See

discussions, stats, and author profiles for this publication at: http://www.researchgate.net/publication/3862593

Calculation of lightning flashover rates of

overhead distribution lines. A comparative study

CONFERENCE PAPER · FEBRUARY 2000

DOI: 10.1109/PESS.2000.867395 · Source: IEEE Xplore

CITATIONS READS

3 68

3 AUTHORS, INCLUDING:

Pritindra Chowdhuri
Tennessee Technological University
68 PUBLICATIONS 569 CITATIONS

SEE PROFILE

Available from: Francisco Gonzalez-Molina

Retrieved on: 09 November 2015
 CAILCULATION OF LIGZ€"lNGE"LASEMIVERRATES OF OVEREEAD
DISTRIBUTION LINES. A COMPARATIVE STUDY
Jum A. Martinez Francisco GonzbMoIina Pritindra Chowdhuri
Departamat d'Enginyeria UMca Center for Electric Power
Universitat Polithica de Catalunya Tennessee Technologid University
Barcelona, Spain Cookeville, TN 38505, USA

Abshnct - The estimation of the lightning performance of

overhead distribution Unes presents many u n c e d t i e s due to
-
strike directty to the line [l] [3]. Fig. 1 shows the
application to both unshielded and shielded lines. For
the random nature of the lightning phenomenon and the lack instance, if the line is unshielded and the download leader
of reliable data. Because of this, simplified methods to calculate is vertical, a stroke with a current peak I will strike the
the number of flashovers in an overhead line can be very us& phase if its jinal path is to the left of the intersectionpoint
and so effective as a rigorous simulation with advanced models.
This paper presents a comparative study of some approaches A, being the arc centered on the phase and having radius rs
proposed for calculating lightning flashover mtes of overhead calculatedby means of the foIlowing expression
distributionlines. q = a.Iy (1)
Keywords: Distribution Systems, Lightning Overvoltages,
Based on this, the mini" distance, ymm, for
Electrogeometric Model, Backflashover, Shielding Failme,
unshielded linesis obtained as follow
Induced Overvoltages. y, = , , k W for P-r, > h
(2)
y,,=q for P - q S h
L INTRODUCTION In this paper cc = 10, fl=0.9, y = 0.65.
Lightning is a major cause of overhead distribution line vertical
stroke
failures. A lightning stroke cm originate an memollage
which can cause either a failure in the affected line or
propagate along the line and damage other equipment
Insulation flashovers of overhead distribution lines can be
produced by one of the following caws
0 the discharge strikes ground close to the line,being able
to induce a voltage greater thanthe dielectric strength
the flash strikes a phase conductor either because of
ineffective shielding of shield wires or because these
cables are not installed
0 the flash strikes a shield wire or a pole; in both cases a
part of the stroke current pqagates to &round where it a) Unshielded line
will originate successive reflections which catl cause a
voltage greater than the dielectric sbrength
The calculationof lightning flashover rates (FOR)relies
on two stages: lirst, a method to distinguish strokes to line
from those to grwnd; second, a method to calculate
overvoltages produced by strokes to phase conductors,
shield wires, and ground. This paper presents a comparative
study of several methods used to calculate flashover rates of
overhead distribution lines. All methods have been applied
to both shielded and unshielded lines.

II. TEIE ELECTROGEOMETRIC MODEL

A stroke close to a line can strike either line conductors or
ground, depending on the peak current ofthe return-stroke. b) Shielded line
The ElectrogeometricModel @GM) allows calculating the Fig. 1. The electtogeometricmodel.
minimum distance below which a lightning stroke will

(c) 2000 IEEE

0-7803-6420-1/00/$10.00 2577

Authorized licensed use limited to: UNIVERSITAT ROVIRA I VIRGILI. Downloaded on June 03,2010 at 08:50:12 UTC from IEEE Xplore. Restrictions apply.
 El.LIGHTNINGPARAMETERS XV. FLASHOVERRATES
In the calculation of lightning owrvoltages, a stroke is FOR calculation will depend on the line configuration
seen as a current source that can have positive, negative, or for unshielded lines, it is obtained from adding failures
both polarities in the same discharge.Waveforms strongly caused by direct strokes to phase conductors the -
depend on the polarity of lightnh~g,and even the presence Direct Lightning Failure Flashover Rate @TATOR) -,
of high objects, mountains or structures, cafl distort and those induced by nearby strokes to ground the -
lightning parameters. Indirect LightningFailure Flashover Rate (ILFFOR)
In general, it is assumed that the probability density for shielded lines, it is obtained from adding failures
function of a lightning variable is given by a log-normal -
caused by strokes to shield wires the Back Flashover
distribution [4], [5] Rate @FOR) -,by strokes to phase conductors - the
Shielding Failure Flashover Rate (SFFOR) and
-y

