You are on page 1of 111

C4 TECHNICAL BROCHURE

Power system technical


performance

Procedures for Estimating


the Lightning Performance of
Transmission Lines – New Aspects

Reference: 839

June 2021
Procedures for Estimating
the Lightning Performance
of Transmission Lines –
New Aspects
WG C4.23

Members

C. ENGELBRECHT, Convenor NL I. TANNEMAAT, Secretary NL


B. BESKE US W. CHISHOLM CA
M-T. CORREIA DE BARROS PT G. DIENDORFER AT
M. HANNIG DE F. H. SILVEIRA BR
J.L. HE CN M. MIKI JP
S. PACK AT T. SHINDO JP
S. VISACRO BR R. ZENG CN

Corresponding Members

M. ISHII JP J. LUNDQUIST SE
I. MOHAMED RAWI MY P MIKROPOULOS GR
MOHD ZAINAL ABIDIN AB-KADIR MY C.A. NUCCI IT
S. OKABE JP S. PODKORITNIK SI
D. PROCTOR US F. RIZK CA
U. SCHMIDT DE L. SCHWALT AT

Copyright © 2021
“All rights to this Technical Brochure are retained by CIGRE. It is strictly prohibited to reproduce or provide this publication in any
form or by any means to any third party. Only CIGRE Collective Members companies are allowed to store their copy on their internal
intranet or other company network provided access is restricted to their own employees. No part of this publication may be
reproduced or utilized without permission from CIGRE”.

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 : 978-2-85873-544-0
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

Executive summary
Lightning remains one of the main causes of outages on transmission lines even though it has been
studied extensively since the beginnings of power transmission. Over the years several approaches have
been formalized to estimate the lightning outage rates of transmission lines. The principles and procedures
used in this regard are now well established throughout the power industry, but there is a need to update
these to benefit from the latest lightning research and computational techniques. The following
developments may be mentioned:
 Many countries have now access to detailed information on lightning activity from lightning location
systems (LLS), which provide improved input data to performance studies.
 Significant new research has been done to understand the attachment process of lightning flashes
to structures. This improved understanding can be applied to obtain better designs of shielding to
protect phase conductors from direct lightning strikes.
 Advanced computational techniques, such as the Method of Moments (MoM) and Finite-Difference
Time Domain (FDTD) formulations, are finding increasing use for calculating the electromagnetic
transients (EMT) in electrical power networks. Results from these advanced techniques can be
used to improve high-frequency models of ground electrodes and tall transmission line structures
for more widely available and computationally efficient circuit-based methods such as the various
EMT implementations.
 Finally, modern personal computers provide greatly improved computing power which allow for
more sophisticated modelling of the interaction of lightning with transmission line structures,
thereby reducing the need for simplified methods.
The purpose with this Technical brochure is to complement and update CIGRE brochure 63 “Guide to
procedures for estimating the lightning performance of transmission lines”, by modernizing the information
and by describing the latest techniques and assumptions used to calculate the lightning outage
performance of overhead transmission lines. A number of important CIGRE brochures form the basis for
this revision:

 No. 72: Guidelines for the evaluation of the dielectric strength of external insulation,
 No. 376: Cloud-to-Ground Lightning Parameters derived from Lightning Location Systems. The
Effects of System Performance,
 No. 440: Use of Surge Arresters for Lightning Protection of Transmission Lines,
 No. 549: Lightning Parameters for Engineering Applications,
 No. 704: Evaluation of lightning shielding analysis methods for EHV and UHV DC and AC
transmission lines,
 No. 781: Impact of soil-parameter frequency dependence on the response of grounding
electrodes and on the lightning performance of electrical systems,
 No. 785: Electromagnetic computation methods for lightning surge studies with emphasis on the
FDTD method.
Summaries of the relevant information are included together with guidance on how to apply this knowledge
to estimate the lightning performance of transmission lines. An important contribution in this regard is the
proposal of a new simplified lightning stroke attachment model, which can be used as an alternative to the
Electro-Geometric Method, to improve the prediction of lightning incidence when upward leaders play a
significant role in the attachment process.
The introductory chapter provides the background, scope, and approach for this brochure. This is followed
in Chapter 2 with descriptions of the physical characteristics and development of a lightning flash as it
pertains to the performance of overhead transmission lines. This chapter further introduces the basic
assumptions and methodology used to calculate the lightning performance of transmission lines.
Chapter 3 provides a summary of lightning observation techniques in use today, together with pertinent
characteristics related to frequency and severity of lightning events, which may be useful for calculating the
lightning outage rate of transmission lines. It summarizes previous CIGRE and IEEE reference documents
which should be consulted for more details on the state of the art and background information.

3
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

Lightning stroke attraction models are discussed in Chapter 4. These models are used to calculate
lightning incidence to transmission lines and to design the shielding arrangements to protect phase
conductors from direct strikes.
An overview of modelling concepts is provided in Chapter 5 with a view of providing guidance when using
high-frequency circuit-based models for calculating the lightning overvoltages on transmission lines. This
chapter is written as a complement to Chapter 5 of brochure 63 [15], therefore only focusing on recent
developments in representing the lightning stroke, transmission line towers, the frequency dependence of
ground electrodes, and in the modelling of insulator flashover. The chapter closes with specific guidance in
developing models for the calculation of the overvoltages during backflashover.
This brochure closes with Chapter 6, where proposals for future research needs are listed. This includes a
request for further benchmarking of the proposed simplified stroke attachment model. It is further noted
that there is a continuing need for improved statistics on the lightning stroke parameters relevant to the
prediction of outage rates on transmission lines.
Several appendixes are attached to this brochure in which supporting materials are presented:

 Appendix A lists the abbreviations and variables used in this document.


 Appendix B provides a summary of the parameters of negative downward stroke current as
obtained through direct measurements on short structures.
 Appendix C gives the analytical description for representing the wave shape of first and
subsequent strokes in time domain simulations. Both the CIGRE and double peaked first stroke
current waveshapes are explained.
 Appendix D describes two methods to calculate the lightning stroke attraction to transmission lines.
The first is the modified Electro-Geometric Method as applied in Japan and the second is the
proposed simplified implementation of the Rizk lightning attachment model.
 Appendix E provides a list of references.

4
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

Contents
Executive summary............................................................................................................. 3

Contents .............................................................................................................................. 5

1 Introduction................................................................................................................ 8
1.1 Background ................................................................................................................................................ 8
1.2 Scope .......................................................................................................................................................... 9
1.3 Approach .................................................................................................................................................... 9
1.4 Overview of the brochure ........................................................................................................................ 10

2 Basics ........................................................................................................................11
2.1 Basic lightning terminology .................................................................................................................... 11
2.2 Development of a negative downward flash .......................................................................................... 11
2.3 A summary of typical cloud-to-ground flash parameters ..................................................................... 12
2.4 Lightning effects on transmission lines ................................................................................................ 13

3 Characteristics of lightning......................................................................................16
3.1 Introduction .............................................................................................................................................. 16
3.2 Lightning occurrence .............................................................................................................................. 16
3.2.1 Ground based lightning detection networks ....................................................................................... 17
3.2.2 Lightning flash counters ..................................................................................................................... 17
3.2.3 Satellite based lightning detection ...................................................................................................... 18
3.2.4 Thunderstorm days (keraunic level) and thunderstorm hours ............................................................ 19
3.2.5 Considerations for calculating the line outage performance ............................................................... 19
3.3 Lightning current parameters ................................................................................................................. 19
3.3.1 Introduction ........................................................................................................................................ 19
3.3.2 General characteristics of strokes in negative downward CG flashes ................................................ 20
3.3.3 Considerations for obtaining lightning current parameters ................................................................. 23
3.3.4 The return stroke peak current ........................................................................................................... 23
3.3.5 Front time parameters and steepness ................................................................................................ 25
3.3.6 Duration (time to half-value) ............................................................................................................... 27

4 Lightning stroke attraction ......................................................................................28


4.1 Attachment of the negative downward stroke to a grounded object .................................................. 28
4.2 Electro-Geometric Method ...................................................................................................................... 31
4.2.1 Introduction ........................................................................................................................................ 31
4.2.2 Flash incidence .................................................................................................................................. 32
4.2.3 Shielding analysis............................................................................................................................... 33
4.2.4 Applicability of the EGM ..................................................................................................................... 34
4.3 The Leader Progression Method ............................................................................................................ 34
4.4 Simplified lightning attachment method ................................................................................................ 36
4.4.1 Background and overview .................................................................................................................. 36
4.4.2 Flash incidence .................................................................................................................................. 37
4.4.3 Shielding analysis............................................................................................................................... 37
4.4.4 Validation ........................................................................................................................................... 37

5 Review of modelling concepts.................................................................................39


5.1 Introduction .............................................................................................................................................. 39
5.2 Representation of the lightning stroke .................................................................................................. 39
5.2.1 Source impedance ............................................................................................................................. 39
5.2.2 Lightning current wave shape ............................................................................................................ 40

5
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

5.3 Representation of the transmission line towers ................................................................................... 43


5.3.1 Introduction ........................................................................................................................................ 43
5.3.2 Generalised tower surge impedance .................................................................................................. 44
5.3.3 Tower as a simple inductor ................................................................................................................ 46
5.3.4 Tower as a constant-impedance transmission line ............................................................................. 46
5.3.5 Tower as a variable-impedance transmission line .............................................................................. 46
5.3.6 Ground plane surge response ............................................................................................................ 47
5.3.7 Multistorey tower model ..................................................................................................................... 47
5.3.8 Representation of the tower surge impedance in TL lightning-performance studies .......................... 48
5.4 Earth electrode impulse response.......................................................................................................... 49
5.4.1 Introduction ........................................................................................................................................ 49
5.4.2 Low frequency response .................................................................................................................... 50
5.4.3 High frequency response ................................................................................................................... 51
5.4.4 Frequency dependence of soil parameters ........................................................................................ 51
5.4.5 Soil ionization ..................................................................................................................................... 54
5.4.6 Simplified representation of tower-footing electrodes in TL lightning-performance studies ................ 55
5.5 Insulator flashover modelling ................................................................................................................. 57
5.5.1 Lightning impulse strength of line insulation ....................................................................................... 57
5.5.2 Lightning surge strength of line insulation under non-standard conditions ......................................... 58
5.6 Backflashover model ............................................................................................................................... 66
5.6.1 Background ........................................................................................................................................ 66
5.6.2 The general concept........................................................................................................................... 66
5.6.3 Overhead transmission line ................................................................................................................ 68
5.6.4 Corona effect ...................................................................................................................................... 68
5.6.5 Transmission line towers .................................................................................................................... 68
5.6.6 Tower earthing electrodes .................................................................................................................. 68
5.6.7 Power frequency source..................................................................................................................... 69
5.6.8 Line insulation and flashover .............................................................................................................. 69
5.6.9 Lightning stroke .................................................................................................................................. 70
5.6.10 Backflashover rate calculation ............................................................................................................ 71

6 Summary and suggestions for further work ...........................................................72

7 References ................................................................................................................73

Appendix A Definitions, abbreviations, and symbols .................................................80


A.1 Acronyms ................................................................................................................................................. 80
A.2 Defined Variables ..................................................................................................................................... 81

Appendix B Updated stroke current parameters from direct measurements............84

Appendix C Analytical representation of the current shape .......................................87


C.1 General...................................................................................................................................................... 87
C.2 CIGRE concave wave front ..................................................................................................................... 87
C.2.1 The current front ................................................................................................................................. 87
C.2.2 The current tail ................................................................................................................................... 88
C.2.3 Example ............................................................................................................................................. 88
C.3 The double peaked wave ......................................................................................................................... 89
C.3.1 Mathematical description.................................................................................................................... 89
C.4 Subsequent strokes ................................................................................................................................. 90
C.4.1 Mathematical description.................................................................................................................... 90
C.5 Application of the waveshape ................................................................................................................. 90
C.5.1 Pre-defined scaling............................................................................................................................. 90
C.5.2 Self-correcting routine ........................................................................................................................ 91

Appendix D Stroke attraction methods ........................................................................93


D.1 EGM in Japan (TEPCO)............................................................................................................................ 93
D.1.1 Overview of calculating lightning strokes to transmission lines .......................................................... 93
D.1.2 Calculation of lightning stroke rate and lightning outage rate ............................................................. 93

6
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

D.1.3 Comparison of calculation results with observations .......................................................................... 95


D.2 Rizk simplified lightning attachment model .......................................................................................... 97
D.2.1 Determination of attractive distance of each conductor L ds0 ............................................................... 98
D.2.2 Calculation of internal variables: ksp0 ................................................................................................ 100
D.2.3 Calculation of internal variables: Ulc0 ................................................................................................ 101
D.2.4 Calculation of internal variables: kl ................................................................................................... 101
D.2.5 Calculation of internal variables: U0.................................................................................................. 102
D.2.6 Determination of Fpr.......................................................................................................................... 103
D.2.7 Determination of Feg ......................................................................................................................... 104
D.2.8 Determination of Fov ......................................................................................................................... 105
D.2.9 Example ........................................................................................................................................... 106
D.2.10 Comparison ...................................................................................................................................... 108
D.2.11 Validation cases ............................................................................................................................... 109

7
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

1 Introduction
1.1 Background
Lightning remains one of the main causes of outages on transmission lines even though it has been
studied extensively since the beginnings of power transmission. Over the years several approaches have
been formalized to aid transmission line designers to optimize their designs with respect to its lightning
performance. Important landmarks in the development of procedures for estimating the lightning
performance of transmission lines are:
 1950: The basic principles for calculating the lightning performance of transmission lines are
introduced by the AIEE [1].
 1982: Publication of a step-by-step simplified method for hand or programmable calculators by
EPRI [2]
 1985-1997: Adoption and refinement of the EPRI simplified method by the IEEE [3][4] which has
subsequently been published as the IEEE standard 1243 [5], and has been implemented as an
open-source tool, IEEE-Flash, for estimating the lightning performance of transmission lines [8]. At
publication, the latest version is 2.05 (2013) which is made available on SourceForge under the
General Public License (GPL) version 3 open-source license.
 1991: Publication of the CIGRE brochure 63 “Guide to procedures for estimating the lightning
performance of transmission lines” [6]
 1999: Publication of the book Insulation coordination for power systems, by A.R. Hileman [7]
providing details and background information on the simplified methods proposed within CIGRE
and the IEEE.
The principles and procedures introduced in these references are now well established throughout the
power industry, but these need to be updated to benefit from the latest findings in lightning research and
computational techniques.
Many countries have now access to detailed information on lightning activity from lightning location
systems (LLS) [14]. Such systems have seen continual development in terms of detection efficiency (i.e.
the proportion of the total number of lightning strokes, or flashes detected), location accuracy (i.e. the
distance between the LLS reported location and true ground striking point), classification accuracy (i.e. if
all cloud-to-ground and intercloud strokes are correctly identified) and lightning peak current estimates (i.e.
the ability of the LLS to estimate the peak current of a stroke).
Significant new research has been done to understand the attachment process of lightning flashes to
structures [10]. This enables an improved estimate of how many lightning flashes will hit a line and it will
also allow for a better placement of lightning shielding air terminals or conductors to reduce the number of
direct lightning flashes to phase conductors. Furthermore, video evidence of lightning flashes to overhead
lines and towers suggests that standard shielding calculation methods, such as the Electro-Geometric
Method (EGM) may fail to accurately predict stroke attraction when upward leaders play a significant role
in the attachment process, such as is the case of tall structures and Extra-High and Ultra-High Voltage
conductors.
Advanced computational techniques, such as the Method of Moments (MoM) and Finite-Difference Time
Domain (FDTD) formulations, are finding increasing use for calculating the electromagnetic transients
(EMT) in electrical power networks [12]. These so-called Numerical Electromagnetic Analysis (NEA)
methods, which rely on a direct numerical solution of Maxwell’s equations, allow for a more accurate
simulation of the response to lightning strokes of relatively short structures with non-uniform parameters
such as transmission line towers and ground electrodes. Results from these advanced techniques can
then be used to develop improved high-frequency models of ground electrodes and tall transmission line
structures for more widely available and computationally efficient circuit-based methods such as the
various EMT program implementations.
Finally, modern personal computers provide greatly improved computing power over those used in the
1990s, to allow for more sophisticated modelling of the interaction of lightning with transmission line
structures, thereby reducing the need for simplified methods. This ability is certainly needed when

8
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

evaluating the use of line arresters as a way to improve the lightning performance of transmission
lines[13].

1.2 Scope
The purpose with this Technical brochure is to complement and update CIGRE brochure 63 “Guide to
procedures for estimating the lightning performance of transmission lines”, by modernizing the information
and by describing the latest techniques and assumptions used to calculate the lightning outage
performance of overhead transmission lines. In addition, the brochure includes application examples to
serve as benchmarks and to allow for a sensitivity analysis of the selected modelling
techniques/parameters.

1.3 Approach
CIGRE brochure 63, published in 1991 [6], is a comprehensive document still used by many as a basis for
performing lightning performance calculations. Importantly, it is recognized that most of the information
contained in the brochure is still relevant and applicable. It was therefore decided to issue this brochure as
a complementary document to brochure 63, in which new or updated aspects are presented. Brochure 63
is also republished in the new CIGRE format. When necessary, cross-references to this document are
included.
Since the publication of brochure 63, CIGRE has published several reference works which form the basis
for this document:

 No. 72: Guidelines for the evaluation of the dielectric strength of external insulation, [15]
 No. 376: Cloud-to-Ground Lightning Parameters derived from Lightning Location Systems. The
Effects of System Performance, [14]
 No. 440: Use of Surge Arresters for Lightning Protection of Transmission Lines, [13]
 No. 549: Lightning Parameters for Engineering Applications, [9]
 No. 704: Evaluation of lightning shielding analysis methods for EHV and UHV DC and AC
transmission lines, [10]
 No. 781: Impact of soil-parameter frequency dependence on the response of grounding
electrodes and on the lightning performance of electrical systems, [11]
 No. 785: Electromagnetic computation methods for lightning surge studies with emphasis on the
FDTD method, [12]
The main task of the working group was to distil the relevant information from these brochures to produce
recommendations and guidelines for interpreting and applying the latest findings to the problem of
estimating the lightning performance of transmission lines. Rather than replicate the information of these
documents, summaries are provided of salient facts which are applicable when estimating the lightning
performance of transmission lines.
This working group also worked to identify a practical alternative to the presently used Electro-Geometric
Method (EGM), because the results from the EGM becomes unreliable when upward leaders play a
significant role in the attachment process. This is typically the case for tall structures or on energized
conductors when the power system AC or DC voltage bias may influence the formation of an upward
leader, as is the case on Extra-High Voltage, Ultra-High Voltage and HVDC lines. This was done in
recognition that more physically reasonable models, such as those loosely grouped under the term
“Leader Progression Models” (LPM), are more computationally intensive and consequently less likely to be
implemented for generalized lightning performance calculations. On the initiative of this working group
possibilities have been explored of simplifying a LPM with a view of easing the calculation burden, but
without a significant loss in the accuracy in the outcome. This resulted in the simplified Rizk lightning
attachment model [90].

9
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

1.4 Overview of the brochure


In Chapter 2 the physical characteristics and development of a lightning flash is described as it pertains to
the performance of overhead transmission lines. This chapter further introduces the basic assumptions
and methodology used to calculate the lightning performance of transmission lines.
Chapter 3 provides a summary of lightning observation techniques in use today, together with pertinent
characteristics related to frequency and severity of lightning events, which may be useful for calculating the
lightning outage rate of transmission lines. It summarizes previous CIGRE and IEEE reference documents
which should be consulted for more details on the state of the art and background information.
Lightning stroke attraction models are discussed in Chapter 4. These are used to calculate lightning
incidence to transmission lines and to design the shielding arrangements to protect phase conductors from
direct strikes. A new simplified attachment model based on the Rizk lightning attachment model is
proposed as an alternative to the presently used Electro-Geometric Method, to improve the prediction of
lightning incidence when upward leaders play a significant role in the attachment process.
An overview of modelling concepts is provided in Chapter 5 with a view of providing guidance when using
high-frequency circuit-based models for calculating the lightning overvoltages on transmission lines. This
chapter is written as a complement to Chapter 5 of brochure 63 [13], therefore only focusing on recent
developments in representing the lightning stroke, transmission line towers, the frequency dependence of
ground electrodes and in the modelling of insulator flashover. The chapter closes with specific guidance in
developing models for the calculation of the overvoltages during backflashover.
This brochure closes with Chapter 6, where proposals for future research needs are listed.
Several appendixes are attached to this brochure in which supporting materials are presented:

 Appendix A lists the abbreviations and variables used in this document.


 Appendix B provides a summary of the parameters of negative downward stroke current as
obtained through direct measurements on short structures.
 Appendix C gives the analytical description for representing the wave shape of first and
subsequent strokes in time domain simulations. Both the CIGRE and double peaked first stroke
current waveshapes are explained.
 Appendix D describes two methods to calculate the lightning stroke attraction to transmission
lines. The first is the modified Electro-Geometric Method as applied in Japan and the second is the
proposed simplified implementation of the Rizk lightning attachment model.

10
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

2 Basics
2.1 Basic lightning terminology
Lightning, which is a high current, atmospheric electrostatic discharge, is usually identified as a “lightning
flash”, and if it involves an object on the ground or in the air, it may also be referred to as a “lightning
strike” [58]. Most lightning flashes do not involve ground and occur between charge pockets inside the
same cloud (Intra-Cloud), between clouds (Inter-Cloud) and even between cloud and air. These are often
grouped under the identifier “IC”. Only about a quarter of the total lightning flashes occur to ground and
they are identified as cloud-to-ground or by the identifier “CG”. For ground-based objects such as
transmission lines, only CG flashes are of concern and they are accordingly the focus of this brochure.
A cloud-to-ground lightning flash is categorized by the development direction of the initial breakdown
through the air, i.e. the direction of leader development, and by the polarity of the charge transfer from
cloud to ground. Consequently, four types of lightning flash are identified (a) Downward Negative, (b)
Upward Negative, (c) Downward Positive, and (d) Upward Positive as illustrated in Figure 2-1. As indicated
in the figure, upward flashes are typically only observed from tall structures with an effective height of over
100 m and are therefore unlikely to occur on common transmission line structures. They are therefore
generally not considered when assessing the lightning performance of transmission lines. A fifth category,
bi-polar lightning, transfers both negative and positive charges during the same flash. Bi-polar flashes are
usually initiated as upward flashes from tall structures and are, therefore, also of no consequence to typical
transmission lines.

(a) Downward Negative (b) Upward Negative (c) Downward Positive (d) Upward Positive
Figure 2-1 – The four types of cloud-to-ground lightning flashes.
Of the downward lightning flashes, those of negative polarity (Type a) are the most prevalent and it is
conservatively estimated to account for about 90 % of the global cloud-to-ground flashes. Those of positive
polarity (Type c), make up about 10 % of the total. For this reason, only downward negative (Type a)
flashes are generally considered for estimating the lightning outage performance of transmission lines.
A negative CG flash comprises one or more return strokes (RS), usually simply referred to as strokes. A
distinction is made between the first stroke, which is the first high current pulse that occurs after the
electrical breakdown of the air is complete, and subsequent strokes which often (but not always) follow the
same breakdown path as the first stroke. Notably, the parameters of the first stroke current are very
different from those of subsequent strokes, which is a distinction that must be considered when
considering lightning performance of transmission lines. In this context it is recognized that the median
current of first strokes is significantly higher than that of subsequent strokes, which identifies first strokes
as the primary cause of lightning outages on transmission lines.

2.2 Development of a negative downward flash


The formation of a negative cloud-to-ground flash begins by the initial breakdown of the air which is
followed by the growth of a stepped leader from the cloud downwards to the ground. The stepped leader
establishes, in an intermittent fashion (hence its name), a conducting path to ground and charge from the
cloud is also deposited along the leader channel during this process. The first return stroke occurs when
the formation of the conducting path to ground is completed. This high current discharge neutralizes the
leader charge and propagates at 30 % to 50 % of the speed of light from the ground to the cloud, which
rapidly heats up the leader channel, resulting in the bright light flash and thunder noise normally
associated with lightning.

11
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

After this first stroke, several subsequent strokes may follow down, approximately the same path –
although this is not always the case. Each subsequent stroke is initiated by a leader which appears as a
continuously downward moving dart, from which it derives the name “dart leader”. The differences in the
behaviour of the stepped and dart leaders are related to the fact that stepped leaders form in virgin air
while dart leaders propagate along previously ionized channels left by preceding strokes. Examples of (a)
a stepped including multiple branches and (b) a dart leader following a previously formed channel are
shown in Figure 2-2.
The initial high-current pulse of a stroke may be followed by a “continuing current”, which typically has a
current magnitude in the range of 10 to 1000 Ampere and may last for up to hundreds of milliseconds.
Over 50 % of all negative cloud-to-ground flashes comprise more than one ground strike point. In such
cases, the subsequent stroke will be preceded partly by a dart leader along the existing channel and
change to a stepped leader attaching to the new ground strike point. These are therefore identified as “dart
stepped leaders”.

a) Stepped Leaders (before first stroke) b) Dart Leader (before subsequent stroke)
Figure 2-2 – Examples of a stepped and dart leader of a cloud-to-ground lightning flash.
(Licensed from: Tom Warner / WeatherVideoHD.TV)
Positive cloud to ground flashes are more likely to comprise of a single stroke, have high stroke currents
and are generally more energetic. They should, therefore, be considered when evaluating the risk for
damage to, for example, overhead ground wires and line surge arresters. They can be the prevalent type
of CG lightning during the cold season (Winter lightning) in temperate regions, or when lightning storms
dissipate.

2.3 A summary of typical cloud-to-ground flash parameters


Some general facts about CG flashes include (summarized from [9]):

 CG discharges make up only about 25 % of the global lightning activity, the rest being IC.
 Upward flashes occur mainly from very tall structures, such as double circuit UHV towers, or
mountain-top installations[49][58],. Furthermore, only a few of them include return strokes and the
peak currents of these are generally low and comparable to that of subsequent return strokes in
CG flashes.
 Negative polarity flashes make up at least 90 % and positive polarity less than 10 % of all
downward lightning flashes. The percentage of positive polarity downward flashes may, however,
be significantly higher for individual thunderstorms, especially during winter.
 More than 80 % of CG flashes comprise multiple strokes [9], which is significantly higher than the
range of between 45 to 55 % previously given in brochure 63 and closer to the assumption of
100 % implied in IEEE Standard 1243 [5].
 A typical negative cloud to ground flash comprises of 3 to 5 strokes. The largest number of strokes
observed and reported in a flash was 26 [9].
 The mean inter-stroke interval is about 60 milliseconds. Inter-stroke time intervals may be shorter
than 1 ms and be as long as several hundreds of milliseconds. For example, in flashes which
contain strokes with long continuing currents (i.e. long continuing current is defined as having a
duration of longer than 40 ms, but they can last for up to some hundreds of milliseconds).
 From direct current measurements, it is found that the first stroke current is typically 2 to 3 times
higher than subsequent strokes. However, about a third of CG flashes contain at least one

12
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

subsequent stroke with a higher peak current than the first stroke. The conditional probability that
one of several subsequent strokes will exceed a weak first stroke shielding failure current
approaches unity [5].
 Each cloud-to-ground lightning flash involves a total energy of the order of 109 J (1 GJ or
278 kWh). Most of the energy is dissipated in producing superheated air, thunder, visible and
ultraviolet light. Only one hundredth to one thousandth of the total is available as electrical energy,
with peak power on the order of 1012 W (1000 GW).

2.4 Lightning effects on transmission lines


Lightning strikes to overhead lines may cause the line insulation to flash over, thereby resulting in a power
outage or, in severe cases it may even cause damage to line components. The focus of this brochure is
however, primarily on the first aspect, although some information provided in Chapter 3 can be helpful in
assessing the risk for damage.
Three distinct interactions need to be considered when evaluating the lightning outage performance of
overhead lines:
 Shielding failure: Lightning strikes to phase conductors. This usually happens on unshielded
lines or if the lightning stroke bypasses any shield wires installed for protecting the phase
conductors from direct flashes. If the lightning current is high enough, it may cause a flashover due
to the voltage build-up on the phase conductors. The shielding failure rate is determined by
considering the presence and position of the overhead shield wires with respect to the phase
conductors. For a given arrangement, the shielding failure outage rate is equal to the number of
flashes terminating on the phase conductor with sufficient current to cause flashover.
 Backflashover: Lightning strikes to shield wires or supporting towers. The bulk of the lightning
current is conducted to ground, but the resulting raise in potential of the structure is sufficient to
cause a flashover from the structure to the phase conductors. The backflashover rate is primarily
determined by the combined impedance of the structure and the earthing system (including the
shield wires) and is given by the total number of lightning flashes to earthed components of the
line that has a sufficient magnitude and rate of rise to cause a flashover of the line insulation.
 Induced overvoltage: Lightning strikes to earth, or ground objects, in the immediate vicinity of the
line. In this case voltages are induced onto the phase conductors by the lightning current.
However, these lightning-induced overvoltages seldom exceed 250–300 kV and are therefore not
a concern for transmission lines which typically have lightning impulse withstand levels exceeding
350 kV. This type of overvoltage is therefore not considered further in this brochure, but is
addressed in CIGRE brochure numbers 287 (2006), [25], 441 (2010) [26] and 550 (2013) [27]
dealing with MV and LV systems.
An overview of a general methodology for calculating the lightning outage rate of a transmission line is
presented in Figure 2-3. This process is followed separately for shielding failures and backflashovers and
the sum of the two outage rates gives the total lightning outage rate of the line. In very simple terms the
methodology can be described as follows:
The outage rate of the line (Box “A” in -Figure 2-3) is, essentially, the number of lightning flashes to the line
which will cause a flashover of the line insulation. To calculate this, one first needs to determine the total
number of lightning flashes to the line (Box “B” in -Figure 2-3) and then assess how many of these flashes
would result in a flashover of the line insulation (Box “C” in -Figure 2-3). Whether or not the insulation will
flash over is determined, on the one hand by the overvoltages generated during a lightning flash (Box “H”
in -Figure 2-3) and on the other hand by the dielectric strength of the line insulation (Box “I” in -Figure 2-3).
The overvoltages across the insulation during a lightning stroke primarily depends on the magnitude of the
lightning current. Thus, a series of simulation studies are performed to determine the minimum current that
would cause a flashover of the line insulation. This value of current is named the “critical current” (Box “G”
in -Figure 2-3) and the probability that it may be exceeded in a lightning strike (Box “E” in -Figure 2-3) is
determined from the available statistical distribution functions(Box “F” in -Figure 2-3). Multiplying this
probability with the total number of lightning strikes to the line, gives the number of outages that may be
expected (Box “A” in -Figure 2-3).

13
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

Figure 2-3 – Overview of the methodology used to calculate the lightning performance of an overhead line.

The input data and subprocesses of this methodology can be described as follows:
 The number of flashes to the line (Box “B” in -Figure 2-3) is determined by calculating an attractive
area, within which lightning flashes will hit the line. This collection area is then multiplied with the
lightning ground flash density to determine the total number of flashes to the line.
o The attractive area is calculated by applying stroke attraction models (Box “D” in -Figure 2-3).
In Chapter 4, stroke attraction models and their application are explained. Historically the so-
called electro-geometric methods (EGM) – see Section 4.2 – have commonly been used
because they offer the simplest way to determine the number of flashes to the shield wires and
phase conductors. Leader progression methods (LPM) – see Section 4.3 – are more detailed
attempts to simulate the development of the stepped leader and the electric fields that govern
it, to determine the striking point. In this brochure, a simplified leader progression method is
proposed in Section 4.4 to overcome the shortcomings of the EGM, but without the heavy
calculation burden associated with the LPM.
o The frequency of occurrence of lightning applicable to transmission systems is expressed in
terms of the lightning ground flash density (Box “C” in Figure 2-3). In Section 3.2, sources of
information on the frequency of occurrence of lightning are provided, including Lightning
Location Systems (LLS) and other historical sources of data.
 The probability of exceeding the critical current (Box “F” in -Figure 2-3) is calculated from lightning
current statistics and the determined critical current (Box “G” in -Figure 2-3). Statistical

14
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

distributions that describe the probability of exceeding a specified current magnitude are presented
in Section 3.3.4 of this document.
 Several steps are required to calculate the critical current (Box “G” in -Figure 2-3):
o Flashover models are used to evaluate whether, or not, the line insulation will flash over for the
calculated insulation stress. It is not possible to directly apply laboratory test results for this
purpose since the lightning surge has a very different wave shape from that applied during
testing – i.e. the standard lightning impulse. The lightning impulse strength of the line
insulation (Box “I” in -Figure 2-3) and the various flashover models available, are discussed in
Section 5.5.
o The insulation stress (Box “H” in -Figure 2-3) is calculated by utilizing high-frequency electro-
magnetic transient modelling techniques. The lightning stroke is considered a current source
and its amplitude and waveshape determine the response of the transmission line. These
characteristics are discussed in Section 3.3. An overview of the specialized high-frequency
models necessary for lightning performance calculations is provided in Chapter 5. This
includes sections on the modelling of the supporting structure (Section 5.3), as well as the
earth electrode (section 5.4).

Estimates of the lightning performance of high-voltage transmission lines (TL) are typically based on the
response of the line to the first return stroke currents of negative cloud-to-ground flashes, as these
represent more than 90 % of cloud-to-ground flashes and their median peak current is typically two to
three times higher than that of subsequent strokes [9]. Subsequent strokes may be a significant contributor
to the outage rate only in very specific cases:

 Shielding failures: In about 15-20 % of multiple-stroke flashes, one of the subsequent strokes will
have a higher peak current than the first stroke. For the low-current first strokes associated with
shielding failures this probability is much higher [47]. It is therefore quite likely that the initial stroke
causing the shielding failure is followed by a higher current subsequent stroke that will flash over
the line insulation.
 Backflashovers: Subsequent strokes may cause backflashovers on lines with thin (e.g. single pole
structures) and relatively tall supporting structures (i.e. > 30 m) in combination with a very low
tower-footing impedance. In these cases, the steep rate of rise associated with subsequent
strokes results in a high voltage drop along the supporting structure, which can be comparable to
that of first strokes [50], [51]. These types of lines, however, have typically a low, and in many
cases an acceptable, backflashover rate.
Furthermore, lightning outages on transmission lines are transient in nature, which means that the line may
be re-energized immediately after the fault has been cleared. It is therefore customary to estimate the
long-term average outage rate of the line. The actual yearly performance may strongly deviate from the
long-term average due to the stochastic and year-to-year climatic variations in the lightning activity [2]. This
has the following consequences:

 The flashover probability characteristics of the line insulation is ignored because its standard
deviation is negligible compared to the statistical dispersion of the lightning generated overvoltage
surges.
 The dielectric strength of the line insulation is defined in terms of the 50 % flashover voltage, U50,
and not the withstand voltage as is customary for other insulation co-ordination studies.
Estimating the lightning performance of transmission lines is a complex task, based on many assumptions
and simplifications. Consequently, the achievable accuracy of the results is relatively limited, and also the
uncertainty limits, which can be used to avoid an overestimation of the model results, is not well defined.
Therefore, the need remains to further improve and update procedures and models based on the latest
research.

