LIGHTNING PERFORMANCE OF
John G. Anderson
William A. Chisholm
Abstract: This chapter reviews issues related to the occurrence of lightning on compact
transmission lines, including the types of lightning flashover that can occur, the factors
that govern flashover, and the performance of shielded and unshielded compact lines.
John G. Anderson is one of the original authors of the EPRI Transmission Line Reference
Book, including the chapter on line lightning performance. He has had more than 50 years
of high-voltage engineering experience, and is a Life Fellow of the IEEE and an elected
member of the National Academy of Engineering. He is a former manager of the General
Electric High Voltage Laboratory and served as a consulting engineer for General
Electric and also as a senior consultant for Power Technologies, Inc. He was one of the
original researchers at the Lenox, Massachusetts EPRI Project EHV and later Project
UHV, and also carried out lightning research at the Empire State Building in New York
City. He is the author/coauthor of more than 40 technical papers and coauthor of three
books concerned with high-voltage transmission, lightning, and insulation performance.
Dr. William A. Chisholm is an internationally acknowledged expert in lightning
protection, insulation, and thermal rating of power systems. He is a Senior Research
Project Manager in the Transmission and Distribution Technologies group of Kinectrics,
the former Research Division of Ontario Hydro, now a division of AEA Technologies
PLC. In this capacity he has completed research contract and project work for more than
40 electric utilities, manufacturers and research organizations. Dr. Chisholm was
recognized as the editor of the “Best Standard” award in 1999 for the IEEE Standard
1243, Guide to Improving the Lightning Performance of Overhead Transmission Lines.
He is a corresponding member of CIGRÉ Study Committee 33 working groups on
lightning and insulator icing test methods. He is the chairman of the IEEE Power
Engineering Society Lightning and Insulator Subcommittee and a member of the PES
Dr. Andrew Phillips is a Senior Program Manager in the Transmission Lines and
Increased Power Flow Research program area of the Power Delivery and Markets Sector.
His current research activities focus on the Overhead Transmission, Underground
Transmission, Increased Power Flow, and HVDC (high-voltage direct current) programs.
Dr. Phillips’ special areas of interest are nonceramic insulators (NCI), lightning and
grounding, inspection and assessment of components, sensor development, and daytime
Before joining EPRI, Dr. Phillips worked at J. A. Jones Power Delivery, where he was a
Project Manager and lead researcher in the fields of insulation, aging equipment, and
lightning. Prior to that, Dr. Phillips worked at the University of the Witwatersrand
performing research for the South African electric power industry.
Dr. Phillips received BSc, MSc, and PhD degrees in electrical engineering from the
University of the Witwatersrand in Johannesburg, South Africa.
Dr. Phillips holds three U.S. Patents and is the author of over sixty journal and
Compaction of transmission lines can have significant benefits to lightning performance
in many cases. If one or more overhead shield wires exist, a reduced spacing to nearby
phases creates increased coupling, resulting in reduced insulator voltages. Also reduced
structure height makes any compact line a smaller lightning target. Finally, the shorter
spans of most compact lines—when accompanied by shield wires—can also reduce the
voltage challenges to insulators. Conversely, possible reductions in insulator lengths and
clearances increase the risk of flashover unless appropriate countermeasures are taken,
and elimination of shield wires in the compaction process exposes the phases to more
direct lightning hits.
The technology involved in voltage upgrading of lines is similar to that of compaction.
However, the risks that might be acceptable in voltage upgrading are driven by economic
issues that may not exist in new line designs. Thus reductions in design tolerances in an
upgrading would not be considered acceptable for a new line. For example, eliminating
shield wires in a voltage upgrade of an existing line might be beneficial in order to
minimize the outage time and reuse of the structures and make or break the upgrading.
There may also be times when construction of a new compact line without shield wires is
desirable, perhaps to make it look like a distribution line or to fit in an overall pole
Lightning phenomena are no different for compact lines, and the same computer design
programs are used. Compaction may improve or degrade performance, and this chapter
concentrates on the beneficial aspects. Lightning performance is part of the design
process with its own set of tradeoffs.
The principal strategies reviewed in this chapter for ensuring good lightning performance
of compact lines include:
• Minimizing support structure footing resistance.
• Increasing line insulation.
• Installing shield wires to prevent lightning hits to the phases.
• Installing transmission line surge arresters (TLSA) to prevent excessive phase
overvoltages during lightning hits.
• Installing underbuilt shield wire(s).
Estimating improvements in transmission line lightning performance gained by any of
these strategies is an evolving art, limited in accuracy by annual variations of local
weather patterns, as well as by changes in lightning flash magnitudes and ground flash
densities from year to year. If detailed and reliable data are available, median flashover
rates can be calculated fairly well by line lightning performance prediction software such
as EPRI TFlash (EPRI 2005b). These software programs generally use median lightning
ground flash density (GFD) data that provide estimated median flashover rates for the
line designer. However, extreme value flashover probability estimates based on year-toyear variations are not yet reported by most programs.
This chapter does not discuss the fine details of line backflashover voltages, leader
progression, grounding nonlinearities, counterpoise propagation, volt-time curves, and
other arcane subjects related to the technology. These topics are explained in many
references (EPRI 1982; IEEE 1997b; EPRI 2005a), and will not be repeated here.
Section 6.2 of this chapter starts by defining and reviewing the various types of lightning
flashovers that occur on transmission lines and the importance of each. By their smaller
size, a compact line will offer a smaller target to lightning than will a conventional line,
but the act of compaction minimizes wire spacings wherever possible, and smaller
spacings can then increase risk of flashover for the fewer lightning incidents predicted to
Section 6.3 details the major factors that govern when flashovers will occur, starting with
the mean number of lightning hits per year that can be expected, the expected
characteristics of the lightning flashes that will challenge the line’s insulation, the
mechanisms of shielding failures, midspan flashovers and backflashovers, and the
influence of ground resistances on the latter. Alternative methods of estimating line
lightning performance are briefly covered.
Section 6.4 reviews performance of compact transmission lines having shield wires,
stressing the benefits of proper grounding and providing numeric examples of the effects
of various footing resistance magnitudes and the number of overhead or underbuilt shield
Significant economic and compaction benefits can result from omitting shield wires.
Section 6.5 examines the pros and cons of this option. Unless transmission line surge
arresters (TLSA) are properly selected and connected to vulnerable phases, each lightning
hit to these unshielded lines will cause one or more flashovers. Thus care must be taken
in selection of arrester locations and selection of arrester energy capabilities, otherwise
excessive maintenance and reduced line reliability may result.
For line designers who do not have access to line lightning simulation programs, such as
EPRI TFlash, Section 6.6 suggests some alternative methods of estimating arrester
energy challenges on unshielded compact lines. The electrical effects can be calculated
using one of the Electromagnetic Transients Programs (EMTP, ATP or equivalent), but
all the associated failure probabilities must be computed separately. Appendix 6.1
expands these procedures. Appendix 6.2 covers possible acceptance tests on qualitycontrol samples of proposed TLSA.
Section 6.7 summarizes important points in the chapter.
6.2 LIGHTNING FLASHOVER ISSUES
This section briefly describes the characteristics of lightning flashovers in a simple form,
the factors that influence each of the mechanisms and their relevance to compact line
Induced Flashovers occur when a ground flash terminates near, but not directly striking,
an energized phase conductor. The intense electromagnetic field from the return stroke
current causes large, short-duration overvoltages that can be calculated using (Rusck
1958; Agrawal et al. 1980; Chowdhuri 1996; Rachidi et al.1997). A simplified model has
recently been described by (Darveniza 2007). Influencing factors in induced flashovers
relate to the distance between the stroke and the line, the effective line height of the
phases, and the insulation strength. Like all lightning flashovers, the number of induced
flashovers will scale linearly with local ground flash density. Experience has shown that
compact lines with insulation strengths of 400 kV BIL or higher and conductor-to-tower
clearances of 0.5 to 0.8 m will be largely immune to induced flashovers.
Shielding Failures occur when lightning terminates directly on an energized phase
conductor. Lightning current is injected directly into the phase conductor. Typically, at
least one stroke of a random flash has sufficient current and a fast rate of current rise,
together with a high conductor surge impedance, to cause a local overvoltage that may
exceed the lightning impulse strength of the struck phase insulator or the dielectric
strength of air gap between phases and results in a flashover to ground or mid-span to a
Influencing factors in shielding failures relate mainly to the parameters that influence the
number, rather than the nature, of the surges. These include the relative height of the line
above local terrain and surrounding objects and the line width. Overhead shield wire
protection (located above the phase conductors) eliminates more than 90% of the
shielding failures on typical lines.
Compact lines have reduced height and width, and these parameters tend to reduce the
number of lightning flashes. Compact lines that do not use overhead groundwires must
absorb shielding-failure surges in some other way, such as line surge arresters.