failures induced by nearby strokes to ground

(ILFFOR).
where: x and C J are ~ the mean value and the standad AU calculations have been made taking into account
deviation of the logarithm of the variable. the following assumptions
If two or more lightning variables are involved, a joint lightniq-stroke peakvalues range from 1to 200 kA
probability density function has to be used. For two time-to-crestvalues range fiom 0.5 to 10.5 ps
variables, this function has the followingexpression the withstand dielectric strength of line insulators
corresponds to the 2-)ls point on the volt-time w e ;
thisvalue is approximated by 1.5 CFO [3]
the power frequency voltage is neglected
a stroke is representedas a double-ramp current source
the line is in open ground
where p is the d c i e n t of correlatioa If the variables all direct strokes to phase conductors will cause
are assumed independently distributed then p = 0, and flashover.
P(X, Y) =P(X) ’P O (5) A short slltnmary ofthe procedures developed to obtain
In this work, FOR are calculated using the probability FOR for both types of linesis presented below.
density function of the peak ament, Po, and the joint
probability density W o n of peak current and time-to- 4.1. Unshielded lines
crest, P a t& assumingthe followingparameters [11- 141
1.33, f = 61.1 kA; OM= 0.55, T; ~ 3 . 8 3p~ 4.1.1. Procedure A1
for current peaks equal or smaller than 20 kA It is based on the application of the EGM and the
CTM 0.6, = 33.3 kA;~ ~ = 0 . 5 5 , = 3.83
-
method presented in [SI I7J for calculathg overvoltages
i n d u c e d by nearby strokes to ground. The procedure can be
for currentpeaks greater than20 kA. summarized as follows [8], [9]
AJl probability density h c t i o n s are n&ted to the first a) The “um stroke cunent is varied from 1 to 200
stroke, which usually presents a higher peak value. A kA in incremats of 0.5 kA
coeflicient of correlation p = 0.47 has been chosen. b) For each current peak, the mini” distance from
The cumulative probability of the current peak, I, which lightning will not divert to the line, ymino, is
exceeding a given value, b, is derived from Po,and calculatedby applying the EGM, see Fig. la, DLFFOR
approximatedby the following equation [11 [3] - is then increased using the following expression
P(I2 io) =
1 DLFFOR(1) =DLFFOR+,, +
(6)
1+ (i&31)26 y,,,i,,(I) .P(I)*AI.2Ng110 (9)
An important parameter is the ground flash density c) For each current peak, the value of the time to crest, tr,
(GFD) per km* and year, N8’If this quantity is not kaown, is varied from 0.5 to 10.5 ps in incrementS of 0.5 p.
it can be estimated from the number of thunderstorm days d) For each pair (I, t& the maximum distance beyond
per year, Td,or the number of thunderstorm hours per year, which lightning will not cause flashover, ymaxKt9,is
Thyby UShg the fOfiOWhg e q d O l l S [3] c a l a using the method presented in [5]; ILFFOR
N, = 0.04 TiB (7) is then i n d as follows
N, = 0.054 Ti1- (8) IL^FPOR(I,tf)= ~ O R ~ I . t f - * t+f )
6-(,,tf) -Y~(l~).P(I,tf).bI.Atf 110 .” (10)

(c) 2000 IEEE

0-7803-6420-1/00/$10.00 2578

Authorized licensed use limited to: UNIVERSITAT ROVIRA I VIRGILI. Downloaded on June 03,2010 at 08:50:12 UTC from IEEE Xplore. Restrictions apply.
 4.2. Shielded lines
Procedures developed to analyze the performance of
shielded lines are based on methods similar to those used for
d e l d e d lines. However, insulation flashovers can now
be caused also by backflash A short explanation of the
principles used with each procedure are presented below.