15
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

3 Characteristics of lightning
3.1 Introduction
The lightning performance of transmission lines is directly related to frequency and severity of lightning
events:

 The frequency of occurrence of lightning flashes is typically expressed in terms of the average
lightning ground flash density (GFD in flashes per km 2 per year) [28]. The GFD is preferably
determined from lightning observations over a large spatial area and over a long period of time.
This may include collecting data on the maximum and average flash rates per unit area, as well as
variations in flash characteristics as a function of location, time of year, time of day and storm type.
A consolidation of such data may provide a useful statistical basis for engineering calculations. For
example, past performance of a line can be assessed by correlating stroke data with line outages,
or an estimate of the prospective performance of the line can be extrapolated from historical data,
such as is done when doing design calculations.
 The severity of a lightning event is generally expressed in terms of one or more parameters that
describe the current or charge of the lightning stroke. Lightning current parameters are preferably
obtained from direct current measurements on structures that resemble transmission line towers.
In general, this means obtaining measurements from masts, chimneys or transmission lines, which
are less than 60 m tall to avoid including data from upward lightning strokes. CIGRE brochure 549
provides a comprehensive review of all parameters that characterize the severity of a lightning
flash [9].
In this Chapter a summary of lightning characteristics pertinent to the lightning outage rate of transmission
lines is given. It relies heavily on previous CIGRE and IEEE publications and it is recommended that these
be consulted for more details and background information.

3.2 Lightning occurrence


A primary requirement for most lightning studies is to have an estimate of the number of lightning flashes
per unit time per unit area, or as it is better known as, the regional ground flash density (GFD – usually
expressed as an annual average, Ng) [28]. Several alternative techniques of observation have been, or are
currently, applied in characterizing the regional ground flash activity. Some of these methods also provide
additional information on the recorded lightning strokes, including time of occurrence, stroke peak current
and wave shapes, which are useful for designing the transmission line lightning protection, or to perform
fault diagnosis and performance analyses.
The confidence in the data obtained from GFD mapping depends on the number of events per grid cell,
which in turn depends on the grid cell size and observation period [30]. For obtaining usable estimates, the
recommended number of events per grid cell should be at least 80 [30] and preferably be 400 [31]. These
correspond to an uncertainty of ±20 % or ±5 % respectively. Consequently, for a grid size of 1 km2, 16
years of data will be required for areas of moderate thunderstorm activity ( Ng = 5 Fl/km2/yr) to obtain the 80
counts necessary whereas in an area with Ng = 1, data over an 80-year period would be required for the
same level of certainty [30]. In such cases the grid size should be increased to reach the necessary counts
in a reasonable time, with the drawback that this will make the spatial resolution less accurate.
As LLSs record strokes, not flashes, estimates of Ng obtained from LLS data will, therefore, depend on the
method that was used to group the strokes into flashes [28]. In this regard, it is important to consider that
some strokes in a lightning flash may not follow the same path to ground, resulting in multiple ground
striking points [32]. Observations indicate that this may be true for 30 % to 50 % of all lightning flashes.
While the distance between the striking points is usually below 2 km, it can be as large as 8 km and in
extreme (i.e. Megaflashes) cases reach 700 km. IEC 62858 [28] recognizes this phenomenon in the
algorithms suggested for grouping strokes into flashes. Within this framework, the LLS data can be made
available in three different levels of abstraction. The lowest level represents the raw (ungrouped) stroke
data, for the middle level, strokes are grouped into flashes according to the branches that have a common
ground strike point and for the highest level all the strokes comprising the flash, irrespective of the ground
strike point, are grouped together [33]. When assessing the lightning risk for ground-based objects, such
as transmission lines, it is important that the ground flash density is based on the middle level of data

16
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

extraction that separately identifies all ground strike points in a flash [9]. If only the overall ground flash
density (i.e. highest level of abstraction) is available, a correction factor of about 1.5 to 1.7 should be
applied to compensate for the fact that each flash may have more than one ground strike point.

3.2.1 Ground based lightning detection networks


Over the years a variety of methods have been developed to detect and locate cloud-to-ground (CG)
lightning strokes. Detection of CG strokes is important for assessing the lightning risk for ground-based
objects. Ground-based systems exploit the electro-magnetic (EM) waves radiated from lightning
discharges for this task and are generally optimized for detecting CG signal energy in the very low
frequency (VLF) to low frequency (LF) range. The location of CG strokes is determined either through
magnetic direction finding (MDF), time-of-arrival (TOA), and combinations thereof [14]. The strength of the
signals received are analysed together with distance to stroke information, to infer the peak of the lightning
stroke current. This enables lightning detection networks to provide more information than just a count of
the strokes detected. Additional stroke data typically include, date and time, location, polarity and peak
stroke current, with some systems also providing recorded field waveforms. Such detailed data are an
attractive source for extracting lightning parameters for engineering applications which, besides the ground
flash density (GFD) and ground stroke density (GSD), may include peak current distribution, flash
multiplicity, and polarity ratio. However, there are many factors that may significantly impact the accuracy
of these derived parameters [14], which make their use unsuitable for lightning performance studies - see
section 3.3.3.
Today many countries have one or more lightning location systems in operation, with several providers
active in the market. Two of the more established networks are the North American Lightning Detection
Network (NALDN) and the European Cooperation for Lightning Detection (EUCLID) [14]. For countries
without a dedicated location system, updated global flash density information is available from the
GLD360, which is a ground-based network detecting lightning worldwide which is active for about 10 years
and is operated by Vaisala [17][16][19]. A GFD map derived from 5 years of data from the GLD360 LD
network is presented in Figure 3-1.

Figure 3-1 – Worldwide cloud to ground flash densities (flashes/km2/yr) based on 5 years of data (2013-2017)
from the GLD360 LD network. Courtesy of Vaisala [17].

3.2.2 Lightning flash counters


Until the late 1980’s, most of the specific information available on ground flash density had been acquired
through the application of 10-kHz lightning flash counters (LFC) in various countries [6] [45]. These devices
aggregate all first-, subsequent- and multi-point termination strokes within a 1-s time window and 20-km
detection radius as a single flash [18]. Flash counters may be readily applied over a suitable time period to
determine regional variations in the average ground flash density and its dispersion, provided that a
rigorous basis of calibration and ground flash discrimination is available and if sufficient data is collected
[45]. Although the distributions of flash density developed from networks of LFC exhibit very low spatial

17
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

resolution (grids larger than 1200 km2), they are considered very robust. The relevance of such systems is,
however, much reduced since the advent of modern LLS, whose spatial resolution is in the order of 1 km2.
Historical data collected through these counters may still provide a valuable input for lightning performance
calculations and these LFC can provide useful ground truthing data for the GFD obtained from other
sources such as satellite or global detection networks. The flash grouping algorithm for a specific LFC,
such as the 1-s / 20 km grouping model in a CIGRE 10-kHZ LFC, should, however, be considered in any
intercomparison with advanced LLS providing individual stroke data.

3.2.3 Satellite based lightning detection


Starting in 1995 until March of 2000, a low earth orbiting satellite equipped with the Optical Transient
Detector (OTD), was used to track lightning activity around the world [20]. The OTD was designed and
positioned such that it surveyed virtually all areas of the globe where lightning normally occurs. The OTD
was able to detect lightning flashes during both daytime and night-time conditions with a detection
efficiency ranging from 40 % to 65 %, depending upon external conditions such as glint and radiation.
From the OTD data, it is now estimated that over 1.2 billion lightning flashes (intracloud plus cloud-to-
ground) occur around the world every year. Most of the detected lightning activity was in the Inter-Tropical
Convergence Zone (ITCZ) and it was found that there is far more lightning over the land masses than over
the oceans.
In 1997 a second type of optical sensor, the Lightning Imaging Sensor (LIS), was placed in orbit to study
both day and night cloud-to-ground, cloud-to-cloud and intra-cloud lightning and its distribution around the
globe. The LIS is three times more sensitive than its predecessor, the OTD [21], and thus able to achieve a
90 % detection efficiency. The LIS is positioned such that it observes lightning where it occurs most, which
is between +/-35 degrees of latitude. The instrument records the time of occurrence, measures the radiant
energy, and determines the location of lightning events within its field-of-view.
While the OTD and LIS were able to survey most of the world’s lightning activity, there is an area known as
the South Atlantic Anomaly (SAA) that decreased the relative detection efficiency of the OTD by as much
as 50 %. The SAA is a high-radiation environment in which the adaptive software noise filters had to be
run more stringently to reject false events. Due to the lower orbit of the LIS satellite, it was much less
affected by the SAA [36].
These satellites provide about 106 flash observations per year, and sample most of the world’s surface.
[37], which allowed for creating a map of lightning activity spanning most of the globe. The v2.3 gridded
satellite lightning flash density map shown in Figure 3-2 was produced by the NASA LIS/OTD Science
Team [38]. It shows the total lightning activity, including both intra-cloud and cloud-to-ground lightning, and
it is therefore not directly applicable for lightning studies of ground-based objects such as transmission
lines.

Figure 3-2 – World map of lightning activity produced by the NASA LIS/OTD Science Team. Sum of intercloud
and cloud-to-ground flashes per km2 per year [39].

18
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

The compiled satellite data can, however, be used as a basis for obtaining a first estimate of the ground
flash density. This was especially useful in regions not covered by ground-based lightning location systems
prior to the deployment of global detection networks. A comparison of data from the satellite-based
lightning detectors to other ground flash density measurements around the world showed that the average
ratio of the satellite lightning flash density to GFD was 4:1, but it could vary between 3:1 to 5:1 [40].
The local GFD may be estimated from satellite data by first determining an appropriate conversion factor
and then applying this factor to the specific region of interest. The conversion factor may be determined by
comparing the satellite lightning flash density shown in Figure 3-2, to either the corresponding CG flash
density presented in Figure 3-1 or an average flash density obtained from lightning flash counters (see
section 3.2.2).
3.2.4 Thunderstorm days (keraunic level) and thunderstorm hours
One of the earliest methods used to quantify thunderstorm activity was “thunderstorm days” (Td) or in some
regions, “thunderstorm hours” (Th). A thunderstorm day is defined as a calendar day on which thunder is
heard irrespective of whether the lightning flashes are nearby or some distance away, and thunderstorm
hours give the yearly duration of thunderstorms.
Due to the subjective nature for acquiring the data, the strong local variations, and the resulting possibility
for inaccuracies, thunderstorm day and thunderstorm hour data should no longer be used to estimate the
GFD. In areas without LLS coverage, an indication of the GFD can be obtained by other means, such as
from satellite-based observations of lightning – see previous section [41].

3.2.5 Considerations for calculating the line outage performance


Each of the detection and measurement methods have benefits and limitations for determining the
lightning parameters for a particular geographical location but, using multiple data sources, readily
available to much of the world, reasonable values can be chosen for estimating the lightning performance
of transmission lines. The following considerations apply when choosing an appropriate value of the GFD:

 The long-term average of the ground flash density is used as a basis for performance calculations.
 The long-term average of the GFD should be based on data spanning at least 10 years, to obtain
a consistent calculation of the GFD of a region.
 Preferential sources for obtaining lightning GFD data ranked from best to worst are: lightning
location systems, lightning flash counters, satellites and, finally when no other options are
available (such as in the Arctic and Antarctic), keraunic levels.
The GFD of any individual year may differ significantly from the long-term average. This will induce similar
variations in the yearly outage rate so that the actual annual line performance may differ significantly from
the calculated long-term average. This may be due to the natural fluctuations from year to year in the GFD
and also because of the stochastic nature of the distribution of lightning strokes in a given area [30]. This
latter effect can easily be quantified by applying a Poisson distribution to obtain the confidence interval of
the average number of strokes to the line. Unfortunately, there is presently not much information publicly
available on the year-to-year variations in GFD.

3.3 Lightning current parameters


3.3.1 Introduction
The severity of lightning flashes can either be described in terms of the physical damages to transmission
line components or by flashovers of the line insulation causing line outages. Both aspects are related to
the magnitude of the lightning current and the associated flash and stroke parameters. Flash parameters
include: the polarity, number of incident strokes and time interval between strokes. Stroke parameters
include: peak current, maximum rate of rise (i.e. steepness), average rate of rise, wave front duration,
overall duration, charge transfer, and specific energy (action integral). An extensive review and
background information on these and other lightning parameters may be found in CIGRE brochure 549 [9].
Damages, such as the melting and breaking of optical ground wires, are typically associated with strokes
involving a large amount of transferred charge, but with a low peak current. This usually corresponds to
long continuing currents in return strokes or to the initial continuous current (ICC) of upward lightning.

19
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

Flashover of the line insulation is typically associated with the high peak current and short duration pulses
at the beginning of the negative first and subsequent return strokes (RS) striking the line. Return strokes
affecting transmission lines may also occur in positive flashes and in about 20 % to 50 % of upward
lightning flashes, following the initial continuous current. Positive flashes have generally been ignored for
lightning performance calculations, because make up only about 10% of all CG flashes. The return strokes
in upward flashes have also been ignored because their peak currents are similar to subsequent strokes
which reduce their significance to the performance of overhead lines [58][9].
Since this brochure focuses mainly on the lightning outage rate of lines, the emphasis here is to provide a
summary on the main parameters of the first and subsequent RS currents of negative downward CG
flashes.
3.3.2 General characteristics of strokes in negative downward CG flashes
Lightning stroke characteristics may vary considerably from stroke to stroke and are therefore usually
expressed in terms of statistical variables. However, some general trends may be identified from the
typical examples presented in Figure 3-3 of the first and subsequent return stroke current obtained from
direct measurements on a short, instrumented tower [106]. It shows the general features of, and
differences between first and subsequent stroke currents of negative downward flashes. Essentially these
are the same as those reported earlier [6], and include:
 A concave wavefront with the maximum steepness of the wave occurring close to the first peak.
 The wavefront of the first stroke (≈ 10 μs) is markedly longer than that of the subsequent stroke
(< 1 μs).
 Multiple peaks of which the highest do not necessarily correspond to the first peak.
 The median magnitude of subsequent strokes is significantly lower than that of first strokes.
 The tail time (as expressed by the time to half value th) of the first stroke (≈ 100 μs) is longer than
that of the subsequent stroke (≈ 50 μs).

20
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

(a) first return stroke of a multiple-stroke flash measured on December 11, 2013

(b) subsequent return stroke of the same flash (second stroke)


Figure 3-3 – Illustrative original unfiltered return stroke currents of a negative CG lightning flash measured at
Morro do Cachimbo Station (MCS). Adapted from IEEE Trans. Industry Applications, Vol. 51, No. 6,
November/December 2015 [106].
Several parameters have been defined to characterize the waveshape of lightning currents. An overview of
these is given in Figure 3-4 and in Table 3-1.

21
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

Figure 3-4 – Definition of front parameters for a lightning current impulse of negative polarity.
See Table 3-1 for descriptions of the parameters.

Table 3-1 – Description of front parameters for a lightning current impulse of negative polarity shown in Figure
3-4.
Parameter Description
IF Final (overall) peak of current (i.e. peak current)
I100 = II Initial peak of current
I10 10 % intercept along the stroke current waveshape
I30 30 % intercept along the stroke current waveshape
I90 90 % intercept along the stroke current waveshape
T10/90 Time between I10 and I90 intercepts on the wavefront
T30/90 Time between I10 and I90 intercepts on the wavefront
S10 Instantaneous rate-of-rise of current at I10
S10/90 Average steepness (through I10 and I90 intercepts)
S30/90 Average steepness (through I10 and I90 intercepts)
Sm Maximum rate-of-rise of current along wave-front, typically at I90
td 10/90 Equivalent linear wave-front duration derived as T10/90 / 0.8
td 30/90 Equivalent linear wave-front duration derived as T30/90 / 0.6
tm Equivalent linear wave-front duration derived from IF / Sm
th Tail time to half value
QI Impulse charge – time integral of current
W Specific energy (action integral) – time integral of the square current

22
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

3.3.3 Considerations for obtaining lightning current parameters


Lightning current parameters, such as those defined in the previous section, are preferably derived from
direct measurements on grounded structures that resemble transmission line towers, which may include
masts, chimneys, and transmission lines. Such structures should ideally be less than 70 m tall, to avoid
including data from upward lightning flashes and to reduce the corruption of the wave shape by wave
reflections from the structure base and top.
Data from lightning location systems (LLS) are not considered sufficiently accurate for this purpose,
because of the indirect nature of these estimates. These systems use simplified expressions to derive
estimates of the stroke peak current from the remotely captured electro-magnetic waves, which are
primarily used for locating the lightning stroke. These expressions relate peak fields to peak current, based
on mean values for the return stroke velocity, but have not been validated extensively for first-return
strokes and stroke peak currents greater than 45 kA [14].
Based on these criteria, there are presently only 3 datasets of lightning current measurements applicable
to lightning performance studies of transmission lines. They are:

 Mount San Salvatore station in Switzerland (MSS): Berger et al (1975) compiled the largest
database for temperate regions from the direct measurements of lightning currents on
instrumented towers Mont San Salvatore Station (MSS), in Lugano Switzerland [52]. This dataset
was later revised by Anderson and Eriksson (1980) [53].
 Transmission line towers in Japan (TLJ): Takami and Okabe (2007) compiled another important
database from measurements performed on instrumented towers of transmission lines in Japan
with tower heights from 60 to 140 m.[56]
 Morro do Cachimbo Station (MCS): Silveira and Visacro (2019) compiled data from direct
measurements of lightning currents on a 60-m high instrumented tower at the Morro do Cachimbo
Station, in Brazil. To date, this is the only dataset, with statistical significance of RS currents in
tropical regions [55] [54].
These datasets are summarized in 9. The first two data sets were considered in CIGRE brochure 549 [9],
and the latter has been updated with the latest available stroke data. The recommendations of this
chapter, therefore, follows those of brochure 549, but with updates from the datasets summarized in 9 as
appropriate.

3.3.4 The return stroke peak current


The return stroke peak current is the lightning parameter that most influences the amplitude of the lightning
overvoltages on the line. Due to the wide range and skew of this parameter, it is best described in terms of
a cumulative distribution function, which in this case, is a Log-Normal distribution function. In brochure 549
it is recommended that the MSS dataset is used as a basis for lightning protection studies [9]. The
parameters describing the cumulative distributions for the peak current of negative first, negative
subsequent and positive strokes of this dataset, are provided in Table 3-2 and the distributions are shown
graphically in Figure 3-5 [9]. This figure also includes a calculated peak current distribution for all first
strokes (including both polarities). This curve has been calculated based on an assumption of a 90/10 %
split between negative and positive polarity CG flashes.

Table 3-2 – Parameters of the cumulative statistical distributions of return stroke peak currents based on the
MSS dataset [9].
Standard deviation of ln/Log of % of cases exceeding
Median
the variable tabulated values [kA]
M [kA] σlnI [β] (base e) σlgI (base 10) 95 % 50 % 5%
Negative First strokes 30 0.610 0.265 11 30 80
Negative Subsequent strokes 12 0.610 0.265 4.6 12 30
Positive strokes 35 1.253 0.544 4.6 35 250

23
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

Probability [%]
99.99 Negative First
99.95 Negative Subsequent
99.9
Positive
Combined Negative and Positive
99
98
95
90
80
70
50
30
20
10
5
2
1

0.1
0.05
0.01

1 10 100 1000
Ipeak [kA]
Figure 3-5 - Cumulative statistical distributions of return stroke peak currents based on the MSS dataset [9].
The cumulative distributions based on the MSS dataset are compared to the previous recommendations of
CIGRE [6] and IEEE [31] in Figure 3-6. The following observations may be made:

 There is good agreement between the recommended distribution and that of CIGRE-63 for first
return stroke current above 20 kA. It has been suggested that the change in slope below 20 kA in
the CIGRE-63 distribution was because data from different populations were combined [9].
 The IEEE distributions for negative first and subsequent strokes [31] [5] are described by the
equations of Table 9-4, and gives the peak current distribution of first stroke current of both
polarities [57] [9]. This explains the good agreement in Figure 3-6 between the IEEE first stroke
current distribution and the calculated peak current distribution for both polarities.
Probability [%]

Probability [%]

99.99 Negative First 99.99 Negative Subsequent


99.95 99.95 CIGRE-63 - Negative Subsequent
99.9 CIGRE [63] - Negative First 99.9
IEEE - Negative First IEEE - Negative Subsequent
99 99
98 Combined Negative and Positive 98
95 95
90 90
80 80
70 70
50 50
30 30
20 20
10 10
5 5
2 2
1 1

0.1 0.1
0.05 0.05
0.01 0.01

1 10 100 1000 1 10 100 1000


Ipeak [kA] Ipeak [kA]
Figure 3-6 – Comparison of frequently used statistical distributions of peak currents.
Recent measurements in Brazil (MCS) and in Japan (TLJ) – see 9 – Table 9-1 & Table 9-2 – show that
there may be a dependence of the lightning parameters on the geographical location. Figure 3-7
graphically depicts the cumulative peak current distributions of negative first and subsequent strokes
obtained from direct measurements at MSS (Switzerland), TLJ(Japan) and MCS(Brazil). The parameters
of the cumulative statistical distributions of these datasets, presented in Table 3-3, show that the median
peak values of both first and subsequent return strokes measured at MCS (43.3 kA and 17.3 kA) are about
40% higher than those measured in temperate regions. Thus, users of this guide are encouraged to apply
lightning current statistics from local sources as they become available.

24
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

Probability [%]

Probability [%]
99.99 MSS (Switzerland) 99.99 MSS (Switzerland)
99.95 99.95
99.9 TLJ (Japan) 99.9 MCS (Brazil)
MCS (Brazil)
99 99
98 98
95 95
90 90
80 80
70 70
50 50
30 30
20 20
10 10
5 5
2 2
1 1

0.1 0.1
0.05 0.05
0.01 0.01

1 10 100 1000 1 10 100 1000


Ipeak [kA] Ipeak [kA]
Figure 3-7 - Cumulative statistical distributions of peak currents obtained from direct measurements at MSS,
TLJ, and MCS stations. (left) First strokes, (right) subsequent strokes.

Table 3-3 – Parameters of the cumulative statistical distributions of return stroke peak currents for the TLJ
and MCS datasets [9].
Standard deviation % of cases exceeding
Median
of ln of the variable tabulated values [kA]
M [kA] σlnI [β] (base e) 95 % 50 % 5%
TLJ 29.3 0.64 10.1 29.3 84.9
Negative First strokes
MCS 43.3 0.47 20 43.3 93.6
Negative Subsequent strokes MCS 17.3 0.54 7.1 17.3 42

Based on the present understanding of lightning stroke attraction to grounded structures it may be
expected that the distributions of peak currents determined from direct measurements on towers may be
biased towards higher currents compared to strokes to flat terrain [9]. Depending on which stroke attraction
model (i.e. striking distance equation) is used, this increase should be in the order of 20 to 40 % [63],
although, this increase may be mitigated to some extent by the influence of neighbouring objects such as
buildings and trees. However, to date no such height bias is evident from measurements taken on towers
of different heights [53] [64].

3.3.5 Front time parameters and steepness


Depending on the criterion adopted for assessing the occurrence of flashover across insulator, the risetime
of the current can become an important influencing factor. Basically, the parameters of the applied wave in
the simulation are selected to be equal to one of the equivalent front times (i.e. td10/90, or td30/90) which are
listed in Table 3-4 for first strokes and Table 3-5 for subsequent strokes. The most conservative (i.e. worst-
case) approach is to set the front time equal to the “equivalent linear wave-front duration”, tm, which is
derived as IF / Sm. For the ramp-slope waveshape used in simplified calculations, this will result in a
wavefront steepness equal to the maximum steepness, Sm.

The steepness of the lightning current is an important parameter to replicate when calculating lightning
overvoltages. The parameters of the Log-Normal distribution for the lightning current wave-front steepness
are listed in Table 3-4 for first strokes and for subsequent strokes in Table 3-5. In general, selecting a
higher steepness will result in higher overvoltages.

25
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

Table 3-4 – Lightning current front parameters of log-normal distribution for negative first stroke currents.

Approximation by log-normal Percent of cases exceeding


Sample distribution tabulated values
Parameter Dataset
size
Geometric mean  95 % 50 % 5%
MSS 80 5.63 0.58 2.25 5.63 14.125
td 10/90
TLJ 120 4.8 0.39 2.5 4.8 9.0
[µs]
MCS 51 6.4 0.41 3.3 6.4 12.5
MSS 80 3.83 0.55 1.5 3.83 9.67
td 30/90
TLJ 120 3.2 0.41 1.6 3.2 6.2
[µs]
MCS 51 4.2 0.50 1.9 4.2 9.6
MSS 75 24.3 0.60 9.1 24.3 65
Sm
TLJ 120 18.9 0.60 7.0 18.9 51.2
[kA/µs]
MCS 51 20.8 0.30 12.7 20.8 34.1
MSS: Mount San Salvatore [53] [52]; MCS: Morro do Cachimbo Station [55]; TLJ: Transmission Line in Japan [56]
 corresponds to the standard deviation for base e.

Table 3-5 – Lightning current front parameters of log-normal distribution for negative subsequent stroke
currents.
Approximation by lognormal Percent of cases exceeding
Sample
Parameter Local distribution tabulated values
size
Geometric mean  95 % 50 % 5%
MSS 114 0.75 0.92 0.125 0.75 3.5
td 10/90
TLJ - - - - - -
[µs]
MCS 77 0.82 0.75 0.24 0.82 2.8
MSS 114 0.67 1.01 0.17 0.67 3
td 30/90
TLJ - - - - - -
[µs]
MCS 77 0.62 0.70 0.2 0.62 1.96
MSS 113 39.9 0.85 9.9 39.9 161.5
Sm
TLJ - - - - - -
[kA/µs]
MCS 77 35.1 0.70 11.1 35.1 111
MSS: Mount San Salvatore [53] [52]; MCS: Morro do Cachimbo Station [55]; TLJ: Transmission Line in Japan [56]
 corresponds to the standard deviation for base e.

The complexity of performing lightning performance studies may be reduced by utilizing correlations
between the peak current and wave front parameters to reduce the number of cases that need to be
considered. In CIGRE brochure 63 the relations in Table 3-6 are suggested.
Table 3-6 – CIGRE correlations between the peak lightning current and wave front parameters for negative
downward flashes [6].
Steepness Front time
Input Parameter IF
Sm σlnI [β] (base e) tm (IF/Sm) td 30/90 σlnI [β] (base e)
Shielding failure –3 < IF ≤ 20 kA 12. 0𝐼𝐹0.171 0.554 2.51 1. 77𝐼𝐹0.188 0.494
Backflashover – IF > 20 kA 6.50𝐼𝐹0.376 0.554 1.37 0. 9060.411
𝐹 0.494

Subsequent strokes 4.17𝐼𝐹0.900 0.706 0.240𝐼𝐹0.998 -- 0.706

From more recent measurements of lightning currents to transmission lines in Japan the relationships in
Table 3-7 were derived [65]. Those derived from Brazilian dataset are listed in Table 3-7 [55]. This latter
dataset showed no correlation between the peak current and maximum steepness.
Table 3-7 – Correlations between the peak current and lightning current front parameters for negative
downward flashes based on measurements in Japan (TLJ) [65].
Input Parameter IF Steepness Sm Front time td 30/90
Negative first stroke 1.40𝐼𝐹0.77 1.39𝐼𝐹0.25

26
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

Table 3-8 – Correlations between the peak current and lightning current front parameters for negative
downward flashes based on measurements in Brazil (MCS) [55].

Input Parameter IF Steepness Sm Front time td 30/90


Negative first stroke No correlation 0.21𝐼𝐹0.796

As shown in Figure 3-8, these relationships give similar values and fall within the normal spread of
measurements, in the range of often occurring lightning currents with peaks from 20 to 80 kA. More
pronounced differences occur in the high current region, especially for the relationships derived from the
Brazilian data set. Since each of these datasets comprise a modest number of measurements, (51 flashes
from Brazil, approximately 90 negative downward flashes from Switzerland and 120 negative downward
flashes from Japan), care should be taken when extrapolating these relationships.
100 100
Cigre - Sm Cigre td 30/90

Equivalent Front Time td 50/90 [µs]


Maximum Steepness Sm [kA/µs]

TLJ - Sm TLJ td 30/90


MCS td 30/90
10

10

1 0.1
1 10 100 1000 1 10 100 1000
Peak Current IF [kA] Peak Current IF [kA]

Figure 3-8 – Relationships between maximum steepness, equivalent front time and peak current.
The relationships between peak current and maximum steepness derived from the Japanese data resulted
in an equivalent linear wave-front duration (tm = IF/Sm) of between 1.8 and 2.2 µs for lightning current
amplitudes in the range 50 kA to 130 kA which is typical for backflashover analyses. The equivalent linear
wave-front duration for the CIGRE relationships is longer, and in the range 2.0 to 3.0 µs over the same
current range. The Japanese relationships in Table 3-7 is therefore more conservative, and comparable to
the 2 µs linear front historically used for simplified calculations [2] [3] [4] [5].

3.3.6 Duration (time to half-value)


The “time to half-value” or “tail time” (th) is commonly referred to as “duration”. This parameter is important
when assessing surge arrester energy in the case of shielding failure. Also for the backflashover case, this
parameter may be important especially when integration or leader progression methods are used to
evaluate insulator flashover on the wave tail. Cumulative distributions of th are listed in Table 3-9 for first
strokes and in Table 3-10 for subsequent strokes. In most studies, the median value is adopted for
duration. Note that the median value of duration of MCS (Brazil) is about 25 % shorter than that of MSS
(Switzerland).
Table 3-9 - Statistics of duration (time to half-value th) of first return stroke currents.
Percent of cases exceeding tabulated values
Dataset
95 % 50 % 5%
MSS [52] 30 75 200
TLJ [56] 9.5 36.5 139.7
MCS [55] 18.4 56.2 171.5

Table 3-10 - Statistics of duration (time to half-value th) of subsequent return stroke currents.
Percent of cases exceeding tabulated values
Dataset
95 % 50 % 5%
MSS [52] 6.5 32 140
MCS [55] 1.84 15.1 123.9

27
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

4 Lightning stroke attraction


An essential part of estimating the lightning performance of transmission lines, is to determine how many
lightning flashes will hit the line and how these flashes are distributed between the shield wires (if present)
and the phase conductors. This is done by modelling the stroke attachment process. Protrusions from the
earth, like transmission lines, towers or wind turbines, will result in local electric-field (E-field)
enhancements, which make them likely locations for the formation of upward leaders. Given sufficient time,
upward leaders can connect with a portion of the downward leader to establish the high current first return
stroke. Such protrusions therefore seem to attract lightning and its effectiveness in doing so is expressed
in terms of its “stroke attraction”. This stroke attraction is evaluated by “simulating” the development of the
downward stepped leader, the initiation and growth of the upward leader and the attachment process.
In order to discuss the calculation methods used to model the stroke attraction process, a basic
understanding of the formation of the negative downward stroke and its attachment to grounded objects is
necessary.

4.1 Attachment of the negative downward stroke to a grounded object


As explained earlier, the negative downward stroke starts with the initiation of a “Stepped Leader” at one of
the negative charge concentrations in the cloud. It grows to earth in a stepwise fashion while the
background electric field exceeds about 100 kV/m [66]. Initially the steps are in the order of 50 m to 100 m
but they become shorter, approaching 10 m, as the stepped leader approaches ground [66]. It takes about
1 μs for each step to form, but the time interval between the steps vary from about 50 μs during the initial
stages of the leader development, down to about 10 μs when it approaches ground [66]. The speed by
which the leader grows, especially when it approaches ground, is important to the stroke attraction of
grounded structures. Measurements indicate that this may vary between 0.3 x 105 m/s and 39 x 105 m/s
with average values ranging from 1 x 105 to 6 x 105 m/s [68].
As shown in Figure 4-1, the stepped leader itself comprises a hot core which is surrounded by a corona
sheath [7]. The core diameter is estimated to be about 0.2 to 1 m and the diameter over the corona sheath,
2R0, is about 6 m.

Figure 4-1 – Components of the stepped leader. Drawing adapted from [7].
Photograph licensed from: Tom Warner / WeatherVideoHD.TV

The stepped leader acts as the conducting path which carries the charge from the cloud to ground. It has
been shown that there is a good relationship between the current of the first return stroke and the charge
stored in the stepped leader channel. It is still, however, unclear, how exactly the charge is distributed
along the stepped leader. For modelling purposes, a uniform, linearly increasing or exponentially
increasing charge distributions have been assumed. For the linearly and exponentially increasing charge
distributions it is assumed that the charge density increases towards the leader tip.