Mid-span Flashovers occur when lightning terminates directly on an overhead shield
wire or a phase conductor at mid-span. The lightning stroke current is injected into the
shield wire or phase conductor. For long span lengths, the stroke current magnitude and
rate of rise, together with the surge impedance, cause a local voltage rise that may exceed
the lightning impulse strength of the conductor-to-conductor (or shield wire) gap and
result in a mid-span flashover.
Influencing factors in midspan flashovers relate mainly to the span length and the line
geometry (wire spacing). The insulation strength between two parallel conductors for
conventional lines is usually much greater than the impulse strength of the insulators, so
midspan flashovers are not common. Hileman 1999 explains the mid-span flashover
mechanism in detail.
Backflashovers occur when lightning terminates directly on a transmission structure or
overhead shield wire. The surge current, together with the impedance of the structure and
the structure footing resistance, causes a transient potential rise on the structure and at the
structure support end of every insulator (Figure 6.3-6). For high surge currents or highresistivity soil, these local overvoltages may exceed the lightning impulse strength of the
structure-to-conductor or ground wire-to-conductor gaps.
Influencing factors in backflashover are the local structure footing resistance, the surge
current magnitude and rate of rise, and conductor-to-structure (or ground wire) clearances
and impulse strength of the insulators. Overhead shield wires couple a fraction of shield
wire voltage onto the local (upper) phase conductors, reducing stress and improving
performance. Adding an underbuilt shield wire(s) couples the lower phase conductor(s),
reducing stress in these conductors and further improving performance. The number of
flashes to the line and the tower surge response are affected by the line height.
Compact lines typically have reduced conductor-to-tower clearances and small towerbase dimensions. Both of these factors can degrade backflashover performance to such an
extent that alternatives, including unshielded operation and use of line surge arresters,
may be more practical.
Lightning Flash vs. Lightning Stroke A lightning flash consists of one or more
lightning strokes traveling down the same channel in less that a second. The median
number of strokes in a flash is three, but as many as 10 have been observed. Figure 6.2-1
shows that each stroke consists of a rapidly-rising high current peak, which can, in some
cases, exceed 100 kA in a few microseconds. After peak stroke current is reached, the
current decays in roughly 100 microseconds or more to 50-400 amperes, which can flow
for hundreds of milliseconds before the next stroke current begins (EPRI 1982; EPRI
2005a). Most of the charge delivered by a stroke does not exist near the high current peak
but is delivered during the continuing current (Component C in Figure 6.2-1). The peak
and rate of rise are important in insulator or air gap design, while total charge delivered is
of great importance in arrester design.
Figure 6.2-1 Components of lightning flashes (MIL-STD 1757).
6.3 FACTORS INFLUENCING LIGHTNING PERFORMANCE
6.3.1 Flashes to a Line
A useful equation developed by Eriksson (Eriksson 1987) can be used to approximate the
average expected number of lightning flashes to a line per 100 km per year:
28h 0.6 + b
N = GFD
N = number of line hits per 100 km per year.
GFD = Ground flash density at the line location, flashes to earth per km2/year.h = height
of the topmost wire at the structure, meters.
b = spacing between topmost shield wires, meters (if no shield wires are present, b =
maximum horizontal spacing between outermost phases).
In North America, regional GFD is displayed on GFD maps published by various entities
(EPRI 2005a). If GFD data are not available, thunder-day (TD) data can be used for
estimating purposed. The relationship between TD and GFD is approximately (EPRI
GFD = 0.04TD1.25 6.3-2
TD = average number of days that thunder is heard per year in the line vicinity.
TD maps are published by the World Meteorological Organization (WMO 1953), and a
United States map is published in McGorman et al. 1984. Also, NASA (Boccippio 2001)
published a climatology of optical transient density (OTD), consisting of both cloud and
ground lightning flashes observed from space with optical sensors. Cross-checks suggest
that these OTD observations can be used to predict ground flash density with a
calibration factor in Equation 6.3-3.
GFD = 0.3 ⋅ OTD 6.3-3
OTD = average number of optical flashes per km2 per year (Boccippio 2001).
Figure 6.3-1 shows the variables in Equation 6.3-1. The number of hits to each
configuration is directly proportional to the lightning activity (which can vary by as much
as 3:1 from year to year). The expected number of hits varies roughly by the 0.6 power of
the line height, and, for a given height, linearly as the line width. Wires with lower
heights will receive fewer strokes and have better lightning performance, if all other
variables remain unchanged.
Figure 6.3-1 Line profile for line hit and shielding failure calculations.
In Figure 6.3-1, the compact line with no shield wires will receive fewer lightning flashes
than the other configurations, but they will all be shielding failures. For unshielded lines
at any voltage level, most of the line hits will result in line flashovers if the line is
unprotected by arresters (TLSAs).
Adding a shield wire adds a meter or two of height above the phase conductors and
thereby collects a few more lightning strikes. However, if the shield wire takes the strikes
as it is intended to, shielding failures or strikes to the phase conductor are avoided, which
greatly enhances the overall line performance. If a second shield wire is added, the
structure height (h) may be slightly reduced, but b is larger in Equation 6.3-1. With one
or two shield wires, there is not much difference in the susceptibility to strikes, but two
wires increase the coupling effect and reduce backflashovers, resulting in an
improvement in overall performance.
Smaller lines receive fewer lightning hits because of their relatively smaller size. Trees
and other nearby objects along the right-of-way boundary and having heights comparable
to the line can intercept lightning that would otherwise be attracted to the line. Lightning
to some trees with deep internal sap ducts (such as pines) tend to detonate from the
pressures of interior arcing and scatter debris as a result. Lightning also has a propensity
to start fires in trees during dry seasons, and tree trimming along rights-of-way can be
difficult to access, so again there are tradeoffs. Usually, however, the benefits in allowing
flashes to strike to nearby trees or other tall objects rather than to a compact line can be
substantial, particularly in reducing backflashovers. Note, however, that lightning hitting
a tree creates essentially the same electromagnetic radiation as mentioned in Section 6.2.
This electromagnetic radiation couples into nearby line conductors, causing flashovers
for lines with BIL ratings lower than approximately 400 kV.
6.3.2 Stroke Characteristics
The essential lightning stroke characteristics that initiate line flashovers are usually
represented by cumulative probability curves (IEEE 1985). Figure 6.3-2 shows a stroke
peak current magnitude probability. Conservatively, 50% of any first strokes to a line
exceed 31 kA crest current, and about 0.8% exceed 200 kA. Three or more strokes can be
expected in any flash over one second in duration.
Figure 6.3-2 Cumulative distribution of peak of negative first stroke current (Anderson
and Eriksson 1980).
Figure 6.3-3 shows the relationship between stroke peak amplitudes and maximum rate
of rise in the current surge for flashes to transmission towers (Takami and Okabe 2007).
Figure 6.3-3 Relationship between peak of negative first stroke current (peak amplitude
(kA)) and maximum rate of rise (kA/μS).
6.3.3 Insulator Flashover Strengths
The dielectric strength of a line insulator varies with the waveshape of the applied
voltage, atmospheric pressure, and humidity, but two rough rules of thumb for lightning
transients derived from (EPRI 1982; Hileman 1999) are:
BIL+ = 125 + 539 S 6.3-4
BIL- = 164 + 469 S 6.3-5
BIL+ = basic impulse insulation level (crest kV) for a positive polarity wave.
BIL- = basic impulse insulation level (crest kV) for a negative polarity wave
(shielding failure calculations).
S is the striking distance (insulator dry-arc distance) in meters.
Basic impulse insulation level is that voltage that will cause flashover 10% of the time for
standard conditions of atmospheric pressure and humidity.
The flashover path around a line insulator exhibits a “volt-time” effect that is important in
fixing the line flashover performance (IEEE 1985). Normally, insulation strength is
evaluated just before the return of traveling waves from adjacent towers. At 2 µs, only the
towers within 300 m of the flash termination share any fraction of the surge current. For
shielded compact lines with minimum-height towers and short span lengths, this time
comes sooner, and the corresponding insulation strength is higher than for conventional
lines. Also, the overvoltage waveshape across insulators has a fast-falling time to half
value of 3-10 µs, compared to the 50-µs value in the standard test waveshape (IEEE
1998). The tail times for overvoltages induced from nearby lightning will be even shorter,
at 2-5 µs. The use of a volt-time curve approach for these cases will usually be too
conservative for accurate estimates. Advanced methods for these cases are found in EPRI
6.3.4 Shielding Failures
A shielding failure occurs when a lightning flash gets by any overhead shield wire
protection and hits a phase directly. The outboard phases are invariably involved, and the
stroke current is usually under 20 kA. Simulation programs such as EPRI TFLASH
(EPRI 2005b) or the L-1 and L-2 applets accompanying EPRI 2005a can be used to
evaluate expected shielding failures for any proposed shield wire – phase geometry.