4.2.1. procedure B1
It is based on the same principles that Procedure Al.
The EGM is used to distinguish lightning strokes to poles
or shield wires h m those to phase conductors or ground,
see Fig. lb. The induced overvoltages originated by a
nearby stroke to ground are calculated using Chowdhuri’s
method, taking into account shield wire effects [6].

r- 4.2.2. procedure B2
The EGM is used again to distinguish lightning strokes
to a pole or shield wire from those to phase conductors or
to ground The induced overvoltages originated by a
nearby stroke to ground are calculated using Chowdhwi’s
method for unshielded lines, and reducedby the protective
xatio proposed by R w k , as discussed in [6].

4.2.3. Procedm B3
Eriksson’s equation is used to obtain the number of
flashes to poles and shield wires, and the EGM is applied
to deduce which flashes will hit phase conductors or
ground. Overvoltages induced by nearby strokes to ground
are evaluatedfrom Rusck’s equation [3], see also [6].
The BFOR calculation differs a little from one
procedure to other; however, for all cases, a shield wire
Fig 2. Procedure for unshieided lines. performance is estimated using the approach presented in
[2] and [3], and pole effects are simplified due to their
short heights. For each current peak value and each shield
4.1.2.Procedure A2 wire,the following steps are used if the EGM is chosen
It is also based on the EGM to obtain the mini” a) The minimm distance from which lightning will not
distance for which a stroke will not divert to the line, and divert to the shield wire line, y d , is calculated by
the maximum distance beyond which a stroke to ground applying the EGM
will not cause flashover. However, the maximum voltage b) For each phase conductor, the probability of having an
which is induced in a line will be now calculated by means insulation flashover, F, is c a l c u l a using the
of the Rusck’s equation [3]. As this calculation does not algorithm implemented in FLASH program [2], and
depend on the time-to-crest of the discharge, the internal BFFOR is increased using the expression
loop shown in Fig. 2 will not be needed. BFOR(I,pIl, = BFOR(I.pt4, +
(12)
4.1.3. Procedure A3 k * Y ~ ( ~ ) * P ( I ) - A I * F , */10
~N,
The DLFFOR is estimated by Eriksson’s equation [101 where k is a factor to take account of the span effect. In
N, (28. ho.6+b) this paper k = 0.6 [I].
DLFFOR = (11) The flowchart of the procedure B1 is presented in Fig.
10
being h the pole height and b the structure width, both in 3. The flowchart of the procedure B2 would be similar, but
without the internal loop in which the timetocrest of a
meters. The LFFOR is estimated by applying again the stroke is vafied. For procedure B3, the application of
EGM and the Rusck’s equation. Therefore, with this Eriksson’s equation will introduce some minor differences
procedure the current loop peak is applied to obtain only in the BFOR calculationtoo.
the ILFFOR.

0-7803-6420-1/00/$10.00 (c) 2000 IEEE 2579

Authorized licensed use limited to: UNIVERSITAT ROVIRA I VIRGILI. Downloaded on June 03,2010 at 08:50:12 UTC from IEEE Xplore. Restrictions apply.
 Furthermore, the results are similar for both horizontal and
vertical configurations. The horizontal configurationhas a
wider structure, but the vertical line is taller.
EFFOR values are very dependent on the calculation
method,but it is evident that the rem-stroke velocity is the
parameter which has the strongest influence. An important
aspect is the trend of the ILFFOR with respect to this
panmeter : the magnitude of the induced overvoltages
decrease, as well as ILFFOR values, when the return-stroke
velocity hcreases if Chowdhuri’s method is used, but the
trend is the contrary when Rusck’s equation is used. For an
average velocity of !NO00 W s all methods provide similar
&S.

5.2 Shielded lines

AU methods provide values of similar order for BFOR.
The differences between line configurations is mainly due

&
I I
1
Ycu
-
to the height ofthe shield wire.
I 1 An important parameter in BFOR calculations is the
pole footing resistance. To estimate its influence, a
I - I + Al
sensitivity analysis has been performed. Table 2 shows the
results for three Werent values. AU calculations were
made assuming that the return-stroke velocity was 90000
W s . By comparing these results and those shown in
Table 1, it is obvious that zhis parameter is the most
important one in lightning performance of shielded
distribution lines.

Table 2 - Effect of the Pole footinn resistance

no

,&,
Fig 3. Procedure for shielded lines.