28
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

The geometry of, and charge concentration at the leader tip, results in a very high electric field which then
results in the ionization of the air due to electron avalanches. These ionized filaments are called
“streamers” and they form a streamer zone in front of the leader tip. Streamers are the precursors of the
leader and they supply the current to heat up the air to form the leader. The streamer is not as conductive
as the leader, but it maintains a constant voltage gradient along its length, which is approximately
500 kV/m and for positive polarity and about 1 000 kV/m for negative streamers. Initially the gradient along
the leader channel is about the same as that for streamers, but it quickly reduces to an ultimate value of
about 1 kV/m [69] as the leader channel heats up. The streamer zone comprises both positive and
negative streamers. The positive streamers reach backward from the edge of the streamer zone to the
leader tip and the negative streamers extend from the edge of the streamer zone a short distance into the
remaining gap.
As the stepped leader approaches ground, the electric field around grounded objects will be increasing,
with the highest fields occurring at the extremities of the grounded structures. At a certain point, the electric
field becomes high enough to ionize the air, triggering the formation of streamers in the high field regions.
As before, the streamers consolidate into a leader channel as the streamer currents heat up the air [67]
and so the upward leader is established. It will grow towards the downward stepped leader if the streamers
at the leader tip supply enough energy to maintain the hot leader channel.
Important parameters [70] [67] that influence the inception of the upward leader are (1) the amount of
charge in the tip of the downward leader, (2) the distance between the downward leader tip and the
structure, (3) the height of the structure. An additional factor that influences the initiation and formation of
the upward leader on overhead lines, is the instantaneous value of the operating voltage on the conductor.
The resulting E-field around the conductor either enhances (in case of a positive polarity voltage) or
inhibits (in case of a negative polarity voltage) the formation of the upward leader. Factors that influence
the growth of the upward leader is (a) the impedance of the structure and its ground connection and (b) the
speed of propagation of the downward leader.

Figure 4-2 – Depiction of a leader approaching a grounded object, just before the final jump.
Photograph licensed from: Tom Warner / WeatherVideoHD.TV.

Both the downward and upward leaders grow generally in the direction of the highest field, but there is
significant randomness in the growth pattern so that they do not exactly follow the path along the highest
potential gradient. This is caused by the shielding effect provided by the corona and streamers in front of
the leader tip. However, during the last few steps of the leader development before the final jump, the

29
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

background electric field becomes strong enough to overcome the local potential grading effect of the
streamers and the leaders may turn towards each other [67]. This situation is illustrated in Figure 4-2.
Conditions for the “final jump” are reached when the streamer zone bridges the remaining distance
between the tips of downward and connecting leader [67]. At this stage flashover is unavoidable, even if
the voltage driving the process would be removed, because so much charge has accumulated along the
leaders. The length of the final jump can be determined from the estimates available for the potential of the
leader channel (which is a function of the prospective return stroke current) and by assuming an average
voltage gradient of 500 kV/m across the streamer zone. Several upward leaders may compete to connect
with the downward stepped leader and the winner is the first one for which the final jump conditions are
fulfilled.
An extensive review of the lightning stroke attachment to transmission lines and methods for evaluating it
can be found in CIGRE brochure 704 [10]. In general, two approaches are taken to evaluate the
attractiveness of transmission lines:
1. The Electro-Geometric Method (EGM) is a simple method and easy to implement. It uses the
“Striking Distance” associated with first return stroke peak current as a basis for evaluating the
attractiveness of transmission lines.
2. The Leader-Progression Method is more computationally intensive as it attempts to simulate the
attachment process of lightning to grounded structures. It uses the “Attractive Distance” as a basis
for evaluating the attractiveness of transmission lines. Note, the “attractive distance” is defined
similarly to the more commonly used “attractive radius”, which describes lightning attachment to
towers, masts and generally slender ground objects.
The striking distance and attractive distance are shown in Figure 4-3 and are defined as follows:
 Striking Distance: The distance between the tip of the downward leader and the future termination
point on the conductor (or structure) at the time that the conditions for the final jump are fulfilled.
 Attractive Distance: The horizontal distance between the vertical shaft of the downward leader and
the conductor (or structure) at the time that the conditions for the final jump are fulfilled.

Figure 4-3 – Definition of the striking distance and attractive distance at the time that the conditions for the
final jump are fulfilled.

30
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

These striking or attractive distances can then be used to estimate the number of flashes to the shield
wires or supporting structures (for evaluating the backflashover performance) and to the phase conductors
(for evaluating the shielding performance). A measure for the adequacy of a model for the lightning stroke
attachment process is how well its estimates of the flash incidence match the observed rates.

4.2 Electro-Geometric Method


4.2.1 Introduction
At present, the electro-geometric method (EGM) is still the most common way to evaluate the risk of
lightning incidents to overhead transmission lines [71], primarily because it is easy to implement into
computer routines. The concept of the EGM was first introduced by Young et al [72]. It is based on the
relationships between the electric potential of the downward leader channel and the magnitude of the
return stroke current, from which striking distance equations were derived, usually in the form of Equation
4-1 [7].

r = A 𝐼𝑝b [m] Equation 4-1

Where:
r the striking distance
Ip the peak of the prospective lightning current.
A, b constants that are dependent on the specific strike equation used.

The general approach of the Electro-Geometric method is illustrated on the right-hand side of Figure 4-4.
For each prospective first stroke current, the exposed distances Dg and Dc are calculated. These
represents the zones within which the first return stroke will terminate on the shield wire and phase
conductor, respectively.

Figure 4-4 - Lightning attachment to overhead transmission line conductors based on the attractive distance
(left) and striking distance (right) [10].

31
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

Several different values were proposed for “A” and “b” of the striking distance equation, as is summarized
in Table 3.2.1 of CIGRE brochure 704 [10], which conveniently, also includes equations for evaluating the
exposure width (DC) and the maximum shielding failure current (Imsf).

None of these striking distance equations take directly account of the fact that taller objects attract more
lightning than shorter ones, as is evident in Equation 4-1 that lacks a height term. Rather, in some models,
this is done indirectly by reducing the attractive distance to ground relative to that of the conductor by the
application of a correction factor that is dependent on the conductor height. However, this approach has
been criticised as not strictly correct because the height adjustment is made to the striking distance to
ground which is not physically justified [7] [74].
In CIGRE brochure 63 [6] it was suggested that the Brown-Whitehead striking distance equations
(Equation 4-2) be used while in the more recently published IEEE [5] Equation 4-3 is recommended.
Equation 4-3[4], which is also used in the source code for the IEEE Flash program V1.8 and higher, is an
empirical approximation for the ratio of ground to conductor striking distance as a function of phase
conductor height [85].

𝑟𝑐 = 7.1 𝐼𝑝0.75 𝑟𝑔 = 6.4 𝐼𝑝0.75 [m] Equation 4-2

𝑟𝑔 = 0.55 𝑟𝑐 for hs > 40 m


𝑟𝑐 = 10 𝐼𝑝0.65 [m] Equation 4-3
3.6+1.7∙𝑙𝑛𝑒 (43−ℎ𝑠 )
𝑟𝑔 = 𝑟𝑐 for hs ≤ 40 m
10

Where:
rc the striking distance to the phase conductors and shield wires
rg the striking distance to the surface of the ground
hs the height of the shield wire.

4.2.2 Flash incidence


The annual flash incidence to the shield wires, NL. It can be calculated as follows:

𝑁𝑔 𝑙
𝑁𝐿 = [𝑆 + 2 ∫ 𝐷𝑔 (𝐼)𝑓(𝐼)𝑑𝐼 ] [Flashes/year] Equation 4-4
1000 𝑔
𝐼=0

Where:
Ng the regional ground flash density. [flashes / km2 / year].
l the length of the line [km].
Sg the distance between the shield wires [m].
Dg(I)the exposed distance of the ground wires [m] as a function of
the prospective lightning current, I [kA].
f(I) the probability density function of the occurrence of the peak lightning current.
The factor 2 in the equation assumes a bilateral symmetry of the line.

The exposed distance is geometrically determined based on the striking distance according to the EGM
model employed.
This has been simplified to express the flash incidence in terms of the average attractive distance, ra [6]:

𝑁𝑔
𝑁𝐿 = (2 ∙ 𝑟𝑎 + 𝑏) [Flashes /100 km /year] Equation 4-5
10

Where:
ra the average attractive distance of the line [m].

32
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

b the width of the line [m].


On shielded lines this is the separation distance between the shield wires, Sg, and on
unshielded lines, it is the separation distance between the outermost phase conductors.
The factor 2 in the equation assumes a bilateral symmetry of the line.
The equation of the attractive distance published in brochure 63 [6] is still widely accepted as a good
estimate:

𝑟𝑎 = 14𝐻𝑇0.6 [m] Equation 4-6

Where:
HT the average tower height [m].
4.2.3 Shielding analysis
The total shielding failure rate (SFR) is the sum of the flash incidences to each phase conductor. The flash
incidence to each of the phase conductors can be calculated as follows:
𝐼𝑚𝑠𝑓
𝑁𝑔 𝑙
SFR = ∫ 𝐷𝐶 (𝐼)𝑓(𝐼)𝑑𝐼 [Flashes per year] Equation 4-7
1000
𝐼=0

Where:
Ng the regional ground flash density [flashes / km2 / year].
l the length of the line [km].
DC(I) the exposed distance of the phase conductors [m] as a function of
the prospective lightning current, I [kA].
f(I) the probability density function of the occurrence of the peak lightning current.
The upper integration limit is the maximum shielding failure current [Imsf, kA].

Likewise, the shielding failure flashover rate (SFFOR) is the sum of the shielding failure flashovers to each
phase conductor and the SFFOR for each phase conductor is calculated as follows:
𝐼𝑚𝑠𝑓
𝑁𝑔 𝑙
SFFOR = ∫ 𝐷𝐶 (𝐼)𝑓(𝐼)𝑑𝐼 [Flashovers per year] Equation 4-8
1000
𝐼=I𝐶

Where:
The lower integration limit is the critical current, IC, which is the minimum current required for a
flashover of the line insulation.

For the usual case where there are several subsequent strokes in a flash, SFFOR approaches SFR. It is,
namely, quite likely that the initial stroke causing the shielding failure, which is often less than the median
peak subsequent stroke amplitude of 12 kA, will be followed by a higher current subsequent stroke that
flash over the line insulation [9].
The following observations may be made with respect to the input parameters:

 The shielding analysis is usually based on the average heights of the conductors.
 It is recommended by Hileman [7] and endorsed by IEEE [5], that the choice of shielding angle be
based on a small non-zero value for the shielding failure flashover rate; for example 0.05
flashovers per 100 km of line length per year.
 A conservative estimate of the critical stroke current is given by[76]:

𝑈50
𝐼𝑐 = 2 [kA] Equation 4-9
𝑍
Where:

33
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

U-50, the negative polarity full-wave lightning impulse strength of the line insulation which is taken
as the voltage with a 50 % probability for flashover [kV]
– see Section 5.5.1 for guidance on the estimation of the U50.
Z the surge impedance of the phase conductor [Ω]

 Power frequency voltage bias on energized conductors is neglected for calculating the critical
current as it is assumed that the average value of this voltage is zero and that any polarity effects
between the phases would average out over time. Furthermore, if the polarity effect is included for
the calculation of the critical current, it would predict that most shielding failures would occur on
the negative polarity half cycle. This is contrary to some field experience from both AC and DC
systems that show a bias towards more shielding failures to conductors which carry a positive
polarity voltage [77] [78] [10].

4.2.4 Applicability of the EGM


Verification of methods for designing the lightning shielding of transmission lines remains difficult, although
field experience on lines with a tower height below 50 m indicate that the EGM generally result in effective
shielding angles [7]. More recent experiences suggest, however, that this method becomes increasingly
inaccurate for lines of above 500 kV or those having tall towers with a span length exceeding 400 m.
Especially in the cases when inter-conductor spacing approaches 20 m, the observational results differ
significantly from those calculated by the conventional method. It was reported that the lightning stroke
rate, particularly to the ground wires of ultra-high voltage (UHV) designed transmission lines and 500 kV
transmission lines was 5.1 and 2.7 times larger, respectively, than the results calculated by the
conventional method [79]. Reasons for this include:
1. The line voltage, which is neglected in the EGM, plays an increasing role in the development of the
upward leader for higher system voltages.
2. The striking distance equations do not fully account for the formation of upward connecting leaders
originating from tall grounded or energized structures like a transmission line or mast.
Thus, the application of the EGM is only justified for lines with an operating voltage of below 500 kV and
relatively short structures (i.e. less than 50 m) where upward leaders play a minor role (See Section 4.1).
For tall structures utilized for EHV and especially UHV lines, Tokyo Electric Power Company (TEPCO)
proposed an improved EGM based on their observations of lightning to 500 kV and UHV lines [10]. This
approach is described in Appendix 11.1.

4.3 The Leader Progression Method


The deficiencies of the EGM were recognized in the 1990s and many methods have been developed to
numerically simulate the process of leader development and attachment to structures based on the
understanding of the underlying physics. Specifically, the so-called Leader Progression Method (LPM) was
developed to address the main shortcomings of the EGM by taking account of the initiation and growth of
the upward leader in determining attractive distances to the line. The attractive distance is described by an
equation in the form of Equation 4-10 [10].

L = ξℎ𝐸 𝐼𝑝F [m] Equation 4-10

Where:
L the attractive distance
Ip the peak of the prospective lightning current.
h the height of the conductor
E, F constants that are dependent on the specific LPM used.

The calculated attractive distances are then used to determine the first stroke incidence to the shield wires
and phase conductors, as illustrated on the left-hand side of Figure 4-4. Again, Dg and DC represents the
exposed distances of the shield wire and phase conductor, respectively.
The LPM involve a step-by-step simulation of the approaching downward stepped leader, at each step
solving the quasistatic fields, to determine the inception of the upward leader and the direction that both

34
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

the leaders would take on the next step. Schematic depictions of two leader progression models are
shown in Figure 4-5.
Different sets of assumptions underpin the various leader progression models, which may result in
significant differences in the calculated attractive/striking distances. These differences include:
For the Downward leader:
 Whether the trajectory of the downward leader is attracted by the upward leader or continues to
extend straight down to the ground.
 How the charge is distributed in the leader channel. Assumptions made in the past include.
o Uniform charge distribution along the leader channel.
o A linearly increasing charge distribution from the cloud base to the leader tip.
o An exponential charge distribution with a distinct charge concentration at the leader tip.
 How the total charge and the voltage of the leader is related to the peak prospective current.
 The speed by which the leader tip propagates down to earth.
For the Upward leader:
 Criteria for the inception of the upward leader. This could be defined as an inception E-field
gradient, a threshold value for the space potential at the structure or it could be determined
through a detailed model of the ionization process.
 How the path of the propagation of the upward leader is determined.
 Criteria for determining when the final jump occurs and when the upward leader fails to connect
with the downward leader.

Figure 4-5 – Schematic representation of two leader progression models [10].

Further details on the various leader progression models, including main parameters defining them, are
provided in CIGRE brochure 704 [10]. This includes descriptions of the downward leader model, the
upward leader model, leader inception model, and final jump model. A review and recommendations are
also provided on the selection of important input parameters, such as the cloud potential, charge
distribution in the leader channel, the charge of upward leader, velocities of upward and downward
leaders, as well as the initial velocity of upward leader.
Several studies have been performed to demonstrate the ability of the LPM to correctly predict shielding
failures as observed on UHV lines where the EGM fails. This is in part because these methods can

35
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

account for the effect of phase conductor voltage magnitude and polarity on the inception conditions of the
upward leader. However, they have not seen widespread application, despite presenting a more physically
sound and rigorous account of the attachment process, because they are more complex and
computationally intensive.

4.4 Simplified lightning attachment method


4.4.1 Background and overview
On the initiative of this working group a simplified attachment model was developed [90] which is based on
the Rizk lightning attachment model originally described in [85] and further updated in [86] [87] [88] [89].
The intension with this simplification is to overcome the computational complexity of the original model
without compromising the results too much. Specific complexities addressed included:
1. The necessity to use the charge simulation technique to calculate the electric field and changes to
it in response to the descending negative downward leader, and the initiation of the upward
positive connecting leader, while taking account of the mutual proximity effect among conductors.
2. The necessity to solve the differential equations defining the leader propagation trajectories,
including effects of variable leader speeds, before the final jump.
3. Assessment of the effect of ambient ground field due to cloud charge, before downward leader
descent, on upward connecting leader initiation and subsequent motion. A particular complication
relates to the effect of corona on ground wires.
4. Assessment of the effect of the line operating voltage, particularly during the positive half-cycle,
which is most significant for UHV lines.

The basic method, simplifications and validation are described in Appendix 11.2 [90]. It greatly reduces the
calculation effort to apply the Rizk lightning attachment model, without a significant loss in the accuracy. At
present the simplified method is only applicable to lines on flat ground.
The basic principle of the method is illustrated in Figure 4-6. The attractive distances to all conductors are
calculated as a function of the instantaneous voltage and the prospective first stroke current, I. The
instantaneous voltage is defined by the phase angle, θ, of the power frequency voltage at the instant of the
first return stroke.
For this method, Lds_s(I, θ) is the attractive distance to the shield wire and Lds_p(I, θ) is that to the phase
conductors.

Figure 4-6 – Lightning attractive distance to overhead transmission line conductors as defined in the Rizk
lightning attachment model.

36
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

4.4.2 Flash incidence


This method allows for the calculation of the total annual number of flashes to the line by integrating all
flashes that fall within the attractive distance to the shield wires and the separation distance between the
shield wires, Sg. It is calculated as follows:
2𝜋 ∞
𝑁𝑔 𝑙 ∆𝜃
N𝐿 = {𝑆𝑔 + 2 ∑ [ ∫ 𝐿𝑑𝑠_𝑠 (𝐼, 𝜃)𝑓(𝐼)𝑑𝐼 ]} [Flashovers per year] Equation 4-11
1000 2𝜋
𝜃=0 𝐼=0
Where:
Ng the regional ground flash density [flashes / km2 / year].
l the length of the line [km].
Lds_s the attractive distance to the shield wire [m] as a function of the prospective lightning current,
I, and power frequency phase angle, θ.
Δθ the step size in phase angle. If twelve phase angles are selected, then Δθ is π/6.
f(I) the probability density function of the occurrence of the lightning peak current.
The factor 2 in the equation assumes a bilateral symmetry of the line.

4.4.3 Shielding analysis


As before, the number of lightning flashes to the uncovered area DC in Figure 4-6 represents the shielding
failures. The shielding failure rate must be calculated for a range of phase angles of the power frequency
voltage because DC is influenced by the instantaneous value of the phase conductor voltage. It is usually
sufficient to select twelve or more representative phase angles.
The calculation of the SFFOR to each phase conductor is now modified to:
𝐼𝑚𝑠𝑓 (𝜃)
2𝜋
𝑁𝑔 𝑙 ∆𝜃
SFFOR = ∑[ ∫ 𝐷𝐶 (𝐼, 𝜃)𝑓(𝐼)𝑑𝐼 ] [Flashovers per year] Equation 4-12
1000 2𝜋
𝜃=0 𝐼=𝐼𝑐 (𝜃)

Where:
Δθ the step size in phase angle. If twelve phase angles are selected, then Δθ is π/6.
DC(I) the exposed distance of the phase conductors as a function of the prospective lightning
current, I, and the power frequency phase angle, θ [Radians].
IC (the lower integration limit) the critical current, which is the minimum current required for a
flashover of the line insulation. A sufficient estimate of the critical stroke current is given by
Equation 4-13.
The total shielding failure is the sum of the SFFOR of all the phase conductors.
A conservative estimate of the critical stroke current is given by [76]:


𝑈50 2
𝐼𝑐 (θ) = 2 + √ 𝑈𝑚 ∙ 𝑠𝑖𝑛(𝜔𝑡 + 𝜃) [kA] Equation 4-13
𝑍 3
Where:
U-50, the negative polarity lightning impulse strength of the line insulation which is taken as the
voltage with a 50 % probability for flashover [kV]
(note that the negative polarity of this voltage should be correctly inputted into the equation)
– see Section 5.5.1 for guidance on the estimation of the U50.
Z the surge impedance of the phase conductor [Ω]
Um the maximum system voltage [kV]

4.4.4 Validation
The simplified method was validated for several case studies for which the line shielding performance is
known, of which the results are presented in Appendix 11.2.11. The accuracy of the simplifications was

37
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

demonstrated in [90]. In general, it is found that the simplified method adequately predicts the number of
strokes to the line and the shielding failure rate for lines traversing flat terrain.
Not all examples fulfilled the “flat-terrain” criterion which resulted in a discrepancy between the predicted
and actual shielding failure rates. On sloping terrain, the phase facing away from the slope will be more
vulnerable to shielding failures and those facing the slope will be more protected. This effect can be
approximated by considering that the downward negative leader during the attachment phase will descend
not vertically but perpendicularly to the ridge. Conductor heights and lateral displacements must be
recalculated accordingly [90].
Finally, it must be realized that the data available is too few to make firm conclusions, also considering the
effect of geography on the performance of real lines which was not considered in the calculations. Further
benchmarking is therefore still required.

38
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

5 Review of modelling concepts


5.1 Introduction
Detailed models of the transmission line, supporting structures, insulators, and earthing are required to
evaluate the effects of lightning flashes to the overhead ground wires and phase conductors. This is
especially so if the placement of line surge arresters is considered as a mitigation option [13]. An extensive
overview of modelling concepts was published in CIGRE brochure 63 [6]. Although this overview is still
relevant today, there have been several developments that provided new insights for improving modelling
approaches based on classical circuit and transmission line theories. These developments refer primarily
to the application of electromagnetic computation methods, such as the finite-difference time-domain
(FDTD) method, the method of moments (MoM), partial element equivalent circuit (PEEC) method, and
hybrid electromagnetic model (HEM), which allows a self-consistent full-wave solution for both the
transient current distribution in a three-dimensional (3D) conductor system and resultant electromagnetic
fields [24] [22]. These methods are however computationally demanding, and their use is still largely
limited to research groups. However, results from these advanced techniques can be used to develop
improved high-frequency models for the more widely available and computationally efficient circuit-based
methods such as the various EMT implementations.
For these reasons, this chapter is written as a complement to Chapter 5 of brochure 63 [13], only focusing
on developments in representations of the lightning stroke, transmission line towers, frequency
dependence of ground electrodes and the modelling of insulator flashover.

5.2 Representation of the lightning stroke


5.2.1 Source impedance
For lightning transient studies, it is not necessary to model the actual current distribution in the lightning
channel so the lightning stroke may simply be represented as a Norton equivalent current source, as is
shown in Figure 5-1. It comprises an ideal current source, I(t), in parallel with an equivalent source
impedance, Zch. The following parameters need to be specified:

 The magnitude of the current source is equal to the prospective stroke current that would be
injected into the ground if it is perfectly conducting.
 The waveshape is chosen to reflect the essential characteristics of the first or subsequent stroke
current, depending on which is considered for the study. This aspect is further discussed in the
next section.
 The impedance, Zch, which is usually referred to as the equivalent impedance of the lightning
channel, is typically assumed to be a constant value [9][143].

Figure 5-1 – Equivalent current source representing the return stroke channel.
A limited number of studies were done to estimate the equivalent impedance of the lightning channel from
direct measurements of lightning current [9] or to infer the channel impedance from theoretical models
[144]. It was found that the channel impedance is a stochastic parameter that is dependent on the return
stroke peak current. Reported estimates of the channel impedance varies from approximately 460 Ω up to
9000 Ω [142], which is, in most cases, significantly higher than the impedance of the transmission line
earthing system presented to the stroke when it hits the line. Consequently, the lightning stroke may be
approximated as an ideal current source, i.e. Zch = ∞, without significantly affecting the accuracy of the
calculation.

39
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

5.2.2 Lightning current wave shape


5.2.2.1 General considerations
Typical examples of first return stroke waveshapes captured at different stations are presented in Figure
5-2. This figure shows that the first stroke has a pronounced initial concave wave front which lasts until
approximately half of the initial current peak. This is followed by a steep, and more-or-less linear, rise of
the current to the first peak. The maximum steepness of the wavefront typically occurs just before to the
first peak is reached. A few microseconds later, the wave reaches a second peak which is typically higher
than the first. Further subsidiary peaks may follow as the current decays [58] [59] with a time to half value
of about 78 μs [6].
Time [μs] Time [μs]
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
0 0
MSS Concave part
MCS
Concave part -10
-10

-20

Current [kA]
Current [kA]

-20
Linear part -30
Linear part
-30
-40
-40
First peak -50
First peak
-50 -60
Second peak Second peak
-60 Time [μs] -70 Time [μs]
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
0 0
-5 Concave part SAS Concave part TLJ
-10 -5
Current [kA]

-15
Current [kA]

-20 -10
-25 Linear part Linear part

-30 -15
-35
-40 -20 First peak
First peak
-45 Second peak Second peak
-50 -25

Figure 5-2 – Typical current wave shapes for first downward negative stroke. MSS- Mount San Salvatore [155],
MCS- Morro do Cachimbo[55], SAS- South Africa stations [156] and TLJ – Japanese Transmission lines [56].
Note that the shapes of the measured first stroke currents in Figure 5-2 are quite different from the well-
known normalized median waveshape obtained from the Berger measurements [52] which is shown in
Figure 5-3. This is because the ‘‘averaging process’’ adopted for developing this median curve resulted in
a loss of the original distinct patterns [59].
Time [μs]
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80
0

-0.2
Current [p.u. of final peak]

-0.4

-0.6

-0.8

-1

-1.2

Figure 5-3 – Normalized median current wave shape of first downward negative strokes derived from the
Berger measurements [52].

40
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

5.2.2.2 Analytical representations of the first stroke current waveform


Different waveforms have been devised for representing the essential features of first return stroke
currents in lightning performance studies. These include:

 A triangular, ramp-slope type function with a 2 µs front has been recommended by CIGRE and
IEEE for simplified lightning performance calculations [2] [3] [4] [5]. The linear rising wave-front
time was selected so that the wave front has the maximum steepness, Sm.
 A double-exponential waveform, also with a 2 µs front, has been used, mainly to avoid additional
transients caused by the discontinuity at the peak of the ramp-slope type wave. The double
exponential is also easily reproduced in high-voltage laboratories [61]. However, the peak rate of
rise Sm of this waveform occurs at t = 0 and falls to zero at the current peak which is not a realistic
representation of the first stroke current wave front.
 CIGRE-concave waveform [6], which utilizes separate expressions to independently represent the
current wave-front and wave-tail. This waveform has found widespread application.
 Special waveforms, such the Heidler function [60] have been developed for an improved
representation of the wave-front, while maintaining a single equation implementation [58]. It is
continuously differentiable and Sm for a single Heidler function occurs at 50% of the peak.
 A Double-peaked waveform [59], is a summation of seven Heidler functions to reproduce the
median value of all relevant amplitude and time parameters of first return stroke currents (i.e. first
and second peaks, front times Td10 and Td30, time-to-half peak and maximum steepness).
The above-mentioned waveforms are compared in Figure 5-4 with the wave parameters selected to
reproduce the median parameters of Mont San Salvatore (MSS) dataset. Note that the peak of single-
peaked waveforms, which corresponds to the initial peak of the typical return stroke current – see Figure
5-2, occurs much earlier than the second, or final, peak. This is to a certain extent the outcome of selected
maximum steepness. For the double-peaked waveform these parameters are decoupled, and more
realistic front times can be achieved.
35
Absolute Lightning Current [kA]

30

25

20

15 Cigre

10 Double peak

Heidler (dbl exp.)


5
Triangular
0
0 5 10 15 20 25 30
Time [µs]
Figure 5-4 – Representations of first return stroke currents:
Triangular, Double-exponential, CIGRE-concave and Double-peaked.
It is important to utilize an accurate representation of return-stroke currents to obtain realistic estimates of
the lightning overvoltages. Especially the wave-front parameters should be selected carefully, as these
determine, to a large extent, tower top voltages and thus the voltages appearing across the line insulation.
This can be illustrated by considering an electrical model whereby the transmission tower is approximated
as an inductance (LT) of between 0.5 to 1.0 μH per meter of height, and the earth electrode at the base of
the tower, as a resistance (RE). The tower top voltage (VTT) is then given by in terms of the lightning current
I(t):
𝑑𝐼(𝑡)
𝑉𝑇𝑇 = 𝐿 + 𝐼(𝑡)𝑅𝐸 Equation 5-1
𝑑𝑡 𝑇

Equation 5-1 shows that the maximum voltage across the tower body occurs when the injected lightning
current reaches its maximum steepness and that the maximum voltage across the ground electrode occurs

41
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

at the peak of the injected lightning current. With typical values of LT = 30 H for a 30-m single pole tower,
dt = 2 s and RE = 20 Ω, the ratio VTT / I(t) becomes (15 + 20) V/A so both terms in Equation 5-1 are
important. The waveform chosen to represent the first stroke should therefore have the maximum
steepness of the wavefront occurring just before the first peak, as this is the worst case. The waveshapes
fulfilling this criterion are (1) the triangular ramp-slope function, (2) the CIGRE curve and (3) the Double-
Peaked wave. Equations for implementing the CIGRE and Double-Peaked wave shapes are provided in
10. Double-exponential and single Heidler functions do not have Sm occurring close to the peak current and
should thus not be used in backflashover calculations.
The front-time of the current wave is also important. On shielded lines, a part of the lightning current
travels along the shield wires to adjacent structures where it is conducted to ground via the foundations
and earth electrodes. This sets up a negative reflected wave, which returns to the stricken tower, that
reduces the tower top voltage when it reaches it. The time it takes for these negative reflections is at
minimum twice the travel time of one span and for typical transmission lines is in the order of some micro-
seconds. For a stroke with a long front time, reflections from neighbouring towers may arrive back at the
stricken tower before the lightning current reaches its peak and thus contribute to lowering the peak tower
top voltage. Wave shapes with a short front time may not arrive in time and have less of a mitigation effect.
The characteristics of first strokes, necessary for selecting the waveshape parameters, are provided in
Section 3.3.
5.2.2.3 Analytical representations of the current waveform of subsequent return strokes
Examples of typical subsequent return stroke waveshapes captured at different stations are presented in
Figure 5-7. These wave shapes are characterised by a fast rise in the current to a single peak value, with a
front duration of about 0.2 to 5 μs, after which it decays with some oscillations back to zero with a time to
half value in the order of 30 μs [9].

Time [μs] Time [μs]


0 5 10 15 20 25 30 0 5 10 15 20 25 30
0 0
MSS MCS
-5
-5
Current [kA]
Current [kA]

-10
-10
-15
-15
-20

-20
-25

-25 -30
Time [μs]
0 5 10 15 20 25 30
0

-10 SAS
-20
Current [kA]

-30

-40

-50

-60

-70

-80

-90
Figure 5-5 – Typical current wave shapes for subsequent negative strokes. MSS- Mount San Salvatore [155],
MCS- Morro do Cachimbo[55], and SAS- South Africa stations [156].

42
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

Time [μs]
0 5 10 15 20 25 30 35 40 45 50
0

-0.2
Current [kA]
-0.4

-0.6

-0.8

-1

-1.2

Figure 5-6 – Normalized median current wave shape of subsequent downward negative strokes derived from
the Berger measurements [52].
The median parameters of amplitude and time derived from the currents measured at Mount San
Salvatore Station, as shown in Figure 5-6, [52] adequately represents the general features of subsequent
strokes. It can be represented in lightning performance studies by the sum of two Heidler functions [62], as
explained in Appendix 10.4 and with the parameters are selected according to the characteristics provided
in Section 3.3. The resulting wave shape representing the median subsequent stroke current is illustrated
in Figure 5-7.

14

12

10
Current (kA)

0
0 2 4 6 8 10
Time (s)
Figure 5-7 – Representation of subsequent return stroke currents by Heidler functions considering median
peak current parameters of subsequent return stroke currents measured at MSS [52].
II = 11.8 kA, td 30/90 = 0.4 μS, Sm 39.9 kA/μS, th = 32 μS.

5.3 Representation of the transmission line towers


5.3.1 Introduction
The surge response of transmission line towers is best analysed with electromagnetic computation
methods which can solve electromagnetic transient phenomena of three-dimensional structures and
earthing systems accurately [24]. However, these methods are not considered practical for general
purpose lightning performance studies because of their heavy computational burden. Circuit based
models, such as the tower inductance term in Equation 5-1, may not be as accurate but they are much
more accessible and easier to understand. Models with surge impedance and travel time also offer a good
compromise between simplicity and accuracy.
CIGRE brochure 63 provides the background and a basis for representing transmission line towers in
lightning performance calculations [6] either as an inductance connecting shield wires to ground, as a
constant-impedance transmission line, or as a variable-impedance transmission line. These approaches
will be briefly summarized here with some additions, such as the multistorey model, which is based on the
latest research.

43
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

5.3.2 Generalised tower surge impedance


5.3.2.1 Equations for calculating tower surge impedance
Equations for estimating the surge impedance (ZT) of four idealized tower shapes for lightning studies are
provided in Table 5-1. The following abbreviations apply:
h1: Height from base to midsection [m]
h2: Height from midsection to top [m]
Ht: Total height of the tower (h1 + h2) [m]
r: Tower radius [m]
req: Equivalent radius of the tower section [m]
r1: Tower top radius [m]
r2: Tower midsection radius [m]
r3: Tower base radius [m]
d: Separation distance of the tower legs [m]
tt: Tower travel time [s]
c: Speed of light, 3 × 108 m/s

Table 5-1 – Tower surge impedance for 4 idealized tower shapes.


Cylindrical [94] Conical [94] Waist [6] Portal [93]

𝜃 𝜃 𝜃 𝑍1 ∙ 𝑍2
𝑍𝑇 = 60 {𝑙𝑛 [cot ( )] − 1} 𝑍𝑇 = 60 {𝑙𝑛 [cot ( )]} 𝑍𝑇 = 60 {𝑙𝑛 [cot ( )]} 𝑍𝑇 =
2 2 2 𝑍1 + 𝑍2

𝑟𝑎𝑣 2𝐻𝑡
𝜃 = tan−1 ( ) 𝑍1 = 60 ∙ [𝑙𝑛 (√2 ) − 1]
𝐻𝑡 𝑟
𝑟
𝜃 = tan−1 ( )
𝐻𝑡 2𝐻𝑡
𝑟1 ℎ2 + 𝑟2 𝐻𝑡 + 𝑟3 ℎ1 𝑑 ∙ 60𝑙𝑛 ( ) + 𝐻𝑡 ∙ 𝑍2
𝑟𝑎𝑣 = 𝑍2 = 𝑟
𝐻𝑡 𝐻𝑡 + 𝑑

𝐻𝑡 𝐻𝑡 1 𝐻𝑡 ∙ 𝑍1 ∙ (𝑑 + 𝐻𝑡 ) ∙ 𝑍2
𝑡𝑡 = : towers with crossarms or 𝑡𝑡 = : towers without crossarms 𝑡𝑡 =
0.85∙𝑐 𝑐 𝑐 ∙ 𝑍𝑇 𝐻𝑡 ∙ 𝑍1 + (𝑑 + 𝐻𝑡 ) ∙ 𝑍2

All surge impedance equations presented in this section are valid for round shapes, as shown in the
figures. For box shaped structures, such as typical for lattice towers, an equivalent radius must be
calculated so that the circumferences (or perimeter) of the two shapes are the same. For example, for a
lattice tower having a rectangular cross section with width “ a” and depth “b” the equivalent radius, req, is:

(𝑎 + 𝑏)
𝑟𝑒𝑞 = [m] Equation 5-2
𝜋

For practical purposes, the travel time of surges along towers without crossarms, as those shown in Table
5-1, is equal to the tower height divided by the speed of light. For towers with crossarms, it was found
experimentally that the apparent propagation along the tower is slowed down to approximately 85 % of the
speed of light, because of the additional mean path length presented by the crossarms. On more detailed

44
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

models, this effect may be included by representing the crossarms as short stub transmission lines with its
surge impedance calculated with Equation 5-3 [37].