Figure 6.3-4 Shield angles for line with overhead groundwires.
The shield angles Ө shown in Figure 6.3-4 between groundwires GW and phases A and C
are particularly important in mitigating shielding failures. Figure 6.3-5 demonstrates how
the ratio of phase hits to total line hits can vary with shield angle Ө for different line
heights for a single shield wire and two outboard phases. The lower the line height, the
greater the permitted shield angle should be for the same performance. The greater the
line height, the less the shield angle should be for good performance. Compact lines that
are double circuit or have distribution underbuild will generally be taller and wider than a
single circuit line and may require two shield wires and a shield angle of 20° or less to
attain infrequent shielding failures. Compact, single-circuit lines at lower voltage may
perform adequately with a 30° shield angle.
Figure 6.3-5 Hits to outboard phases for various shield angles and line heights.
For lines with no overhead groundwires, nearly every hit terminates on a phase conductor
rather than a pole, and is thus a shielding failure. For practical purposes, every shielding
failure results in a flashover on at least one unprotected insulator. Some shielding failures
can cause severe insulator damage. First return-stroke currents of 100 kA have
corresponding steepness of 50 kA/µs in Figure 6.3-3. These currents are injected into the
high surge impedance of the phase conductors Zphase of about 300 Ω in corona, in each of
two directions. The steepness of the resulting transient voltages, given by dI/dt times
Zphase/2, gives voltage rates of change of more than 7.5 MV/µs. This puts a current of
more than 20 A through a typical 3-pF insulator string capacitance, increasing risk of
insulator damage by puncture or shed fragmentation in weak regions of the insulation,
whether porcelain or polymer. The same extreme dV/dt voltage steepness also stresses
any parallel line surge arresters, installed to limit the overvoltage level across insulators
to prevent shielding failure flashovers.
6.3.5 Midspan Flashovers
For most lines in areas of high lightning activity, at least 40% of lightning hits occur on
the spans between the structures, and particularly for compact lines with closer spacings
between shield wires and phase conductors, adequate shield wire-spacing must be
maintained to prevent flashover from shield wires and a phase conductor. EPRI 2005a
and Hileman 1999 explain this flashover process in detail. If an overhead groundwire is
struck, the shorter the span length and the lower the tower footing resistances, the less
likely a midspan flashover is to occur. This is because current reflections from the nearby
towers reduce the voltage stress at midspan, and can interrupt a midspan flashover that
would otherwise occur (Hileman 1999).
6.3.6 Backflashovers and Structure Ground Resistance
A low tower ground resistance is of extreme importance in regions of high lightning
activity if lightning backflashovers are to be minimized on shielded lines or lines with
line surge arresters. A backflashover occurs when a lightning flash strikes a structure top
or overhead shield wire, and the resulting potential differences across one or more of the
insulators exceed the insulator dielectric strength. Figure 6.3-5 shows a simplified
equivalent single phase circuit for a hit to a tower top. The stroke current splits, part IS
going out into the shield wire impedances ZS, and part IT flowing into the structure (or
ground wire) inductance LT and the ground resistance RG. It should be clear that—for a
given stroke current—the higher the ground resistance RG, the higher the voltage at the
tower top and the greater the voltage VP across the insulator. There is also a voltage
induced on the phase conductor by “coupling” from all overhead shield wires and from
any underbuilt distribution circuits. This coupling is shown in Figure 6.3-6 as
capacitances CSP and CPG, although in reality, magnetic coupling also exists.
Figure 6.3-6 Simplified single-phase circuit for a hit to a tower.
Reflections from adjacent towers are not shown in Figure 6.3-6. This situation thus
persists for two span travel times, given by twice the span length divided by 90% of the
speed of light (EPRI 1981; IEEE 1985).
6.3.7. Alternate Methods of Estimating Flashover Performance
For those who do not have access to the EPRI TFlash program (EPRI 2005b), other
alternatives—though not as comprehensive—can be used for rough estimates of line
lightning performance. Reference IEEE 1997 is an IEEE Standard and contains a small
computer program diskette called FLASH. This program is essentially a digital version of
an estimating method published in EPRI 1982. It does not handle arresters. However,
some success has been attained by regarding each phase protected by arresters at every
tower as if they were underbuilt shield wires, then using FLASH or other programs to
evaluate the voltages on the unprotected phases.
6.4 LIGHTNING PERFORMANCE OF SHIELDED COMPACT
6.4.1 General Insulation and Spacing Considerations
A compact line is defined as a line that takes advantage of reduced phase-to-phase
spacing, and controls overvoltages through other means such as surge arresters or closing
resistors in circuit breakers, controls conductor motion through rigid insulators and short
spans, and controls audible noise and EMI as described in other chapters of this book. It
can be argued that any extra-high-voltage line is “compact,” since most designs at 345
kV and above have higher phase-to-phase voltage stress per meter of phase separation
than lines built for system voltages below 300 kV.
One direct consequence of reduced phase spacing is that the insulator dry-arc distance in
compact lines may also be reduced. In contaminated areas, additional sheds or deeper
skirts can add the necessary leakage distance, but large-diameter sheds do not increase
the metal-to-metal dry arc distance significantly. This means that compact lines built with
1-m dry arc distances have lightning performance that is significantly worse than
conventional lines built with 2-m electrical clearances in the same lightning and
The experience of a utility in Minnesota, Otter Tail, described in Chapter 2—although
not a typical case—is an instructive example. It is a particularly good example of the
sorts of problems that can arise when a utility is required to be extremely aggressive in
modifying lightning performance of a compact line that does not meet initial
expectations. A 41.6-kV subtransmission line was initially upgraded to a 115-kV line
with compact dimensions and no overhead groundwires. As Otter Tail gained experience
with this compact 115 kV design, they began to make many modifications, since their
lightning outage rates proved unacceptable to their customers. Based upon breaker
operation data, Otter Tail determined that 290 km of these compact115-kV lines had a
nine-year average of 18.3 trip-outs per 100 km/yr during lightning storms. To reduce
these outages, Otter Tail added overhead groundwires and line arresters in some sections
to improve the outage performance. Photos of their two alternatives are shown in Figure
Figure 6.4-1 Retrofit of overhead groundwire (left) and center-phase transmission line
surge arrester (right) to Otter Tail 115-kV compact line.
The top phase conductor with transmission line surge arrester (TLSA) protection can be
closer to the unprotected phases than an overhead groundwire. This means that more of
the lightning surge voltage appears on the unprotected phases—and the stress on the
insulators is the difference between these voltages. Additional details of TLSA protection
are found in Section 6.5.
While the reduced insulation level degrades lightning performance, closer phase-tooverhead groundwire spacing of compact lines generally improves the electromagnetic
coupling, indicated by Csp in Figure 6.3-6. This means that a higher fraction of the towertop impulse voltage appears on the floating phase conductors. Since the insulator impulse
strength must withstand the difference in voltage between the tower voltage and the phase
conductor, any increase in the coupled voltage on the parallel phase conductors reduces
impulse voltage appearing across the phase conductor insulators.
An example of the increase in coupled voltage with compact geometry is given by
Table 6.4-1 Effects of Compact Line Insulation and Phase Spacing on Lightning
Performance (Outages per 100 km per year) of Line with Horizontal Configuration
±2 m at 14-m height
11-m phase conductor height at tower
2 m (14 discs) dry-arc distance
1 m (7 discs) dry-arc distance
Rfooting = 20 Ω
Rfooting = 30 Ω
Rfooting = 50 Ω
Rfooting = 100 Ω
Ng = 6 flashes / km -year.
Both phase configurations have good shielding from direct flashes. The improved
coupling with close phase-to-phase spacing offers a significant benefit of between 10 and
20% better lightning backflashover performance. However, this alone cannot make up the
deficit in backflashover rate related to the reduced insulation.
A number of options can be considered to improve the dismal lightning performance of
the compact transmission line with 2-m phase spacing and 1-m dry arc distance. Some
possible options would be:
Improve the grounding resistance
Improve the coupling by bundled pairs of overhead groundwires together to
increase the geometric mean radius
Increase the insulation strength
Add line surge arresters
Install an underbuilt shield wire
Install twin bundled ground lead (unless steel pole or tower)
Figure 6.4-2 illustrates the structure configurations assumed for these examples.
Performance improvement for arranging the conductors in a more compact delta
configuration is given in Table 6.4-2.
The relative merits of each option listed above are ranked in Table 6.4-2.
