Journal of Electrostatics 68 (2010) 275e283

Contents lists available at ScienceDirect

Journal of Electrostatics
journal homepage: www.elsevier.com/locate/elstat

Shielding failure rate calculation by means of downward and upward lightning

leader movement models: Effect of environmental conditions
M.R. Bank Tavakoli, B. Vahidi*
Department of Electrical Engineering, Amirkabir University of Technology, No. 424, Hafez Avenue, Tehran 15916-34311, Iran

a r t i c l e i n f o a b s t r a c t

Article history: In this paper the effects of environmental conditions on shielding failure rate (SFR) of transmission lines
Received 3 June 2009 are investigated. The study utilizes a previously published work in which leader progression model for
Received in revised form lightning upward and downward leaders are used to calculate the SFR. Taking into account the effects of
26 January 2010
reduced air density and humidity on the parameters of upward leader model and wind pressure on the
Accepted 23 February 2010
movement of lightning leaders and wires, SFR and maximum lightning stroke current causing shielding
Available online 11 March 2010
failure are computed. The electric field in all simulations is calculated by means of charge simulation
method. The results of simulation show that the effects of relative air density and height of installation
Keywords:
Relative air density
are quite higher than that of the wind pressure and humidity while the humidity has the lowest impact
Wind pressure on the SFR of investigated transmission line.
Humidity Ó 2010 Elsevier B.V. All rights reserved.
Lightning
Leaders
Simulation

1. Introduction
The well-known Electro-Geometrical Model (EGM) has long
played the role of a simple and effective method for lightning
performance analysis of transmission lines [1e3]. More compli-
cated methods of leader progression models which consider the
details of lightning leader movement from cloud to line and
upward leader inception criteria have been proposed by investi-
gators [4,5]. The procedure of shielding failure rate calculation in
three dimensional configuration for modeling the sag, tower and
wires with maximum possible details are also introduced [5]. The
lightning performance of overhead transmission lines is usually
expressed by a factor called shielding failure rate (SFR). It is
important to know the effect of environmental conditions on SFR
factor of transmission lines. Due to the fact that testing the
environmental conditions for a real transmission line is not an
easy job, simulation tools would be quite helpful to create insight
into the problem. Actually, the upward lightning leader parame-
ters (which influence the inception condition) are sensitive to
environmental conditions. Different laboratory tests and theo-
retical modeling were performed to investigate the breakdown
* Corresponding author. Tel.: þ98 21 64543330.
E-mail address: vahidi@aut.ac.ir (B. Vahidi).
characteristics for large air gaps and also the influence of
humidity and reduced air density on the parameters of discharge
[6e10]. The most important effect of humidity is that it decreases
the charge per-unit length of the lightning upward leader
necessary to achieve thermal transition from diffuse glow to
leader channel [6]. Also it will increase the streamer zone electric
field intensity which is kept constant during propagation in the
upward model [7,11]. Wind pressure changes the route of light-
ning upward and downward leaders and tilts the phase and
ground wires and consequently it changes shielding conditions
along the span [12].
In this paper, based on previous work of three dimensional
leader movement model, the effects of humidity, reduced air
density and wind on the shielding performance of overhead lines
are investigated. The effects of humidity and air density are
modeled by tuning the parameters of positive upward connecting
leader. Shield and phase wires movement is also modeled in
simplified conditions to estimate SFR changes due to conductor
movement under wind pressure.
The rest of this paper is organized as follows. In Section 2, the
leader movement model for shielding failure analysis is
reviewed. The principle of modeling and effects of humidity,
reduced air density and wind are described in Sections 3, 4 and
5 respectively. Section 6 is also dedicated to some discussions
and conclusions.

0304-3886/$ e see front matter Ó 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.elstat.2010.02.003
276 M.R. Bank Tavakoli, B. Vahidi / Journal of Electrostatics 68 (2010) 275e283