5.3.2.2 Tower sections and crossarms


The surge impedance of a round tower section with radius, r, and a non-zero height above ground, h, is
given by [37]:


𝑍 = 60 cosh−1 ( ) [Ω] Equation 5-3
𝑟

The travel time, t, of the section is given by:

𝑡 = 𝑙⁄𝑐 [s] Equation 5-4

Where:
l the section length [m]
c the speed of light [3 x 108 m/s].
2ℎ
The approximation to Equation 5-3, 𝑍 = 60 𝑙𝑛 ( ), is valid when h >> r.
𝑟

Equation 5-3 is valid for both horizontal and vertical directions of propagation and can therefore be used
for both tower sections and crossarms.
For a box shaped crossarms, the equivalent radius is calculated with Equation 5-2. For a triangular
tapering crossarm, with a width “a” and depth “b” at the base of the crossarm, the equivalent radius
becomes:

(𝑎 + 2𝑏)
𝑟𝑒𝑞 = [m]
2𝜋

Equation 5-5

(𝑎 + 𝑏)
𝑟𝑒𝑞 = [m]
2𝜋

5.3.2.3 Insulating towers with an earth lead


The surge impedance of transmission line poles made of insulating materials such as wood, fibre-
reinforced polymer (FRP) or composite materials is basically determined by any earth lead(s) running
down the pole – since the pole itself does not carry any current. Earth leads will be in full corona when
conducting lightning current. This corona envelope increases the lead capacitance which reduces the
surge impedance. The radius of the corona envelope, rC, is determined iteratively with Equation 5-6 [2]:

𝑉
𝑟𝐶 = 2ℎ
𝐸0 𝑙𝑛 ( ) [m] Equation 5-6
𝑟𝑐

Where:
V: an estimate of the insulation flashover voltage under backflashover, which may be taken as
1.8 times its positive polarity 50 % flashover voltage [kV].
E0: the limiting corona gradient below which the corona will not extend further [kV/m]. A value of
1500 kV/m is suggested.
h: the height of the conductor which should be taken as the pole height [m].

The initial value of rC is usually taken as 0.01 m and the recursive substitution converges quickly.

Presently, insulating support structures are mostly used on systems of 145 kV or lower, which means that
most downleads are used on lines with an insulation distance of 2 m or less. Based on these parameters,
the upper limit for the corona radius is estimated to be 0.25 m.

45
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

The corona envelope around the earth wire modifies the wire capacitance but not the inductance. Thus,
the surge impedance equation is modified as follows:

2ℎ 2ℎ
𝑍𝑐 = 60 [√𝑙𝑛 ( ) 𝑙𝑛 ( ) − 1] [Ω] Equation 5-7
𝑟𝐶 𝑟

Where:
r the radius of the earth wire [m].

The corona will also slow down the propagation of the surge down the wire and a conservative estimate of
the surge travel time is given by:

𝐻𝑡
𝑡𝑡 = [s] Equation 5-8
0.85 ∙ 𝑐

5.3.3 Tower as a simple inductor


The simplest representation of a tower for lightning studies is to model it as a pure inductor with the
inductance calculated as:

𝑍𝑇
𝐿𝑇 = ∙ 𝐻𝑡 [H] Equation 5-9
𝑐

This representation is, however, only suitable for simplified hand calculations or in case of short towers
with a high footing resistance. Equation 5-9 may also be used to simplify the representation of towers more
than two spans away from the stricken point to reduce the complexity and size of the transient model.

5.3.4 Tower as a constant-impedance transmission line


The traveling wave behaviour of the tower can be incorporated in the transient model by representing it as
a constant parameter lossless transmission line. The characteristics of this modelling element are specified
in accordance with Table 5-1. If there are crossarms, the propagation velocity may be specified to 85 %
the speed of light.
The insulator voltages are defined by the induced voltage on the phase minus the voltage on the tower at
the height of the phase conductor.

5.3.5 Tower as a variable-impedance transmission line


The tower model may be divided in several series elements, each with different surge impedance
corresponding to segment height and average radius as sketched in Figure 5-8.

Figure 5-8 - Tower as a variable-impedance transmission line.

46
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

In this example, the tower is divided into four sections with the surge impedance in each section calculated
as follows:
 The surge impedance of the tower base (ZTd in Figure 5-8) is calculated with the equation given for
a waist tower in Table 5-1. The values of r1, r2 and r3 are selected so that the tower base
resembles a truncated cone.
 The surge impedance of the upper sections is calculated with Equation 5-3, based on the
equivalent radius and height of each section.
When multiple reflections and transmissions among segments are included, better estimates of voltages at
each phase conductor height are developed than with a model using constant surge impedance. The
variable impedance model also allows for adding crossarms, as explained in section 5.3.2.2.

5.3.6 Ground plane surge response


Studies of reflections from the base of towers, using scale models, triggered lightning and theory, have
shown that the initial reflection coefficient at the tower base is less than the theoretical coefficient based on
a lumped terminal resistance [95] [96]. It was believed that this effect may be neglected in simplified
calculations because of the mitigation effect of the relatively long front times of the first return stroke
current. A comparison of the calculated insulator voltages under backflashover conditions using FDTD,
Moment of methods and circuit theory [95] [96] [97] [98] [101], has shown however that neglecting the
apparent ground plane surge response may result in significant errors in estimating insulator voltages. This
is especially true in cases where the voltage across the tower is a significant contributor to the tower top
voltage – as is the case for lines with low tower footing resistances.
It is possible to model the initial surge response of a perfectly grounded tower as an inductance at the
tower base. Its value is a function of tower height and can be estimated as follows [95]:
𝑡𝑓
𝐿𝐺𝑃 ≈ 60 ∙ 𝑡𝑡 𝑙𝑛 ( ) [H] Equation 5-10
𝑡𝑡

Where:

tt : the travel time of the tower [s]


tf : the front time of the lightning current [s]

Detailed electro-magnetic analyses [96][97][98] confirm this effect but the cause was identified as a
dependence of the tower surge impedances, ZTa…ZTd in Figure 5-8, on the direction of the travelling wave.
For example, a conical tower has a constant surge impedance for waves travelling down the tower, and
highly variable surge impedance for waves travelling up the tower, as defined by Equation 5-3 with height
measured up from ground level. Circuit based models cannot support this formulation and should rely on
the approximation introduced above (Equation 5-10).

5.3.7 Multistorey tower model


The multistorey model was developed to better match measurements of the lightning surge response of a
UHV tower [100]. It is an extension of the variable-impedance transmission line model discussed in section
5.3.5, with the addition of parallel connected R-L circuits between each tower section originally to
represent effects of crossarms. The equivalent tower model is shown in Figure 5-9. The surge impedances
of the individual sections are calculated according to the guidance given in section 5.3.5.

47
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

Figure 5-9 – The multistorey tower surge impedance model.


The series resistors (Ra …Ri… Rd) can be selected to attenuate the negative voltage reflection from the
tower footing [101] and to include the ground plane surge response. They are calculated as follows [102]:

𝑅𝑖 = ∆𝑅𝑖 ∙ 𝑙𝑖

2 ∙ 𝑍𝑇𝑑 1
For the tower base: ∆𝑅𝑑 = 𝑙𝑛 ( )
ℎ𝑑 𝛼𝑑 [ohm] Equation 5-11
2 ∙ 𝑍𝑇𝑖 1
For the upper sections: ∆𝑅𝑖 = 𝑙𝑛 ( )
(ℎ𝑎 − ℎ𝑑 ) 𝛼𝑖

Where:
i identifies the tower section from “a” to “d”
li the length of tower section “i”
ZTi the surge impedance of tower section “i”
α the attenuation coefficient, which is set to 0.89 for all tower sections
c the speed of light – 3 x 108 m/s
Other variables are defined in Figure 5-9.

The inductors in parallel with the attenuation resistors are added to damp the effect of the resistors after
the initial current transient [101] and are selected so that the time constant of the R-L branch is about two
times the travel time of the tower [102].
ℎ𝑎
𝐿𝑖 = 2 ∙ 𝜏 ∙ 𝑅𝑖 with 𝜏 = [H] Equation 5-12
𝑐

This multistorey model has been verified against field measurements and electromagnetic computation
methods as a good alternative to predict insulator voltages over a wide range of tower footing resistances
[101].
5.3.8 Representation of the tower surge impedance in TL lightning-performance studies
The use of more sophisticated tower models, such as those in sections 5.3.5 to 5.3.7, becomes more
important in situations where it may be expected that the voltage build up along the tower will be a
significant contributor to the insulator voltages. Equation 5-1 provides a simple insight into the problem.
Detailed models should be considered when:

(𝐿 𝑇 + 𝐿𝐺𝑃 ) ∙ 𝑆𝑚 ≥ 𝑅𝐸 [H] Equation 5-13

Using:

48
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

LT the equivalent inductance of the tower, which can be estimated as:


LT of 1.0 H per meter of tower height for single pole structures (constant ZT = 300 ).
LT of 0.5 H per meter of tower height for lattice structures (constant ZT = 150 ).
LGP calculated from Equation 5-10
RE resistance of the earth electrode at the base of the tower
Sm Maximum steepness (rate of rise) of the imposed lightning stroke current

5.4 Earth electrode impulse response


5.4.1 Introduction
In brochure 63 the impulse response of the earth electrode was primarily discussed in terms of its
ionization behaviour [6]. In this section the focus will be more on its high-frequency behaviour with special
reference to a recently published CIGRE brochure on this matter [23]. Furthermore, as a simplification only
uniform soil is considered, as the “apparent resistivity”, derived from low-frequency resistance
measurements on earth electrodes, may be used to represent stratified soils [23].
The electrical response of an earth electrode can be explained by the simplified circuit shown in Figure
5-10 [106] [103]. This circuit represents a buried horizontal conductor, also known as a counterpoise.

Figure 5-10 - A Simplified circuit representing ground electrodes as a series of connected distributed circuits
(RL, LL, are the per-unit-length longitudinal Resistance and Inductance, RE, CE are the shunt Resistance and
Capacitance of the electrode).
The electrode is modelled as a series connection of a number of T-elements. Here, the electrode wire
impedance is represented by RL (internal resistance of the wire) and LL (Inductance of the wire) while RE
and CE represent the dissipation resistance and capacitance, respectively. RE is shown as a non-linear
resistor, indicating that it may change due to soil ionization when high currents are discharged into the soil.
Expressions for RE, LL and CE in uniform soil are provided in Table 5-2 [112].
Table 5-2 - Expressions for RE, LL and CE of earth electrodes.
Conductor orientation RE [Ωm] LL [H/m] CE [F/m] With
Vertical 𝜌 𝜇0 2𝜋𝜀0 𝜀𝑟 4𝑙
𝑅𝐸 = 𝐴 𝐿𝐿 = 𝐴 𝐶𝐸 = 𝐴1 = 𝑙𝑛 ( ) − 1
2𝜋 1 2𝜋 1 𝐴1 𝑟
Horizontal 𝜌 𝜇0 𝜋𝜀0 𝜀𝑟 2𝑙
𝑅𝐸 = 𝐴2 𝐿𝐿 = 𝐴 𝐶𝐸 = 𝐴2 = 𝑙𝑛 ( )−1
𝜋 2𝜋 2 𝐴2 √2𝑟𝑑
In the equations above:
l the conductor length [m]
r the conductor radius [m] (r << l)
d the burial depth [m] (d << l)
ρ the uniform soil resistivity [Ωm]
μ0 the permeability of free space (4𝜋 × 10−7 H/m)
ε0 the permittivity of free space (8.85 × 10−12 F/m)
εr the permittivity of the soil (typically ≈ 10 - 20)

The total impedance of the equivalent circuit in Figure 5-10 will be frequency dependent due to the
presence of the inductive and capacitive elements. The capacitive elements will tend to lower the electrode
impedance as the frequency goes up and at the same time the inductive elements will tend to increase the
overall impedance. This frequency dependency is illustrated in Figure 5-11 which shows the overall
impedance Z(ω) of a counterpoise wire as a function of frequency. Four cases are shown including two
lengths of counterpoise and two levels of soil resistivity. The impedance Z(ω) curves were calculated with
an electromagnetic model, HEM [109].

49
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

ρ0 = 300 Ωm

(a) l =10 m (b) l =30 m.


ρ0 = 1000 Ωm

(c) l =10 m (d) l =30 m


Figure 5-11 – Frequency dependence of the earthing impedance of a single horizontal electrode (Burial depth
0.6 m) [109].

5.4.2 Low frequency response


The impedance plots in Figure 5-11 show for frequencies below about 1 kHz, that all four electrodes
appear as a pure resistance, as is evident from the zero-degree phase angle. This is because at low
frequencies, the wire resistance (RL), and reactive impedance (ωLL) is small compared to the dissipation
resistance (RE) and may therefore be neglected. The electrode capacitance may also be neglected due to
the large value of the capacitive impedance (1/ωCE) at this low frequency [111]. Essentially this means that
the whole electrode is at the same potential and the overall impedance is, therefore, given by the
dissipation resistance, RE.

For typical transmission line earth electrodes, including vertical and horizontal buried wires, the low-
frequency dissipation resistance, RLF, can adequately be estimated from [104]:

𝜌 1 11.8 ∙ 𝑔2 1 𝐴
𝑅𝐿𝐹 = [ 𝑙𝑛 ( )+ 𝑙𝑛 ( )] [Ω] Equation 5-14
2𝜋 𝑔 𝐴 𝐿𝑤 𝐹 ∙ 𝐴𝑤

Where:

ρ the soil resistivity, assuming uniform soil [Ωm].


g the geometric radius of the electrode, which is equal to √𝑟𝑥2 + 𝑟𝑦2 + 𝑟𝑧2 [m]
A the is the surface area of electrode volume in contact with earth [m2]
Lw the total length of the wires making up the electrode [m]
Aw the wire surface area. When all wires in the electrode have the same wire radius rw:
Aw is equal to 2𝜋𝑟𝑤 𝐿𝑤 [m2]
F a constant, F = 8/n2 for n radial wires meeting at a point and otherwise F = 2.
Equation 5-14 is valid for A/(FAw) > 1.

Equation 5-14 shows that there is a linear relationship between the electrode resistance and the soil
resistivity. This is also reflected in Figure 5-11, which shows that increase in soil resistivity of 3.3 times
from 300 Ωm to 1000 Ωm results in the same 3.3 times increase of the electrode resistance. The ratio of
resistivity to RLF defines the effective perimeter, P = /RLF of a hemispheric electrode with the same RLF.

Further inspection of Equation 5-14 shows that there are two terms within the square brackets. The first
term, named “Geometric” resistance, is the resistance of an equivalent solid electrode of the same

50
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

dimensions and the second, named the “Contact” resistance, corrects for the fact that the electrode is not
a solid but is comprised of wires. The importance of the electrode size and conductor length is also
illustrated in Figure 5-11 which shows that RLF of the 30 m electrode is about 2.5 times lower than that of
the 10 m one. The effective perimeter, P, of the 30 m electrode is about 17 m and P for the 10-m horizontal
buried wire is 7 m.
5.4.3 High frequency response
5.4.3.1 General observations
Figure 5-11 shows that the electrode impedance starts to deviate from the low-frequency resistance at
frequencies above 1 to 10 kHz. At these higher frequencies, the reactive (ωLL) and capacitive (1/ωCE)
impedance of the buried wires become more prominent, and this results in a significant voltage drop along
the earth conductor. Since the whole electrode is no longer at the same voltage – as was the case at low
frequencies – its impedance can no longer be approximated by RE ≈ RLF. A further contribution to the
change in impedance is the frequency dependence of the soil parameters, which results in a reduction in
the dissipation resistance [23].
Other observations that may be made from the results in Figure 5-11 include:

 For all electrodes there is a tendency at increasing frequencies for the impedance to first become
capacitive (negative phase angle) before becoming inductive (positive phase angle).
 In the MHz range, the impedance of the four example electrodes, converge into a small range of
60-80 Ω which corresponds roughly to the ground plane surge impedance.
 For frequencies associated with lightning currents, i.e. 100 kHz and higher, the earth electrode can
either be capacitive or inductive, depending on the electrode size and soil characteristics.
 In high resistivity soils the capacitive effects tend to be stronger which results in a larger reduction
of the impedance as the frequency increases. This can be explained in terms of the longer RECE
time constant.
 Both the electrode length and the soil resistivity influence the frequency at which the transition
between the low- and high-frequency impedance values occur.
 The resistivity of soil tends to decrease with an increase in frequency. This is apparent in the
graphs shown in Figure 5-11 as the reduction in impedance from low frequency up to the transition
frequency.
In the time domain, the counterpoise wire can be considered as a non-uniform and lossy transmission line.
A current pulse impressed onto one end of the wire will propagate along it, and while propagating, will be
attenuated and distorted due to the current dissipation into the surrounding soil. This attenuation is
greatest in low resistivity soil and it also becomes greater at higher frequencies because of the frequency
dependence of the soil resistivity [113]. This behaviour manifests itself as a change in the apparent
electrode impedance over time. For a steeply rising impulse, the initial impedance will be the surge
impedance of the electrode which then changes over a few micro-seconds to the low frequency impedance
[105]. This transition is usually complete after a few round-trip travel times to the far end of the conductor
[112]. For long buried horizontal wires, the initial impedance is usually higher than the low frequency
impedance, but in very high resistivity soil (e.g. ρ >1000 Ωm) the reverse may be true.

5.4.4 Frequency dependence of soil parameters


There is a substantial body of empirical evidence in support of the frequency dependence of soil
parameters [116] which has been reported on in detail in CIGRE brochure 781 [23]. It is now known that
both, the soil resistivity and permittivity, ε, decrease with increasing frequency. The importance of this
effect when considering the high frequency response of the ground electrode is demonstrated in Figure
5-12. It shows a comparison of the measured and simulated ground potential rise of an earth electrode
(GPR) in response to an injected impulse current (solid black line). Two simulation results are shown. The
first simulation, which was based on constant soil parameters (red curve), considerably overestimates the
GPR whereas the simulation based on frequency dependent soil parameters (dashed blue line) accurately
replicates the measured GPR (green line).

51
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

Figure 5-12 - Measured and simulated earth potential rise of a 9.6 m long electrode buried 0.5m deep in 1400
Ωm low-frequency resistivity soil due to the impression of lightning-patterned current with front time of 0.4 μs.
Adapted from [23].
The following causal expressions, which are based on many field measurements, are proposed to describe
the resistivity and relative permittivity of soils, in the range of 100-Hz to 4-MHz [23].
𝜌0
𝜌= [Ωm] Equation 5-15
1 + (4.7 × 10−6 ∙ 𝜌00.73 ) ∙ (𝑓)0.54

𝜀𝑟 = 9.5 × 104 ∙ 𝜌0−0.27 𝑓 −0.46 + 1.2 Equation 5-16

In these so-called Alipio-Visacro expressions,


ρ the soil resistivity at frequency f in Hz
εr the relative permittivity at frequency f in Hz
ρ0 the apparent soil resistivity [Ωm] at 100 Hz
5.4.4.1 The impulse impedance
For lightning performance calculations on transmission lines, it is important to know the earth electrode
impedance at the time of the peak lightning current as it is when the insulators will be subjected to the
highest voltage. This voltage peak governs the risk for flashover, even if it occurs at a later time. The
impedance at this time is defined as the impulse impedance ZP, which is given by the ratio of peaks of the
applied impulse current and resulting impulse voltage (ZP = VP/IP). The impulse impedance falls
somewhere between the surge impedance and low frequency resistance of the electrode. Depending on
the soil resistivity, Zp of extended electrodes can therefore be either larger (in the case of low resistivity
soil) or smaller (in case of high resistivity soil) than the low-frequency resistance. This differs from the
common, and erroneous, assumption that Zp is always larger than RLF.

A first estimate of Zp for a four-legged counterpoise as shown in Figure 5-13 may be obtained from [110]:

𝑍𝑃 = 0.16 ∙ 𝜌0 ∙ 𝑙 −0.687 [Ω] and for 100 ≤ ρ0 ≤ 600 Ωm


Equation 5-17
𝑍𝑃 = 0.4 ∙ 𝜌00.89 ∙ 𝑙 −0.75 [Ω] and for 600 ≤ ρ0 ≤ 4000 Ωm

Where:
ρ0 the low frequency soil resistivity, assuming uniform soil [Ωm].
l the average length of the counterpoise arms [m]

Figure 5-13 – A typical four-legged counterpoise defining the parameters of Equation 5-17 [110].

52
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

For other arrangements of counterpoise wires a first estimate of Zp can be obtained by applying a factor of
4/n to the result from Equation 5-17, where “n” is the number of counterpoise wires in the arrangement and
less than 4. Alternatively, a more accurate estimate may be obtained from the impulse coefficient IC
(defined in section 5.4.4.3 below) and the low-frequency grounding resistance RLF as follows:

𝑍𝑝 = 𝐼𝐶 × 𝑅𝐿𝐹 Equation 5-18

Where:
IC The impulse coefficient - given by Equation 5-21 below.
RLF The low frequency resistance of the electrode which could either be measured or calculated.
The expressions above are valid for electrode lengths shorter than or equal to the effective length (see
next section).
5.4.4.2 The effective length of counterpoise electrodes
It has long been recognized that counterpoise wires longer than 60 – 100 m become less effective when
considering high-frequency effects such as lightning [105]. This is primarily because of the dissipation of
the surge as it propagates along the wire, which means that the extremities of the electrode, where most of
the surge is already dissipated, have negligible current density and contribute therefore little to the rise of
the ground potential at the tower. It is therefore desirable, when designing earth electrodes, to know what
the maximum length the counterpose should be to achieve a cost-effective electrode with a good lightning
performance. This limiting length is defined as the effective length, lEF.

The effective length is defined in terms of the impulse impedance ZP as shown in Figure 5-14. In this
example the impulse impedance is given by the peak of the voltage wave divided by the applied current
peak of 11.8 kA. The calculated impulse impedance for this series of tests is shown on the right.
The results in Figure 5-14 show that ZP decrease for an increase in counterpoise length up to about 15 m
after which it remains constant. This means that the part of the counterpoise further away than 15 m does
not contribute to reduce the peak electrode voltage at the tower. For this example, the effective length (lEF)
of the counterpoise is therefore 15 m.
160 12
L=5m Zp
140 L = 10 m
10
L = 20 m
Impulse Impedance [Ω]

120
L = 40 m
8
100
Voltage [kV]

80 6

60
4
40
2
20
lEF
0 0
0 2 4 6 8 10 0 10 20 30 40 50
Time [μs] Length of Counterpoise [m]

Voltage rise at tower Corresponding impulse impedance


Figure 5-14 – Examples of the voltage rise at the tower and corresponding impulse impedances for various
lengths of counterpoise. IP = 11.8 kA, and a soil resistivity of 300 Ωm. Adapted from [114].
In more general terms, the value of lEF depends on several factors. It decreases with decreasing soil
resistivity (greater attenuation in low resistivity soils). It also decreases for impulses with a shorter front-
time since they represent higher frequencies (greater attenuation at higher frequencies) [113]. The
following equations are proposed for providing a first estimate of lEF as a function of the low-frequency soil
resistivity ρ0 [Ωm] [110]:

𝑙𝐸𝐹 = 17 + 0.042 ∙ 𝜌0 − 2 × 10−6 ∙ 𝜌02 [m] and for first strokes


Equation 5-19
𝑙𝐸𝐹 = 9 + 0.021 ∙ 𝜌0 − 1 × 10−6 ∙ 𝜌02 [m] and for subsequent strokes

53
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

5.4.4.3 The impulse coefficient


The ratio of the impulse impedance to the low-frequency resistance is known as the Impulse Coefficient IC
and is defined as:

𝑍𝑝
𝐼𝐶 = Equation 5-20
𝑅𝐿𝐹

IC is a useful parameter to determine ZP from a measured low frequency resistance RLF [115]. A typical
profile of IC as a function of electrode length is shown schematically in Figure 5-15.

Figure 5-15 - Typical profile of the impulse coefficient IC as a function of the electrode length.
It is generally found that the impulse coefficient is a constant value if the electrode is shorter than the
effective length. This is because both Zp and RLF have approximately the same dependency on the length
of the electrode. Considering the frequency dependence of the soil resistivity, it is found that Zp is usually
smaller than RLF. For electrode lengths less that lEF, IC may vary from 0.4 in high resistivity soils to 1.0 in
low resistivity soils. For electrodes longer than lEF, Zp remains constant while RLF further reduces with
increased electrode length. Consequently, IC shows an increasing trend as shown in Figure 5-15. Further
details may be found in [111].
The following expression is proposed for obtaining a first estimate of the impulse coefficient ( IC) for
counterpoise electrodes shorter than lEF as a function of the low-frequency soil resistivity ρ0 [110]:

𝐼𝐶 = 0.89 − 5 × 10−5 ∙ 𝜌0 Equation 5-21

Equation 5-21 shows that IC is only governed by the soil parameters. Thus it also provides a valid
estimate of IC for other tower footing electrode arrangements, provided that the extent of the electrode
does not exceed lEF.

5.4.5 Soil ionization


The soil around small electrodes may break down electrically, or ionize, when the electric field in the soil
exceeds its critical breakdown field. This may happen when the electrode conducts high current, such as
during a lightning strike, to ground. The critical breakdown E-field is dependent on many parameters, but
usually falls in the range of 300 to 18,500 kV/m [95] [121]. Values of between 300 [120] and 400 kV/m [6]
are usually assumed for simulation studies.
Soil ionization develops in the form of several highly conducting breakdown channels that extends from the
electrode surface into the surrounding soil along the voids between the soil particles. This envelops the
electrode with an ionization zone which has a low soil resistivity. The ionization zone essentially enlarges
the earth electrode geometric radius and surface area which results in a reduction of the earth electrode
resistance [95] [108]. The ionization zone around the earth electrode changes shape with increasing
current [95], to become a hemi-sphere at very high current. Since it takes time for the soil to ionize, there
are significant delays in both the ionization and deionization processes [107]. This means that the
maximum extent of the ionization, and associated minimum value of electrode resistance, occur after the
peak injected current.
This subject is further explored in brochure 63 [6] which includes the introduction of the Weck simplified
approach, which is still widely used. It provides an estimate for the electrode resistance when ionized
based on the injected current, I, the low frequency resistance, RLF, and the soil resistivity, ρ:

54
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

𝑅𝐿𝐹
𝑅𝐼 = 𝐼
for I ≥ IPI
√1+𝐼 [Ω] Equation 5-22
𝑃𝐼

𝜌𝐸0
𝐼𝑃𝐼 = 2 [kA] Equation 5-23
2𝜋𝑅𝐿𝐹

Where:

ρ the soil resistivity, assuming uniform soil [Ωm].


RI the ionized resistance of the electrode [Ω]
RLF the low frequency resistance [Ω]
I the injected current [kA]
IPI the peak current required for the inception of ionization process [kA]
E0 the critical inception electric field for ionization which is usually assumed to be 400 kV/m.

Utilising the equations above it can be estimated that the maximum extent of the soil ionization in high
resistivity soil will rarely exceed 10 m [99]. The effect of ionization on backflashover rate using Equation
5-22 and Equation 5-23 is strong for an effective electrode perimeter /RLF = 10 m, corresponding to a
hemisphere with 1.6 m radius, but negligible for /RLF = 40 m [7].

The general expression for RLF (Equation 5-14) of wire-frame electrodes, such as a buried horizontal wire
approximating a vertical metal plate, can be is utilized to obtain a second estimate of the electrode
resistance when partly or fully ionized. An assumption is made that rw and Aw in Equation 5-14 will
increase as the electrode ionizes, to the extent that the contact resistance term can be neglected. Thus,
the estimate of ionized resistance is based only on the geometric resistance term in Equation 5-14:

𝜌 1 11.8 ∙ 𝑔2
𝑅𝐼 = [ 𝑙𝑛 ( )] [Ω] Equation 5-24
2𝜋 𝑔 𝐴

Where:

ρ the soil resistivity, assuming uniform soil [Ωm].


g the geometric radius of the electrode, which is equal to √𝑟𝑥2 + 𝑟𝑦2 + 𝑟𝑧2 [m]
A the is the surface area of electrode volume in contact with soil [m2]

Equation 5-24 generally overestimates the effect of the soil ionization for large sized earth electrodes
typically found on transmission lines compared to detailed numerical models found in [24] that have a
different value of rw for each segment of a buried wire exhibiting ionization.

Neither of these simplified models include the time delays measured in of the ionization and de-ionization
process, leading to overestimation of the effect of soil ionization at the peak of the lightning current and
thus an underestimation of the peak insulator voltages.

5.4.6 Simplified representation of tower-footing electrodes in TL lightning-performance


studies
5.4.6.1 Basis for this approach
The best way to accurately represent the response of an earth electrode to lightning current is to develop
an advanced model which is based on a physical representation of all conductors in the electrode.
However, these models are generally more complex and computationally intensive, which hinders their
widespread application. A simplified representation that provides a reasonable accuracy is therefore
preferred.
In this respect it was found through extensive studies [114], that the earth electrode may be represented
by their first-stroke impulse impedance ZP without a great loss in accuracy as is demonstrated in Figure
5-16. It shows that practically the same outage rates are predicted for the simplified approach, based on
the impulse impedance, and an advanced model, which was based on a physical representation of the

55
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

earth electrode. Figure 5-16 also includes a case where the earth electrode was represented by its low-
frequency resistance, RLF. The results from this case deviated, however, strongly from that obtained with
the physical representation of the electrode.
Similar results were obtained for other typical transmission lines [114]. These studies covered lines with a
range of operating voltages and considered different electrode arrangements, electrode lengths, as well as
a wide range of soil-resistivities.
Another advantage of using the impulse impedance is that it is practically invariant in the range of front-
times associated with first return strokes [23], and therefore relatively easy to determine.

Figure 5-16 - Outage rates of a 138-kV transmission line obtained under ZP, RLF and physical representations
of electrodes, using an electromagnetic model and the DE method (flashover criterion). Soils of 600 and 2000
Ωm. Electrodes: counterpoise wires of length L. Towers: 30 m high and single shielding wire. N g of 2.4
flashes/km2/year [114].

5.4.6.2 Guidelines for implementing the simplified approach


Based on the considerations above, using the first-stroke impulse impedance ZP is recommended in the
studies of the lightning performance of transmission lines. This parameter can be determined in different
ways:

 ZP can be measured by using earthing impedance meters which utilize an impulse excitation of the
ground electrode [118] [119], though some of these measurements are still somewhat complex
and requires that specific procedures be followed to prevent surge impedance coupling effects.
 Determining ZP by using analytical expressions, as those provided earlier in this section, for ZP for
counterpoise wires (Equation 5-17), as they take the frequency dependence of soil parameters
into account.
 Determining ZP directly from the product RLF × IC (Impulse coefficient measured or calculated from
the low frequency resistance RLF) because on real transmission lines, the earth electrodes are in
most cases shorter than the effective length. This method can be extended to longer electrodes,
by using the RLF value determined for an electrode with the length equal to lEF.
The first-stroke impulse impedance, determined according to the guidance above, does not account for soil
ionization. Several authors have recommended that ionization be ignored when evaluating the lightning
performance of transmission lines, claiming that it has a limited effect on the backflashover performance
because the effective perimeter of foundations is large. The argument rests on the fact that the extent of
soil ionization is dependent on the size of the earth electrode and that it has the greatest effect on
concentrated (short) electrodes. Such short electrodes with an effective perimeter less than 10 m are
typically only utilized in low resistivity soil where it is easy to achieve electrode impedances of below 10 Ω.
The backflashover of such a line will already be acceptable and a further reduction in the electrode
impedance, due to ionization, would only result in a modest further reduction in the backflashover rate.
Conversely, in high resistivity soils, an acceptable backflashover rate can only be achieved with large
electrodes, which may extend to about 50 m. The effect of ionization on these electrodes will be limited, if
present at all, because the current density and associated electric field may not be high enough to ionize
the soil. Thus, also in this case soil ionization may be ignored without significantly affecting the calculated
backflashover rate.
It is, however, possible to include the favourable effect of soil ionization in the simplified approach
proposed above. As a first approximation, it is suggested to replace RLF in Weck’s simplified approach with

56
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

the impulse impedance ZP. It is further suggested that a higher value for critical breakdown E-field is used
in high resistivity soil, for instance estimated E0 = 1000 kV/ m [95] or preferably measured on local soil
samples, to estimate the true effect of soil ionization in these conditions.

5.5 Insulator flashover modelling


5.5.1 Lightning impulse strength of line insulation
Overvoltages generated by lightning flashes to transmission lines are characterized by very short durations
and fast front times. A typical overvoltage, as shown in Figure 5-17 has a front time (T1) ranging between
0.1 μs and 20 μs and an overall duration – as defined by the time to half value, (T2) – of less than 300 μs
[122]. For testing purposes, lightning overvoltages are represented by the standard lightning impulse,
which is a double exponential wave, with a front time of 1.2 μs and a time to half value of 50 μs.