Table 6.4-2 Options for Improving Compact Line Lightning Performance (Outages
per 100 km per year)
OGHW -2 and 2 m at 14-m height
11-m phase conductor height at tower
1 m (7 disc) dry-arc distance
Horizontal, 2-m phase spacing
Twin bundled shield wires
Instead, cut Rfooting by 50%
Arrange conductors in a delta shape
Instead, increase to 1.5-m insulation
Instead, convert to delta, single shield
wire on pole, 1.5-m braced posts, with
bottom phase TLSA for coupling
Improving coupling by adding more OHGW gives a modest improvement in lightning
performance compared to the reference compact line. However, it seems to be more
practical to convert the horizontal circuit to a delta arrangement using braced posts, and
then to convert the bottom phase to a shield wire by fitting it with a suitable transmission
line surge arrester.
Figure 6.4-2 Options for improving compact line lightning performance.
Hoffmann (Hoffmann et al. 2003) offers a design alternative, also discussed in Section
2.3.7, that takes further advantage of coupling to ensure acceptable lightning performance
for urban 230-kV lines in Brazil. Figure 6.4-3 shows the vertical orientation of twoconductor bundles with a phase separation of 3 m. With a maximum 200-m span and a
pole height of 22 m, and using the same ground flash density Ng of 6 flashes per km2 per
year as in Tables 6.4-1 and 6.4-2, the lightning tripout rate of this line is computed to be:
• 4.6 outages per 100 km per year with a single overhead groundwire
• 2.4 outages per 100 km per year if two overhead groundwires were located
conventionally, above the phase conductors
• 1.9 outages per 100 km per year with the above- and below- arrangement of shield
wires shown in Figure 6.4-3.
This example illustrates that it is feasible to obtain a good lightning performance from a
compact transmission line using conventional overhead shield wire protection, and also
that the use of an underbuilt horizontal (messenger) ground wire makes a further
improvement. The underbuilt ground wire can contribute a performance similar to adding
a surge arrester to the bottom phase conductor.
Figure 6.4-3 Example of 230-kV compact line with underbuilt shield wire protection.
6.4.2 Wood and Fiberglass as Insulating Materials for Compact Lines
Wet wood or wet fiberglass (without sheds) exhibits very little power frequency dielectric
strength and can be subject to fires from leakage currents at voltages at 230 kV or above
if lengths are inadequate. Prudence requires that no power frequency dielectric strength
be attributed to these materials when used as crossarms or poles. If unbonded, they
should always have polymer, porcelain, or glass insulators as leakage current suppression
resources. However, these crossarms do exhibit significant lightning impulse dielectric
strengths, and wet wood crossarms are also generally regarded as having some power
frequency follow-current suppression capabilities. A seasoned wet wood crossarm alone
can have an impulse flashover strength of over 300 kV per meter. Darveniza (1980)
tested the lightning impulse strengths of many combinations of wet wood and porcelain
insulators in series, and Figure 6.4-4 shows interpolated values from his report.
Some deterioration in lightning impulse strength has been reported when porcelain or
glass insulators have been used with unbonded fiberglass crossarms or structural
components because of the much different dielectric characteristics of the two materials.
Figure 6.4-4 Impulse flashover strengths of wet wood crossarms and porcelain insulators
Additional dielectric strength benefits can sometimes be achieved by using wood or
fiberglass standoff struts to locate pole ground wires away from the surface of the pole in
the vicinity of the pole top to add additional air gap spacing to the crossarm-insulator
6.5 LIGHTNING PERFORMANCE OF UNSHIELDED COMPACT
6.5.1 General Engineering Considerations
Any unshielded compact line—unprotected by TLSAs—will essentially flash over each
time lightning strikes the line, and these flashovers can involve several towers and phases
for the same event. The only benefits of reducing footing resistances for unshielded lines
without arrester protection are increased fault currents for relaying purposes and possibly
reduced step-and-touch potentials at a faulted structure. Increasing insulation on an
unshielded line may also have little effect. Even a strike directly to the low impedance of
an EHV phase conductor bundle usually creates such enormous voltages that something
is almost certain to flash over regardless of insulation level. Thus the number of lightning
flashovers per year for any line with an insulator BIL of 400 kV or higher and without
TLSA protection is essentially equal to the number of lightning hits to the line per year,
and an estimate of the median value per year can be found by Equation 6.3-1. For
unshielded lines—unprotected by TLSA—with insulator BILs under 400 kV, additional
flashovers may be created by lightning hits to the earth near the line. The powerful
electromagnetic fields created by these hits in the line vicinity induce high-voltage
transients onto the phases that can exceed the phase insulation strength, as discussed in
Modern compact line technology is undergoing a fundamental change, specifically in
application of TLSAs to replace or augment shield wires for lightning protection. This
application affects almost every other aspect of compact line engineering, including
phase spacing, clearances, foundations, grounding, safety, maintenance, and general line
performance. A compact transmission line does not automatically require a TLSA.
However, a properly applied TLSA application can:
• Absorb an appropriate amount of surge energy without failure.
• Limit overvoltages, providing a major improvement in line compaction.
• Provide better lightning performance than shielded lines with conventional
dimensions. However, direct exposure of arrester-protected phases with no overhead
shield wires for lightning diversion has—in some cases—required line rebuilds with
addition of shield wires to avoid excessive arrester failures, breaker operations, and
maintenance problems. Energy capabilities and mechanical capabilities of TLSAs can
vary widely, as can ground flash densities, so care must be taken.
• Supplant shield wires completely in some applications.
On the other hand, improperly applied TLSAs can:
• Result in an excessive number of arrester failures by poor arrester selection.
• Waste money with little or no improvement in lightning performance by improper
TLSA location strategies.
Transfer overvoltages to insulation that would otherwise be un-stressed.
Degrade lightning performance by arrester mechanical failures that reduce striking
• Degrade general line reliability under other adverse-weather conditions such as fog or
galloping, in some cases to levels worse than if no TLSA were applied.
• Create excessive radio noise by poor connections or blown disconnectors or
improperly configured gaps.
• If disconnectors do not function, the failed arrester becomes a momentary short that is
difficult to locate unless fault-locating relays are available to assist with pinpointing
the failed arrester.
Modern polymer-housed TLSAs are designed and tested to have nonexplosive failure
modes that allow application in urban environments, where they have the greatest benefit
to customers. The goals of improved lightning performance are usually reached with
arresters selected for a Transient Overvoltage (TOV) rating that is well above the system
voltage, and most manufacturers of TLSAs specify coordination that forces station-class
arresters to absorb this energy.
The use of TLSAs rather than shield wires also exposes aluminum phase conductors to
lightning arc damage. Test standards such as IEC 60794, based on component C in Figure
6.3-1, have been developed to reproduce lightning damage to aluminum-strand optical
fiber groundwires (Chisholm et al. 2001). These test standards are also appropriate for
selection of phase conductors on unshielded compact transmission lines, with or without
TLSAs have two principal applications:
• Reducing insulator backflashovers in local areas of high footing resistance
• Reducing shielding failure flashovers on lines with no overhead groundwires
In many cases, installation of shield wires adds little additional initial cost to a
transmission line, but additional structure mechanical loads from these wires require
more costly structures and foundations to cope with increased overturning moments and
increased ice loading. Continuous and peak overhead groundwire losses and line
maintenance issues can sometimes be of concern. However, if shield wires are removed
and arresters substituted to manage lightning overvoltages, care must be taken to install
and maintain them properly and to be certain that each arrester has adequate energy
capacity to survive severe lightning discharges. This is not to say that TLSA applications
are necessarily a better alternative than shield wires—each has its place according to
lightning exposure and grounding conditions.
Selection of the proper operating voltage of any gapless arrester continuously connected
to a phase must recognize that the arrester must withstand the maximum continuous
phase to ground operating voltage (MCOV) without overheating. The manufacturer will
provide this voltage rating for each production arrester, but the designer should also
specify to the manufacturer the maximum expected voltage and duration on any unfaulted
phase during a phase-ground fault. A conventional MCOV rating can be supplemented
with an arrester maximum coulombs rating or maximum energy capability of each block
in Joules per cm3, assuming adiabatic heating close to the melting point of the zinc oxide
material. For arresters in series with a gap, there is no MCOV, except the maximum
arrester resistance drop during the conduction period.
A number of developments have completely shifted the paradigm for lightning protection
of compact lines. In the previous edition (EPRI 1978), expensive reductions in footing
resistance were needed to obtain the marginal improvements described in Section 6.4.
The alternative of improving coupling and limiting overvoltage levels using TLSAs was
introduced there, and is expanded in detail in this section.