V. EXAZMPLES
The above procedures have been applied to the
evaluation of two configurations, considering both
unshielded and shielded lines. Table 1 shows line and
lightning parameters, as well as the performance of each
line. Calculations for unshielded lines were performed
assuming that the rms line voltage was 25 kV, and the
lines were built with an average span length of 30 meters. Although a perfect shielding is not provided for any
line configuration, the SFFOR is very small in all cases.
The most important &ect of shield wires is the
5.1 Uloshielded lines
significant reduction of ILFFOR whatever method used
From the results presented in the table, it is obvious This agrees with many previous results presented in the
that the DLFFOR is practically independent on the litffature [6],[12].
method, since all procedures give similar rates.

0-7803-6420-11001$10.00 (c) 2000 IEEE 2580

Authorized licensed use limited to: UNIVERSITAT ROVIRA I VIRGILI. Downloaded on June 03,2010 at 08:50:12 UTC from IEEE Xplore. Restrictions apply.
 Table 1 - Lightninguerformance of overhead distributionlines

Line Procedure A1 Procedure A2 Procedure A3

30000
I

1
12.399
74.661 1 12.399
0.855 I 11.419
0.855

1 1
87.060 13.254 12.274
I
Direct I
12.399 12.399 11.419
90000 2.173 1.577 1.577
Total 1 14.572 13.976 12996
4 p h = 1 0 ~
CFO = 150 kV
Ng= 1 fykm2tyear
150000
Direct

Total
12.399
0.013
12,412
I 12.399
2.619
15.018
I 11.419
2.619
14.038
Direct 12.594 12.594 12.409
30000 Indirect 85.314
Total 97.908 14.378 14.193
Direct 12.594 12.594 12.409
90000 Indirect 2.952 2.952
Total 15.729 15.546 15.361
10 mm
bph = Direct , 12.594 12.594 12.409
CFO = 150 kV 150000 Indirect 0.024 4.482 4.482
NR= 1 fl/lcm2/year Total ~ 12.618 17.076 16.891
t
Return-stroke Flashover Rate
Line Procedure B1 Procedure B2 Procedure B3
velocity @U/ s) (fl/lOOkm/yr)
0.042 0.042 0.042
' 6.382 6.382 5.700
1 47.215 28.9% 0.138
53.639 35.420 5.880
0.042 0.042 0.042
6.382 6.382 5.700
' 0.613 0.067 0.349
Total 7.037 6.491 6.091
ShieldingFailure 0.042 0.042 0.042
Baclcflikhover 6.382 6.382 5.700
150000
0.002 2x4 0.725
Total 6.426 6.424 6.467
ShieldingFailure 0.003 0.003 0.003
6.933 6.933 6.727
52.110 39.276 0.342
Total 59.046 I 46.212 7.072

90000 1 BacFkfl
ShieldingFailure
I
1
0.003
6.933
0.801
0.003
6.933
0.164
0.003
6.727
0.723
1 7.737 7.100 7.453
I
ShieldingFailure ~ 0.003 0.003 0.003
Backflashover 6.933 6.933 6.727
150000 ~ 0.002 8.E-5 1.339
1 6.938 6.936 8.069

&,, = phase conductor diameter , +BIR = shield (ground) wire diameter , R = pole footing resistance