Lightning surge on transmission line Laboratory generated Lightning Impulse


Figure 5-17 – Fast-front transient overvoltage and laboratory test wave.
Guidelines for estimating the lightning impulse strength of transmission lines are provided in CIGRE
brochure 72 [15]. These are based on the so-called “gap factor” method whereby the insulation strength is
determined from that of a rod-plane gap with the same insulation length, and by applying a suitable
correction factor, or gap factor, to adjust for the differences in the insulation configuration. The gap factors
are the same as those determined from the switching impulse strength of line configurations. According to
this method, the flashover strength of the line insulation at sea level may be estimated as follows [7] [15]:
 50 % flashover strength of a rod-plane gap (U50_rp) of length d [m]:
Positive polarity:
+
𝑈50_𝑟𝑝 = 525 ∙ 𝑑 [kV] and for 1 m ≤ d ≤ 8 m Equation 5-25

Negative polarity:

𝑈50_𝑟𝑝 = 950 ∙ 𝑑0.8 [kV] and for 1 m ≤ d ≤ 8 m Equation 5-26

 The gap factor (kg) is applied as follows to estimate the lightning impulse strength of air gaps:
Positive polarity:
+ +
𝑈50 = (0.75 + 0.25𝑘𝑔 ) ∙ 𝑈50_rp [kV] Equation 5-27

Negative polarity for gaps less than 7 m:


− −
𝑈50 = 𝑈50_rp (1.51 − 0.51𝑘𝑔 ) [kV] and for 𝑘𝑔 ≤ 1.44
Equation 5-28
− − (0.776)
𝑈50 = 𝑈50_rp [kV] and for 𝑘𝑔 > 1.44

Typical gap factors applicable to transmission lines are:


 Conductor – crossarm: kg ≈ 1.45.
 Conductor – window: kg ≈ 1.25.
The presence of insulators may influence the breakdown strength of the gap and this is strongly dependent
on the insulator type (i.e. disc, longrod or polymer), because of differences in the capacitance between
units, and differences in the dry arc distance of insulators with the same axial length [15]. No general rules
can be given but a conservative estimate may be obtained as follows [7]:

57
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

Positive polarity:
+
𝑈50 = 560 ∙ 𝑑 [kV] Equation 5-29

Negative polarity:

𝑈50 = 605 ∙ d [kV] Equation 5-30

for an insulator length of d [m]

These estimates are valid for conventional transmission lines not utilizing wood or insulating materials as
part of the supporting structure.
The insulation strength of the insulation of lines at altitude may be estimated by the application of the
altitude correction factor as defined in IEC 60071-2[123]:

𝑈50
𝑈50_𝑎𝑙𝑡𝑖𝑡𝑢𝑑𝑒 = [kV]
𝐾𝑎
Equation 5-31
𝐻
𝑚×
𝐾𝑎 = 𝑒 8150 [p.u.]

Where:
Ka the altitude correction factor
H the height above sea level [m]
U50 the flashover strength at sea level
U50_altitude is the flashover at an altitude of H
m a constant and equal to 1.0 for lightning impulse voltages

When calculating outage rates, it is recognized that the statistical dispersion of the overvoltages generated
by lightning to transmission lines is much larger than the standard deviation of the flashover probability of
the line insulation. The latter may therefore be neglected when evaluating the flashover performance of the
line [123].

5.5.2 Lightning surge strength of line insulation under non-standard conditions


In the previous section, the strength of air gaps in transmission line towers was defined in terms of the
standard 1.2/50-s double exponential lightning impulse. The shape of this impulse is, however, very
different from that which may be expected across the insulation during a lightning stroke of which
calculated examples are shown in Figure 5-18 [139].

Figure 5-18 – Calculated insulator voltages on lines with different rated voltages [139].
In brochure 63 [6] it is suggested that the duration of the wave tail may be estimated from the line
parameters as follows:

58
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

𝑍𝑔 ∙ 𝑙𝑠𝑝
t ℎ = 0.693 [µs] Equation 5-32
𝑅𝐸 ∙ 𝑐

Where:
th the tail time [μs]
Zg the shield wire surge impedance [Ω]
RE the structure footing impedance [Ω]
lsp the length of the span, [m],
c the speed of light equal to 300 m/µs

For lines with a single ground wire th typically varies between 5 and 20 µs, and for lines with double shield
wires values are in the range from 2 to 10 µs. These estimates agree well with results from detailed
backflashover studies, such as those presented in Figure 5-18 [139]. It shows the voltage across insulators
of transmission lines with different rated voltages and line configurations. The tail times determined from
these and other studies ranged from 2 to 10 μs [134] [135] [136].
Experimental studies have shown that the shape of the applied impulse significantly influences the
flashover strength of the insulation. Specifically, a shorter time to half value (i.e. less than 50 μs) results in
an increase in the flashover voltage [15] [15] and an upper bound on this increase is the flashover voltage
of a 1/5 μs impulse [134]. It is, therefore, necessary to adjust the U50 values determined through
standardized testing, for the different shapes of the lightning surge when evaluating flashover of the line
insulation.
A second aspect that needs to be accounted for is the time it takes for the flashover to develop across the
insulation. It is dependent on the applied voltage stress, with shorter times to flashover occurring when
higher voltages are applied. This characteristic is known as a volt-time curve (or V-T curve). Examples of
volt-time curves derived from experimental data on four common gap types are presented in Figure 5-19
[15].
2 2
Negative Upper Limit Negative Upper Limit
Peak of the applied voltage [p.u. of U50]
Peak of the applied voltage [p.u. of U50]

1.8 Negative Lower Limit 1.8 Negative Lower Limit


Positive Upper Limit Positive Upper Limit
1.6 Positive Lower Limit 1.6 Positive Lower Limit

1.4 1.4

1.2 1.2

1 1

Rod-Plane Air Gap Rod-Rod Air Gap


0.8 0.8
0 2 4 6 8 10 12 0 2 4 6 8 10 12
2 Time to flashover: tbr [μs] 2 Time to flashover: tbr [μs]
Negative Upper Limit Negative Upper Limit
Peak of the applied voltage [p.u. of U50]

Peak of the applied voltage [p.u. of U50]

1.8 Negative Lower Limit 1.8 Negative Lower Limit


Positive Upper Limit Positive Upper Limit
1.6 Positive Lower Limit 1.6 Positive Lower Limit

1.4 1.4

1.2 1.2

1 1

Longrod Insulator String Disc Insulator String


0.8 0.8
0 2 4 6 8 10 12 0 2 4 6 8 10 12
Time to flashover: tbr [μs] Time to flashover: tbr [μs]

Figure 5-19 –Voltage – time to breakdown (V-T) curves for standard lightning impulses and gap lengths
between 1 m and 6 m. Adapted from [15].

59
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

The timing of the insulation flashover is important when considering remedial measures to improve the
lightning performance of the line. Most of these measures are applied some distance away from the
insulation (for example, the ground electrode is typically at least 20 m away), which means that there is a
time delay before it can influence the voltage across the insulation. This could mean that the voltage
mitigation effect arrives too late to prevent flashover.
The strength of line insulation under non-standard conditions have been extensively studied [125] [126]
[127] [128] [129] [130] [131] [132] [139] with an aim of developing predictive methods which can be
programmed into simulation programs, such as the various EMT implementations. An overview of these is
given in this section.

5.5.2.1 Voltage-Time curve penetration


The simplest model available is to compare the voltage across the line insulation with the V-T curve.
Flashover is deemed to occur if the insulator voltage exceeds the V-T curve at any time in the simulation.
The volt-time characteristics are obtained by testing air gaps or insulators. It may be described with the
following relationship [10]
𝑏
𝑈𝑏𝑟 (𝑡𝑏𝑟 ) = (𝐴𝑡𝑏𝑟 + 𝑈0 ) ∙ 𝑑 [kV] Equation 5-33

Where:
Ubr the peak of the applied voltage [kV]
tbr the time at which break down of the gap occurs [μs]
d the gap length [m]
U0 the voltage below which no breakdown is possible [kV]
A, b constants determined experimentally

An often-used equation for the voltage-time curve is [132] [2] [125]:


−0.75
𝑈𝑏𝑟 (𝑡𝑏𝑟 ) = (710 𝑡𝑏𝑟 + 400)𝑑 [kV] Equation 5-34

Equation 5-34 can, however, not be applied to all insulation configurations because it assumes a fixed
flashover gradient for all configurations. A more generally applicable formulation may be obtained by
normalizing Equation 5-33 to the determined U50, which allows it to be scaled to reflect the actual flashover
strength of the line insulation.
From lightning tests on transmission line configurations, it has been established that the breakdown
voltage at 2 μs is about 1.4 to 1.73 times U50, and at 3 μs this ratio falls between 1.24 to 1.45 [7]. Using
these relationships, Equation 5-34 may be normalized to the 50 % flashover voltage as follows:

𝑈𝑏𝑟 −0.75
(𝑡 ) = 1.455 𝑡𝑏𝑟 + 0.82 for tbr < 16 s Equation 5-35
𝑈50 𝑏𝑟

Hileman [7] proposed a different version of this equation (also shown in Figure 5-19):

𝑈𝑏𝑟 −0.5
(𝑡 ) = 1.39 𝑡𝑏𝑟 + 0.58 for tbr < 11 s Equation 5-36
𝑈50 𝑏𝑟

Voltage-Time curve penetration was used in the earlier simplified methods [2] [3] [4] . The critical flashover
voltage is evaluated at times to breakdown (tbr) of 2 and 6 μs.
The simple volt-time curve flashover model has significant deficiencies. The equations provided are
normalized only to the insulation length without a corresponding adjustment to the time to breakdown,
relying on the constants found in [132] for dry arc distance from 5.6 to 11.2 m. After subtracting a fixed
1 s time observed to form streamers [125], the volt-time curve in Equation 5-34 suggests that the average
leader velocity increases dramatically as average stress approaches 1000 kV/m. The V-T curve methods
also do not consider the variation of voltage before the intersection point, which is inconsistent with the
physical process of flashover development. Despite these problems, the V-T curve approach provides
convenient and accurate estimates of backflashover rates, even with hand calculations [2].

60
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

5.5.2.2 Integration methods


Another well-known method for considering the strength of insulation under non-standard conditions is the
integration model of Witzke and Bliss, which is illustrated schematically in Figure 5-20.

Figure 5-20 – Schematic representation of the integration method [15].


With this method, the voltage appearing across the insulator is integrated and break down occurs when the
accumulated integral value exceeds a certain threshold, DE. In its generalized form this method is
described as follows [134]:
𝑡𝑏𝑟 𝑛
𝑈(𝑡) − 𝑈0
𝐷𝐸 = ∫ [max ( , 0)] 𝑑𝑡 Equation 5-37
𝑡0 𝑈50

Where:
U(t) the voltage across the insulation [kV]
U0 an onset voltage across the insulation [kV]
U50 the 50 % flashover voltage across the insulation [kV]
n a constant
DE a function of U50, n, dielectric material, and other factors

The two coefficients are determined so that the integration model fits the V-T curve when considering a
standard lightning impulse. The coefficients have no physical meaning and may vary for different types of
insulators or gaps. A review of parameters Equation 5-37 may be found in references [124] and [134]. It is
endorsed that a preference should be given for selecting the lowest value of “n” that still results in an
adequate fit of the chosen V-T curve [134]. Integration method modelling using high values of “n” is
unreasonably sensitive to narrow, high peak voltages associated with tower surge impedance effects.

5.5.2.3 Physical models


Physical models consider the different phases of flashover development and their dependence on the
applied voltage to determine if and when the insulation flashes over [128] [129] [130] [131]. As shown in
Figure 5-21, the discharge development under a lightning impulse comprises several sequential physical
phases. These are corona inception, streamer propagation and finally, leader propagation. The model
sums the developing time of these physical phases to determine the time when the breakdown occurs
[129] [130] [131]:

𝑡𝑏𝑟 = 𝑡𝑖 + 𝑡𝑠 + 𝑡𝑙 Equation 5-38

Where:
ti time to corona inception,
ts the time necessary streamer development, and
tl the leader development time.

61
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

Figure 5-21 – Overview of the discharge development under a lightning impulse [15].

Figure 5-22 – A schematic representation of the evaluation of flashover with a physical model.
With reference to Figure 5-22, flashover is evaluated according to the following steps:

 The insulation configuration is modelled together with the source supplying the voltage across the
insulator. It is important to accurately model the source impedance as these physical models
include an estimate of the pre-discharge current which may influence flashover development in
cases of a weak source or if there is an intentional impedance (such as a surge arrester) in the
flashover circuit.
 The corona initiation time (ti) is normally neglected because of the fast rate of rise of lightning
impulses and because the corona inception voltage is typically far below the breakdown voltage.
 The time duration of the streamer phase (ts) is evaluated, but as a simplification it is often assumed
that the streamer phase is complete when the instantaneous voltage stress across the insulation

62
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

reaches a fixed value E0. As a rough approximation, E0 may be assumed to be equal to the 50 %
flashover stress – i.e. U50/d. The streamer current is normally neglected.
 Finally, the leader development time is evaluated based on the assumption that leader
development starts after the streamer spans the length of the gap. The length of the leader,
l = (d - x) is calculated by evaluating the speed of the leader growth, v, as a function of its length
and the voltage across the insulation, U(t). For this phase the pre-discharge current is also
calculated as a function of the leader velocity. Flashover occurs when the leader spans the
complete gap length, d. Leader progression stops when the instantaneous voltage stress in the
remaining gap falls below E0.
The model has found several different implementations of which a few are presented here:

Pigini et al [131]:

 Streamer development time:


−1
𝐸𝑝
𝑡𝑠 = [1.25 ∙ − 0.95] [μs] Equation 5-39
𝐸50

Where:
Ep the peak voltage stress across the insulation [kV per meter of insulation length]
E50 the flashover stress across the insulation [kV per meter of insulation length]
E50 = U50/d

 Leader velocity:
𝑑𝑙 𝑈(𝑡) 𝑈(𝑡)
= 170 ∙ 𝑑 ∙ 𝑒 (0.0015 𝑑 ) [ − 𝐸0 ] [m/s] Equation 5-40
𝑑𝑡 𝑥

Where:
E0 the inception voltage stress across the insulation [kV per meter of insulation length]
E0 is assumed to equal to E50
U(t) the applied voltage across the insulation [kV]

 Leader current:
𝑑𝑙
𝑖𝑙 = 𝑞 ∙ [A] Equation 5-41
𝑑𝑡

with q = 400 μC per meter of leader length

CIGRE [6]:

 Streamer development time:


The streamer is considered fully developed when:

𝑈(𝑡𝑠 ) = 𝐸0 ∙ 𝑑 [μs] Equation 5-42

with E0 determined from Table 5-3.

 Leader velocity:
𝑑𝑙 𝑈(𝑡)
= 𝑘1 ∙ 𝑈(𝑡) ∙ [ − 𝐸0 ] [m/s] Equation 5-43
𝑑𝑡 𝑥

with k1 and E0 determined from Table 5-3.

63
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

Table 5-3: Constants for the CIGRE flashover model.


Configuration Polarity k1 [m2/V2/s] E0 [kV/m]
−6
+ 0.8 × 10 600
Air gaps, post and long rod insulators
- 1.0 × 10−6 670
+ 1.2 × 10−6 520
Disc insulators
- 1.3 × 10−6 600

 Leader current:
𝑑𝑙
𝑖𝑙 = 𝑞 ∙ [A] Equation 5-44
𝑑𝑡

with q = 400 μC per meter of leader length.

Motoyama et al [137], Mozumi et al [138]:


 Streamer development time:
The streamer development is complete when:

1 𝑡𝑠
∫ 𝑈(𝑡) > 𝑘2 ∙ 𝑑 + 𝑘3 [μs] Equation 5-45
𝑡𝑠 𝑡=0

with k2 and k3 determined from Table 5-4.

 Leader velocity:
This model assumes leader growth from both gap electrodes therefore x = d - 2l
𝑑𝑙 𝑈(𝑡) 𝑑
= 𝑘4 ∙ [ − 𝐸0 ] [m/s] and for 0 ≤ 2𝑙 ≤
𝑑𝑡 𝑑 − 2𝑙 2
Equation 5-46
𝑑𝑙 𝑈(𝑡)
= 𝑘5 ∙ [ − 𝐸′] + 𝑣 ′ [m/s] and for
𝑑
≤ 2𝑙 ≤ 𝐷
𝑑𝑡 𝑑 − 2𝑙 2

with k4 and k5 determined from Table 5-4 and E’ and v’ determined as follows:

𝑈(𝑡)
𝐸′ =
𝑑 − 2𝑙
𝑑
When 2𝑙 = Equation 5-47
𝑑𝑙 2

𝑣 =
𝑑𝑡

 Leader current:
𝑑𝑙
𝑖𝑙 = 2 ∙ 𝑞 ∙ [A] Equation 5-48
𝑑𝑡

with q = 410 μC per meter of leader length.


Table 5-4: Constants for the Motoyama/ Mozumi flashover model.
Configuration Polarity k2 [kV/m] k3 [kV] k4 [m2/kV2/s] k5 [m2/kV2/s] E0 [kV/m]
+ 400 50 2500 420 750
Rod - Plane
- 460 150 2500 420 750

64
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

Wang et al [139]:
 Streamer development time:
The streamer development time is evaluated with Equation 5-45, but with k2 and k3 determined
from Table 5-5, which is based on experimental results.

 Leader velocity:
𝑑𝑙 𝑈(𝑡)
= 𝑘6 ∙ [ − 𝐸0 ] [m/s] Equation 5-49
𝑑𝑡 𝑥

with

𝑥 =𝑑−𝑙 For disc insulators


Equation 5-50
𝑥 = 𝑑 − 2𝑙 For polymer insulators
and k6 and E0 determined from Table 5-5.
Table 5-5: Constants for the Wang flashover model.
Configuration Polarity k2 [kV/m] k3 [kV] k6 [m2/kV2/s] E0 [kV/m]
+ 430 190 2900 580
Disc insulators
- 490 290 2500 640
+ 360 90 1500 620
Polymer insulators
- 500 140 1300 750

 Leader current:
𝑑𝑙
𝑖𝑙 = 𝑞 ∙ [A] Equation 5-51
𝑑𝑡

with q = 500 μC per meter of leader length.

Considerations for choosing a LP model:


The difference between the different LP models lies primarily in which part of the voltage-time curve they
reproduce accurately. This is illustrated in Figure 5-23 where the volt-time curves of a 1.86 m insulator
generated with several flashover models are compared[140].
6
Equation 5-33
Peak of the applied voltage [p.u. of U50]

Equation 5-34
5
CIGRE
Pigini
4 Motoyama / Mozumi
Shindo [E120]
3

0
0 2 4 6 8 10 12 14
Time to flashover: tbr [μs]

Figure 5-23 – A comparison of the V-T curves produced by various LP insulator flashover models for an
insulator string length of 1.86 m adapted from [140].

65
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

Figure 5-23 shows that the Motoyama and CIGRE models fits Equation 5-35 for times to breakdown in the
range 0.5 to 3 s. The CIGRE, Pigini and Shindo [130] models replicate the V-T curve from about 6 s.
Some experimentation may therefore be required to choose the model which best reproduce the voltage-
time curve in the range of times to breakdown expected for the line insulation. In practise, this could result
in utilizing models that accurately replicate relatively long times to flashover on lines with short squat
towers and long spans,
e.g. (LT + LGP) x Sm < RE, and those more accurate in reproducing short times to flashover on lines with
short spans and tall, slender towers, e.g. (LT + LGP) x Sm ≥ RE.
It was further shown to be important to [141] to include the predischarge current flowing in the gap during
the leader phase (il ) in the model as it generally slows down the leader development, thereby resulting in
higher value of the critical current (and therefore a better backflashover rate) than when the predischarge
current is not included in the model [141].
5.5.2.4 Calculation of insulator flashover in TL lightning-performance studies
Both the integration and physical models are presently used to evaluate flashover of the line insulation
under non-standard lightning overvoltages. Physical models are shown to provide an acceptable accuracy
over a wide range of air gap configurations covering gap factors ( kg) from 1.0 to 1.4 [15] and gap distances
of over 1 m [124]. The models have the additional advantage of providing an estimate of the pre-discharge
current which is important to account for on longer gap lengths and gaps containing a series impedance
such as those found on externally gapped line arresters. On the other hand, it was found that these
physical models did not have a good accuracy in cases where the discharge developed along insulators
such as is the case on rod-plane gap containing a disc insulator string [15].
The use of integration methods is recommended for air gaps smaller than 1,2 m [124], and where the
flashover develops across insulators – provide that the parameters of the method are selected from tests
on a similar configuration [15].

5.6 Backflashover model


5.6.1 Background
When lightning hits the shield wires or towers, the injected current propagates along the wires and towers
to be dissipated into the soil via the earth electrodes at the base of the towers. Because of the impedances
involved, this gives rise to potential differences across the line insulation which can lead to a flashover if
the strength of the line insulation is exceeded. This event is called a backflash (or backflashover). The
performance of the line under this condition is usually expressed as the number of flashovers per 100 km
per year, which is defined as the backflashover rate or BFR.
CIGRE brochure 63 [6] provides essential background information and a detailed description of the general
concepts involved when evaluating the backflashover performance of transmission lines. This is aided by
equations and curves which permits a simplified estimation of the required insulation strength and footing
resistance for new designs, or alternatively, when assessing the performance of existing designs.
The advent of fast personal computers and greater availability of powerful simulation software, enabled a
wider use of detailed models which makes designers less reliant on the analytical methods introduced by
CIGRE [6], IEEE [5] and others [7]. In this section the focus is therefore on providing general guidelines for
circuit-based models and methods for calculating the backflashover performance of transmission lines.
Analytical procedures continue to play important roles as benchmarks for validating the more detailed
simulation models and as teaching resources.

5.6.2 The general concept


A lightning stroke terminating on the shield wire, as shown in Figure 5-24, creates waves of current and
voltage which travel along the shield wire. The voltage and current waves are coupled by the surge
impedance of the wire over ground, computed using Equation 5-3. At the tower TT, and other points of
impedance discontinuity, the incident waves are reflected back toward the stroke terminating point and are
also transmitted down the tower and onward via shield wire to adjacent towers. The incident voltage Up on
the shield wire also appears with reduced magnitude, C.Up, on parallel running phase conductor(s). The
coupling coefficients C are computed from the self and mutual surge impedance matrix, assuming current
Ip flows in the shield wire and no impulse current flows in the (insulated) phases. A phase protected by a

66
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

line surge arrester will take up a voltage of Up minus the arrester voltage drop. A process of reflection and
transmission continues, and dynamic voltages are produced across (1) the air insulation within the span as
well as (2) across the air insulation or insulators at the tower.

Figure 5-24 – Voltages generated when lightning strikes a shield wire. Up is the voltage on the shield wire and
C is the coupling factor between the shield wire and phase conductor.
The highest shield wires voltage occurs at the stroke location and its magnitude will decrease with every
transmission past an impedance discontinuity. On the other hand, the largest clearance and insulation
strength are typically found midspan and the lowest at the tower. For a complete evaluation it is, therefore,
required that the voltages across all insulation points be evaluated against the insulation strength at each
point to determine where the flashover occurs. It has been shown [7], however, that most flashovers will
occur at the tower, even for lightning flashes within the span. This is true provided that the span is not too
long, and that the clearance between the shield wire and conductor is much larger than the insulation
distances in the tower. With important differences between sags of steel shield wires and aluminium/steel
phase conductors, this is nearly always true on transmission lines. Thus, for most cases flashovers in the
span may be neglected when evaluating the backflashover performance of a line.
Furthermore, it has also been shown [7] that the highest voltages across the tower insulation occur when
the lightning stroke directly hits the tower. Strokes within the span generally result in lower tower insulation
voltages because of the better current sharing between the neighbouring towers. Consequently, a
conservative estimate will be obtained if it is assumed that all strokes to the line occur at the tower, and
none of them occur in the span. A comparative analysis [7] [6] has shown that the BFR when all possible
stroke locations are considered, is about 60 % of that when all strokes are assumed to hit the tower.
Thus, the calculation of the BFR may be simplified considerably by only considering strokes at the tower.
This leads to the simplified model shown in Figure 5-25.

Figure 5-25 – A simplified schematic diagram of a transient model for calculating backflashover performance.
The aim with this model is to determine the minimum lightning current which will result in a flashover of the
line insulation (i.e. the critical current, IC). Variables that must be considered in this process include (1) the
instantaneous power frequency voltage on the phase conductors, (2) variations in the span length and (3)

67
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

variations in the tower footing impedance. Guidelines for building the backflashover model are provided in
the following paragraphs.

5.6.3 Overhead transmission line


Each span of the overhead line included in the model should be represented with an un-transposed,
distributed parameter, multi-phase model which includes all the phase conductors and shield wires. In
most cases it is sufficient to utilize a constant parameter with the line parameters calculated at the
dominant transient frequency. The dominant transient frequency is the sine wave that has the same peak
current and maximum derivative as the desired lightning surge current. This is between 80 kHz and
120 kHz for first return strokes and 500 kHz for backflashover studies considering subsequent strokes.
Frequency dependent models may also be applied.
For backflashover studies only a relatively short section of the line needs to be modeled as flashovers are
usually concentrated around the peak of the overvoltages and rarely occur past 10 μs. The mitigation
effect of the neighbouring towers on insulator voltage also needs to be included in the model. This leads to
the following minimum guidelines:

 The model should include the central, stricken tower plus at least four neighbouring towers on
either side as shown in Figure 5-25.
 The travel time to the termination of the detailed line model should be at least 5 μs which means
about 1.5 km from the stricken tower.
 The detailed line model is terminated with resistances equal to the equivalent characteristic surge
impedance of the line in the phase domain, to eliminate reflections from the ends of the detailed
line model.
The variation of span length, lsp, along the line should be analysed utilizing descriptive statistics to obtain
the mean span length. A log-logistic cumulative distribution function (CDF) – Equation 5-52 – suits this
purpose well as it can easily be fitted to transmission line data using the least squares method [146].

1
𝐶𝐷𝐹 = 𝑙𝑠𝑝 𝛼
Equation 5-52
1+( )
𝑚𝑒𝑑𝑖𝑎𝑛

For most purposes it is sufficient to model the line with equal span lengths set to the mean span length of
the line [146].

5.6.4 Corona effect


The effect of corona around the shield wires may be neglected as a conservative approach. The simplified
method to compute downlead corona radius in Section 5.3.2.3 provides realistic voltage dependence of
surge impedance compared to observations [4][37]. The effects of corona on the surge impedance
coupling coefficients are more significant. For more details on the modelling of corona may be found in
brochure 63 [6].

5.6.5 Transmission line towers


The central (stricken) tower is modelled in detail with crossarms according to the guidance provided in
section 5.3. In cases of tall towers with a low tower footing resistance it may be necessary to represent the
imperfect ground plane surge response as described in section 5.3.6. This can be done by either utilizing
the multistorey model or by including an additional small inductance in series with the footing impedance.
Simpler models may be used for the neighbouring towers which can be modelled either as a lossless
propagation line with constant characteristics or as an equivalent inductance and earthing resistance for
towers further away.

5.6.6 Tower earthing electrodes


The earth electrode at the central (i.e. stricken) tower is represented by its first-stroke impulse impedance,
ZP, according to the guidance provided in Section 5.4. Ionization may be neglected as a conservative
approach. A statistical distribution for soil resistivity  at tower earth electrodes is recommended [146],
using Equation 5-52 along with a value of   1.9 or fitted to results of tower-by-tower earth resistance
measurement programs. The variation of the tower footing along the line under study should be accounted

68
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

for by applying descriptive statistics. It has been shown that the dispersion of tower footing resistances well
represented by a log-normal distribution function [146].
A good estimate of the total line performance may be obtained by combining the results from at least 10
tower footing impedance values (R1 … R10). These values are selected from the cumulative probability
curve describing the variation in tower footing impedance [146] of the line under study. The most accurate
results are obtained if the resistance values are selected at a uniform interval of probability (i.e. 0.95,
0.85, … 0.15, 0.05) as shown in Figure 5-26. In this example ten resistance values are selected, namely:
3.9, 6.4, 8.6, 11, 14, 17, 21, 26, 35 and 58 Ω. Each of these resistance values represents the performance
of one tenth (1/10) of the line length.
1
Probability of exceeding the absyssa [p.u.]

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0
1 R1 R2 R3 10 R4 R5 R6 R7 R8 R9 R10 100
Tower Footing Resistance [Ω]

Figure 5-26 – Selection of footing impedances for a backflashover performance.

5.6.7 Power frequency source


The model should include at least one 3-phase power frequency source to represent the system voltage at
the time of the lightning flash. The magnitude of the applied voltage is set to the maximum operating
voltage of the line and the phase angle of the voltage is varied to determine the effect of the instantaneous
voltage when the lightning flash occurs. It is usually sufficient to select 12 or more representative phase
angles. As a time saving measure, a good approximation of the backflashover rate may be obtained by
using only, 6 representative phase angles which are selected to correspond to the zero-crossings of the
power frequency voltage, as shown in Figure 5-27 [151].

[1] [2] [3] [4] [5] [6]


Voltage

Time / phase angle

Figure 5-27 - Optimized selection of the phase angle of the power frequency voltage
The 3-phase power-frequency voltage is modelled as a voltage source and for backflashover studies it is
not necessary to include a Thevenin network equivalent impedance.

5.6.8 Line insulation and flashover


5.6.8.1 Initiation of flashover
The initiation of flashover is calculated according to one of the flashover models provided in Section 5.5.

5.6.8.2 Flashover
In the simplest case, flashover of the line insulation may be represented by an ideal switch across the
insulation gap in the tower. The switch closes as soon as the conditions for flashover has been reached as

69
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

determined by the flashover model. The fast closing of the switch may, however, result in an unrealistically
steep voltage collapse, which could in turn, lead to excessively high overvoltages on un-faulted phases.
This is specifically a problem when using small time steps. A more realistic chopped wave voltage collapse
may be achieved by including a small inductance in series with the switch, to represent the inductance of
the arc (approximately 1 μH/m of arc length).
The insulator may be represented by a lumped capacitance in parallel with the switch and arc impedance.
A typical cap-and-pin disc has a capacitance of 40 pF and a string of ten discs would have 4 pF.
Capacitance of polymer insulators is negligible and can be ignored.
It is good practice to monitor the current through the modeled insulator as well as the leader current during
the simulation, to confirm modeling assumptions and to check for excessive porcelain insulator current
(i.e. > 13 A) [157] which may suggest the possibility of steep-wave puncture damage.
5.6.9 Lightning stroke
The backflashover calculation is based on the characteristics of the first return stroke of the negative cloud
to ground flash. The stroke is either modeled as an ideal current source or with a Norton equivalent
impedance of 1000 Ω as described in section 5.2. The current magnitude and wave shape parameters are
selected according to the first return stroke characteristics provided in Chapter 3. In this regard it is
important to take regional effects into account, as far as they are available.
The lightning stroke is applied to the central tower in the model.
The purpose of the model is to successively increase the magnitude of the lightning current until flashover
of one or more insulators is achieved. The minimum current that will result in a flashover is called the
critical current, IC, and is further used in the statistical calculation of the backflashover rate. The critical
current is usually determined as a function of the instantaneous value of the power frequency voltage as
designated by the phase angle of the power frequency voltage source.
For single circuit lines and in simplified calculations the determination of the critical current is stopped after
the first flashover is identified. On double circuit lines, the calculation is usually continued to determine
additional values of the critical current where multiple phases flash over [152] [153]. An example of the
calculation of the critical currents per phase of a double circuit line is presented in Figure 5-28. From this
information the backflashover on a per circuit bases can be calculated as is presented in Figure 5-29.
For these calculations it is important to account for the changes in the circuit parameters introduced by the
flashover of the line insulation. This includes a change in the impedance presented to the lightning stroke,
improved voltage coupling of the lightning current injected into the faulted phases to the remaining un-
faulted phases, and the introduction of new reflection points introduced by the flashover [152].
350
Critical Lightning Current [kA]

300
Phase ID
250 Flashovers on
Circuit 1 A
Phase C and C'
Flashovers on
200
Phase B and B'
Circuit 1 B
FO on Ph A' Flashover on Ph C FO on Ph A'
150 Circuit 1 C
Circuit 2 A'
100
No Flashovers Circuit 2 B'
50
Circuit 2 C'
0
0 30 60 90 120 150 180 210 240 270 300 330 360
Power Frequency Voltage Phase Angle [Degrees]

Figure 5-28 - Calculated critical lightning current per phase for a 380 kV double-circuit line, low-reactance
phase configuration.

70
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

350

Critical Lightning Current [kA]


300
Flashovers on Circuits 1 and 2
250

200 Circuit ID
FO on Ckt 2 Flashover on Circuit 1 FO on Ckt 2
150 Circuit 1
Circuit 2
100
No Flashovers
50

0
0 30 60 90 120 150 180 210 240 270 300 330 360
Power Frequency Voltage Phase Angle [Degrees]

Figure 5-29 - Calculated critical lightning current per circuit for a 380 kV double-circuit line, low-reactance
phase configuration.

5.6.10 Backflashover rate calculation


When the simulation studies are complete a two-dimensional array of critical current values is obtained as
a function of the power frequency phase angle and the tower footing impedance. These values can then
be used with the number of flashes to the line (Equation 4-11) and the probability of the lightning current
being exceeded (Section 3.3.4) to estimate the backflashover rate as follows:
𝑛 2𝜋
1 ∆𝜃
BFR = K𝑆𝐹 𝑁𝐿 ∑ {∑ [ 𝑃⌈𝐼𝑐 (𝑅𝑖 , 𝜃)⌉]} [Flashovers per year] Equation 5-53
𝑛 2𝜋
𝑖=1 𝜃=0
Where:
NL the number of flashes to the line as determined by the stroke attraction model employed –
Equation 4-11 or Equation 4-4.
n the total number of tower footing impedance values (R1 ..Ri ..Rn) for which the critical current
is calculated (usually n = 10).
Δθ the step size in phase angle. If twelve phase angles are selected, then Δθ is π/6.
IC the critical current as a function of the footing impedance Ri and phase angle θ of the power
frequency voltage.
P(IC)the probability that the critical current IC will be exceeded.
KSF the span factor which is the ratio of the backflashover rate when considering all stroke
terminating points along the span to the backflashover rate when all strokes are assumed to
hit the tower.