6.5.2 Recent TLSA Applications on Compact Lines
With energy absorption capability at destruction in excess of 1500 Joule per cm3, and
corresponding charge levels of 10-20 C for IEC Class C (62-63 mm diameter) zinc oxide
blocks, there is now a possibility that overhead groundwires can be completely
eliminated. This reduces the height and visual impact of lines, giving a new axis of line
A partial list of recent TLSA developments includes:
• Adoption of metal oxide varistors as the nonlinear element
• Production experience with a wide range of MOV formulations
• Control of the V-I characteristics that affect energy sharing
• Adoption of sealed polymer housings to improve reliability
• Introduction of external series gaps as an alternative to explosive disconnects and
hinged installations to reduce cost and improve reliability
• Proliferation of suppliers, leading to a competitive market
TLSAs have proved to be quite reliable in both shielded and unshielded applications if
they are properly applied. Local area reduction of lightning backflashovers by TLSA is
described in detail in References (EPRI 1997; Koch et al. 1985; Shih et al. 1985; EPRI
2005a), and this application has had considerable success. Arresters have also been
applied on unshielded 69- to 230-kV transmission lines, with a 400-kV application
considered for a compact design in Norway (Loudon et al. 1998) and a 500-kV
application in Japan (Ishida et al. 1992). Williamson (2007) reports application of gapless
TLSAs on 138-kV lines of New Brunswick Power Transmission Corporation. They were
quite successful in improving lightning performance in high ground resistance areas, but
did encounter mechanical and hardware problems that constituted valuable experience.
6.5.3 Precautions for TLSA Use on Compact Lines
Calculations based on the V-I characteristics of line surge arresters suggest that the
sharing of energy among parallel arresters on protected unshielded circuits could be a
problem. The EPRI TFLASH program, for example, contains algorithms to calculate the
effectiveness of all reasonable TLSA applications and estimates the total energy
absorption as heat for a statistical distribution of stroke currents. TFLASH compares
calculated energy values with manufacturers’ ratings, and “red flags” excessive events.
However, to date, there have been few failures from excess energy. Experience has
instead shown that manufacturers’ ratings tend to be conservative. Arrester energy testing
performed by EPRI on a specific design of TLSA showed that less than 4% of the
varistor blocks failed at energies lower than twice their rated values (EPRI 2000).
Each ungapped arrester has a MCOV (maximum continuous operating voltage) and a
manufacturers’ designated set of discharge voltages for various arrester surge currents.
This discharge voltage should be at least 10% below the insulator BIL for the highest
expected discharge current, usually negative for most lightning hits to phases on
unshielded lines. A gapped arrester has no MCOV since no voltage exists across the
arrester until a lightning hit occurs.
Most problems with arrester applications have been related to misapplication beyond
manufacturers’ mechanical ratings—for example, when arresters are used to restrain
conductor motion. In the modern compact line, restraint of conductor motion is important
for maintaining minimum phase-to-phase and phase-to-ground clearances under highwind and galloping conditions. These clearances affect the choice of tower head
geometry and insulator configurations, as described in other chapters. The fact that
conductor motion is restrained at the tower also makes application of line surge arresters
more practical. Arresters can be installed in parallel with one of the rigid elements, and
the use of braced post designs with two-section posts can automatically provide the
necessary series gap for consistent operation of gapped TLSAs.
Figure 6.5-1 shows a gapless TLSA installation, with seven discs on a 115-kV line. The
action of wind on the conductor puts tension or compression loads onto the body and
fittings of the arrester, which is a polymer-encapsulated fiberglass tube with a limited
mechanical rating. Mechanical failure of the arrester or fittings by tension or compression
is likely in high winds.
Figure 6.5-1 Polymeric TLSA restraining conductor motion on a 115-kV Line.
Arresters tend to fail shorted, and some means must be applied to disconnect a failed
gapless arrester from the line to prevent a permanent fault. One method utilizes a
disconnector at the line end of the arrester, shown by the small knob in Figure 6.5-1
(EPRI 1997). The power frequency fault current through a failed arrester heats a small
cartridge that blows open the arrester connection to the line, allowing the line breakers to
clear the fault in the newly formed air gap. The hanging connection or arrester supplies a
visible signal that the arrester has failed, and any subsequent lightning flashes to this
newly unprotected location will cause shielding failure flashovers. When arresters are
added without an existing shield wire, arrester disconnectors become another failure
mode. Disconnector operation from direct lightning strikes can sometimes occur even if
the arrester survives.
Figure 6.5-2 shows a gapless arrester application on a compact,
unshielded 138-kV line (Williamson 2007). With five years of operation,
mechanical defect rates of 28% and 0.6% were found with two different
TLSA suppliers, and no arresters were reported to have failed from
excessive energy dissipation.
Figure 6.5-2 138-kV arrester and insulator post assembly.
The second means for preventing permanent faults employs an air gap in series with the
arrester. The simple external air gap sparks over for lightning overvoltages, connecting
the arrester to the protected phase. The air gap has enough dielectric strength to interrupt
the power-frequency follow current at the first zero crossing for normal operating
voltage. This method is popular in Asia and Europe. Older engineers in North America
are familiar with the series gaps needed for successful application of silicon carbide
arresters, capable of interrupting about 100 A of ac power-follow current. The new metaloxide technology performs much better than silicon carbide, with an ac follow current of
less than 1 A. This means that the complex magnetic coils and grading circuits are no
If the arrester fails, breakers clear the faulted arrester, and the gap recovers its dielectric
strength and disconnects the arrester. The use of an external series gap has the advantage
that the arrester varistor blocks and housing are under stress only when the arrester is
operating. This allows the length of the arrester column to be reduced while still meeting
all interruption requirements. The short time lag for the gap sparkover does not
significantly interfere with the TLSA protective characteristics. Since gap sparkover for
an operating arrester limits power follow current, electrode ablation is not usually a
problem, and sparkover directly to the phase conductor is often possible, or to an armor
rod around the conductor in some cases. Care should be taken that either gapped or
gapless TLSAs do not at any time inject significant radio noise into the protected phase.
For a gapped arrester that is not an integral part of the arrester housing, the manufacturer
will specify the minimum gap dimensions to clear an arc to a phase. For unrestrained
conductor motion and a gap that is not an integral part of the arrester housing, it is usually
not necessary to be concerned about lightning performance under maximum wind swing
conditions since experience has shown that most lightning hits occur with little or no
wind swing. However, the gap must always be sufficient to withstand power frequency
voltages, and—for that reason—gapped installations are usually applied to restrained
TLSA housings are invariably polymeric, making them lighter to install and less likely to
fragment in case of a fault than porcelain housings. The housings can be designed to
radially vent high-pressure gases during an internal fault, limiting fragmentation.
Handling issues are particularly important for long TLSAs, and care must be taken that
they are not bent during installation.
6.5.4 Case Study: TLSA Application on Existing 115-kV Compact Line
One applicable and instructive compact phase arrangement uses a three-phase delta wire
configuration on single poles, as shown in Figure 6.5-3.
Figure 6.5-3 Typical 115-kV compact line geometry from 1980, using polymer post and
semiconductive glaze bell insulators (courtesy Ontario Hydro 1980).
The line in Figure 6.5-3 was designed to be a 115-kV upgrade of a 44-kV
subtransmission line, with reliability achieved by meshing of multiple supply lines rather
than through good lightning and contamination performance of any individual line.
With a height h at the tower of 11.3 m, bottom-phase conductor separation of b = 2 m and
a ground flash density of 2 flashes per km2 per year, Equation 6.3-1 suggests that there
will be 24 shielding failures per 100 km per year. The rate of 24 outages per 100 km per
year would not be suitable for adequate power quality in a dual-supply system, given the
number of customers affected by the voltage dips associated with each 115-kV
An unshielded compact transmission line designed today could treat the center phase,
supported on a polymer post, with a line surge arrester. This would make it function as a
shield wire under lightning conditions. The arrester voltage will add to the tower-top
voltage, and this will actually improve the lightning performance of the unprotected
phases by increased coupled voltage to the phases.
It is also possible to stack a pair of post insulators in series to provide the recommended
air gap from central flange to phase conductor, given by those manufacturers that supply
externally-gapped TLSA rather than, or as well as, TLSA with explosive disconnects.
This configuration has been used extensively, with more than a million “Current Limiting
Arcing Horns” used for distribution systems in Japan.
With a top-phase arrester in parallel with the central post insulator at every pole, there
will be no more shielding failures. Instead, there will be some backflashovers on the
unprotected 0.914-m three-bell insulator strings in Figure 6.5-3. Figure 6.5-4 shows how
the backflashover rate of this line is predicted by the FLASH program (IEEE 1997) to
vary with footing resistance, treating the arrester-protected phase as a groundwire.
Figure 6.5-4 Calculated lightning performance of compact 115-kV line with top-phase
A target performance of four outages per 100 km per year can be achieved in this area of
moderate ground flash density if it is possible to maintain the footing resistance of each
tower at less than 20 Ω.