2. Leader movement model for shielding failure rate

calculation

The leader movement model calculates the SFR of transmission

lines based on dynamic simulation of downward lightning leader
movement toward earth and checking the stable upward leader
inception from towers, wires and ground [5]. Using the configura-
tion of one span (two towers, conductors and a perfectly-con-
ducting ground), the space above the span is divided into meshes,
see Fig. 1. The width of area in Y direction, where the simulation
should be performed, is selected so that out of this area no flash to
wires and towers occurs. Owing the symmetry which exists in the
problem, it is enough to perform the simulation for 1/2 of the span
length. The downward lightning leader will start from each mesh,
descends toward the ground in continuous steps until a stable
upward leader is incepted from test points on towers, wires or
ground.
The field observations of negative downward lightning leaders
revealed that these leaders approach the earth from cloud in
consecutive steps [13]. The downward leader propagation is
represented by the steps with a length of 1/k of the leader tip
distance from ground, where k is the initial altitude of downward
leader divided by 100 m. This model ensures the step length of
downward lightning leader become always lower than 100 m in
accordance with field observations [13]. Similarly, if the downward
leader approaches near the earth and the model results in a step
length lower than 10 m, the step length is set to 10 m. To find the
next jump point of the downward leader, a hemisphere is drawn
with a radius of step length around the tip of the downward leader.
Because the potential is always negative (assuming negative
downward leaders), next jump point of the leader is a point on the
hemisphere where the absolute value of potential is the lowest (i.e.
the voltage gradient along the line connecting the leader tip to the
target point become maximum), see Fig. 2. The steps of downward
leader are modeled by horizontal and vertical line charges. The
charge density of downward leader (which is a function of lightning
peak return stroke current) is computed by use of following
equation based on electrostatic considerations of measured wave-
form of return stroke current [11]:
(  
z  z0  z 
rðzÞ ¼ Ip m0 1  1 0
H  z0 H

Fig. 2. Lightning leader movements models, (a) downward leader, (b) upward leader.