In Appendix 3 of CIGRE brochure 63 [6] it was indicated that the span factor, KSF, typically varies between
0.63 and 0.42 depending on the calculated critical current at the tower. On this basis, the use of KSF = 0.6
has been recommended [2]. This recommendation may not be justified in all cases, however, as recent
studies [149][150] have shown that KSF is influenced by several parameters such as the span length, the
insulation strength, and tower-footing grounding impedance. Indications are that KSF may be significantly
higher than 0.6 for TLs with an operating voltage below 230 kV (e.g. 138 kV) and significantly lower than
0.6 for lines with an operating voltage above 230 kV (e.g. 500 kV). Further studies are required to provide
more detailed guidance on how to select appropriate values for KSF.

Alternatively, the use of KSF may be avoided by extending the backflashover calculation to also consider
lightning strokes to the span. In such a case the BFR is calculated for a stroke to the tower and for strokes
to number of span sections, and then summed to obtain the total BFR rate.

71
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

6 Summary and suggestions for further work


In this brochure an update is given on models and procedures that can be used to estimate the lightning
outage rates of transmission lines. Estimating the lightning performance of transmission lines is a complex
task and the achievable accuracy of the results is still relatively limited. There remains a need to improve
and update procedures and models based on the latest research. Important input in this regard is studies
where the calculated outage rates is benchmarked against the long-term average outage rates from lines
that have been in service for a long time.
It is further noted that the line outage performance may vary widely from year to year due to statistical and
climatological variations. For system operators it may be useful if estimations could be obtained of this
variation. However, this will only become possible if information becomes publicly available on the year-to-
year variations of the ground flash density.
An important contribution in this document is the proposed simplified lightning attachment model,
developed by Rizk. The WG has made a start in validating this method but further benchmarking is still
required.
Some utilities operating EHVDC systems report a strong influence of polarity on backflashover, with the
positive pole flashing over for large negative flashes and the negative pole flashing over for large positive
flashes in roughly equal measure. The role of ±500 to 800 kV dc voltage bias on shielding failures is not so
clear and important to validating the predictions of the lightning attachment models.
Modern LLS can now resolve the weak first return-stroke currents that cause shielding failures, separated
by tens of milliseconds from subsequent strokes that cause line-to-ground faults. LLS data, combined with
travelling wave recorder observations, can resolve the question of whether every shielding failure leads to
a flashover.
Global lightning detection networks (LDN) now report results near the North Pole, and thus provide
instrumented readings of flash density at all locations. However, the transfer function between Ng and the
number of first return strokes to an overhead conductor was obtained based on a specific flash counting
method, specifically the CIGRE 10 kHz LFC. Studies assisted by drone inspection should be conducted to
establish the true physical relation between stroke incidence to overhead wires (indicated by burn marks)
and the locally observed stroke or flash density using global LDN.
Another aspect that requires further investigation is to improve the recommendations regarding the
characteristics of first and subsequent strokes relevant to transmission lines. In this regard, it will be
important to obtain more data from direct measurements on short, grounded structures that resemble
transmission line towers. In this regard, more information is required on the waveshape parameters of
positive first return strokes, which due to their high extreme current values could be a significant
contributor to the overall line backflashover rate.
The performance calculation methodology presented in this brochure still contains many simplifications
which are often represented by correction factors. One of these is the span factor, KSF, which is the ratio
of the backflashover rate when considering all stroke terminating points along the span to the
backflashover rate when all strokes are assumed to hit the tower. Presently a value of 0.6 is recommended
but recent studies have shown that this factor may be significantly higher, which depends on several
factors including the span length, the insulation strength, and tower-footing grounding impedance. Further
studies are required to provide more detailed guidance on selecting appropriate values for KSF.
In this brochure, the focus was primarily on providing guidelines for calculating the lightning performance of
conventional single circuit transmission lines. Remedial measures, such as the application of line surge
arresters [13] and under-build earth wires [154], have not been considered. Advances in the modelling of
TL components and the use of modern simulation tools have led to the development of improved
techniques for determining the performance of more complex configurations such as multi-circuit lines and
to assess the efficiency of the remedial measures such as those mentioned above. Further exploring of
such techniques and their application to assess the performance of complex line configurations is still
needed, as well as investigations into the efficiency and feasibility of novel mitigation measures such as
under-built earth wires.

72
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

7 References
[1] AIEE Committee Report, “A Method of Estimating the Lightning Performance of Transmission Lines”, AIEE
Trans., Vol 69, Pt II, 1950, pp 1187-1196
[2] Anderson J. G., "Transmission Line Reference Book - 345 kV and Above-" Second Edition, 1982, Chapter 12,
Electric Power Research Institute, Palo Alto, California, EL-2500. 1982.
[3] IEEE Transmission and Distribution Committee WG Lightning Performance of Transmission Lines, “A Simplified
Method for Estimating Lightning Performance of Transmission Lines”, IEEE Transactions on Power Apparatus
and Systems, Vol. PAS-104, No. 4, July 1985, 918 – 932.
[4] IEEE Working Group on Estimating the Lightning Performance of Transmission Lines, “Estimating lightning
performance of transmission lines. II. Updates to analytical models”, IEEE Transactions on Power Delivery, Vol.
8, No. 3, July 1993, pp 1254-1267
[5] IEEE Standard 1243-1997 – “IEEE Guide for Improving the Lightning Performance of Transmission Lines”,
1997-12-16, Reaffirmed: 2008-09-26
[6] CIGRE SC 33 (now C4) WG 33.01, “Guide to procedures for estimating the lightning performance of
transmission lines”, Technical Brochure No. 63, 1991. Reissued with this brochure.
[7] Hileman A.R., “Insulation Coordination for Power Systems”, Book, CRC Press, ISBN 9780824799571
[8] McDermott T. E., “Using IEEE Flash to estimate transmission and distribution line lightning performance”, PES
T&D 2012, Year: 2012
[9] CIGRE SC C4, WG C4.407, “Lightning Parameters for Engineering Applications”, Technical Brochure No. 549,
2013
[10] CIGRE SC C4, WG C4.26, “Evaluation of lightning shielding analysis methods for EHV and UHV DC and AC
transmission lines”, Technical Brochure No. 704, 2017
[11] CIGRE SC C4, WG C4.33, “Impact of soil-parameter frequency dependence on the response of grounding
electrodes and on the lightning performance of electrical systems”, Technical Brochure No. 781, 2019
[12] CIGRE SC C4, WG C4.37, “Electromagnetic computation methods for lightning surge studies with emphasis on
the FDTD method”, Technical Brochure No. 785, 2019
[13] CIGRE SC C4, WG C4.301, “Use of Surge Arresters for Lightning Protection of Transmission Lines, Technical
Brochure No. 440, 2010
[14] CIGRE SC C4, WG C4.404, “Cloud-to-Ground Lightning Parameters derived from Lightning Location Systems.
The Effects of System Performance”, Technical Brochure No. 376, 2009
[15] CIGRE SC 33 (now C4) WG 33.07, “Guidelines for the evaluation of the dielectric strength of external
insulation.”, Technical Brochure No. 72, 1992.
[16] Murphy, M. J., & Said, R. K. (2020). “Comparisons of lightning rates and properties from the U.S. National
Lightning Detection Network (NLDN) and GLD360 with GOES‐16 Geostationary Lightning Mapper and
Advanced Baseline Imager data.” Journal of Geophysical Research: Atmospheres, 125, e2019JD031172.
https://doi.org/10.1029/2019JD031172
[17] Schulz, W., & Nag, A., “Lightning Interaction with Power Systems, Volume 1: Fundamentals and modelling”
(Book edited by A. Piantini) Chapter 4: (2020). “Lightning geolocation information for power system analyses.”
[18] Anderson, R. B., H. R. van Niekerk, S. A. Prentice, and D. Mackerras, Improved lightning flash counter, Electra,
No. 66, pp 85 – 98, 1979.
[19] https://www.vaisala.com/en/products/data-subscriptions-and-reports/data-sets/gld360
[20] https://ghrc.nsstc.nasa.gov/lightning/index.html
[21] http://pmm.nasa.gov/node/163
[22] CIGRE SC C4, WG C4.501 “Guideline for Numerical Electromagnetic Analysis Method and its Application to
Surge Phenomena”, Technical Brochure No. 543, June 2013.
[23] CIGRE SC C4, WG C4.33, “Impact of soil-parameter frequency dependence on the response of grounding
electrodes and on the lightning performance of electrical systems”, Technical Brochure No. 781, October 2019.
[24] CIGRE SC C4, WG C4.37, “Electromagnetic computation methods for lightning surge studies with emphasis on
the FDTD method”, Technical Brochure No. 785, Dec 2019.
[25] CIGRE SC C4, WG 4.02, Protection of MV and LV networks against lightning – Part 1: Common topics,
Technical Brochure No. 287, 2006.
[26] CIGRE SC C4, WG 402, Protection of Medium Voltage and Low Voltage Networks Against Lightning, Technical
Brochure No. 441, 2010.
[27] CIGRE SC C4, WG 408, Lightning Protection of Low-Voltage Networks, Technical Brochure No. 550, 2013.

73
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

[28] IEC 62858, “Lightning density based on lightning location systems (LLS) - General principles”, International
Electrotechnical commission, Geneva, 2019.
[29] IEC 62305:2013, “Protection against lightning” - IEC 62305-1:2010 Part 1: “General principles”, IEC 62305-
2:2010: Part 2: “Risk management”, IEC 62305-3:2010: Part 3: “Physical damage to structures and life hazard”,
IEC 62305-4:2010: Part 4: “Electrical and electronic systems within structures.”
[30] Diendorfer G., “Some comments on the achievable accuracy of local ground flash density values” 29 th
International Conference on Lightning Protection 23rd – 26th June 2008 – Uppsala, Sweden
[31] IEEE Standard 1410-2010 – “IEEE Guide for Improving the Lightning Performance of Electric Power Overhead
Distribution Lines”, 2009-11-02.
[32] Pédeboy S., “Identification of the multiple ground contacts flashes with lightning location systems”, 22nd
International Lightning Detection Conference (ILDC), April 2-3, 2012, Broomfield CO, USA.
[33] Pédeboy S., “Review of the lightning dataset and Lightning Locating Systems performances as recommended
by the IEC 62358 standard”, International Colloquium on Lightning and Power Systems, Delft 2019.
[34] Orville, R., Songster, H., "The East Coast Lightning Detection Network,", IEEE Transactions on Power Delivery,
vol.2, no.3, pp.899,907, July 1987
[35] Cummins K. L., et.al. “A Combined TOA/MDF Technology Upgrade of U.S. National Lightning Detection
Network”, Journal of Geophysical Research, Vol. 103, No. D8, pp 9035-9044, April, 1998
[36] Christian, H.J. et al “Global frequency and distribution of lightning as observed from space by the Optical
Transient Detector”, Journal of Geophysical Research, Vol 108, pp 4-1 – 4-15, 2003
[37] EPRI, “AC Transmission Line Reference Book – 200 kV and Above”, Third Edition, EPRI Palo Alto, CA, 2005.
1011974
[38] Gridded lightning climatology from TRMM-LIS and OTD: Dataset description, Daniel J. Cecil, Dennis E.
Buechler, Richard J. Blakeslee, Atmospheric Research 135-136 (2014) pp 404-414
[39] Cecil, D. J., D. E. Buechler, and R. J. Blakeslee. “Gridded Lightning Climatology from TRMM-LIS and OTD:
Dataset Description.” Atmospheric Research. Vols. 135–136. January 2014. pp. 404–414. doi:10.1016/j.
atmosres.2012.06.028.
[40] Chisholm W.A., “Estimates of lightning ground flash density using optical transient density”, 2003 IEEE PES
Transmission and Distribution Conference and Exposition (IEEE Cat. No.03CH37495), Vol. 3, 2003.
[41] IEC 61400-24:2019, Wind energy generation systems - Part 24: Lightning protection, Geneva, 2019.
[42] FAA document 6950.20 – “Fundamental Considerations of Lightning Protection , Grounding, Bonding, and
Shielding”, July 28th, 1978
https://www.faa.gov/regulations_policies/orders_notices/index.cfm/go/document.information/documentID/9894
[43] Anderson R.B., Eriksson A.J., Kroninger H., Meal D.V. and Smith M.A. “Lightning and thunderstorm
parameters” IEE Conference Publication No. 236, “Lightning and Power Systems”, London, June 1984.
[44] IEEE Standard 998-2012 – “IEEE Guide for Direct Lightning Stroke Shielding of Substations”, 2013-04-30.
[45] Anderson R.B. and Eriksson A.J. "Lightning parameters for engineering application" Electra No. 69, 65-102,
March 1980.
[46] Chowdhuri P.; J.G. Anderson; W.A. Chisholm; T.E. Field; M. Ishii; J.A. Martinez; M.B. Marz; J. McDaniel; T.E.
McDermott; A.M. Mousa; T. Narita; D.K. Nichols; T.A. Short, Parameters of lightning strokes: a review, IEEE
Transactions on Power Delivery, Vol 20 No. 1, 2005.
[47] Chisholm W.A. Cummins K.L., Lightning Parameters: A Review, Applications and Extensions, 2005/2006
IEEE/PES Transmission and Distribution Conference and Exhibition, 2006
[48] Eriksson, A. J., "Lightning and Tall structures", Trans SAIEE, vol 69, pt 8, Aug 1978.
[49] CIGRE SC C4, WG C4.410, “Lightning Striking Characteristics to Very High Structures”, Technical Brochure
No. 633, 2015
[50] Visacro, S. Mesquita, C. R. Dias, R. Silveira F. H., and De Conti A., "A class of hazardous subsequent lightning
strokes in terms of insulation stress," IEEE Trans. Electromagn. Compat., vol. 54, no. 5, pp. 1028-1033, Oct.
2012.
[51] Silveira, F. H. Visacro, S. De Conti A., “Lightning Performance of 138-kV Transmission Lines: The Relevance of
Subsequent Strokes”, IEEE Trans. Electromagn. Compat. vol. 55, no. 6, pp. 1195-1200, Dec. 2013.
[52] Berger K., Anderson R. B., and Kroninger H., “Parameters of lightning flashes”, Electra, vol. 80, pp. 223-237,
1975.
[53] Anderson R. B. and Eriksson A. J., "Lightning parameters for engineering application," Electra, vol.69, pp. 65-
102, 1980.

74
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

[54] S. Visacro A. Soares Jr. M. A. O. Schroeder, L. C. L. Cherchiglia, V. J. de Sousa Statistical analysis of lightning
current parameters: Measurements at Morro do Cachimbo Station Journal of Geophysical Research, Vol. 109,
D01105, doi:10.1029/2003JD003662, 2004.
[55] Silveira F.H., Visacro S., “Lightning Parameters of a Tropical Region for Engineering Application: Statistics of 51
Flashes Measured at Morro do Cachimbo and Expressions for Peak Current Distributions”, IEEE Trans.
Electromagn. Compat., vol.62, No.4, Aug. 2020.
[56] Takami, J. and S. Okabe. 2007. “Observational results of lightning current on transmission towers.”, IEEE
Trans. Power Del. 22: 547-56.
[57] Popolansky, F. 1972. Frequency distribution of amplitudes of lightning currents. Electra 22: 139-47.
[58] Rakov V., Uman M.A., "Lightning - Physics and Effects," Cambridge University Press, 2003.
[59] Visacro S., “A representative curve for lightning current waveshape of first negative stroke,” Geophys. Res.
Lett., vol. 31, L07112, Apr. 2004.
[60] Heidler F., “Analytische Blitzstromfunktion zur LEMP-berechnung,” in Proc. 18th Int. Conf. Lightn. Protec.
(ICLP), Munich, Germany, Sep. 1985, pp. 63-66.
[61] Heidler F. Flisowski Z. Zischank W. Bouquegneau Ch. Mazzetti C. “Parameters of lightning current given in IEC
62305 – background, experience and outlook”, 29th International Conference on Lightning Protection 23rd –
26th June 2008 – Uppsala, Sweden
[62] C. A. Nucci, F. Rachidi, M. V. Ianoz, and C. Mazzetti, “Lightning-induced voltages on overhead lines,” IEEE
Trans. Electromagn. Compat., vol. 35, no. 1, pp. 75–86, Feb. 1993.
[63] Borghetti, A., C. A. Nucci, and M. Paolone. 2004. “Estimation of the statistical distributions of lightning current
parameters at ground level from the data recorded by instrumented towers.”, IEEE Trans. Power Del. 19: Page
97 1400-9.
[64] CIGRE TF 33.01.03, Report 118. 1997. “Lightning exposure of structures and interception efficiency of air
terminals”, 86 p.
[65] Takami, J.; Okabe, S., “Observational Results of Lightning Current on Transmission Towers” IEEE Transactions
on Power Delivery, Volume: 22, Issue: 1, 2007, Page(s): 547 – 556.
[66] Cooray V, “The mechanism of the lightning flash”, Chapter 4, “The lightning Flash”, Book, IEEE Power & Energy
Series 34, 2003
[67] Cooray V, Becerra M, “Attachment of lightning flashes to grounded structures”, Chapter 4, “Lightning Protection,
Book”, IEEE Power & Energy Series 58, 2010.
[68] Rakov V.A., Uman M.A., “Downward negative lightning discharges to ground”, Chapter 4, “Lightning Physics
and effects”, Book Cambridge press, 2003.
[69] Rizk F.A.M, Trinh G.N. “Lightning Incidence and Lightning Protection”, Chapter 9, “High Voltage Engineering”,
CRC Press, Book, 2014.
[70] Cooray V. “Introduction to lightning”, Chapters 7, 17 and 18, Book, Springer Science+Business Media Dordrecht
2015.
[71] Armstrong H. R., Whitehead E. R., “Field and Analytical Studies of Transmission Line Shielding”, IEEE Trans.
Power Apparatus Syst., Vol.87, pp.270-281, 1968.
[72] Young F. S., Clayton J. M., and Hileman A. R., "Shielding of transmission lines", IEEE Trans. Power Apparatus
Syst., Special Supplement, Vol.S82, pp.132-154, 1963.
[73] Whitehead E. R., "CIGRE survey of the lightning performance of extra high-voltage transmission lines", Electra,
No.33, pp.63-89, 1974.
[74] Mousa A.M. Discussion of paper: IEEE Transmission and Distribution Committee WG Lightning Performance of
Transmission Lines, “A Simplified Method for Estimating Lightning Performance of Transmission Lines”, IEEE
Transactions on Power Apparatus and Systems, Vol. PAS-104, No. 4, July 1985, 918 – 932
[75] IEEE Working Group on Estimating the Lightning Performance of Transmission Lines: "Estimating Lightning
Performance of Transmission Lines II - Updates to Analytical Models", IEEE Working Group Report, 92 SM453-
1 PWRD, 1992.
[76] Z.G. Datsios et al.: Estimation of the minimum shielding failure flashover current for first and subsequent
lightning strokes to overhead transmission lines,” Electric Power Systems Research, vol. 113 (SI), pp. 141-150,
Aug. 2014.
[77] Hansheng Cai, Lei Jia, Gang Liu, Shangmao Hu, Jian Shi, Hengxin He, Xiangen Zhao, Junjia He, “Lightning
performance of EHV and UHV overhead transmission Lines in China southern power grid”, 33rd International
Conference on Lightning Protection, September, 25-30, 2016, Estoril, Portugal, paper 102-2014, 0091.
[78] EPRI: “EPRI High-Voltage Direct Current (HVDC) Transmission Reference Book”, Chapter 15 Overhead lines
with HVDC Transmission, Electric Power Research Institute, Palo Alto, PID 1024318, 2012.

75
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

[79] S. Taniguchi, T. Tsuboi, and S. Okabe, “Observation Results of Lightning Shielding for Large-scale
Transmission Lines”, IEEE Trans. Dielectr. Electr. Insul., Vol. 16, pp. 552-559, 2009.
[80] AIEE Committee Report, “A Method of estimating lightning performance of transmission lines”, Trans. American
Institute of Electrical Engineers (AIEE), Vol. 69, pt. II, pp. 1187-1196, 1950.
[81] J. Takami and S. Okabe, “Observational Results of Lightning Current on Transmission Towers”, IEEE Trans.
Power Delivery, Vol. 22, pp. 547-556, 2007.
[82] S. Taniguchi, T. Tsuboi, S. Okabe, Y. Nagaraki, J. Takami, and H. Ota, “Improved Method of Calculating
Lightning Stroke Rate to Large-sized Transmission Lines Based on an Electric geometry model”, IEEE Trans.
Dielectr. Electr. Insul., Vol. 17, pp. 53-62, 2010.
[83] S. Taniguchi, T. Tsuboi, S. Okabe, Y. Nagaraki, J. Takami, and H. Ota, “Method of Calculating the Lightning
Outage Rate of Large-sized Transmission Lines”, IEEE Trans. Dielectr. Electr. Insul., Vol. 17, pp. 1276-1283,
2010.
[84] S. Okabe, S. Taniguchi, T. Tsuboi, H. Ohta, E. Zaima, “Observation Results of Lightning Shielding and
Improvement in the Prediction Method for the Lightning Failure Rate with Large-Sized Transmission Lines”,
CIGRE Paris Session 2010, C4-207.
[85] F.A.M. Rizk, “Modeling of Transmission Line Exposure to Direct Lightning Strokes”, IEEE Trans. Power Del.,
vol.5, no. 4, pp. 1983-1997, Oct. 1990
[86] F.A.M. Rizk, “Modeling of proximity effect on positive leader inception and breakdown of long air gaps”, IEEE
Trans. on Power Del., vol. 24, no.4, pp.2311-2318, 2009.
[87] F.A.M. Rizk, “Modeling of lightning exposure of buildings and massive structures”, IEEE Trans. On Power Del.,
vol.24, no.4, pp. 2311-2318, Oct. 2009
[88] F.A.M. Rizk, “Modeling of substation shielding against direct lightning strikes”, IEEE Trans. EMC, vol. 52, no.3,
pp. 664-675, 2010.
[89] F.A.M. Rizk, “Modeling of UHV and double circuit EHV transmission line exposure to direct lightning strikes”,
IEEE Trans. Power Del., Vol. 32, no.4, pp.1739-1747, August 2017
[90] F. A. M. Rizk, A Simplified Approach for Assessment of Exposure of EHV and UHV Lines to Direct Lightning
Strikes, IEEE Transactions on Power Delivery, Vol. 33, No. 5, 2018.
[91] F.A.M. Rizk, “Modeling of Lightning Incidence to Tall Structures. Part I:Theory” , IEEE Transactions on Power
Delivery, Vol.9, No. 1, January 1994, pp. 162-171.
[92] Whitehead E.R. CIGRE survey of the lightning performance of extra-high-voltage transmission lines, (for CIGRE
WG 33.01), Electra No.33, March 1974, pp 63-89,
[93] IEEE PES Lightning Performance of Overhead Lines Working Group: 15.09.08, IEEE FLASH v2.05, Software
available at https://sourceforge.net/projects/ieeeflash/
[94] Chisholm W. A.; Chow Y.l.; Srivastava K.D.; Lightning Surge Response of Transmission Towers, IEEE
Transactions on Power Apparatus and Systems, 1983, Vol. PAS-102, No. 9
[95] Chisholm W. A., Janischewskyj, W. “Lightning Surge Response of Grounding Electrodes.” IEEE Transactions
PWRD, Vol. 4, No. 2, 1989: pp. 1329–1337.
[96] Baba Y.; Rakov V.A., On the interpretation of ground reflections observed in small-scale experiments
Simulating lightning strikes to towers, IEEE Transactions on Electromagnetic Compatibility, 2005, Vol. 47, No.
3.
[97] Baba Y.; Rakov V.A., A Study of Current Waves Propagating Along Vertical Conductors and Their Associated
Electromagnetic Fields, Presented at the 7th International Conference on Power Systems Transients (IPST’07)
in Lyon, France, June 4-7, 2007.
[98] Y. Baha;M. Ishii, Numerical electromagnetic field analysis on lightning surge response of tower with shield wire,
IEEE Transactions on Power Delivery, Vol. 15, No. 3, 2000.
[99] EPRI, Guide for Transmission Line Grounding: A Roadmap for Design, Testing, and Remediation: Part I –
Theory Book. EPRI, Palo Alto, CA: 2006. 1013594.
[100] T. Yamada;A. Mochizuki;J. Sawada;E. Zaima;T. Kawamura;A. Ametani;M. Ishii;S. Kato, Experimental
evaluation of a UHV tower model for lightning surge analysis, IEEE Transactions on Power Delivery, Vol 10, no.
1, 1995
[101] Mikihisa SAITO;Masaru Ishii;Megumu Miki;Kenji Tsuge, On Evaluation of Voltage Rise at Transmission Towers
Hit by Lightning by Electromagnetic and Circuit analyses, IEEE Transactions on Power Delivery, 2020 to be
published
[102] A. Ametani, and T. Kawamura, A Method of a Lightning Surge Analysis Recommended in Japan Using EMTP,
IEEE Transactions on Power Delivery, Vol. 20, No. 2, April 2005, pp 867-875.
[103] Grcev L., Popov M., On High-Frequency Circuit Equivalents of a Vertical Ground Rod, IEEE Transactions on
Power Delivery, Vol. 20, No. 2, April 2005, pp 1598 – 1603.

76
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

[104] Chisholm W. A., Evaluation of Simple Models for the Resistance of Solid and Wire-Frame Electrodes, IEEE
Transactions on Industry Applications, Vol. 51, no. 6, 2015.
[105] Bewley L.V. Travelling Waves on Transmission Systems, book, 2nd Edition corrected, Dover Publications Inc.
1963.
[106] Visacro, S., What Engineers in Industry Should Know About the Response of Grounding Electrodes Subject to
Lightning Currents, IEEE transactions on Industry Applications, vol. 51, no. 6, Nov/Dec 2015, pp 4943-
[107] Jinliang He, Baoping Zhang, Rong Zeng, and Bo Zhang. Experimental Studies of Impulse Breakdown Delay
Characteristics of Soil. IEEE Transactions on Power Delivery, Vol. 26, no.3, July 2011, pp.1600-1607
[108] Jinliang He and Bo Zhang. Progress in Lightning Impulse Characteristics of Grounding Electrodes with Soil
Ionization. IEEE Transactions on Industry Applications, Vol.51, no.6, pp.4924-4933, Nov/Dec. 2015.
[109] Visacro S., Soares-Jr, A., “HEM: A model for simulation of lightning related engineering problems,” IEEE Trans.
Power Del., vol. 20, no. 2, pp. 1026–1208, Apr. 2005.
[110] Visacro S., Silveira F. H., Lightning Performance of Transmission Lines: Requirements of Tower-Footing
Electrodes Consisting of Long Counterpoise Wires, IEEE Transactions on Power Delivery, Vol. 31, No. 4,
August 2016, pp 1524-1532
[111] Visacro S., “Transients on grounding systems,” in Numerical Analysis of Power System Transients and
Dynamics, 1st ed., vol. 1, A. Ametani, Ed. London, U.K.: IET, 2014, ch. XIV, pp. 481–507.
[112] Martinez-Velasco J.A., Ramirez A.I. Dávila M., Overhead Lines, Chapter 2 of Power System transients –
Parameter Determination Edited by J.A. Martinez-Velasco, Book, CRC Press, 2010.
[113] Visacro S., “A comprehensive approach to the grounding response to lightning currents,” IEEE Trans. Power
Del., vol. 22, no. 1, pp. 381–386, Jan. 2007.
[114] Visacro S., Silveira F. H., Lightning Performance of Transmission Lines: Requirements of Tower-Footing
Electrodes Consisting of Long Counterpoise Wires, IEEE Transactions on Power Delivery, Vol. 31, No. 4,
August 2016, pp 1524-1532.
[115] L. Grcev, “Impulse efficiency of ground electrodes,” IEEE Trans. Power Delivery, vol. 24, no. 1, pp. 441–451,
Jan. 2009.
[116] Scott J. H., Electrical and Magnetic Properties of Rock and Soil. Washington, DC, USA: U.S. Geological
Survey, Dept. Interior, 1966.
[117] Visacro S. and Alipio R., "Frequency dependence of soil parameters: experimental results, predicting formula
and influence on the lightning response of grounding electrodes," IEEE Trans. Power Delivery, vol. 27, no. 2,
pp. 927–935, Apr. 2012.
[118] Petrache E., Chisholm W.A. Phillips A., Evaluating the Transient Impedance of Transmission Line Towers,
Proceedings of IX International Symposium on Lightning Protection (SIPDA), Foz do Iguacu, Brazil, 26-30
November 2007.
[119] B. D. Rodrigues and S. Visacro, “Portable grounding impedance meter based on DSP,” IEEE Trans. Instrum.
Meas., vol. 63, no. 8, pp. 1916–1925, Aug. 2014.
[120] A. C. Liew and M. Darveniza, “Dynamic model of impulse characteristics of concentrated earths,” Proc. IEE, vol.
121, no. 2, pp. 123–135, 1974.
[121] E.E. Oettle, A new general estimation curve for predicting the impulse impedance of concentrated earth
electrodes, IEEE Transactions on Power Delivery, Vol. 3, No.4, 1988.
[122] International Electrotechnical Commission. Insulation Co-ordination - Part 1: Definitions, Principles and Rules.
IEC 60071-1. Edition. 8.0. 2006.
[123] International Electrotechnical Commission. Insulation Co-ordination - Part 2: Application Guidelines. IEC 60071-
2. Edition. 4.0. 2018.
[124] International Electrotechnical Commission. Insulation Co-ordination - Part 4: Computational guide to insulation
co-ordination and modelling of electrical networks. IEC 60071-4. Edition. 1.0. 2004.
[125] IEEE Task Force 15.09 on Nonstandard Lightning Voltage Waves, “Review of Research on Nonstandard
Lightning Voltage Waves,” IEEE Trans. Power Del., vol. 9, no. 4, pp. 1972-1981, Oct. 1994.
[126] IEEE Task Force 15.09 on Nonstandard Lightning Voltage Waves, “Bibliography of Research on Nonstandard
Lightning Voltage Waves,” IEEE Trans. Power Del., vol. 9, no. 4, pp. 1982-1990, Oct. 1994.
[127] R. O. Caldwell and M. Darveniza, “Experimental and Analytical Studies of the Effect of Non-standard
Waveshapes on the Impulse Strength of External Insulation,” IEEE Trans. Power Appar. Syst., vol. 92, no. 4,
pp. 1420-1428, Jul. 1973.
[128] C. F. Wagner and A. R. Hileman, “Mechanism of Breakdown of Laboratory Gaps,” AIEE Trans. Power Appar.
Syst., vol. 80, no. 3, pp. 605-618, Oct. 1961.
[129] T. Suzuki and K. Miyake, “Experimental Study of Breakdown Voltage-time Characteristics of Large Air Gaps
with Lightning Impulses,” IEEE Trans. Power Appar. Syst., vol. 96, no. 1, pp. 227-233, Jan./Feb. 1977.

77
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

[130] T. Shindo and T. Suzuki, “A New Calculation Method of Breakdown Voltage-time Characteristics of Long Air
Gaps,” IEEE Trans. Power Appar. Syst., vol. 104, no. 6, pp. 1556-1563, Jan. 1985.
[131] A. Pigini, G. Rizzi, E. Garragnati, A. Porrino, G. Baldo and G. Pesavento, “Performance of Large Air Gaps
under Lightning Overvoltages: Experimental Study and Analysis of Accuracy of Predetermination Methods,”
IEEE Trans. Power Del., vol. 4, no. 2, pp. 1379-1392, Apr. 1989.
[132] M. Darveniza, F. Popolansky, and E. R. Whitehead, “Lightning protection of UHV transmission lines,” Electra,
no. 41, pp. 39–69, Jul. 1975.
[133] IEEE Standard 1243. IEEE Working Group on Estimating the Lightning Performance of Transmission Lines.
“IEEE Design Guide for Improving the Lightning Performance of Transmission Lines.” Chisholm, W. A. (ed.).
Piscataway, NJ: IEEE, 1997.
[134] Chisholm, W.A. New challenges in lightning impulse flashover modeling of air gaps and insulators Electrical
Insulation Magazine, IEEE Volume: 26, Issue: 2, 2010, Page(s): 14 – 25.
[135] M. Ishii, T. Kawamura, and T. Kouno, “Multistory transmission tower model for lightning surge analysis,” IEEE
Trans. Power Del., vol. 6, no. 3, pp. 1327–1335, Jul. 1991.
[136] A. Inoue and S.-I. Kanao, “Observation and analysis of multiple-phase grounding fault caused by lightning,”
IEEE Trans. Power Del., vol. 11, no. 1, pp. 353–360, Jan. 1996.
[137] H. Motoyama, “Experimental study and analysis of breakdown characteristics of long air gaps with short tail
lightning impulse,” IEEE Trans. Power Delivery, vol. 11, no. 2, pp. 972-979, April 1996.
[138] T. Mozumi, Y. Baba, M. Ishii, N. Nagaoka, and A. Ametani, “Numerical electromagnetic field analysis of archorn
voltages during back-flashover on a 500-kV twin-circuit line,” IEEE Trans. Power Delivery, vol. 18, no. 1, pp.
207-213, January 2003.
[139] X. Wang, Z. Yu, and J. He, “Breakdown process experiments of 110- to 500-kV insulator strings under short tail
lightning impulse,” IEEE Transactions on Power Delivery, Vol. 29, No. 5, Oct. 2014, pp 2394 – 2401.
[140] Z. G. Datsios, P. N. Mikropoulos, T. E. Tsovilis, Insulator String Flashover Modeling with the aid of an ATPDraw
Object, UPEC 2011 ∙ 46th International Universities' Power Engineering Conference ∙ 5-8th September 2011 ∙
Soest ∙ Germany.
[141] Z. G. Datsios, P. N. Mikropoulos, Implementation of Leader Development Models in ATP-EMTP Using a Type-
94 Circuit Component, 2014 International Conference on Lightning Protection (ICLP), Shanghai, China
[142] Z.G. Datsios Z.G. et al.: “Effects of lightning channel equivalent impedance on lightning performance of
overhead transmission lines,” IEEE Trans. Electromagnetic Compatibility, vol.61, no.3, pp.623-630, Jun. 2019.
[143] A. Ametani, T. Kawamura, A Method of a Lightning Surge Analysis Recommended in Japan Using EMTP, IEEE
Transactions on Power Delivery, Vol. 20, No. 2, April 2005, pp 867 – 875.
[144] V. Cooray, The Lightning Flash, IET Power and Energy Series 69, 2nd Edition, 2014.
[145] IEEE Task force on Fast front transients, "Modeling guidelines for fast front transients," in IEEE Transactions on
Power Delivery, vol. 11, no. 1, pp. 493-506, Jan. 1996
[146] Chisholm W. A., de Almeida de Graaff S., Adapting the statistics of soil properties into existing and future
lightning protection standards and guides, International Symposium on Lightning Protection (XIII SIPDA), 2015.
[147] A. De Conti, S. Visacro, Analytical representation of singe- and double-peaked lightning current waveforms,
IEEE Transactions on Electromagnetic Compatibility, Vol. 49, No. 2, May 2007
[148] A.J. Oliveira, M.A.O. Schroeder, Adjustment of current waveform parameters for first lightning strokes:
Representation by Heidler functions, 2017 International Symposium on Lightning Protection (XIV SIPDA)
[149] F. H. Silveira et al.: “The impact of the distribution of lightning strikes along the span on backflashover rate of
transmission lines,” in Proc. 34th International Conference on Lightning Protection (ICLP), Rzeszow, 2018.
[150] Z.G. Datsios et al.: “Estimation of the minimum backflashover current of 150 and 400 kV overhead transmission
lines through ATP-EMTP Simulations: Effect of the lightning stroke location along line spans,” in Proc. 21st Int.
Symp. High Voltage Eng., Budapest, Hungary, 2019, paper no. 993.
[151] Z.G. Datsios et al.: “Closed-form expressions for the estimation of the minimum backflashover current of
overhead transmission lines,” in IEEE Transactions on Power Delivery, doi: 10.1109/TPWRD.2020.2984423.
[152] Sargent M.A. Darveniza M., The Calculation of Double Circuit Outage Rate of Transmission Lines, IEEE
Transactions on Power Apparatus and Systems, Vol. PAS-86, No. 6, 1967.
[153] Sargent M.A. Darveniza M., Lightning Performance of Double-Circuit Transmission Lines, IEEE Transactions on
Power Apparatus and Systems, Vol. PAS-89, No. 5, 1970.
[154] VISACRO, S.; SILVEIRA, F. H.; CONTI, A. R., The use of underbuilt wires to improve the lightning performance
of transmission lines. IEEE Transactions on Power Delivery, v. 27, p. 205-213, 2012.
[155] Berger K., Novel Observations on lightning discharges: Results of research on mount San Salvatore, Journal of
the Franklin Institute, Vol. 283, No. 6, June 1967.