6.5.5 Case Study: TLSA Application on Single-Circuit 138-kV Unshielded Line
Figure 6.5-5 displays one example of a single-circuit 138-kV compact line without shield
wires. The phase conductors are shown as “A,” “B,” and “C”. The following dimensional
data and voltage data apply:
Ground flash density: 10 flashes/km2/year
Insulator BIL: -760 kV
Average footing resistance (concrete): 25 Ω Conductors: Drake 795 kcmil
Line length: 100 km
Span distance: 200 m
Conductor sags: 3 m
Arrester MCOV: 106 kV
Arresters on Phase B
Figure 6.5-5 Single-circuit delta post-insulator line without shield wires.
Table 6.5-1 shows TFLASH lightning flashover performance calculations for the cases of
no arrester on the top phase, arresters on the top phase at every pole, and arresters on the
top phase at every other pole.
Table 6.5-1 Single Circuit Lightning Performance with and ithout Arresters per 100
Arresters on Every Pole
On Top Phase B Only
Arresters on Every Other
Pole on Top Phase B Only
With arresters on the top phase at every pole, there is an approximate 20:1 improvement
over the performance with no arresters. With arresters on the top phase at every other
pole there is little improvement. This is because the flashovers are simply pushed over to
the unprotected poles. Again, caution should be observed that adequate energy capability
exists for the top phase arresters for high charge lightning flashes to the line. For arresters
on the top phase, as shown in Figure 6.3-3, either gapped or ungapped arresters can be
6.5.6 Case Study: TLSA Application on Double-Circuit 138-kV Unshielded Line
Figure 6.5-6 displays a double-circuit 138-kV line with no shield wires. The following
dimensional data and voltage data apply:
Ground flash density: 10 flashes/km2/year
Insulator BIL: -760 kV
Footing resistance (concrete): 25 ohms
Conductors: Drake 795 mcm
Line length: 100 km.
Span distance: 200 m.
Conductor sags: 3 m.
Arrester MCOV: 106 kV
No wood bonding, i.e. no electrical conductors along the wood surface.
Figure 6.5-6 Double-circuit compact line without shield wires.
Table 6.5-2 shows TFLASH lightning performance calculations for Figure 6.5-6 for the
• no arresters,
• arresters on the top and bottom phases of Circuit #1,
• arresters on the top phases of Circuits #1 and #2, and
• arresters on the top phase of Circuit #1 only.
In this case study. the arresters are placed on every tower. The results are presented in
Table 6.5-2 Double-circuit Flashovers with and without Arresters
per 100 Kilometers per Year
Arresters – Top
Phase of Ckt 1
Ckt 1 Only
Ckts 1 & 2
In Table 6.5-2, some flashes to the phases also caused backflashover on other phases,
leading to multiphase-ground faults. For this insulation of 760-kV BIL, every flash
causes a flashover. Even if the first stroke is less than the critical current of about 3 kA, it
is likely that one of the subsequent strokes will exceed 12 kA. When a lightning flash
with high peak amplitude causes a flashover of the stricken phase, the lightning current
then enters the pole ground, and this can—in some cases—increase the ground potential
sufficiently to flash over another phase. The “Backflashovers” column shows the
occurrence of simultaneous multiphase flashovers due to this ground potential rise.
The following conclusions can be drawn from Table 6.5-2:
1. With arresters on both top phases at every pole, only a few backflashovers on the
lower phases will occur per 100 kilometers per year, and the top phases essentially act
as shield wires. The coupling effect is usually higher because of the tight phase
spacing on a compact line, and hence the backflashover rate is better than it would be
with larger phase spacing—i.e., not a compact line. Provided arrester energy
capacities are adequate and arrester long-term environmental durability is
satisfactory, this should be an adequate replacement for shield wires.
2. With arresters on the top and bottom phases of only one circuit at every pole,
backflashovers of the protected circuit rarely occur. The unprotected circuit will
frequently flash, primarily from every strike to that side of the line.
3. Arresters connected only to the top phase of one circuit at every pole will perform
almost as well as arresters on both top and bottom phases of that circuit. This result is
dependent on the footing resistance; for high values of footing resistance, flashovers
to the bottom phase will increase. Roughly speaking, any phase with an arrester
connected to it acts somewhat like a shield wire.
6.6 CALCULATION OF ARRESTER ENERGIES
The principal problem in planning TLSA protection on unshielded lines is determination
of the minimum size arrester block to use. The greater the diameter of the block, the more
expensive the arrester, but an inadequate block diameter (or quality) can lead to an
unacceptable failure rate. Blocks can puncture from excessive surge voltage or fail from
excessive heat energy injection. Calculation of maximum expected TLSA energies can
involve substantial errors because of a paucity of field data. Failures can be expected to
occur primarily when very severe flashes strike the line, but very little information about
the probability distributions of very severe flashes exists. Severe flashes constitute only a
few percent of all available data, and will vary with latitude and other factors. Where
transmission lines have shield wires, experience has shown that TLSAs can function very
well in suppressing insulator voltages, and even without shield wires, TLSAs should
function satisfactorily if the quality and diameter of the arrester metal-oxide blocks are
sufficient. Figure 6.6-1 shows energy test results on some earlier model ungapped TLSAs
having 41 mm diameter blocks.
Figure 6.6-1 Example TLSA failure probabilities vs injected energy in kilojoules per
MCOV for an ungapped arrester design with 41-mm diameter blocks.
The arresters began to fail at slightly above the manufacturer’s rating of 2.2 kilojoules per
MCOV, but the 50% failure rate did not occur until 9.8 kilojoules per MCOV was
reached. Appendix 6.1 discusses how these failure probability curves can be used to
calculate TLSA energies during lightning hits on unshielded lines. On lines with shield
wires, computer programs—such as EPRI TFlash—signal conditions wherein TLSA
energies are exceeding their manufacturers ratings. The frequency distribution is not
normal, and resembles a Weibull distribution skewed toward the low end.
1. Lightning Performance. From the standpoint of lightning performance, compact
transmission lines—if designed properly—can have adequate lightning performance.
Compact transmission lines usually collect a smaller number of lightning hits than
conventional lines because of their smaller footprint. However, compaction may
require some sacrifice of air gap clearances and insulator lengths, so that critical
stroke currents for flashover can be less. Design comparisons of lightning
performance of proposed compact lines and conventional lines can only be made with
a comprehensive lightning simulation program such as EPRI TFlash.
2. TLSAs. Since the ultimate in compaction requires removal of shield wires, interest is
increasing in utilization of TLSAs to limit compact line lightning flashovers, and
designs up to 500 kV have been developed. TLSAs can help maximize compaction of
3. Gapped versus Gapless Designs. TLSAs are widely available in both gapless and
externally gapped designs. An early decision must be made as to which is more
applicable for a specific line design. Gapless arresters have an advantage for some
tower geometries, particularly where conductors are allowed to swing. The blown
disconectors clearly signal either arrester failure or disconnector failure. However,
some disconnectors have also proved to have poor mechanical reliability.
Externally gapped arresters can sometimes be positioned on the pole horizontally and
in parallel with the phase conductors, so that sparkover can be made to occur directly
from a restrained conductor to the live end of the arrester. This gives a simple and
convenient installation. Gapped arresters have a further advantage that voltage exists
on the arrester only during the sparkover process, permitting shorter arresters with
reduced leakage distance and no concern over heating from operating voltages.
Factoring in the improved reliability with rigid connections, externally gapped
arresters would seem to be a better overall choice for compact line designs that use
line post insulators, and gapless designs rated for the necessary tension, cantilever,
and compression loads would seem more appropriate for restraining motion with
4. Product Test Data. Since there can be a significant variation in mechanical and
electrical quality and energy capability of TLSAs now being marketed, it becomes
important to have access to good test data of failure probabilities and modes for any
5. Inspection and Maintenance Strategies with TLSAs. If TLSAs are used to
eliminate shield wires, it should be expected that occasional failures due to severe
lightning hits will occur, and that there will be mechanical problems as well.
Inspection, replacement, and maintenance strategies should be factored into the lifecycle cost analysis in comparison to 25- to 50-year, maintenance-free experience with
APPENDIX 6.1 CALCULATION OF ARRESTER FAILURE
PROBABILITIES OF UNSHIELDED LINES
Arrester failure probabilities on an unshielded line depend on a variety of parameters:
• Gapped or direct-connected arrester designs and their ratings and quality
• Distribution of arresters on a struck phase
• Span distances
• Tower footing resistances
• Regional ground flash densities
• Shielding effects of nearby objects (trees, structures, regional topology, etc.)
• Statistical variation of storms and flash intensities from year to year
The following software procedures are suggested to make the calculations in the absence
of dedicated software facilities:
Estimate number of hits to the line per year.
The mean number of hits per year N to a compact line can be estimated using Equation
6.3-1, recognizing that the hit rate can double in some years. Using this number N—if
regional information of the percentage of positive flashes is not known—assume that
10% of N flashes to the line will be positive, and that any positive flash to a line will be
severe enough to fail an arrester. This is an overly simple, but conservative, assumption.