Fig. 1. The meshes for downward lightning leader movement toward ground and
structure of a span.
m1 þ m2 ðz  z0 Þ
þ 
1 þ m3 ðz  z0 Þ þ m4 ðz  z0 Þ2
h  
10z0 z i
0:3e 75 þ 0:7 1  0 (1)
H
where m0 ¼ 1.476  105, m1 ¼ 4.857  105, m2 ¼ 3.9097  106,
m3 ¼ 0.522, m4 ¼ 3.73  103, z0 is the height of the downward
leader tip from earth in m. H is the cloud height in m, Ip is the return
stroke peak current in kA, r is the downward leader charge density
in C/m and z is the variable height of the point on the leader where
charge density is to be calculated, see Fig. 2(a).
According to extensive field observation and measurements,
some probability distribution functions are introduced for lightning
peak current [14]. In this paper, we will use a range of 3e35 kA to
scan the lightning current range causing shielding failure.
For shielding failure analysis of transmission lines, a lightning
stroke with very low peak current does not create high enough
overvoltage to be dangerous, even if they end directly to the phase
 M.R. Bank Tavakoli, B. Vahidi / Journal of Electrostatics 68 (2010) 275e283
wires. Therefore, it is reasonable to limit the lower limit of lightning
current to 3 kA and confine the simulation time. It is a common
approach which is also applied by EGM [15]. Very high lightning
currents are not also relevant for shielding failure analysis because
a stroke having peak current higher than a specified maximum
value would terminate the ground wires and no shielding failure
would happen [1e3]. This maximum stroke current depends on the
line geometry and stroke specifications, and for our investigated
overhead line, this maximum current is in the range of 25e30 kA. In
our paper, we have chosen 35 kA to be in the safe side to ensure
tracking all strokes which may cause shielding failure.
One of the important phases of simulation is the upward leader
inception from ground structures. When negative downward
lightning leader approaches the earth, increasing electrical field
(near 3000 kV/m) on earthed structures may lead to local ionization
near some sharp points which is said to be corona inception time. In
fact, the point of strike is a point on ground structures where a stable
positive upward connecting leader has been incepted. Different
criteria have been proposed for calculating the electric field required
for stable upward leader inception [16e18]. Using the approach of
self-consistent model [11,18] the upward leader inception is checked
in each step of downward leader movement. The principle of
upward leader model is presented in Fig. 2(b). The streamer zone
charge is initially calculated in say zero step:
U0
l0s ¼
Es  Eb
DQ 0 ¼ 0:5$KQ $l2s $ðEs  Eb Þ
where U0 (kV) and Eb (kV/m) are the voltage and field on the fitted
line of background electric field as depicted in Fig. 2(b), Es (kV/m) is
the constant streamer zone electric field which is affected by
pressure, temperature and humidity (see Sections 3 and 4), l0s (m) is
the initial length of streamer zone and KQ (C/kV m) is a geometrical
factor that takes into account the effect of all streamers on total
steamer zone charge. The necessary (not enough) condition for
stable upward leader inception is that DQ0 (C) be high enough
(1 mC) for streamer zone to change into a leaderestreamer system
[11]. If this condition is fulfilled, then by taking initial leader length
of 1 cm, the streamer zone length and streamer zone space charge
is calculated in each advancement step (i  1):
" #
i Einit Einit  Einf li =x0
Utip ¼ Einf $lil þ x0 $Einf $ln  e l
Einf Einf
U0 þ Es $lil  Utip
i
lis ¼
Es  Eb
  h   i
DQ i ¼ KQ $ lsi1  lil $ Es lil  lli1 þ Utip
i1 i
 Utip
DQ i
Dlil ¼
ql
liþ1 ¼ lil þ Dlil
l
i
where Utip (kV) is the leader potential in step i, Einf (kV/m) and Einit
(kV/m) are the initial value and final quasi-stationary values of the
leader gradient, x0 (m) is a constant equal to v, q where v (m/s) is the
ascending leader speed and q (s) is the conductance equation time
constant [17]. q1 (C/m) is the charge per-unit length necessary to
achieve thermal transition from diffuse glow to leader channel and
is essentially affected by humidity (see Section 3) [8]. The
277
procedure continues until DQi is positive. If DQi become negative,
no stable upward leader is incepted. Moreover, this procedure
continues until leader elongates to a predefined lmax (which is
taken to be 3e5 m). There are two conditions for stable leader
inception from any test point: a) initial space charge DQ0 is high
enough to start transition of corona charge to leaderestreamer
system (1 mC) and b) DQi remains always positive until the
streamer tip reaches predefined lmax. Therefore, if these conditions
are fulfilled for a test point on ground structures, the lightning will
finally strike that point. Of course the correct physical explanation
is that the upward leader will incept and connect the downward
leader. Because we are interested in the target point, we can simply
say that lightning will attach the target point. Therefore, it would be
possible to find out that the lightning stroke will attach phase wire,
shield wire or earth.
For each mesh of Fig. 1, a range of return stroke currents are
scanned to find the min and max return stroke current (Imin, Imax) in
which the lightning leaders strike the phase wires. Then, the
shielding failure rate on each mesh can be computed:
dx
SFRij ¼ 0:1  GFD  dy  ½PðImax Þ  PðIc Þ (9)
D
where dx (m) and dy (m) are the mesh length and width according
to Fig. 1, D (m) is the span length and SFRij is the shielding failure
rate of mesh ij in strokes/100 km/year for the currents exceeding Ic.
(2) The term stroke means the number of lightnings which strikes the
phase conductor.
P(I) is the probability distribution function of stroke currents
(3) exceeding I and is given by [14]:
1
PðIÞ ¼ 2:6 (10)
I
1 þ ð31
where I is the return stoke current in kA. GFD is the ground flash
density and is calculated by following equation [19]:
GFD ¼ 0:04  Td1:25 (11)
where Td is the number of thunderstorm days per year (keraunic
level) and GFD is the ground flash density (flash/km2-year). The
data and dimensions are given in Appendix. Total SFR is then
calculated by [5]:
m X
X n
SFR ¼ 4 SFRij (12)
(4)
i¼1 j¼1
The SFR calculation procedure for a specified configuration of
(5) towers, conductors and space charge of upward and downward
leaders requires electric field calculation in the environment of
study and in each step of leader movement. Charge simulation
(6) method is used to compute the electric field in all points of Fig. 1
[20,21]. This method uses a numerical procedure to calculate the
electric field in the environment. Basically, some limited shapes of
charges are placed by this method in arbitrary locations and their
(7)
charge value is computed so that the boundary conditions, which
are defined by problem definition, are ensured. If this set of charges
(8) predicts the other known voltage or field in computational space
appropriately, then the electric field in all points could be calcu-
lated using this set of charges [21]. Fig. 3 shows the ring, and line
charges which are used to calculate the field according to charge
simulation method. The clouds are modeled as a simplified uni-
polar negative charge on the height of 2000 m and with an
extension of 5000 m and are modeled by ring charges. According to
the field tests the storm field are said to produce the electrical field
on ground level in the range of 1e20 kV/m [13]. The problem about
278 M.R. Bank Tavakoli, B. Vahidi / Journal of Electrostatics 68 (2010) 275e283

a Es ¼ 425d
1:5
þ ð4 þ 5dÞg (14)
where Es is the electric field intensity in streamer zone (kV/m), g is
the absolute humidity (g/m3) and d is the relative air density (the
Ring Charges ratio of air density at given condition to the standard condition)
Shield Wires Hcloud representing the pressure and temperature dependency [7]:

P T0
d¼ (15)
P0 T
Phase Wires where d is the relative air density, P is local air pressure (atm) and T
is local temperature (K), P0 ¼ 1 (atm) and T0 ¼ 293 (K) are standard
pressure and temperature at sea level.
Countour The upward inception model is affected by humidity levels
Ring Charges Points based on (13) and (14). A transmission line is then assumed in an
area, at sea level and local temperature of 20  C (293 K).
By modifying the above parameters, all the procedure of lightning
downward and upward leader progression (Section 2) are repeated
in different humidity levels. All other data for simulated overhead
b Ground Level line, clouds and leaders are listed in the Appendix.
z The results of simulation are depicted in Fig. 4. The SFR
according to this figure shows a reducing trend. However, the
A(0,0,H) B(D,0,H) change in SFR is quite small. It could be due to the fact that
the situation for upward leader inception is changed identically for
C(D/2,0,h) phase wires and shield wires and thus, no sharp changes are to be
seen for the strokes attaching the phase wires. For this simulated
Wire
overhead line, a change 0f 50% in absolute humidity from 10 g/m3
will cause a change of w7% in SFR. It is good to mention here that
for this overhead line configuration with two ground wires, the
Vertical & Horizontal
absolute value of SFR is essentially low (i.e. under 0.2 stroke/
Line Charges
Ground Level 100 km/year). Moreover, the changes due to humidity itself are also
x low. This complies with recent experiments which are performed
First Tower Mid-Span Second Tower for model overhead lines to see the effect of humidity on lightning
Position Position performance of overhead lines [22].
y
In Fig. 5, the max current attaching the phase wires in different
Fig. 3. The charges which are used for field calculation according to charge simulation humidity levels is shown. The max current of strokes attaching the
method, a) cloud and tower models and, b) wires and their sags models. phase wires are decreased in higher humidity levels. The max
stroke current which attach the phase wire will decrease due to
our simulation is that for each step of lightning leader movement, increased humidity from 11.5 kA in 10 g/m3 absolute humidity to
the electric field should be computed again in the space and this, by 8.5 kA in 15 g/m3 (i.e. 16% decrease for a 50% increase in absolute
essence, makes the simulation quite time-consuming. humidity). Here also the shielding design of our overhead
The wires are specially modeled by horizontal and vertical lines line confines the stroke currents passing the shield to lower values
of a hyperbolic curve to model the sag of the lines. The wind (i.e. under 14 kA).
pressure will change the positions of wires, depending on their
tension and wind speed. These effects are described more in
4. Effects of reduced air density
Section 5. The parameters values are listed in the Appendix.
The pressure and temperature have also their influence on
3. Effects of humidity upward leader inception and on the shielding behavior of trans-
mission line. A practical example is the overhead line which is
It is known that increasing the absolute humidity in air in non-
homogenous fields will slightly increase the breakdown voltage [6].
Actually, upward leader inception conditions are essentially 0.25
affected by absolute humidity [8]. To take into account these effects P=1 atm T=293 K
in our simulations, the most sensitive parameters of the upward
Shielding failure rate
(stroke/100km/year)

0.2
leader model are to be considered as humidity dependent.
Charge per-unit length of leader channel; ql, shows most 0.15
sensitivity to humidity [6,8]. From the test results in fast rising
times, the charge per-unit length has shown following relation to 0.1
humidity [6]:
0.05
ql ¼ 59  2:3g (13)
where ql is the charge per-unit length of leader channel in (mC/m) 0
0 5 10 15 20 25
and g is the absolute humidity in (g/m3). Moreover, the streamer
Humidity (gr/m^3)
zone field intensity is also affected by humidity and relative air
density [7]. Fig. 4. Shielding failure rate of the line with respect to humidity.
 M.R. Bank Tavakoli, B. Vahidi / Journal of Electrostatics 68 (2010) 275e283 279

16 30
Max Lightning Current (kA)

Max lightning Current (kA)

14 P=1 atm T=293 K
25
12
20
10
15
8
6 10

4 5

2 0
0 0 0.2 0.4 0.6 0.8 1 1.2
0 5 10 15 20 25 Relative air density
Humidity (gr/m^3)
Fig. 7. Maximum stroke current passing the shield and attaching the phase wires in
Fig. 5. Maximum stroke current passing the shield and attaching the phase wires with different relative air density (at 1 atm pressure and 11 g/m3 absolute humidity).
respect to humidity.