78
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

[156] Eriksson A.J., The lightning ground flash: an engineering study, Thesis, University of Natal, December 1979.
[157] Lantz A. D., Measuring the Lightning Strength of High-Voltage Insulators, Transactions of the American Institute
of Electrical Engineers. Part III: Power Apparatus and Systems, Vol. 79, No. 3, 1960

79
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

8 Definitions, abbreviations, and symbols


8.1 Acronyms
Table 8-1 – Definition of acronyms used in this Technical Brochure.
Acronym Phrase Definition
AC Alternating current
CDF Cumulative distribution function
CG Cloud-to-Ground
DC Direct current
EGM Electro-Geometric Method
EHV Extra-High Voltage 315 kV – 765 kV
EMT Electromagnetic transients
EUCLID European Cooperation for Lightning Detection
FDTD Finite Difference Time Domain Advanced computational techniques
GFD Ground flash density
GSD Ground stroke density
HV High-Voltage 69 kV – 287 kV
IC Intra-Cloud and Inter-Cloud
ICC Initial continuous current
ITCZ Inter-Tropical Convergence Zone
LF Low frequency
LFC’s Lightning flash counters
LIS Lightning Imaging Sensor
LLS Lightning location systems
LPM Leader Progression Method
MCS Morro do Cachimbo Station
MDF Magnetic direction finding
MoM Method of Moments Advanced computational techniques
MSS Mount San Salvatore station in Switzerland
NALDN North American Lightning Detection Network
NEA Numerical Electromagnetic Analysis Advanced computational techniques
OTD Optical Transient Detector
RS Return Stroke
SAA South Atlantic Anomaly
SFFOR Shielding failure flashover rate
SFR Shielding failure rate
TL Transmission lines
TLJ Transmission line towers in Japan
TOA Time-of-arrival
UHV Ultra-High Voltage 800 kV and above
VLF Very low frequency

80
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

8.2 Defined Variables


Table 8-2 – Definition of variables used in this Technical Brochure
Variable Unit Definition
A m2 Surface area of overall electrode volume
Aw m2 Electrode wire surface area
α Degrees Shielding angle
β Standard deviation of ln (base e) of the variable
c m/s Speed of light, 3 × 108 m/s
CE F/m Capacitance of a buried wire
d m Separation distance of the tower legs
d m Burial depth
d m Gap length
DC m Unshielded width of phase conductor
DE Disruptive effect
Dg m Exposure width of shield wire
E(t) kV/m Voltage stress across the insulation
E0 kV/m Critical inception electric field for ionization (usually 400 kV/m)
E0 kV/m Onset voltage stress across the insulation
E50 kV/m 50 % flashover stress across the insulation (E50 = U50/d)
Ep kV/m Peak voltage stress across the insulation
ε0 F/m Permittivity of free space (8.85 × 10−12 F/m)
εr -- Relative permittivity of the soil (typically ≈ 10 - 20)
f Hz Frequency f in Hz
g m Geometric radius of the electrode, which is equal to √𝑟𝑥2 + 𝑟𝑦2 + 𝑟𝑧2
h m Height
h1 m Height from base to midsection
h2 m Height from midsection to top
hP m Height of phase conductor
hS m Height of shield wire
Ht m Total height of the tower (h1 + h2)
I kA Lightning stroke current
I10 kA 10 % intercept along the stroke current waveshape
I30 kA 30 % intercept along the stroke current waveshape
I90 kA 90 % intercept along the stroke current waveshape
I100 = II kA Initial peak of current
IC Impulse coefficient (IC = ZP/RLF)
IC kA Critical current
IF, Ip kA Final (overall) peak of current (i.e. peak current)
il A Leader current
Imsf kA Maximum shielding failure current
IPI kA Peak current required for the inception of ionization process
kg Gap factor
KSF Span factor
L m Attractive distance
l km The length of the line
l m The conductor length
l m Leader length
La m Average attractive distance,
LC m Attractive distance to shield wire
lEF m Effective Length of the earth conductor
LL H/m Inductance of a buried wire
LP m Attractive distance to phase conductor
lsp. m Length of the span,
LT µH Tower inductance
Lw m Total length of the wires making up the electrode

81
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

Variable Unit Definition


μ0 H/m Permeability of free space (4𝜋 × 10−7 H/m)
Ng Fl/km2/yr Ground Flash Density (GFD)
NL Fl/100 km/yr Flash incidence to the shield wires
P M Effective perimeter of an earth electrode,  / RLF
q μC/m Leader charge density
QI C Impulse charge – time integral of current
r m Conductor, tower radius
r m Striking distance
r1 m Tower top radius
r2 m Tower midsection radius
r3 m Tower base radius
rc m Striking distance to conductors
rC M Corona radius,
RE Ω Earth electrode resistance
req m Equivalent radius of the tower section
rg m Striking distance to earth surface
RI Ω Ionized resistance of the ground electrode
RL Ω/m Internal resistance of a buried wire
RLF Ω Low frequency ground electrode resistance
rw m Wire radius
ρ Ωm Uniform soil resistivity
s m The maximum extent of the electrode, measured from the centre
S10 kA/µs Instantaneous rate-of-rise of current at I10
S10/90 kA/µs Average steepness (through I10 and I90 intercepts)
S30/90 kA/µs Average steepness (through I30 and I90 intercepts)
Sg m Distance between the shield wires
Sm kA/µs Maximum rate-of-rise of current along wave-front, typically at I90
T1 µs Impulse front time ()
T2 µs Impulse tail time (T1)
T10/90 µs Time between I10 and I90 intercepts on the wavefront
T30/90 µs Time between I30 and I90 intercepts on the wavefront
tbr μs Time at which break down of the gap occurs
Td days Thunderstorm days
td 10/90 µs Equivalent linear wave-front duration derived as T10/90 / 0.8
td 30/90 µs Equivalent linear wave-front duration derived as T30/90 / 0.6
tf µs Front time
Th hours Thunderstorm hours
th µs Tail time to half value
ti μs Time to corona inception,
tI μs The leader development time.
tm µs Equivalent linear wave-front duration derived from IF / Sm
ts μs Streamer development time
tt m Tower travel time
θ Radians Phase angle of the power frequency voltage
U0 kV Voltage below which no breakdown is possible
U+50 kV Positive polarity lightning impulse strength
U-50 kV Negative polarity lightning impulse strength
U+50_rp kV Positive polarity lightning impulse strength of a rod-plane gap
U-50_rp kV Negative polarity lightning impulse strength of a rod-plane gap
Ubr kV Peak of the applied voltage
VTT kV Tower top voltage
W (A2.s) × 104 Specific energy (action integral) – time integral of the square current

82
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

Variable Unit Definition


Z Ω Surge Impedance
Zc Ω Surge impedance of a wire in corona
Zch Ω Equivalent source impedance of a lightning stroke
Zg Ω Shield wire surge impedance
ZP Ω Impulse impedance of the ground electrode
ZP1st Ω First-stroke impulse impedance
ZT Ω Tower Surge impedance

83
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

9 Updated stroke current parameters from direct


measurements
In this Appendix a summary of the statistical distributions derived from three datasets of lightning current
measurements is presented. These datasets originated as follows:
 Mount San Salvatore station in Switzerland (MSS): Berger et al (1975) compiled the largest
database for temperate regions from the direct measurements of lightning currents on
instrumented towers at Mont San Salvatore Station (MSS), in Lugano Switzerland [52]. This
dataset was later revised by Anderson and Eriksson (1980) [53].
 Transmission line towers in Japan (TLJ): Takami and Okabe (2007) compiled another important
database from measurements performed on instrumented towers of a transmission line in Japan
[56]
 Morro do Cachimbo Station (MCS): Silveira and Visacro (2019) compiled data from direct
measurements of lightning currents on a 60-m high instrumented tower at Morro do Cachimbo
Station, in Brazil. To date, this is the only dataset, with statistical significance, of RS currents in
tropical regions [54] [55].
A summary of published parameters for negative first and subsequent return stroke currents, of these
three datasets, are presented in Table 9-1 and Table 9-2 respectively. The following notes apply:
 The statistical variation of the listed lightning parameters is assumed to be log-normal.
 The values for  listed in the tables corresponds to the standard deviation of Ln (I) – base e.
 For the MSS dataset, the updated parameters II, IF, td 10/90, td 30/90 [53] are listed. Parameters th, Q
and W, however, were taken from the original dataset [52], since they were not addressed in the
update [53].
 The standard deviation ( ) of parameters th, Q and W are not included for MSS data, as they were
not provided in either the original [52] or updated [53] dataset.
 Table 9-2 does not include data of subsequent stroke currents measured at TLJ, as they were not
reported by Takami and Okabe (2007) [B39].
The definition of the lightning parameters listed in Table 9-1 and Table 9-2 is provided in Table 9-3.
While the use of the log-normal distribution for stroke current magnitudes is widely implemented [9] [5]
[31], some users prefer the use of a log-logistic distribution for lightning performance calculations of
transmission lines[55][47][46]. The log-logistic distribution was first proposed by Anderson [2] as a
simplification to facilitate the calculation of lightning current probabilities for the EPRI simplified method for
hand or programmable calculators. The same distribution was later adopted by the IEEE [3]. The
cumulative distribution function is in the form:
1
P(𝐼) = [Flashovers per year]
𝐼 𝛼 Equation 9-1
1+( )
𝑀
Where
α a shape parameter which selected to fit the log-normal distribution function.
M the geometric mean of the stroke currents

The log-logistic distribution function approximates the log-normal distribution well except at the extremes,
which is acceptable for calculating the lightning performance of transmission lines. For substation
insulation coordination studies, however, the use of the log-normal distribution function is preferred [7].
The log-logistic equations as applied by the IEEE and those derived from the MCS dataset are provided in
Table 9-4.

84
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

Table 9-1 – Parameters of the first stroke current of negative downward flashes.
Approximation by log-normal Percent of cases exceeding tabulated
Sample distribution values
Parameter Local
size
Geometric mean * 95 % 50 % 5%
MSS 75 27.7 0.46 12.9 27.7 59.5
II TLJ 120 27.7 0.64 9.8 27.7 78.7
[kA]
MCS 51 37.6 0.46 17.7 37.6 80.2
MSS 80 31.1 0.48 14.1 31.1 68.5
IF TLJ 120 29.3 0.64 10.1 29.3 84.9
[kA]
MCS 51 43.3 0.47 20 43.3 93.6
MSS 80 5.63 0.58 2.25 5.63 14.125
td 10/90 TLJ 120 4.8 0.39 2.5 4.8 9.0
[µs]
MCS 51 6.4 0.41 3.3 6.4 12.5
MSS 80 3.83 0.55 1.5 3.83 9.67
td 30/90 TLJ 120 3.2 0.41 1.6 3.2 6.2
[µs]
MCS 51 4.2 0.50 1.9 4.2 9.6
MSS 75 24.3 0.60 9.1 24.3 65
Sm TLJ 120 18.9 0.60 7.0 18.9 51.2
[kA/µs]
MCS 51 20.8 0.3 12.7 20.8 34.1
MSS 90 75 - 30 75 200
th TLJ 120 36.5 0.81 9.5 36.5 139.7
[µs]
MCS 51 56.2 0.68 18.4 56.2 171.5
MSS 90 4.5 - 1.1 4.5 20
QI TLJ - - - - -
[C]
MCS 51 5.7 0.63 2 5.7 16.1
MSS 91 5.5 - 0.6 5.5 55
Specific Energy
TLJ - - - - - -
[A2.s] × 104
MCS 51 9.79 1.19 1.4 9.79 68.95
MSS: Mount San Salvatore [53] [52]; MCS: Morro do Cachimbo Station [55]; TLJ: Transmission Line in Japan [56].

Table 9-2 – Parameters of the subsequent stroke current of negative downward flashes.
Approximation by log-normal Percent of cases exceeding tabulated
Sample distribution values
Parameter Local
size
Geometric mean * 95 % 50 % 5%
MSS 114 12.3 0.53 5.2 12.3 29.2
IF TLJ - - - - - -
[kA]
MCS 77 17.3 0.54 7.1 17.3 42
MSS 114 0.75 0.92 0.125 0.75 3.5
td 10/90 TLJ - - - - - -
[µs]
MCS 77 0.82 0.75 0.24 0.82 2.81
MSS 114 0.67 1.01 0.17 0.67 3
td 30/90 TLJ - - - - - -
[µs]
MCS 77 0.62 0.7 0.2 0.62 1.955
MSS 113 39.9 0.85 9.9 39.9 161.5
Sm TLJ - - - - - -
[kA/µs]
MCS 77 35.1 0.7 11.1 35.1 111
MSS 115 32 6.5 32 140
th TLJ - - - - - -
[µs]
MCS 76 15.1 1.28 1.84 15.1 123.9
MSS 117 0.95 0.22 0.95 4.0
QI TLJ - - - - - -
[C]
MCS 77 1.28 1.21 0.175 1.28 9.37
MSS 88 0.6 - 0.055 0.6 5.2
Specific Energy
TLJ - - - - - -
[A2.s] × 104
MCS 77 0.707 1.58 0.053 0.707 9.51
MSS: Mount San Salvatore [53] [52]; MCS: Morro do Cachimbo Station [55]; TLJ: Transmission Line in Japan [56].

85
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

Table 9-3 – Definition of front parameters for a lightning current impulse of negative polarity.

Parameter Description
IF Final (overall) peak of current (i.e. peak current)
I100 = II Initial peak of current
I10 10 % intercept along the stroke current waveshape
I30 30 % intercept along the stroke current waveshape
I90 90 % intercept along the stroke current waveshape
T10/90 Time between I10 and I90 intercepts on the wavefront
T30/90 Time between I30 and I90 intercepts on the wavefront
Sm Maximum rate-of-rise of current along wave-front, typically at I90
td 10/90 Equivalent linear wave-front duration derived as T10/90 / 0.8
td 30/90 Equivalent linear wave-front duration derived as T30/90 / 0.6
th Tail time to half value
QI Impulse charge – time integral of current
W Specific energy (action integral) – time integral of the square current

Table 9-4 – The log-logistic approximations of the cumulative statistical distributions of return stroke peak
currents.
Negative first stroke Negative subsequent stroke
1 1
IEEE [31] 𝑃(𝐼) = 2.6 𝑃(𝐼) = 2.7
1 + (𝐼⁄31) 1 + (𝐼⁄12)
1 1
MCS [55] 𝑃(𝐼) = 3.8 𝑃(𝐼) = 3.2
1 + (𝐼⁄43.3) 1 + (𝐼⁄17.3)

86
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

10 Analytical representation of the current shape


10.1 General
In accurate calculations of the lightning performance of equipment it may be necessary to simulate the
concave front in the lightning stroke current representation. The fundamental requirement in such a
simulation is that it provides:

 the correct amplitude of the current;


 the highest steepness close to the peak amplitude (taken here as at I90);

 for first strokes the correct average steepness expressed as the front time passing through the
30 % and the 90 % values of the current. This front time must be larger than the current amplitude
divided by the maximum steepness, thus resulting in the concave shape. For subsequent strokes
this parameter may be neglected.
Many mathematical expressions may be suitable to fulfil these requirements and the one given here is only
one proposal. Its disadvantage is that the current front and the current tail are not described by a single
expression, but are separated into two parts, one describing the front up to 90 % of the amplitude, the
other, the amplitude range on the tail.

10.2 CIGRE concave wave front

Ip : Impulse amplitude
Sm: Maximum rate of rise
t f: Equivalent front duration

Figure 10-1 - Analytically derived lightning current impulse, showing the front parameters.

10.2.1 The current front


The current front of the first strokes can be expressed as:

𝐼 = 𝐴𝑡 + 𝐵𝑡 𝑛 Equation 10-1
The basic assumption is that the current shape reaches the instant of maximum steepness (90 %
amplitude) at a time tn dependent on the exponent n. In principle, both variables have to be evaluated by
an iterative solution of the generalised equation.
3𝑥 𝑥(𝑛−1) 3𝑥𝑛
(1 − ) (1 − 𝑥)𝑛 = + (1 − ) (1 − 𝑥) Equation 10-2
2𝑆𝑁 2𝑆𝑁 2𝑆𝑁

with
𝑆𝑚 𝑡𝑓 0.6𝑡𝑓
𝑆𝑁 = 𝑋𝑁 =
𝐼 𝑡𝑛

I current amplitude
Sm maximum steepness
tf front time (td30)

87
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

However, a sufficiently accurate solution is given by:


1
𝑛 = 1 + 2(𝑆𝑁 − 1) (2 + ) Equation 10-3
𝑆𝑁
and
3𝑆𝑁2
𝑡𝑛 = 0.6𝑡𝑓 [ ]
1 + 𝑆𝑁2
The constants then are:
1 𝐼
𝐴=
(0.9 ∙ 𝑛 − 𝑆𝑀 )
𝑛−1 𝑡𝑛
1
𝐵= 𝑛 (𝑆 𝑡 − 0.9𝐼)
𝑡𝑛 (𝑛 − 1) 𝑀 𝑛
For subsequent strokes the current front is given by:

𝐼 = 𝑆𝑀 𝑡𝑓 Equation 10-4

10.2.2 The current tail


The fundamental requirements for the current tail are:

 to have the maximum steepness at its beginning thus providing a steady transition from one part to
the other
 to reach the correct amplitude value
 to describe the current tail. In order to avoid iterative procedures, the last requirement is not
mathematically correctly fulfilled in the following proposal, but absolutely sufficient for practical
purposes.
A suitable expression is:
𝑡−𝑡𝑛 𝑡−𝑡𝑛
−[ ] −[ ] Equation 10-5
𝐼 = 𝐼1 𝑒 𝑡1 − 𝐼2 𝑒 𝑡2

with the constants as follows:


𝑡ℎ −𝑡𝑛 𝐼
𝑡1 = 𝑡2 = 0.1
𝑙𝑛(2) 𝑆𝑀
𝑡1 𝑡𝑠 0.9 𝑡1 𝑡𝑠 0.9
𝐼1 = [𝑆𝑀 + ] 𝐼2 = [𝑆𝑀 + ]
𝑡1 −𝑡2 𝑡2 𝑡1 −𝑡2 𝑡1

t1, t2: time constants


I1, I2: constants
Th: time-to-half value
10.2.3 Example
The average values of a first stroke are:
I = 31 kA
Sm = 26 kA/µs
tf = 3 µs
th = 75 µs
The constants derived for this current is presented in Table 10-1.
Table 10-1 – First stroke current constants.
Parameter Front Parameter Tail
n 8.29 -- T1 105 µs
A 3.24 kA/µs T2 0.12 µs
B 3.65 x 10-5 kA/µs I1 31 kA
tn 4.67 µs I2 3.1 kA

The resulting current shape is shown in Figure 10-1.

88
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

10.3 The double peaked wave


An analytical representation is given for the first stroke of a double peaked lightning current waveshape
[147]. The method sums 7 Heidler functions to approximate the waveshape as measured at short
instrument towers. The parameters that characterize the waveshape are given in the figure below.

Figure 10-2 – Parameters used to characterize return-stroke current, showing the front parameters.

10.3.1 Mathematical description


To create a progressive build up four waveforms with increasingly higher front times and gradually steeper
rates of rise shall be applied. Then the maximum steepness ( Sm) and first peak value (Ip1) can be
approximated by adding the fifth function. The tail time (T50) and second peak (Ip2) are determined by
adding the sixth and seventh function.
𝑚
Equation 10-6
𝑖(𝑡) = ∑(𝐼0𝑘 /𝜂𝑘 )𝑒𝑥𝑝(−𝑡/𝜏2𝑘 ){(𝑡/𝜏1𝑘 )𝑛𝑘 /[1 + (𝑡/𝜏1𝑘 )𝑛𝑘 ]}
𝑘=1

With:
𝜂𝑘 = 𝑒𝑥𝑝[−(𝜏1𝑘 /𝜏2𝑘 )(𝑛𝑘 𝑡2𝑘 /𝑡1𝑘 )1/𝑛𝑘 ] Equation 10-7

Table 10-2 – Parameters for approximation of first stroke currents measured in two different datasets.
Dataset 1 Dataset 2
k 𝐼0𝑘 𝜏1𝑘 𝜏2𝑘 𝐼0𝑘 𝜏1𝑘 𝜏2𝑘
𝜂𝑘 𝜂𝑘
(kA) (µs) (µs) (kA) (µs) (µs)
1 6 2 3 76 3 2 3 76
2 5 3 3.5 10 4.5 3 3.5 25
3 5 5 4.8 30 3 5 5.2 20
4 8 9 6 26 3.8 7 6 60
5 16.5 30 7 23.2 13.6 44 6.6 60
6 17 2 70 200 11 2 100 600
7 12 14 12 26 5.7 15 11.7 48.5

Applying these values leads to a current wave shown in Figure 10-2 and the following comparison of
median (measured) parameters and calculated values.
Table 10-3 – Comparison of measured parameters for two different datasets.
Ip1 Ip2 T10 T30 T50 S10 S30 Sm
Dataset Data
(kA) (kA) (µs) (µs) (µs) (kA/µs) (kA/µs) (kA/µs)
Measured 40.4 45.3 5.6 2.9 53.5 5.8 8.4 19.4
1
Calculated 40.1 45.3 5.2 3.0 53.8 6.2 8.0 20.2
Measured 27.7 31.1 4.5 2.3 75.0 5.0 7.2 24.3
2
Calculated 27.8 31.0 4.6 2.3 75.0 4.8 7.2 24.4

89
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

10.4 Subsequent strokes


10.4.1 Mathematical description
The synthesis of typical subsequent stroke currents is simpler than that of first stroke currents, mostly
because they present a much less pronounced concavity at the front. Only two Heidler functions are
required to construct a waveform reproducing median characteristics of subsequent stroke currents [147].

a b

Figure 10-3 – Normalized double peaked waveshape current for two datasets.

Table 10-4 – Parameters used to synthesize the currents.


Dataset 1 Dataset 2
k 𝐼0𝑘 𝜏1𝑘 𝜏2𝑘 𝐼0𝑘 𝜏1𝑘 𝜏2𝑘
𝜂𝑘 𝜂𝑘
(kA) (µs) (µs) (kA) (µs) (µs)
1 10.7 2 0.25 2.5 15.4 3.4 0.6 4.0
2 6.5 2 2.1 230 7.2 2 4.0 120

10.5 Application of the waveshape


In studies the user might want to apply a scalable format of the waveshape, especially with regard to the
amplitude and steepness. For this the values from Table 9-1 can be modified in the following ways.

10.5.1 Pre-defined scaling


The proposed methodology is based on applying adequate multipliers to the base values that, varying
between certain limits, allow to modify the characteristics of the waveform represented [148]. The values of
the parameters characterizing the seven Heidler functions that fit each lightning current are organized as
arrays:
𝐼0𝑘 = α [6 5 5 8 16.5 17 12 × 𝛿] kA Equation 10-8
𝜏1𝑘 = 𝜷 [3 3.5 4.8 6 7 70 12] µs Equation 10-9
𝜏2𝑘 = 𝜷 [76 10 30 26 23.2 200 26] µs Equation 10-10
𝜂𝑘 = [2 3 5 9 30 2 × 𝜸 14] µs Equation 10-11
Where α, β, γ and δ are the established multipliers. It is noted that multipliers α and β apply to the
respective parameters of the seven Heidler functions, while multiplier δ applies only to the respective
parameter of the 7th function and γ to the 6th.
The multiplier α is responsible for lightning current amplitude adjustment. It is used for setting the first peak
Ip1 and contributes as adjunct to the adjustment of the second peak Ip2. Once α is established the
adjustment of δ is made, which is responsible for fine tuning Ip2, starting from the value previously
established by α. This creates the possibility of changing the values of Ip1 and Ip2 separately.
The multiplier β is intended to act on the wavefront providing the variation of T 30. However, it is used for
both τ1k, and τ2k, although τ2k being the time constant associated with the decay of i(t). This is due to the
fact that the simultaneous multiplication of τ1k and τ2k by β is necessary so that the amplitudes Ip1 and Ip2
are not changed. Therefore, another multiplier was investigated so to adjust the decay time.
The multiplier γ changes the lightning current decay after the occurrence of I p2, thus modifying T50. This
adjustment is indirect, since ηk is not directly related to the decay of i(t). Although the isolated variation of

90
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

τ2k is intuitive and acts directly on T 50, its use is not convenient due to observed influence on the values of
Ip1 and Ip2. The adjustment by ηk also presents the same disadvantage, however, with lower intensity, and
may be useful in situations where the lightning current decay time plays an important role.

10.5.2 Self-correcting routine


As mentioned in the previous section the method using pre-defined scaling has a few limitations, where a
change of parameters is used to change one of the waveshape characteristics, but also has an influence
on other characteristics. To enable some more flexibility to the user there is the option to use a self-
correcting routine, which adjusts the parameters until the current has the required waveshape.
To be able to do this, the user shall provide input parameters to the routine. One option is to use the
characterizing parameters Ip1, Ip2, S10, S30, T10, T30, T50 and Sm, as shown in Figure 10-2. However, these
values are not very convenient for the use in studies. An alternative option is to define the peaks (with Ip1,
Ip2, tp1 and tp2), the maximum steepness (sm) and the half-time of the tail (T50). This last option is easier
since the user is then able to neglect the first part of the current waveshape (between t=0 and the steep
front). Of course the parameters S10, S30, T10, T30 can be determined at any time for comparison.
The following steps are of importance in the self-correcting routine, also refer to the flowchart in Figure
10-5. In this routine also factor γ is applied to adjust the decay after the second peak, however it is used on
the steepness factor for the 7th Heidler function.
1. The normalized current waveshape shall be scaled based on the time and amplitude of the second
peak. This is comparable to the approach in the previous section: multiplication of parameters with
α and β. A factor γ is applied for the half-time of the tail.
2. A sub-routine is necessary to retrieve the actual characteristics (I p1, Ip2, tp1, tp2, Sm and T50) of the
waveshape. Depending on the approach it is recommended to use the normalized waveshape for
dataset MCS, refer to Figure 10-4, since this waveshape has more distinctive peaks.

Figure 10-4 – Normalized double peaked waveshape current for two datasets.

3. The retrieved parameters of the first peak amplitude (Ip1), first peak time (tp1) and maximum
steepness (Sm) are compared to the required values. The input parameters of the Heidler functions
are corrected with a factor equal to the difference, where σ is used for the first peak amplitude, ω
is used for the first peak time and φ for the maximum steepness.
4. Since these changes affect the complete waveshape the sub routine is called again.
5. The retrieved parameters of the second peak amplitude (Ip2) are compared to the required values.
The input parameter of the Heidler function is corrected with a factor equal to the difference, where
λ is used for the second peak amplitude.
Steps 2 to 5 can be repeated several times until the accuracy seems acceptable.

𝐼0𝑘 = 𝛂 [𝐼1 𝐼2 𝐼3 𝐼4 𝐼5 × 𝛌 𝐼6 𝐼7 × 𝝈] kA Equation 10-12

91
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

𝜏1𝑘 = 𝜷 [𝜏11 𝜏12 𝜏13 𝜏14 𝜏15 × 𝝎 𝜏16 𝜏17 ] µs Equation 10-13
𝜏2𝑘 = 𝜷 [𝜏21 𝜏22 𝜏23 𝜏24 𝜏25 × 𝝎 𝜏26 𝜏27 × 𝒙] µs Equation 10-14
𝜂𝑘 = [𝜂1 𝜂2 𝜂3 𝜂4 𝜂5 𝜂6 𝜂7 × 𝜸 ] µs Equation 10-15

a b

Ipk2 Tpk2 T50 Ipk2 Tpk2 Sm

Initial
waveshape c d

Retrieve
shape points

Adjust
Ip1, tp1, sm

Repeat
Retrieve
shape points e f

Adjust
Ip2

Final
waveshape

Figure 10-5 – Flowchart. Figure 10-6 – Output with variation of input parameters.

92
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

11 Stroke attraction methods


11.1 EGM in Japan (TEPCO)
The improvements concern the striking distance and the distribution of lightning current waveforms. The
characteristics based on test data recently obtained by complementing the large gap region and also the
velocity distribution obtained by actual measurements using streak cameras were applied to the striking
distance. Regarding the distribution of lightning current waveforms, the distribution of lightning current
peak values introduced by the American Institute of Electrical Engineers (AIEE) report in 1950 has been
conventionally used [80]. The present study adopts data recently obtained by observations of lightning and
applied the distribution of lightning current peak values in consideration of the distribution of front duration
[81].
11.1.1 Overview of calculating lightning strokes to transmission lines
Figure 11-1 is a conceptual diagram showing the calculation of lightning strokes to transmission lines. The
respective striking distances (absorption radii) of the ground wire, power line, and ground are calculated in
order to calculate the lightning stroke rates of lightning that has approached within the respective striking
distances. For example, in Figure 11-1, lightning that has approached within an arc connecting the
intersection points a and b strikes the ground wire (GW) and lightning that has approached within an arc
connecting the intersection points b and c strikes the upper phase power line (C1).

rg

rc

rc

rc

re

Figure 11-1 – Conceptual diagram of lightning stroke to transmission lines based on the EGM.
Since the striking distance differs depending on the lightning current peak value, the distribution of
lightning current peak values is taken into consideration. In addition, the distribution of lightning stroke
angles is taken into consideration within the range ±90° with a vertical downward approach as standard.

11.1.2 Calculation of lightning stroke rate and lightning outage rate


The basic flowchart is presented in Figure 11-2 for the calculation of the lightning outage rate of
transmission lines, as well as the lightning stroke rate and other variables whose values are required as
part of the calculation [82] [83].
First, the lightning stroke rate to transmission lines is calculated based with the electro-geometric method
(EGM). The new features include considering the lightning stroke current waveform distribution and
revising the lightning stroke distance in consideration of the return stroke velocity distribution.

93
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

Lightning stroke in the area of the transmission lines (flashes / 100 km / year)

Probability density of stroke current


Calculation of lightning stroke rate based on
Electric Geometrical Model Distribution of return stroke velocity
(The striking distance)
*Power lines : rc Angle probability density
*Ground wires : rg = 1.0 × rc F(α) = 0.75 × cos3α
*Ground : re = 1.1 × rc
Lightning stroke ratio between tower and span
Tower : Span = 1 : 4

Stroke to tower Stroke to span

1 / 5 Ratio of stroke 4/5 Striking rate


calculation
Calculation of the stroke Calculation of the stroke Calculation of the stroke
probability to tower probability to ground wire probability to conductor

· Calculation of the tower surge · Calculation of the tower surge · Calculation of the ground wire
impedance ZT by stroke to impedance ZST by stroke to surge impedance ZS by stroke
tower span to span Failure rate
· Calculation of the coupling · Calculation of the coupling · Calculation of the coupling calculation
factor “C” factor “C” factor “C”
· Calculation of the potential · Calculation of the potential
distribution coefficient distribution coefficient

Non-F.O Non-F.O.
Judgment of 1F.O. . Judgment of F.O. Judgment of F.O.
(1) Non-F.O (2) (3)
.
F.O. F.O. F.O.

Judgment Probability of
F.O.
of simultaneous F.O. the shielding
(4) failure (1LG)
Non-F.O.

· Calculation of the tower surge impedance Z’T, Z’ST after 1LG


Probability of the back
· Calculation of the coupling factor “C” after 1LG flashover on span (1LG)

Non-F.O
Judgment of 2F.O. .
(1) Probability of the back
flashover on tower (1LG)
F.O
.
· Calculation of the tower surge impedance Z’’T, Z’’ST after 2LG
· Calculation of the coupling factor “C” after 2LG

F.O.
Probability of the back
flashover on tower (2LG)
Judgment of 3F.O. Non-F.O.
(1)

F.O
· Calculation of . ST after 3LG
Z’’’T, C’’’

F.O Probability of the back


. flashover on tower (3LG)
Judgment of 4F.O.
(1) Non-F.O
Probability of the
.
back flashover on F.O.
tower (over 4LG)
Over 4LG 3LG 2LG 1LG

For Lightning outage rate

space
Figure 11-2 – Flowchart to calculate the lightning outage rate of transmission lines.
(Notes 1 to 4 and explanation of symbols follow on next page).