For each line section of nine contiguous structures, select a phase protected by TLSA,
and determine an equivalent circuit for a lightning hit to the phase at center structure 5.
Figure A6.1-1 shows the conceptual circuit for a hit to a protected phase.
Figure A6.1-1 Conceptual circuit for a TLSA energy analysis.
The volt-ampere characteristics of the arresters “A” (all arresters assumed to have the
same MCOV) can be approximated by:
I A = αV Aβ A6.1-1
IA = arrester current, kA.
VA = instantaneous voltage across the arrester, kV.
α,β = constants to be determines from manufacturer’s specifications. Test values
of VA at IA = 5 kA and 40 kA can be used to calculate α and β.
The resistors RG are tower dynamic footing resistances, and can be assumed to be the
low-frequency values if better data are not available. Note that the tower inductance is
not shown, since at maximum arrester current, the dI/dt and inductive voltage drops are
zero. For gapped arresters, the gap can be represented by an open switch in series with
the arrester. The switch closes when the voltage across it reaches a preset value
determined by the manufacturer’s acceptance test data. Once closed, the switch remains
closed for the duration of the event.
The impedance ZO is the surge impedance of the struck phase, degraded somewhat by
corona. A value of 500 ohms for a single conductor should be sufficient for these
calculations, and somewhere around 300 ohms for a bundle.
Select a flash severity S to be used in the EMTP calculation.
Assigning a “severity” of 8 to a flash to be used in an EMTP calculation means that 8 out
of 10 random flashes will have a charge magnitude less that this flash. A severity of 5
assigned to a flash means that 5 out of 10 random flashes will have a charge magnitude
less that this flash. It is unrealistic to expect TLSA on a compact unshielded line to
withstand all possible lightning flashes, since some flashes will be extremely severe, and
consequently a few TLSA failures should be expected. Figure A6.1-2 is taken from a
flash coulomb probability curve, and provides this “Severity Rating” on a scale of 1-10 in
terms of the corresponding coulombs delivered for a negative flash. For example, the
severity of 8 mentioned above corresponds to a total flash charge of approximately 16
coulombs, whereas a severity rating of 10 corresponds to approximately 100 coulombs
and should be extremely rare.
Using the selected flash severity, find the corresponding flash coulombs delivered.
Figure A6.1-2 is used to find the total charge delivered
Figure A6.1-2 Total negative flash coulombs vs. severity rating.
Calculate the stroke current waveshape and stroke magnitude to be used in the EMTP
Metal-oxide arresters are very nonlinear devices, and their heat absorption for a total
flash coulomb content can vary widely for different current waveshapes, primarily
because an arrester is always in parallel with the constant surge impedance of a conductor
and is influenced by current-sharing with other arresters on the same phase. The
conventional 8 x 20 current waveshape used to test station arresters does not represent a
flash current injected into a phase. This flash current will usually consist of a sequence of
several strokes over a time period of many milliseconds, and the low currents between
the high current peaks can be carried off by the phase surge impedance with very little
flowing into the arresters. Since inductance is absent in Figure A6.1-1, dI/dt becomes of
little importance in adiabatic heating of the arresters, and a simplified stroke current
waveshape similar to that shown in Figure A.6.1-3 is suggested, wherein four of these
strokes in rapid succession can be assumed to constitute a flash.
Figure A6.1-3 Suggested simplified flash current waveshape for TLSA
If four identical strokes with Figure A6.1-3 waveshapes constitute a negative flash, then
the peak current Imax of each stroke is given by:
I max = 2.5QF A6.1-2
QF = total charge (coulombs) delivered by the flash.
Imax = crest current, kA.
Since the strokes are identical, only the arrester energy delivered by one stroke has to be
calculated by EMTP and the result multiplied by four to get the total energy injected into
the arrester at the flash location.
Use EMTP to calculate the total kilojoules injected into each of the arresters of Figure
This calculation is made using one of the stroke current waveshapes of Figure A6.1-3.
Divide the total kilojoules in each arrester by the arrester MCOV and multiply by four
since the lightning flash consists of four of the Figure A6.1-3 transients.
Compare the maximum calculated kJ/MCOV by test data exemplified by Figure 6.6-1.
As a conservative approximation, if the calculated kJ/MCOV equals or exceeds the 50%
failure rate, assume that a failure will occur. This assumption is acceptable since—as
shown by Figure 6.6-1—a substantial number of failures will occur below the 50% value.
A more accurate approach would utilize a distribution of charge magnitudes in a flash
(see Figure A6.1-4), calculate the failure rate for each charge magnitude, and sum all the
failures. However, this will require dedicated software, and the distribution of negative
charge at high values in Figure A6.1-4 is dubious due to lack of field data.
Sum positive and negative failures to find total expected failures for this selected arrester
If the number of failures per 100 km per year is unacceptable, investigate the use of
larger diameter blocks in the arresters. This requires test data similar to Figure 6.6-1 and
volt-ampere test data for the proposed arrester to determine α and β in Equation A6.1-1.
Figure A6.1-4 Distribution of expected charge in a negative flash
APPENDIX 6.2 TESTING TRANSMISSION LINE SURGE
ARRESTERS PRIOR TO INSTALLATION
If TLSAs are to replace shield wires on a compact transmission line, it is vital that they
survive existing regional electrical and environmental challenges over a period of many
years. Challenges will include:
• Weather (icing, solar radiation, internal water accumulation, condensation, etc)
• Mechanical motion
• Continuous and momentary electrical stresses and energy challenges
• Installation abuse
• Corrosion of metallic parts
Arrester blocks can fail for a variety of reasons:
• Thermal runaway from excessive heating when exposed to severe lightning currents
or excessive power frequency voltages
• Insulation breakdown at the insulating collar surrounding one or more arrester blocks
• Insulation puncture through a block leading to a cascading failure
• Block cracking
Most of these failures occur near the maximum energy levels of the blocks, usually
substantially beyond the manufacturers’ ratings, and in the field may not be detectable in
some cases until some sort of visible mechanical disruption occurs. Adequate quality
control should be ensured by electrical and environmental tests prior to line
TLSAs may be subject to environmental challenges not anticipated by the manufacturer
or overlooked in the electrical or mechanical design. Accelerated life tests in
environmental chambers—including voltage, temperature, humidity, UV light, and
mechanical cycling—are recommended. Figure A6.2-1 shows a 500-kV accelerated
aging environmental chamber at the EPRI Lenox, Massachusetts Research Center.
Although mostly utilized for insulator life testing, the facility can evaluate a variety of
arresters. A few weeks can simulate a year or more of normal exposure. The aging
acceleration varies from 7 to 14 depending on the type of test.
Figure A6.2-1 EPRI 500-kV Aging Test Chamber.
High Current Electrical Stresses
Severe lightning transient currents passing through a TLSA create severe electrical
stresses in the arrester blocks, and poor quality or poor design can result in catastrophic
failures. This kind of electrical stress can be applied to a proposed arrester by an impulse
generator. The impulse test can consist of a single current surge test of 40 kA or to the
maximum available current from the generator. Arrester current oscillograms are then
inspected and compared with short-circuit oscillograms and lower arrester current
oscillograms to look for signs of block failure by abrupt increases in current.
Temporary Overvoltage Acceptance Tests
Every gapless TLSA must withstand expected temporary overvoltages (TOV) that will
occasionally be present during faults and abnormal system conditions. The manufacturer
will often supply TOV ratings for various durations of overvoltage. For gapped arresters,
the gap should not flash over during TOV periods accompanied by simulated severe
weather conditions. If a flashover does occur, the gap must clear when the voltage drops
to normal. A gapless test arrester is first energized at normal voltage via a hipot
transformer and the arrester current and normal watts loss observed. Then a temporary
overvoltage from an autotransformer and current limiting resistor is applied to the arrester
via a single-pole double-throw switch for a few seconds—arrester current and watts
measured in the interval—and normal operating voltage then reapplied by the switch. The
decay in arrester current and watts loss during the following normal voltage is then
monitored as the arrester recovers. If desired, this test can be repeated at higher and
higher overvoltages until failure to establish failure probability curves.
TLSA Energy Failure Tests
For line arresters, failure energies can exceed 1000 kilojoules for larger arresters, and this
exceeds even the energy capability of large impulse generators. Under these conditions,
Ringler et al. 1998 showed that TLSA energy failure tests can be carried out by injecting
a 60 Hz overcurrent into a test arrester and oscillographically measuring the total
kilojoules into the arrester until failure occurs. This test takes advantage of the adiabatic
accumulation of heat per cycle. The product of test duration and current amplitude
(charge) was found to be constant over five orders of magnitude, from 0.8 A to 35 kA,
because failures occurred before significant amounts of heat could escape from the
blocks. This test is essentially a continuation of the TOV and transmission line discharge
tests mentioned above but at a higher voltage and current until failure.
