Académique Documents
Professionnel Documents
Culture Documents
Abstract: Lightning is the main cause of transmission line outages affecting reliability of power
supply thus, consequently, resulting in economic losses. A statistical method for lightning
incidence calculations in transmission lines is introduced. Simple expressions for the estimation of
an expected range of lightning strikes to a transmission line depending on interception probability
distribution have been obtained, based on a recently proposed statistical lightning attachment
model derived from scale model experiments. The expected number of lightning strikes depends,
besides transmission line geometry, on lightning stroke current distribution and interception
probability. The results of the statistical method have been compared with those obtained by
employing other models from literature, including that suggested by the IEEE Std. 1243:1997, in
lightning incidence calculations, and with field observation data; a satisfactory agreement has been
shown to exist. Results on lightning incidence calculations are further discussed through an
application to typical 150 kV and 400 kV lines of the Hellenic transmission system.
In the present study, a statistical method for lightning In previous work [4], by using the three electrode
incidence calculations in transmission lines is arrangement shown in Figure 1, at applied voltages
introduced, which yields an expected range of lightning always causing breakdown interception probability
strikes depending on interception probability distributions were obtained, for several electrode
distribution. This task is accomplished by simple configurations differing in D and h, by varying R;
expressions, which have been derived on the basis of typical such distributions are shown in Figure 2. The
scale model experiments [3-6], and take into account interception probability distributions were found to be
besides transmission line geometry, the lightning stroke well approximated with the normal distribution, thus
current and interception probability distributions. The both striking distance and interception radius can be
results according to the statistical method are compared treated as statistical quantities with their distribution
with those obtained from the IEEE Standard [1], described by a mean value and a corresponding standard
previously reported models [2, 7-10] and with field deviation [4]. Actually, the interception radius R is
Pg. 1 Paper G-5
Proceedings of the 16th International Symposium on High Voltage Engineering
ISBN 978-0-620-44584-9 Copyright c 2009 SAIEE, Innes House, Johannesburg
140
Attractive Critical Failure
[2]
120
[10]
[9]
80
60
40
20
k
Rci h Req = R ( I , h ) f ( I ) dI (3)
, = c2 [6] (1) 0
h D
where f(I) is the probability density function of the
where h, D are defined in Figure 1 and values for the lightning stroke current distribution and R is the
coefficients c2, k and for , in formula form, are given in interception radius.
Table I [6]. Equation (1) is valid for 0<h/D 1. For
h/D >1 the asymptotic values of (1) at h/D unity i.e. the According to [12] the probability density function of the
equal coefficient c2 may be adopted [6]. stroke current is lognormally distributed as:
where h is in meters and I is in kA. Expression (5b) dispersion, as was recognized by Eriksson [2]. Apart
has been obtained by solving equation (3) with the aid from [7], all models agree well with the data points TL
of a mathematical software package by considering the from [2], which is regarded as the most reliable
distribution of interception radius. For the lightning according to Eriksson [2], and PA which was derived
crest current distribution with I = 30.1 kA and from [11] where direct flashes were recorded together
i = 0.76, suggested in [12], (5a) and (5b) become: with the local ground flash density.
0.3 0.43 It is important to note that expressions (6) and (7) do not
Reqc = 31h , (%) = 38.2h . (6) take into account the variation of equivalent interception
radius with lightning stroke current distribution. This is
The general expression (7) can be used for the considered in expressions (5a) and (5b) and the result of
estimation of the equivalent interception radius their application for three different lightning stroke
according to the lightning attachment models suggested current lognormal distributions (Table 3) is shown in
for lightning incidence calculations in [14]: Figure 5. From Figure 5 it is obvious that with
increasing I the equivalent interception radius
increases, however this is more evident for lower
Req = rh (7)
interception probabilities and higher conductors.
where factors r and E, derived from [14], are listed in
Table 2: Table 3: Stroke current distribution parameters.
140 140
3
.
.
Failure
120 3 120
Equivalent Interception radius (m)
and Table 2 is shown in Figure 4 together with field 20 Median current 24 kA, = 0.72
Median current 30.1 kA, = 0.76
20 Median current 39kA, = 0.76
Median current 30.1 kA, = 0.76
140
The annual number of lightning strikes to shield wires
[10]
120
per 100 km of a transmission line, NS, is given as:
100
80
TL
[9]
N S = 0.1N g 2R eq +b ( ) (8)
60
PA
40 where Req is in meters, Ng is the ground flash density
20
(strikes km-2yr-1) and b is the separation distance
between the outer shield wires.
0
0 10 20 30 40 50 60 70
Conductor height (m) The expected annual number of lightning strikes to
Figure 4: Equivalent interception radius as a function of shield wires of typical 150 kV and 400 kV lines of the
conductor height; points depict field data. Hellenic transmission system, tower geometries are
shown in Figure 6, according to different models
In all models, the equivalent interception radius tends to employed in lightning incidence calculations is shown
saturate with increasing conductor height; an exception in Tables 4 and 5, respectively. In these calculations Req
to this is model [7], which, as noted in [14], tends to has been estimated by putting the average height of
concave upwards and seems to underestimate the shield wires over the span in equations (6) and (7) and
equivalent interception radius at lower conductor assuming Ng = 4 km-2yr-1. From Tables 4 and 5 it can be
heights. In general, there is satisfactory agreement deduced that there is variability in NS among models,
between the present work and previously reported however all of them yield an increase of NS with
models with the field data in [2] considering also that increasing transmission line height. The range of NS
the latter are relatively uncertain and have a statistical yielded by the statistical method depending on
Pg. 3 Paper G-5
Proceedings of the 16th International Symposium on High Voltage Engineering
ISBN 978-0-620-44584-9 Copyright c 2009 SAIEE, Innes House, Johannesburg
4. DISCUSSION