max stroke current to 25 kA. A relative rate of change of 34% is also

installed in higher altitudes. Therefore, to investigate the effect of
observed here in relative air density of 1.
installation height; pressure, temperature and humidity relation
In higher altitudes, not only the relative air density reduces, but
with respect to height from sea level is modeled by following
also the humidity itself reduces. The effect of higher altitude
equations [23]:
installations on the shielding performance of overhead line is also
  simulated using Eqs. (16)e(18). The sea level conditions are
H
P ¼ P0 exp  l (16) assumed to be 1 atm pressure, 20  C temperature and 11 g/m3
8
humidity. The height of cloud is set to be constant during all
simulations.
T ¼ T0  6Hl (17) Fig. 8 shows the effect of height on the shielding failure rate in
  above mentioned conditions. It is obvious that the failure rate is
Hl increasing, as expected, with respect to height. The range of
g ¼ g0 exp  (18)
3 changes in SFR is a little bit higher than those of relative air
density (Fig. 6) because, now, the humidity is also decreasing
where Hl is the height of structure from sea level (km) and P0, T0 and
which affects the total shielding rate to be higher. In average, an
g0 are pressure, temperature and absolute humidity at sea level
increase of 0.22 strokes/100 km/year can be noticed on each km
respectively. In Fig. 6 the effect of relative air density on the shielding
rise above sea level. It also complies with field observations [24].
failure rate is shown. In this figure, the line is assumed to be at sea
Maximum stroke current attaching the phase wire is also shown
level and constant standard humidity (i.e. 1 atm and 11 g/m3).
in Fig. 9 which shows the increasing trend as well. High value of
Therefore, the pure effect of relative air density (or decreasing the
stroke currents can be seen in high altitudes (i.e. w27 kA in
temperature) on results is shown. Comparing to the effects of
w5 km). Nevertheless, the practical installations of overhead lines
humidity, the relative air density seems to be more influencing the
are naturally limited to 2e3 km above sea level and therefore, the
shielding performance of the line. The relative air density of 0.5
results in Figs. 8 and 9 for very high altitudes are only theoretical
(the value of 0.5 is more or less the most applicable change in relative
values.
air density in overhead line installations [15]) has showed an SFR
It is clear from the results that the relative air density effect on
value of 0.72 strokes/100 km/year which is quite high comparing the
the shielding performance of the line is higher than the humidity
humidity effect. The rate of change of SFR due to reduced air density
effect. The simulations are made for typical ranges of humidity and
is also high where a 10% increase in reduced air density (in 0.9
reduced air density. As it can be seen in Figs. 8 and 9, the results
relative air density) causes more than 40% decrease in SFR.
may differ if parameters change in wider range which is not
In Fig. 7, the maximum stroke current attaching the phase wires
essentially a possible physical solution.
is shown. The maximum stroke current also decreases with respect
It should be noted here that in all simulations the structures of
to relative air density. Relative air density of 0.5 has increased the
the line and leader progression simulations (esp. downward leader
0.8

0.7 0.9
0.6 0.8
Strokes/100km/year

0.7
Strokes/100km/year

0.5
0.6
0.4
0.5
0.3 0.4
0.2 0.3

0.1 0.2
0.1
0
0 0.2 0.4 0.6 0.8 1 1.2 0
Relative air density 0 1000 2000 3000 4000 5000 6000
Height from sea level (m)
Fig. 6. Shielding failure rate of the simulated line in different relative air density
(at 1 atm pressure and 11 g/m3 absolute humidity). Fig. 8. Shielding failure rate of the simulated line installed in higher altitudes.
280 M.R. Bank Tavakoli, B. Vahidi / Journal of Electrostatics 68 (2010) 275e283

30
a
Max lightning current (kA)

25

20 lleader
lleader0
15

10
Hemisphere of the next
jump points
5
lwind
0
0 1000 2000 3000 4000 5000 6000 b lwind
Height from sea level (m)

Fig. 9. Maximum stroke current passing the shield and attaching the phase wires of
the simulated line installed in higher altitudes.
lleader0 lleader
Test point for upward
leaders
approaching the earth) are identical to investigate only the effect of
local environmental conditions on the results of simulations.

5. Effects of wind c Z
The wind blowing around overhead lines will change the z=H
shielding conditions because the shield and phase wires which are
under different tensions will tilt to different degrees under wind α Wind
Locus of wire in mid-span
pressure. Therefore, while at the towers the shielding angle is held
constant, the wires will tilt in the direction of wind and shielding
due to wind pressure ϕ
z=h
conditions in other points of the span will differ. Moreover, in
a lower degree, the wind affects the downward and upward Wire
movement of lightning leaders and changes the normal direction cross section
of leaders. Leader movement model in three dimensions allows
simulating such conditions because all the wires along the span
are modeled with details. To investigate the above mentioned
-Y +Y
effects of wind, a corridor of 500 m is assumed from the earth in
Fig. 10. Modeling the wind effects (a) downward leader jump, (b) upward leader
Fig. 1 where the wind would change the leader movement movement, (c) tilting due to wind pressure.
direction. In each step of downward leader approaching the earth,
the movement vector due to wind speed is added up to the leader
in Fig. 1. To simplify the problem, the effect of nonsymmetrical
jump distance vector when the leaders enter the wind corridor,
shape of wires (e.g. ice, etc.) is neglected and only the drift force is
Fig. 10(a) and (b).
modeled. Based on Fig. 10(c), the wind blowing under the angle of
! ! ! attack of f in positive Y direction, will tilt the wire from the Z axis
l leader ¼ l leader 0 þ l wind (19) up to a degree.