Agrawal, A. K., H. J. Price, and S. H. Gurbaxani. 1980. “Transient Response of
Multiconductor Transmission Lines Excited by a Nonuniform Electromagnetic Field.”
IEEE Transactions on Electromagnetic Compatibility. Vol. EMV-22. May. pp.119-129.
Anderson, R. J. and A. J. Eriksson. 1980. “Lightning Parameters for Engineering
Application.” Electra. No. 69. March. pp. 65-102.
Boccippio, D. J., K. L. Cummins, H. J. Christian, and S. J. Goodman. 2001. “Combined
Satellite and Surface Based Estimation of the Intracloud-Cloud-to-Ground Lightning
Ratio Over the Continental United States.”Monthly Weather Review. Vol. 129. January.
Chisholm, W. A., J. P. Levine, and P. Chowdhuri. 2001. “Lightning Arc Damage to
Optical Fiber Ground Wires (OPGW): Parameters and Test Methods.” Proceedings of
IEEE Power Engineering Society Summer Meeting. pp. 88–93.
Chowdhuri, P. 1996. Electromagnetic Transients in Power Systems. John Wiley & Sons.
CIGRE. 1991. Working Group 01 (Lightning) of Study Committee 33 (Overvoltages and
Insulation Coordination). “Guide to Procedures for Estimating the Lightning Performance
of Transmission Lines.” Brochure #63. Paris. October.
Cummins, K. L., E. P. Krider, and M. D. Malone. 1998. “The US National Detection
Network and Applications of Cloud to Ground Lightning Data by Electric Power
Utilities.” IEEE Transactions on Electromagnetic Compatibility. Vol. 40. No.4.
November. pp. 465-480.
Darveniza, M. 2007. “A Practical Extension of Rusck’s Formula for Maximum Lightning
Induced Voltages that Accounts for Ground Resistivity.” IEEE Trans. Power Delivery.
Vol. 22. No.1. January. pp 605-612.
EPRI. 1978. Transmission Line Reference Book – 115-138 Compact Line Design.
Electric Power Research Institute. Palo Alto, CA.
EPRI.1982. Transmission Line Reference Book – 345 kV and Above. Second Edition,
Electric Power Research Institute. Palo Alto, CA.
EPRI. 1997. Guide for Application of Transmission Line Surge Arresters 42-230 kV.
Report TR-108913. Electric Power Research Institute. Palo Alto, CA.
EPRI. 2000. Transmission Line Surge Arrester Impulse Energy Testing, Progress Report.
Report No. 1000461. Electric Power Research Institute. Palo Alto, CA. December.
EPRI. 2002. Tower Grounding and Soil Ionization Report. Report No. 10011908. Palo
EPRI. 2005a. EPRI AC Transmission Line Reference Book – 200 kV and Above. Third
Edition. Report 1011974. Electric Power Research Institute. Palo Alto, CA..
EPRI. 2005b. TFlash Users Guide – Version 4. Report TR1011388. Electric Power
Research Institute. Palo Alto, CA. May.
Eriksson, A. J. 1987. “The Incidence of Lightning Strikes to Power Lines.” IEEE
Transactions on Power Delivery. Vol.2. No. 3. pp. 859-870.
Hileman, A. R. 1999. Insulation Coordination for Power Systems (book). New York.
Hoffmann, J. N., R. L. de Souza, N. Prosdócimo, I. Moreira, F. V. Swinka. “Linha de
Transmissäo Urbana Compacta Experimental em 230 kV.” XVII SNPTEE. Minas Gerais,
Brazil. October 19-24.
IEC. 2003. 60794-1-2-2003-05. Optical Fibre Cables – Part 1-2: Generic Specification—
Basic Optical Cable Test Procedures. Method H2: Lightning Test Method for Optical
Aerial Cables along Electric Power Lines.
IEEE. 1985. Working Group (J.G. Anderson, W.A. Chisholm, I. S. Grant, A. R. Hileman,
W. Janischewskyj, G. E. Lee, V. G. Longo, D. Parrish, N. Roukos, E. Whitehead, and J.
T. Whitehead). “A Simplified Method for Estimating Lightning Performance on
Transmission Lines.” IEEE Transactions on Power Apparatus and Systems.. Vol.104.
No. 4. April. Pp. 919-927.
IEEE. 1991. “IEEE Standard C62.22. IEEE Guide for the Application of Metal-Oxide
Surge Arresters for Alternating Systems (ANSI).” Institute of Electrical and Electronic
Engineers, Service Center, 445 Hoes Lane, Piscataway, NJ 08855-1331.
IEEE. 1997a. C62.22–1997. “IEEE Guide for the Application of Metal-Oxide Surge
Arresters for Alternating Systems.” IEEE Standards Association. 445 Hoes Lane,
Piscataway, NJ 08855-1331.
IEEE. 1997b. 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. 1998. “IEEE Standard Techniques for High-Voltage Testing.” IEEE Standard 4.
Institute of Electrical and Electronic Engineers, Service Center, 445 Hoes Lane,
Piscataway, NJ 08855-1331.
IEEE. 1999. C62.11 – 1999 “IEEE Standard for Metal Oxide Surge Arresters for AC
Power Circuits.” IEEE Standards Association, 445 Hoes Lane, Piscataway, NJ 088551331.
IEEE. 2005. “Parameters of Lightning Strokes: A Review.” Task Force 15.09 on
Parameters of Lightning Strokes. IEEE Transactions on Power Delivery. Vol. 20. No. 1.
Ishida. K., K. Dokai, T. Tsozaki, T. Irie, T. Nakayama, H. Fujita, K. Arakawa, and Y.
Aihara. et al. 1992. “Development of a 500 kV Transmission Line Arrester and its
Characteristics.” IEEE Transactions on Power Delivery. July. Vol. 7. No. 3. pp. 12651274.
Koch, R. E., J. A.Timoshenko, J. G. Anderson, and C. H. Shih. 1985. “Design of Zinc
Oxide Transmission Line Arresters for Application on 138 kV Towers.”IEEE
Transactions on Power Apparatus and Systems. October. pp. 2675-2680.
Loudon et al “A Compact 420 kV Line Utilizing Line Surge Arresters for Areas With
Low Isokeraunic Levels”, CIGRE 1998 Session, Paper 22/33/36-08, Paris, France/
McGorman, D. R., M. W. Maier, and W. D. Rust. 1984. Lightning Strike Density to the
Contiguous United States from Thunderstorm Duration Records. Report to U. S. Nuclear
Regulatory Commission. NUREG/CR - 3759.
Narita,T, T. Yamada, A. Mochizuki, E. Zaima and M. Ishii. 2000. “Observation of
Current Waveshapes of Lightning Strokes on Transmission Towers”. IEEE Transactions
on Power Delivery. Vol. 15, No. 1. January. Pp 429-435.
Rachidi, F. et al. 1997. “Response of Multiconductor Power Lines to Nearby Lightning
Return Stroke Electromagnetic Fields.” IEEE Transactions on Power Delivery. Vol. 12.
Ringler, K. G., P. Kirkby, , C. Erven, M. Lat, and T. Malkewicz. 1998. “The Energy
Absorption Capability and Time-to-Failure of Varistors Used in Station-class MetalOxide Surge Arresters.” IEEE Transactions on Power Delivery. Vol. 12. No. 1. January.
Rizk, F. A. M. 1990. “Modeling of Transmission Line Exposure to Direct Lightning
Strokes.” IEEE Transactions on Power Delivery. Vol. 5, October. Pp 1983-1997.
Rusck, S. 1958. “Induced Lightning Overvoltages on Power Transmission Lines With
Special Reference to the Overvoltage Protection of Low Voltage Networks.”
Transactions of the Royal Institute of Technology. Stockholm, Sweden, no. 120.
Shih, C. H., R. M. Hayes, D. K. Nichols, R .E. Koch, J. A. Timoshenko, and J. G.
Anderson.1985. “Application of Special Arresters on 138 kV Lines of the Appalachian
Power Company.”IEEE Transactions on Power Apparatus and Systems. October. Pp
Takami, J. and S. Okabe. 2007. “Observational Results of Lightning Current on
Transmission Towers.” IEEE Transactions on Power Delivery. Vol. 22. no. 1. January.
Williamson, J. 2007. “Experience at NB Power with Surge Arresters on Transmission
Lines.” 2007 World Congress & Exhibition on Insulators, Arresters & Bushings, Brazil.
WMO (World Meteorological Organization). 1953. “World Distribution of Thunderstorm
Days.” WMO. No. 21, Part 2. Geneva, Switzerland. Also www.wmo.int.
Zanetta, Jr, L. C. 2003. “Evaluation of Line Surge Arrester Failure Rate for Multipulse
Lightning Stresses.” IEEE Transactions on Power Delivery. Vol.18. No. 3. July. pp 796801.