94
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

(1) Backflashover on tower k, k’ : wave correction factor


𝑘 a : atmospheric correction factor
(𝑘𝑖 − 𝐶𝑖 ) ∙ 𝑍𝑇 ∙ 𝐼 > ( ) ∙ 𝑉50 − 𝐸 ∙ 𝑠𝑖𝑛(𝜔𝑡 + 𝜙)
𝑎 V50: 50% flashover voltage (rod-rod gap)
(2) Back flashover on span V’50: 50% flashover voltage (parallel conductor gap)
𝑘 ′ 𝐸 ∙ 𝑠𝑖𝑛(𝜔𝑡 + 𝜃): AC voltage to ground
𝜂𝑔 ∙ (1 − 𝐶2 ) ∙ 𝑍𝑆 ∙ 𝐼 > ( ) ∙ 𝑉50 − 𝐸 ∙ 𝑠𝑖𝑛(𝜔𝑡 + 𝜙) Ki : factor of tower potential distribution on i phase
𝑎
(3) Flashover of lightning to conductor Ci : coupling factor between ground wire and i phase
𝑍0 ∙ 𝑍𝑐 𝑘′ C2 : coupling factor between ground wire and upper
𝜂𝑐 ∙ ∙ 𝐼 > ( ) ∙ 𝑉50 − 𝐸 ∙ 𝑠𝑖𝑛(𝜔𝑡 + 𝜙) phase
𝑍𝑐 + 2 ∙ 𝑍0 𝑎
(4) Judgment of simultaneous F.O ηg: factor of corona attenuation on ground wire (0.7)
ηc: factor of corona attenuation on conductor (1.0)

Next, for the lightning stroke probability to each conductor, the probability of a flashover across the air gap
insulation of the transmission lines (mainly between arcing horns) is calculated. Here, the chance of a
flashover occurring is judged and determined based on a comparison of the increased potential caused by
a lightning stroke, also taking account of the AC phase of phase conductors, with the air gap withstand
voltage between arcing horns. Subsequently, after a single line-to-ground (1-LG) event, whether flashover
will occur on other phases is judged in consideration of the potential decrease rate after the ground fault
and thus calculation is continued and repeated for 2-LG or a ground fault on additional phases.
The chance of a flashover occurring between conductors (e.g. ground wire – upper phase conductor) is
also judged for lightning strokes to ground wires in mid-span. Whether a flashover caused by a lightning
stroke will occur between arcing horns or between the ground wire and the upper phase conductor is
predicted by calculation of the overvoltage across each insulation gap, based on distributed constant
circuit theory. Simultaneous flashover is now also taken into consideration. The correction factors for
estimating the flashover rate to the ground wire and to the ground have also been revised.

11.1.3 Comparison of calculation results with observations


The modified EGM has been verified against observed line performances of the Japanese Transmission
grid [84]. The results are summarised here.

Calculations of lightning strokes to ground wires


Figure 11-3 shows the representative conductor arrangements of UHV designed and 500 kV transmission
lines used for the calculation. The calculation results of the “Lightning Stroke to Ground Wires” are
compared with actual observations in Figure 11-4. The lightning stroke density NL = 3.0 (strokes / km2 /
year) is a reduced value based on the isokeraunic level (IKL) value (in the design stage); and the values NL
= 6.7 and NL = 4.4 are statistical values based on Lightning Location System (LLS) data during the period
of observation along each of the transmission lines. The lightning stroke rates calculated are larger than
those by the conventional technique and are closer to the observed rates.
UHV designed transmission Line

Ground Wire
38 m OPGW500 mm2  1
32 m
500 kV transmission Line

31 m Ground Wire
19 m 22.6 m OPGW290 mm2  1
Phase Conductor 28 m
ACSR410 mm2  8
32 m
19 m
11 m 16.0 m
Phase Conductor
33 m ACSR610 mm2  4
16.8 m
11 m
17.6 m
70 m
50 m

Figure 11-3 - Typical conductor arrangements of large-scale transmission lines.

95
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

Lightning stroke rate (flash/100km/year)/

Lightning stroke rate (flash/100km/year)/


300 300
Lightning stroke to tower Lightning stroke to tower
250 Lightning stroke to ground wire 250 Lightning stroke to ground wire

200 200

150 150

100 100

50 50

0 0
Observation Calculation Observation Calculation
(NL=6.7) (NL=6.7) (NL=4.4) (NL=4.4)

(a) UHV designed transmission lines (b) 500-kV transmission lines

Figure 11-4 - Lightning stroke rate to ground wires -calculations and observations.

Calculations on lightning strokes to power lines


The calculated predictions of the present method developed are compared with actual observations
regarding the “Lightning Stroke to Power Lines” in Figure 11-5. The present method produces predictions
closer to the actual observations, e.g. showing an increase in the lightning stroke rate to the upper phase,
while the conventional method did not match the observations, particularly in the distribution of strokes by
phase conductor position.
Lightning stroke rate (flash/100km/year)/

5
Lightning stroke rate (flash/100km/year)/

3
Lightning stroke to upper phase Lightning stroke to upper phase
Lightning stroke to middle phase Lightning stroke to middle phase
4 Lightning stroke to lower phase Lightning stroke to lower phase
2
3

2
1
1

0 0
Observation Calculation Observation Calculation
(NL=5.2) (NL=5.2) (NL=4.9) (NL=4.9)

(a) UHV designed transmission lines (b) 500-kV transmission lines


Figure 11-5 - Lightning stroke rate to power lines -calculations and observations.

Calculations on lightning outage rate


For the lightning outage rate, the predictions of the present technique developed are compared with actual
observations in Figure 3.6. The representative conductor arrangements used for the calculation are shown
in Figure 11-3.
Lightning outage rate of transmission line.

Lightning outage rate of transmission line.

2 2 Lightning stroke to upper phase


Lightning stroke to upper phase Lightning stroke to middle phase
Lightning stroke to middle phase Lightning stroke to lower phase
Back-flashover (1LG)
Lightning stroke to lower phase
(cases / 100km / year)

(cases / 100km / year)

1.5 1.5 Back-flashover (2LG)


Back-flashover (1LG)
Back-flashover (2LG)
1 1

0.5
0.5

0
0 Observation Calculation
Observation Calculation
(NL = 4.9) (NL = 4.9)
(NL = 4.7) (NL = 4.7)

(a) UHV designed transmission lines (b) 500-kV transmission lines

Figure 11-6 - Lightning outage rate -calculations and observations.

96
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

Conventional calculations tended to underestimate the actual outage rate in both UHV designed and
500 kV transmission lines. The difference in UHV designed transmission lines is caused by lightning
strokes to the phase conductor, where the prediction is lower than the actuality, as well as the differing
aspect of lightning outages, e.g. fewer predicted outages at upper phase conductors.
In contrast, the present technique presents predictions closer to the observed facts of lightning stroke rates
to transmission lines. Furthermore, the lightning outage rate, which is calculated based on the lightning
stroke rate, is also closer to the reality in terms of both the total number of outages and the occurrence of
outages on different phases.

11.2 Rizk simplified lightning attachment model


The objective of the Rizk simplified lightning attachment model [90] is to calculate the lateral attractive
distance of each conductor of the transmission line on flat ground as a function of the prospective lightning
stroke current and the instantaneous power-frequency voltage. This concept is illustrated in Figure 11-7.
The steps of this simplified method are explained both in flow-chart format and in a table format, where
each equation is briefly described. This Appendix closes with a worked example and a few verification
benchmarks.

Figure 11-7 – Lightning attractive distance to overhead transmission line conductors as defined in the
simplified Rizk lightning attachment model.
Symbols used in the simplified approach include:
Usp space potential
Usp0 space potential in absence of other conductors
Ulc continuous upward leader onset space potential
Ulc0 corresponding value for individual free conductor
ksp space potential proximity factor
ksp0 proximity factor in charge cloud field
qs quasi-stationary induced charge per unit length
R geometric parameter for determination of Ulc
Eg electric field due to cloud charge before negative leader descent
Egm maximum value of Eg
Uci corona onset space potential
Ui magnitude of the space potential at the conductor position due to downward negative leader
charge
Udc magnitude of the space potential due to cloud charge

97
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

Upf power frequency voltage level of the conductor, assumed constant within the short time
interval under consideration
kspi a proximity factor due to the presence of other conductors considering Ui, Udc and Upf
Two important input parameters are:
Hcl height of cloud base
Eg Background electric field at ground level prior to negative leader development

Note on the choice of Hcl:


As mentioned in reference [85] the cloud base height is usually taken between 2 and 3 km.
In this model Hcl =2500m.

Notes on the choice of Eg:


Based on available measurements of Eg [91] is suggested that the ambient ground field is a
statistical variable that can be represented by an exponential distribution of the form:
𝐸𝑔 −𝐸𝑔0
[− ]
𝐸𝑔1
𝑃(𝐸𝑔 ) = 𝑒
Where:
Eg0 a minimum value and Eg1 a shape parameter:
o For flat ground, particularly in bushy regions, Eg0 and Eg1 can be assumed as 2 kV/m. For such
regions, the probability that Eg exceeds 10 kV/m would be approximately 2 %.
o In regions of high altitude, hilly areas, and for winter lightning in Japan, representative values
are Eg0 = 3 kV/m and Eg1 = 6 kV/m. In such regions the probability that Eg exceeds 20 kV/m is
approximately 6 %.
o The choice is left to the user depending on his environment and hopefully some local
measurements of the ambient ground field in lightning storm conditions.

11.2.1 Determination of attractive distance of each conductor Lds0


The lateral attractive distance of each conductor is calculated as follows:

Figure 11-8: Calculation of Lds.

98
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

The calculation of the correction factors Fpr, Feg and Fov is explained in the sections that follow. It should be
noted that the correction factors only apply to systems of 345 kV and above, for which it is assumed that
the upward leaders play a significant role in the attachment process.

In the present approach the ratio Lds/Lds0, comprising a


dimensionless variable, is determined as the product of three
A 𝐿𝑑𝑠 = 𝐿𝑑𝑠0 (𝐾𝑐𝑜𝑟 ∙ 𝐹𝑝𝑟 ∙ 𝐹𝑒𝑔 ∙ 𝐹𝑜𝑣 ) dimensionless functions Fpr, Feg and Fov which account for the
effects of proximity, cloud charge field and operating voltage
respectively.
In addition, it was found that due to the simultaneous effects of
large conductor bundle height and high operating voltage, a
1 𝐾𝑐𝑜𝑟 = 1 + (𝐹𝑒𝑔 − 1) ∙ (𝐹𝑜𝑣 − 1)
second order correction factor Kcor should be introduced.
This correction does not apply to ground wires.
2 Fpr Refer to F.
3 Feg Refer to G and H.
4 Fov Refer to I and J.
The base lateral attractive distance Lds0 of a single conductor
varies primarily with the prospective return stroke current Io
and conductor height h, through a regression formula of the
form:
𝐿𝑑𝑠0 = 𝐴𝐼0 𝛼 ℎ𝛽 (m, kA, m)
Where in general the constants A, α, B vary with the ranges of
the variables Io and h. For the same values of Io and h, Lds0
varies only slightly with the equivalent radius re.

Lds0
5 hc

Here the effects on the lateral attractive distance of mutual


conductor proximity, ground field due to cloud charges prior to
negative leader descent and the line operating voltage are
neglected. For overhead lines it is proposed to have some
correction factors applied to take these aspects into account.

In an overhead line the lateral attractive distance of the


conductors cannot directly be determined with the approach as
Lds
mentioned above. It is proposed to determine the lateral
6
distance per single conductor Lds0 and apply correction factors
hc
to take several influences into account and come to the value
Lds for conductors in an overhead line.

99
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

11.2.2 Calculation of internal variables: ksp0


The space proximity factor is calculated as follows:

Figure 11-9 – Calculation of ksp0.

𝑞𝑠
B 𝑘𝑠𝑝0 = With qs and qs0 from formula's below.
𝑞𝑠0

𝑞𝑠 = 𝐴−1 ∙ 𝑉

or in matrix form:

𝑞𝑠1 𝐴11 ⋯ 𝐴1𝑛 −1 𝑉1


[ ⋮ ]=[ ⋮ ⋱ ⋮ ] ∙[⋮] Calculate the induced charge qs per conductor from
𝑞𝑠𝑛 𝐴𝑛1 ⋯ 𝐴𝑛𝑛 𝑉𝑛 the induced voltage (V) and the inverted potential
1 or coefficient matrix A-1, with dimension F/m. The
𝑞𝑠1 𝐶11 ⋯ 𝐶1𝑛 𝑉1 inverted potential coefficient matrix is the capacitance
[ ⋮ ]=[ ⋮ ⋱ ⋮ ]∙[ ⋮ ] matrix of the line.
𝑞𝑠𝑛 𝐶𝑛1 ⋯ 𝐶𝑛𝑛 𝑉𝑛

where
𝑉𝑛 = 𝐸𝑔 ∙ ℎ𝑐_𝑛

The induced charge for a single conductor with


Eg = ground field prior to negative leader decent
2 𝑞𝑠0 = 𝐸𝑔 ∙ ℎ𝑐 ∙ 𝐶
(e.g. 10 kV/m)
hc = height of conductor
2𝜋𝜖0 Capacitance of a conductor, with input:
3 𝐶= 2ℎ𝑐 hc = height of conductor (m)
𝑙𝑛( )
𝑟𝑒
re = equivalent radius of the conductor(bundle) (m)

100
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

11.2.3 Calculation of internal variables: Ulc0


The continuous upward leader onset space potential for each individual free conductor is calculated as
follows:

Figure 11-10 – Calculation of Ulc0.

2247
The basic positive leader onset potential Ulco of a single conductor
𝑈𝑙𝑐𝑜 = 5.15−5.49∙𝑙𝑛(𝑟𝑒 ) above ground can be calculated based on this formula, with input:
C 1+ 2ℎ
ℎ𝑐 ∙𝑙𝑛( 𝑟 𝑐 ) hc = height of conductor (m)
𝑒
re = equivalent radius of the conductor(bundle) (m)

11.2.4 Calculation of internal variables: kl


The positive leader onset proximity factor kl is obtained from the lookup table shown below. Background on
how it is calculated is also provided.

Figure 11-11 – Calculation of ki.

It is found that contrary to ksp0, the values of kl vary within relatively


narrow limits.
This allows the recommended values of kl in the table making in
𝑈𝑙𝑐
D 𝑘𝑙 = general charge simulation of line configuration unnecessary.
𝑈𝑙𝑐0

2247
𝑈𝑙𝑐 = 10.30−10.98∙𝑙𝑛(𝑟𝑒 ) The positive leader onset space potential for configurations other than
1 1+ 2ℎ
𝑅∙𝑙𝑛( 𝑟 𝑐 ) a single conductor above ground is expressed by Ulc.
𝑒
The parameter R can be found by placing an arbitrary charge Q o at the
conductor position, with the conductor in question removed, and
determining the induced potential due to the ground plane and the
2 𝑅 is not calculated
other grounded conductors in the vicinity of the position of Q0. With
proximity effects, it is found that R < 2h, so that Ulc
becomes lower than Ulc0.
3 𝑈𝑙𝑐0 Refer to C.

101
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

11.2.5 Calculation of internal variables: U0


The leader tip potential at the height of the conductor is calculated as follows:

Figure 11-12 – Calculation of U0.

The negative leader tip potential U0 at the level


of the conductor height above ground
𝑞0 ×103 4ℎ𝑐 2ℎ𝑐 𝐻𝑐𝑙
E 𝑈0 (ℎ𝑐 ) = {𝑙𝑛 ( )− [𝑙𝑛 ( ) + 1]} determines the final jump by streamer
111.26 𝑟𝑠 𝐻𝑐𝑙 √𝑟𝑠 ℎ𝑐
breakdown between positive and negative
leader tips.
2𝑄𝑓𝑝 The charge q0 is based on the charge of the first
1 𝑞0 =
𝐻𝑐𝑙 peak and a cloud base height.
The charge of the first peak is based on the
2 𝑄𝑓𝑝 = 76 × 103 ∙ 𝐼00.68
prospective return stroke current peak (I0)
𝑞0 ℎ𝑐
The radius of the space charge region (rs)
3 𝑟𝑠 = (1 − ) around the negative leader (at the level of the
111.26 𝐻𝑐𝑙
leader tip height, is found proportional to q0.

102
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

11.2.6 Determination of Fpr


The factor Fpr takes account for the effect of proximity of other conductors and is calculated as follows:

Figure 11-13 – Calculation of Fpr.

Extensive application of the model demonstrated that the


𝑘𝑙 𝑈𝑙𝑐𝑜
𝐹𝑝𝑟 = 1 − 2.0 [( − 1) ( )] proximity function Fpr depends on two dimensionless
𝑘𝑠𝑝𝑜 𝑈0
variables: the proximity ratio kl/ksp0 and the ratio between the
F
𝑘𝑙 𝑈𝑙𝑐𝑜
basic positive leader onset space potential Ulc0 and the
For ( − 1) ( ) < 0.5 negative leader tip potential U0, for the current I0 in question,
𝑘𝑠𝑝𝑜 𝑈0
at the level of conductor height hc (or hgw)
1 𝑘𝑠𝑝0 Refer to B
2 𝑈𝑙𝑐0 Refer to C.
3 𝑘𝑙 Refer to D.
4 𝑈0 Refer to E.

103
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

11.2.7 Determination of Feg


Due to the large equivalent radius of the bundle, it is assumed that there is no corona on the phase
conductor because of the ground field Eg. For ground wires the combination of small diameter and large
height above ground makes corona inception quite likely and requires formulation of Feq differently below
and above corona onset. Refer to G and H below. For this calculation it is assumed that all phase
conductors for system voltages of 345 kV and above will be bundled conductors and that single shield
wires are used.

Figure 11-14 – Calculation of Feg.

𝑘𝑠𝑝0 ∙𝐸𝑔 ∙ℎ𝑐 0.303 𝐸𝑔 ∙ℎ𝑐 0.846 Below corona onset this formula is applicable.
G 𝐹𝑒𝑔 = 1 + 3.818 ( ) ∙( )
𝑘𝑙 𝑈𝑙𝑐𝑜 𝑈0 (Eg < Eci)
𝑘𝑠𝑝0 ∙𝐸𝑔 ∙ℎ𝑐 0.620 Above corona onset this formula is applicable
H 𝐹𝑒𝑔 (𝐸𝑔 ) = 1 + [𝐹𝑒𝑔 (𝐸𝑔𝑐𝑖 ) − 1] ∙ ( )
𝑈𝑐𝑖 (Eg ≥ Eci)
The corona onset space potential for a single
conductor or bundle conductor becomes Uci
1 𝑈𝑐𝑖 = 𝐴 𝑛 𝑞𝑐𝑖 according to the presented formula, with:
Potential A in m/F
Number of sub-conductors n, for shield wires n = 1
At corona onset the conductor charge becomes qci
2 𝑞𝑐𝑖 = 2𝜋𝑟𝑐 𝜖0 𝐸𝑐𝑖 according to the presented formula, with: radius rc ,
in meters
The corona onset gradient for a single conductor is
0.24
3 𝐸𝑐𝑖 = 33.7𝑥102 𝑚 (1 + ) given by the Whitehead Formula radius rc, in
√100𝑟𝑐
meters, and roughness factor m = 0.6
𝑈𝑐𝑖
Egci is the corona onset ground field, determined by
4 𝐸𝑔𝑐𝑖 = dividing the corona onset space potential by the
𝑘𝑠𝑝0 ℎ𝑐
height of the conductor hc multiplied with ksp0.
𝑘𝑠𝑝0 ∙𝐸𝑔𝑐𝑖 ∙ℎ𝑐 0.303 𝐸𝑔𝑐𝑖 ∙ℎ𝑐 0.846
5 𝐹𝑒𝑔 = 1 + 3.818 ( ) ∙( ) Calculate Feg at Egci - 𝐹𝑒𝑔 (𝐸𝑔𝑐𝑖 )
𝑘𝑙 𝑈𝑙𝑐𝑜 𝑈0
6 𝑘𝑠𝑝0 Refer to B
7 𝑈𝑙𝑐0 Refer to C.
8 𝑘𝑙 Refer to D.
9 𝑈0 Refer to E.

104
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

11.2.8 Determination of Fov


The factor Fov takes account for the effect of the (instantaneous) operating voltage. Details on how to
calculate the equivalent radius of bundled conductors and the potential coefficient matrix can be found in
textbooks dealing with the calculation of transmission line parameters [112].

Figure 11-15 – Calculation of Fov.


I 𝐹𝑜𝑣 = 1 For shield wires
𝑈𝑝ℎ𝑒 𝑈𝑙𝑐𝑜 0.667
J 𝐹𝑜𝑣 = 1 + 2.62 ∙ ( )( ) For phase conductors
𝑘𝑙 ∙𝑈𝑙𝑐𝑜 𝑈0
1 𝑈𝑝ℎ𝑒 = 𝑈𝑝ℎ ∗ 𝑘𝑠𝑣 With Uphe equal to Uph multiplied by ksv.
The instantaneous voltage is the real part of the
voltage that is present on the lines at a certain
2
2 𝑈𝑝ℎ = √ 𝑈𝑚 ∙ 𝑠𝑖𝑛(𝜔𝑡 + 𝜃) moment. With:
3 Um = the phase voltage (phase to phase, rms)
θ = the phase angle of a certain conductor
𝑞𝑘𝑠𝑣
3 𝑘𝑠𝑣 = With ksv equal to qksv divided by qksv0.
𝑞𝑘𝑠𝑣0
𝑞𝑘𝑠𝑣 = 𝐴−1 ∙ 𝑈𝑝ℎ

or in matrix form:
Calculate the induced charge qksv per conductor from
𝑞𝑘𝑠𝑣1 𝐴11 ⋯ 𝐴1𝑛 −1 𝑈ph_1 the applied AC voltage (Vph) and the inverted potential
4 [ ⋮ ]=[ ⋮ ⋱ ⋮ ] ∙[ ⋮ ] coefficient matrix A-1, with dimension F/m. The inverted
𝑞𝑘𝑠𝑣𝑛 𝐴𝑛1 ⋯ 𝐴𝑛𝑛 𝑈𝑝ℎ_𝑛 potential coefficient matrix is the capacitance matrix of
or the line.
𝑞𝑘𝑠𝑣1 𝐶11 ⋯ 𝐶1𝑛 𝑈ph_1
[ ⋮ ]=[ ⋮ ⋱ ⋮ ]∙[ ⋮ ]
𝑞𝑘𝑠𝑣𝑛 𝐶𝑛1 ⋯ 𝐶𝑛𝑛 𝑈𝑝ℎ_𝑛
The charge qksv0 is equal to the instantaneous voltage
5 𝑞𝑘𝑠𝑣0 = 𝑈𝑝ℎ_𝑛 𝐶
multiplied by the conductor capacitance.
2𝜋𝜖0 Capacitance of a conductor, with input:
6 𝐶= 2ℎ𝑐 hc = height of conductor (m)
𝑙𝑛( )
𝑟𝑒
re = equivalent radius of the conductor (bundle) (m)
7 𝑈𝑙𝑐0 Refer to C.
8 𝑘𝑙 Refer to D.
9 𝑈0 Refer to E.

105
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

11.2.9 Example

Input parameters
Diameter No. Bundle Voltage Angle Horizontal Heigth
conductor conductors separation ph-ph distance
cm -- cm kV ° m m
1 3,33 6 66 1150 90 -24 22
2 3,33 6 66 1150 -30 0 22
3 3,33 6 66 1150 210 24 22
4 3 1 0 0 0 -19 37
5 3 1 0 0 0 19 37

Eg = 10 kV/m
I0 = 28kA
Hcl = 2500m

Equivalent
diameter
m
1 0,964
Equivalent diameter 2 0,964
3 0,964
4 0,030
5 0,030
Uinst

kV
1 939
Instantaneous voltage 2 -469
3 -469
4 0
5 0
Conductor
capacitance
F
1 1.23E-11
Conductor capacitance 2 1.23E-11
3 1.23E-11
4 6.54E-12
5 6.54E-12

Potential matrix [A] in (m/F)


1 8.11E+10 1.32E+10 5.48E+09 2.37E+10 8.48E+09
2 1.32E+10 8.11E+10 1.32E+10 1.69E+10 1.69E+10
3 5.48E+09 1.32E+10 8.11E+10 8.48E+09 2.37E+10
4 2.37E+10 1.69E+10 8.48E+09 1.53E+11 1.41E+10
5 8.48E+09 1.69E+10 2.37E+10 1.41E+10 1.53E+11
-1
Inverse potential matrix [A ] in (F/m)
1 1.318E-11 -1.648E-12 -3.352E-13 -1.815E-12 -3.299E-13

2 -1.648E-12 1.329E-11 -1.648E-12 -1.027E-12 -1.027E-12

3 -3.352E-13 -1.648E-12 1.318E-11 -3.299E-13 -1.815E-12

4 -1.815E-12 -1.027E-12 -3.299E-13 6.991E-12 -3.786E-13

5 -3.299E-13 -1.027E-12 -1.815E-12 -3.786E-13 6.991E-12

106
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

qs qs0 V ksp0
1 1,67E-09 1 2,71E-09 1 220 1 0,616
2 1,44E-09 2 2,71E-09 2 220 2 0,530
3 1,67E-09 3 2,71E-09 3 220 3 0,616
4 1,75E-09 4 2,42E-09 4 370 4 0,722
5 1,75E-09 5 2,42E-09 5 370 5 0,722

qksv qksv0 ksv Uphe


1 1.331E-05 1 1.157E-05 1 1.1500 1 1079.84
2 -7.013E-06 2 -5.786E-06 2 1.2121 2 -569.08
3 -5.729E-06 3 -5.786E-06 3 0.9902 3 -464.87
4 -1.067E-06 4 0.000E+00 4 0.0000 4 0.00
5 1.025E-06 5 0.000E+00 5 0.0000 5 0.00

Ulc Ulco kl rs
1 1431 1 2057 1 0,990 1 5,221
2 1431 2 2057 2 0,990 2 5,221
3 1431 3 2057 3 0,990 3 5,221
4 1902 4 2062 4 0,950 4 5,190
5 1902 5 2062 5 0,950 5 5,190

U0 Uci Feg Feg(Eg)


1 14281 1 0,0 1 1,049 1 1,049
2 14281 2 0,0 2 1,047 2 1,047
3 14281 3 0,0 3 1,049 3 1,049
4 16684 4 308,5 4 1,098 4 1,098
5 16684 5 308,5 5 1,098 5 1,098

Intermediate results
Lds0 x Fpr x Feg x Fov x Kcor = Lds(28kA)
1 57.394 0.825 1.049 1.381 1.019 69.913
2 57.394 0.750 1.047 0.799 0.991 35.687
3 57.394 0.825 1.049 0.836 0.992 41.182

4 70.771 0.922 1.098 1.000 1.000 71.663


5 70.771 0.922 1.098 1.000 1.000 71.663
Unprotected lateral distance
DC ( 28 kA)
1 3,25 m
2 <0 m

3 <0 m
4 19 + 71.66 = 90.66
Note: separation width of the shield wires plus the attractive distance
5 19 + 71.66 = 90.66

107
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

11.2.10 Comparison
In the table below the results from the calculation method as described in this annex is compared with the
numerical examples in the reference paper [90].
Niagara Mohawk 345 Cigre 109 Niagara Mohawk 765 Russian UHV
Rizk Cigre Rizk Cigre Rizk Cigre Rizk Cigre
Bundle Bundle Bundle Bundle Bundle Bundle Bundle Bundle
Conductor
GW GW GW GW GW GW GW GW
10.21 10.21 43 43 24.28 24.28 22 22
hc,m
17.79 17.17 50 50 40.07 40.07 37 37
0.086 0.086 0.145 0.145 0.218 0.218 0.482 0.482
re,m
0.00556 0.00556 0.006 0.006 0.00556 0.00556 0.015 0.015
10 10 9 9 15 15 28 28
I0, kA
10 10 9 9 15 15 28 28
Hcl, m 2500 2500 2500 2500 2500 2500 2500 2500
q0, μC/m 291 270.9 383.4 586.1 586.1
rs, m 2.605 2.393 3.412 5.22 5.221
21.97 21.97 38.72 38.75 41.48 41.48 57.39 57.39
Lds0, m
28 28 41.14 41.18 50.76 50.76 70.77 70.77
na 0 na 0 na na
Uci, kV
130.2 130.2 154.5 154.5 142.3 142.3 309 308.5
1685 1685 2125 2125 2037 2037 2057 2057
Ulc0, kV
1848 1848 2176 2103 2066 2066 2061 2062
7061 7045 9863 9863 11109 11096 14240 14281
U0,(I0,hc) kV
8402 8402 10156 10156 12583 12584 16711 16684
0.59 0.59 0.508 0.518 0.569 0.569 0.616 0.616
Ksp0, pu
0.7 0.7 0.549 0.554 0.699 0.699 0.722 0.722
0.92 0.92 0.896 0.92 0.97 0.97 0.98 0.99
Kl, pu
0.86 0.86 0.84 0.86 0.94 0.94 0.96 0.95
20 20 10 10 20 20 10 10
Egm, kV/m
20 20 10 10 20 20 10 10
323 323.3 472 487.5 750 749.5 1065 1079.8
Uphe, kV
na 0 na 0 na 0 na 0
21.59 21.6 36.31 36.32 48.89 48.87 69.65 69.91
Lds(I0), m
27.87 27.87 34.99 35.72 50.57 50.58 70.68 71.66
Fpr 0.733 0.666 0.741 0.825
Feg 1.088 1.14 1.149 1.049 1.049
Fov 1.21 1.235 1.321 1.377 1.381
Kcor 1.019 1 1.048 1.018 1.019

All the deviations noted in the table are considered insignificant and the result of rounding off errors, which
may be accumulated due to the sequence of calculations. However, the impact of these errors on the
outcome is minor and are therefore not a concern.

108
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

11.2.11 Validation cases


Validation of the predicted versus actual line shielding performance is made complicated by the
assumption of perfectly flat terrain used in the calculation model versus the actual terrain which may be
sloping that could result in a higher than predicted shielding failure rate. Not all the examples presented
here fulfil the “flat” terrain requirement.
Case 1: CIGRE 109 [92]:

Double Circuit – Low reactance


System Voltage: 400 kV
LI Strength: 1690 kV
Shield wires: Single
Shielding Angle [α]: 38o
Diameter of Phase conductor: 3.01 cm
Number of Sub-conductors: 3
Bundle separation: 45 cm
Diameter of Shield wire: 1.2 cm
Sag: No sag
GFD: 4 Fl/km2/yr

Figure 11-16 – CIGRE 109: An overview of line parameters.

Table 11-1 - CIGRE 109: Comparative outage rates.


Flashes to the line Shielding Failure Rate Shielding Failure Flashover Rate
[NL] Fl/100 km/yr [SFR] Fl/100 km/yr [SFFOR] Fl/100 km/yr
Observations N/A -- 3.55
EGM
7.62 7.06
(CIGRE/Hileman)
Rizk - simplified 68.2 4.5 4.5

Case 2: CIGRE 31 [92]:


Double Circuit – Super Bundle
System Voltage: 345 kV
LI Strength: 1600 kV
Shield wires: Double
Shielding Angle [α]: 22o
Diameter of Phase conductor: 3.2 cm
Number of Sub-conductors: 1
Diameter of Shield wire: 1.0 cm
Sag: Average height
GFD: 4 Fl/km2/yr

Figure 11-17 – CIGRE 31: An overview of line parameters.


The terrain of GIGRE 31 line is given as: 50% flat ground, 50% “rolling” terrain and lightly wooded, which
may explain the higher than predicted shielding failure rate.
Table 11-2 - CIGRE 31: Comparative outage rates.
Flashes to the line Shielding Failure Rate Shielding Failure Flashover Rate
[NL] Fl/100 km/yr [SFR] Fl/100 km/yr [SFFOR] Fl/100 km/yr
Observations N/A -- 3.44
EGM
0.98 0.70
(CIGRE/Hileman)
Rizk - simplified 67 0.00 0.00

109
TB 839 - Procedures for estimating the lightning performance of transmission lines – new aspects

Case 3: Japan 500 kV [10]:

Double Circuit – Low Reactance


System Voltage: 500 kV
LI Strength: 2000 kV
Shield wires: Double
Shielding Angle [α]: -7o
Diameter of Phase conductor: 2.42 cm
Number of Sub-conductors: 6
Bundle separation: 66 cm
Diameter of Shield wire: 2.52 cm
Sag: Average height
GFD: 4.9 Fl/km2/yr

Figure 11-18 – Japan 500 kV: An overview of line parameters.


This line traverses both flat ground and 30 degree ridge terrain, which could explain higher than predicted
shielding failure rate.
Table 11-3 - Japan 500 kV: Comparative outage rates.
Flashes to the line Shielding Failure Rate Shielding Failure Flashover Rate
[NL] Fl/100 km/yr [SFR] Fl/100 km/yr [SFFOR] Fl/100 km/yr
Observations 130.8 1.28 0.51
EGM
228 0.00 0.00
(CIGRE/Hileman)
Rizk - simplified 121.6 0.02 0.02

Case 4: Russian UHV [90]:


Double Circuit – Low Reactance
System Voltage: 1150 kV
LI Strength: -- kV
Shield wires: Double
Shielding Angle [α]:
Diameter of Phase conductor: 3.33 cm
Number of Sub-conductors: 6
Bundle separation: 66 cm
Diameter of Shield wire: 3.0 cm
Sag: Average height
GFD: 2.5 Fl/km2/yr

Figure 11-19 – Russian UHV: An overview of line parameters.

Table 11-4 - Russian UHV: Comparative outage rates.


Flashes to the line Shielding Failure Rate Shielding Failure Flashover Rate
[NL] Fl/100 km/yr [SFR] Fl/100 km/yr [SFFOR] Fl/100 km/yr
Observations 0.1
EGM
0.21 0
(CIGRE/Hileman)
Rizk - simplified 49.3 0.26 0.14

110
CIGRE
21, rue d'Artois
75008 Paris - FRANCE

© CIGRE

ISBN : 978-2-85873-544-0

You might also like