1 D cosð4Þ
! a ¼ tan1 (23)
l wind ¼ !
v wind $ttravel (20) 2 W  D sinð4Þ
where D is the drift force (N) and W is the weight force of conductor
! ! ! !
v wind ¼ vwind x i þ vwind y j þ vwind z k (21) in the span (N). The drift force is also calculated as follows [12].
Angle of attack 4 is the direction of wind blow with respect to the
! reciprocal axis to the span, see Fig. 10(c).
j l leader 0 j
ttravel ¼ (22)
vleader 1
D ¼ r v2 d$lspan (24)
! 2 air wind
where l leader is the final vector of distance on each step of
!
downward or upward movement, l leader 0 is the vector of distance where rair is the air density (kg/m3), v2wind is the wind speed (m/s),
which the downward or upward leader should have been traveled if d is the conductor diameter (m) and lspan is the conductor length
!
no wind blows, l wind is the vector of movement due to wind blow, trough the span (m). With a common assumption of hyperbolic
!v wind is the wind speed vector, ttravel is the approximate time shape of the wire, the length of the conductor in the span would be
required for each jump of leaders and vleader is the speed of upward calculated easily [25].
or downward leader.
8ðH  hÞ2
Please note that although (19)e(22) are written in general, lspan ¼ D þ (25)
the vectors and speed of downward and upward leaders are 3D
different and the effect of wind would be different. Also, the and the parameters are defined in Fig. 3.
downward leader is only affected if it enters the wind corridor of The overall effect of wind on the phase and shield conductors
500 m. would be the movement and changing the shielding conditions
Another effect of the wind, which is actually more important, is through the span. Therefore, on a wind blowing from the left to the
the drift pressure forcing the wires to move. To take this effect into right in Fig. 1, the set of phase and shield wires would be tilted
account, we should know the wire tilting formulation due to wind through the span to the right (with different degree of a) and it
 M.R. Bank Tavakoli, B. Vahidi / Journal of Electrostatics 68 (2010) 275e283 281

0.3 on the SFR. Obviously, the increase in SFR due to changes in angle of
0.28 Angle of Attack 0 attack in constant wind speed is quite small (i.e. w2% in full range
Shielding Failure Rate
(Strokes/100km/year)

0.26 Angle of Attack 10 from 0 to 30 ). Due to the fast upward and downward lightning
Angle of Attack 20 leader speeds (Eq. (20)), the changes in length vector by the wind in
0.24
Angle of Attack 30
0.22 short duration of travel is quite small. Therefore, most contributing
0.2 effect of wind would be conductor tilting and changes in shielding
0.18
condition through the span. Fig. 12 also shows maximum lightning
current attaching the phase wires. The increasing trend is also seen
0.16
in this figure. The maximum peak current is confined under 22 kA
0.14
for the simulated range of wind speed. Moreover, the change of
0.12
lightning peak current in constant wind speed due to angle of
0 10 20 30 40
Wind Speed (m/s)
attack is less than 1%.
Comparing the results of reduced air density, it seems that the
Fig. 11. Shielding failure changes of the line subjecting to different wind speeds effect of height of installation is higher than the effect of typical
blowing in þY direction and having different angle of attack. wind pressure. Although, the changes in SFR due to humidity
changes in ordinary ranges is lower than the effect of typical wind
means that the symmetry in the simulations without wind effect do speeds through the span.
not exist anymore and a full half-span should be simulated, see
Section 2 [5]. This fact will increase the simulation time dramatically.
Moreover, the inclination of conductors in Y direction are mathe- 6. Discussion and conclusion
matically represented by following equations, see Fig. 3(b):
In this paper based on the leader progression model for light-

   
b$D ning performance calculations of overhead lines, the effects of
z ¼ H  H  h cos h  bx cosðaÞ (26)
2 humidity, reduced air density and wind on the shielding failure rate
(SFR) of an overhead line were investigated. The dependency of the

