rentz, R. A., Elroy, A.J., "An Improved Technique t o Measure Transmission System
Switching Surge Phenomena," IEEE Trans. PAS-92, 1973, pp. 1987-1991.

Page 147
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

APPENDIX A - FICTITIOUS TRANSFORMER: TRANSIENT SIMULATION RESULTS

A.1 Lightning Impulse – Nodes Maximum Voltage Values

Page 148
Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Page 149
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

A.2 Lightning Impulse – Branches Maximum Voltage Values

Page 150
Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Page 151
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

A.3 Variant FT1 - Lightning Impulse – Temporal Wave Shapes

Page 152
Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Page 153
Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Page 154
Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Page 155
Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Page 156
Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Page 157
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

A.4 Variant FT2 - Lightning Impulse – Temporal Wave Shapes

Page 158
Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Page 159
Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Page 160
Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Page 161
Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Page 162
Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Page 163
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

A.5 Variant FT3 - Lightning Impulse – Temporal Wave Shapes

Page 164
Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Page 165
Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Page 166
Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Page 167
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

A.6 Variant FT4 - Lightning Impulse – Temporal Wave Shapes

Page 168
Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Page 169
Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Page 170
Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

Page 171
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

A.7 Variant FT5 - Switching Impulse – Temporal Wave Shapes

Response to damped oscillatory wave shape with D = 0.6

Transformer model without taking into account internal damping effects

Response to damped oscillatory wave shape with D = 0.6

Transformer model taking into account internal damping effects

Page 172
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

A.8 Variant FT6 - Switching Impulse – Temporal Wave Shapes

Response to damped oscillatory wave shape with D = 0.7

Transformer model without taking into account internal damping effects

Response to damped oscillatory wave shape with D = 0.7

Transformer model taking into account internal damping effects

Page 173
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

A.9 Variant FT7 - Switching Impulse – Temporal Wave Shapes

Response to damped oscillatory wave shape with D = 0.8

Transformer model without taking into account internal damping effects

Response to damped oscillatory wave shape with D = 0.8

Transformer model taking into account internal damping effects

Page 174
 Electrical Transient Interaction between Transformers and the Power System – Part 1: Expertise

A.10 Variant FT8 - Switching Impulse – Temporal Wave Shapes

Response to damped oscillatory wave shape with D = 0.9

Transformer model without taking into account internal damping effects

Response to damped oscillatory wave shape with D = 0.9

Transformer model taking into account internal damping effects

Page 175