0-7803-6420-1/00/$10.00 (c) 2000 IEEE 2581

Authorized licensed use limited to: UNIVERSITAT ROVIRA I VIRGILI. Downloaded on June 03,2010 at 08:50:12 UTC from IEEE Xplore. Restrictions apply.
 VI. CONCLUSIONS IEEWorking Group on Lightning Performance of
" m i s s i o n Lines," I E E Guide for improving the
Lightning is a very random phenomenonand one ofthe lightning p e r f o m of transmission lines", IEEE
major threats for the operation of distribution systems. Stmdard 1243-1997.
Several procedures have been applied to evaluak the IEEE W o b g Gmup on Lightning Performance of
llglurting performance of overhead distribution lines, and Distribution Lines," lEEE Guide for improving the
several methods for obtaining overvoltages have been lightning performance of electric power overhead
compared. distribution lines",IEEE Standard 1410-1997.
The most importaut conclusions from this study can be RB. Anderson and AJ. Eriksson, "Lightning
suinnlarized as follows parameters for engineering applications," Elecfra, no.
induced overvoltages are a problem only for lines with 69, pp. 65-102, 1980.
ve:ry low insulation levels, i.e CFO 5 75 kV P. Chowdhuri, "Estimation of flashover rates of
the installation of shield wires can signiscanfly reduce overhead power distribution lines by lightning
the magnitude of overvoltages induced by nearby strokes to nearby ground," IEEE Trans. on Power
strokes to ground Delivery, vol. 4, no.3, pp. 1982-1989, July 1989.
0 the parameter which has a strongest influence on the P. Chowdhuri, "Lightning-induced voltages on
backflashover rate is the pole footing resistance multiconductor overhead lines," IEEE Trans. on
0 differences between methods to calculate the number Power Delivery, vol. 5 , no. 2, pp. 658-667, April
of failures originated by direct strokes to either phase 1990.
or shield wires are rather small; however, it is P. Chowdhuri, Electromagnetic Transients in Power
important to note that this is only true for realistic System, RSP Press, John Wiley and Sons, 1996.
values of conductor heights, as those analyzed here, JEEE Working Group Report, "Calculating the
since for either shorter or taller heights differences can lightning performance of distribution lines," IEEE
be sigruficant [9] Trans. on Power Delivery, vol. 5, no. 3, pp. 1408-
differences between methods to obtain the number of 1417, July 1990.
insulation flashovers originated by nearby strokes to P. Chowdhwi and S. Mehairjan, "Alternative to
ground are important, being highly dependent on the Monte Carlo method for the estimation of lightning
return-stroke velocity. incidence to overhead lines," IEE Proc.-Gener.
Some of these conclusions have been known for many Trans" Distrib., vol. 144, no. 2, pp. 129-131, March
years 131, [6], [ll]; however, the influence of the return- 1997.
stroke velocity needs of a further study and clarification. [lo] A.J. Eriksson, "The incidence of lightning strikes to
Calculations have been performed assuming the power lines," ZEEE Trans. on Power Delivey, vol. 2,
statistical nature of some lightning paramem, but a more no.3, pp. 859-870, July 1987.
accurate and deeper analysis is needed Basically, only two [ll] E. Cinieri and F. Muzi, "Lightniug induced
stroke parameters, the peak current and the the-to-crest, ovewoltages. Improvement in quality of services in
llave been assumed of random nature. Results presented in M V distribution lines by addition of shield wires,"
Table 1 are only a first approach, as some other m e t e r s IEEE Trans. on Power Delivery, vol. 11, no. 1, pp.
should be considered random, for instauce the return-stoke 361-372, January 1996.
velocity or the stroke polarity. Power frequency voltage is [12] F. Rachidi, C A Nucci, M. Ianoz and C. Mazzetti,
mother raudom variable which can have some influence "Influence of a lossy ground on lightning-induced
on the performance of distribution lines. In addition, it is voltages on overhead lines," IEEE Trans. on
important to keep in mind that actual statistical parameters Electromagnetic Compatibility, vol. 38, no. 3, pp.
lo be used for a given line could be very Merent from 250-264, August 1996.
those used in this study. Finally, all calculations have been Juan A. Martins was bom in Barcelona (Spain). He is a h f w r
iriade assuming perfect ground, but it is recognized that Titular at the bparhneat d'Enginyeria J!ll&rica of the Universitat
ground conductivity can have a significantinfluence [12]. Pofitkdca de Catalunya. His teaching and mearch interests include
"k&on and Distrihtion, Power System Analysis and EhlTp
applications.
VIL REFERENCES F "GO1!~&5Molina was bom in Barcelona (Spain). He is
cuccently a Ph.D. candidate at the Univemitat Politkdca de Catalunya.
His meach interests indude Transient Analysis of Power Systems, and
[I I Working Group 01 (Lightning) of Study Committee Lightdug perfonnameof " i s i o n and Distribution Ovexhead Lies.
33 (Overvoltages and Insulation coordination), PtItindrp Clwwdhd is a professor of electrical engineering at the
"Guide to procedures for estimating the lightning Ceater for Electric Power, Tennessee Technological University,
performance of transmission lines," CIGRE Brochure Cookeville, TN.He is a Fellow of the IEEE,the Institution of Electrical
Engineers (U.K.), the American Association for the Advancement of
63, October 1991. Soience and the New York Academy of Sdences. He has received four
U.S. patents.

0-7803-6420-1/00/$10.00( c ) 2000 IEEE 2582

Authorized licensed use limited to: UNIVERSITAT ROVIRA I VIRGILI. Downloaded on June 03,2010 at 08:50:12 UTC from IEEE Xplore. Restrictions apply.