    parameters of upward lightning leader to humidity and reduced air
b$D
y ¼ H  h$cos h  bx sinðaÞ (27) density is taken into account. Downward and upward lightning
2
leader directions and changes in shielding conditions through the
 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi span by wind are also modeled in details. The simulations show
2 H  H2  h2  that the effect of humidity is not that high while reduced air density
b ¼ ln  (28)
D  h  and similarly the height of installation has most influence on the
shielding failure rate of the line. It was also shown that the wind
The parameters are already defined in Fig. 3(b) and a is shown in increases the SFR but the effect is predicted to be lower than
Fig. 10(c). reduced air density due to the height of installation.
The simulation with the modeling approach as stated above is One major problem to study the shielding performance of
performed for the tower configuration of Fig. 1, wind blowing in þY overhead lines due to changes in environmental conditions is the
direction, sea level condition and humidity of 11 g/m3. Fig. 11 shows lack of precise field observation. Even for utilities, discriminating
the calculated SFR of the line. the shielding failure from back flashover for a faulty overhead line
The shielding failure rate increases when the wind speed is too hard [14]. The laboratory works are also rare in this regard.
increases because the higher speed of the wind causes higher drift However, the shielding performance of a model overhead line has
pressures and higher inclinations of phase and shield conductors. been recently investigated in laboratory [22]. Those tests have
For our simulated overhead line the SFR values is under 0.3 even for shown that the effect of humidity on the shielding failure rate is
high wind speed of 30 m/s. Near 20% change in SFR value can be negligible. Our simulation also shows small changes in shielding
seen by increasing the wind speed from 10 to 15 m/s. However, the performance of overhead line due to humidity. The height of
total value of SFR is still small. The angle of attack has similar effects installation (reduced air density), however, has been seen in

22
Max Lightning Current (kA)

20 Angle of Attack 0
Angle of Attack 10
18 Angle of Attack 20
Angle of Attack 30
16

14

12

10
0 5 10 15 20 25 30 35
Wind Speed (m/s)
Fig. 12. Maximum stroke current passing the shield and attaching the phase wires of the line subjecting to different wind speeds blowing in þY direction and having different angle
of attack.
282 M.R. Bank Tavakoli, B. Vahidi / Journal of Electrostatics 68 (2010) 275e283

already installed overhead lines to have influence on overhead line 4.2 m 7.35 m 7.35 m
shielding performance and the increase in SFR is also predicted by
EGM [2,3]. The increase in SFR is also observed in mountainous 1.7 m
installations as well [24]. The lack of laboratory tests to investigate
the shielding performance of model overhead lines in windy 6m 8.3 m
conditions makes it hard to compare the simulation results in
windy conditions in this paper with those coming from the tests or
1.9 m
observation. While the simulation has shown that the wind effect
would not too harsh, the prediction remains to be verified by the 4.3 m
results of wind tunnel tests for model overhead lines.
Nowadays overhead line shielding design procedure is essen-
tially based on EGM as stated in introduction. Although it is not 12 m 0.4 m
directly mentioned, this engineering method inherently assumes
specific standard environmental conditions (which are relative air
density 1 and no wind to blow). The EGM is not capable to model
the environmental conditions like the lightning leader progression
approach which was explained through this paper. While the
5.5 m
leader progression approach improves the models of lightning
leaders and overhead line, the long simulation time required to
investigate the shielding performance may make the approach too
exhaustive. Coming back to the simulation results, the humidity
effect was seen to have small effect on the shielding performance.
The wind effect is affecting the line more. However, the height of
installation has a relatively considerable effect. The EGM is too fast
and, while neglecting the humidity and wind effects, considers the
height of installation [1e3] and therefore seems to be still the 19.8 m
engineering method to chose. Nevertheless, the lightning leader
movement approach under different environmental conditions is
to be used for specific high voltage overhead lines crossing the
areas with high keraunic levels and requirement for very low
outage due to lightning.

Acknowledgment 0.3 m

The authors would like to sincerely thank Dr. H. Gharagozloo for

his kind discussions. 12.1 m

Fig. 13. Tower dimensions for simulated overhead line.

Appendix

In Table 1 and Table 2 the simulation data are listed. Fig. 13 also References
shows the dimensions of the towers of simulated 230 kV overhead line.
[1] F.S. Young, J.M. Clayton, A.R. Hileman, Shielding of transmission lines. IEEE
Table 1 Trans. Power App. Syst. S82 (1963) 132e154.
Parameters of the positive upward leader model. [2] IEEE, Guide for improving the lightning performance of transmission lines.
IEEE Std. 1243 (1997).
Parameter Value Unit [3] A.J. Erikson, An improved electrogeometric model for transmission line
shielding analysis. IEEE Trans. Power Deliv. 2 (1987) 871e886.
Einf 30 kV/m
[4] L. Dellera, E. Garbagnati, Lightning stroke simulation by means of the leader
v 2  104 m/s progression model, part I: description of the model and evaluation of
q 50 ms exposure of free shielding structures. IEEE Trans. Power Deliv. 5 (1990)
KQ 3.5  1011 C/V 2009e2022.
[5] B. Vahidi, M. Yahyaabadi, M.R. Bank Tavakoli, S.M. Ahadi, Leader progression
Table 2 analysis model for shielding failure computation by using charge simulation
method. IEEE Trans. Power Deliv. 23 (2008) 2201e2206.
Parameters of conductors, towers and environment.
[6] Les Renardières Group, Positive discharges in long air gaps. Les Renardières
Parameter Value Unit (1975) Results: Electra 53(1977).
[7] A.J. Eriksson, B.C. Roux, H.J. Geldenhuys, D.V. Meal, Study of airgap breakdown
Air density at sea level 1.29 kg/m3 characteristics under ambient conditions of reduced air density. IEE Proc. A
Shield wire weight force 3.821 N/m 133 (1986) 485e492.
Phase wire weight force 19.381 N/m [8] I. Gallimberti, The mechanism of the long spark formation. J. Physique Coll. 40
D 430 m (1979) 193e250.
Shield wire sag 5 m [9] P. Ortega, R.T. Waters, A. Haddad, R. Hameed, A.J. Davies, Impulse breakdown
Phase wire sag 9 m voltages of air gaps: a new approach to atmospheric correction factors
dx 43 m applicable to international standards. IEEE Trans. Dielectr. Electr. Insul. 14
dy 20 m (2007) 1498e1508.
Td 15 days per year [10] N.L. Aleksandrov, E.M. Bazelyan, Temperature and density effects on the
Shield wire radius 0.0049 m properties of a long positive streamer in air. J. Phys. D: Appl. Phys. 29 (1996)
2873e2880.
Phase wire radius 0.0158 m
[11] M. Becerra, V. Cooray, A simplified physical model to determine the lightning
Downward leader speed 5  105 m/s
upward connecting leader inception. IEEE Trans. Power Deliv. 21 (2006)
Upward leader speed 2  104 m/s
897e908.
 M.R. Bank Tavakoli, B. Vahidi / Journal of Electrostatics 68 (2010) 275e283
[12] Electric Power Research Institute, EPRI Transmission Line Reference Book:
Wind-induced Conductor Motion. EPRI Inc., New York, 2008.
[13] E.M. Bazelyan, Y.P. Raizer, Lightning Physics and Lightning Protection. Institute
of Physics Pub., Bristol, 2000.
[14] J.G. Anderson, Lightning Performance of Transmission Lines, Transmission Line
Reference Book 345 kV and above. EPRI Inc., New York, 1982, (Chapter 12).
[15] A.R. Hileman, Insulation Coordination for Power Systems, Power Engineering
Series. Marcel Dekker Inc., New York, 1999.
[16] U. Kumar, P.K. Bokka, J. Padhii, A macroscopic inception criterion for the
upward leaders of natural lightning. IEEE Trans. Power Deliv. 20 (2005)
904e911.
[17] F.A.M. Rizk, Modeling of transmission line exposure to direct lightning strokes.
IEEE Trans. Power Deliv. 5 (1990) 1983e1997.
[18] N. Goelian, P. Lalande, A. Bondiou-Clergerie, G.L. Bacchiega, A. Gazzani,
I. Gallimberti, A simplified model for the simulation of positive-spark devel-
opment in long air gaps. J. Phys. D: Appl. Phys. 30 (1997) 2441e2452.
283
[19] A.J. Erikson, The incidence of lightning strikes to power lines. IEEE Trans.
Power Deliv. 2 (1987) 859e870.
[20] H. Singer, H. Steinbigler, P. Weiss, A charge simulation method for the calcu-
lation of high voltage fields. IEEE Trans. Power App. Syst. PAS-93 (1974)
1660e1668.
[21] N.H. Malik, A review of the charge simulation method and its application. IEEE
Trans. EI. 24 (1989) 3e20.
[22] S. Taniguchi, S. Okabe, T. Takahashi, T. Shindo, Air-gap discharge characteris-
tics in foggy conditions relevant to lightning shielding of transmission lines.
IEEE Trans. Power Deliv. 23 (2008) 2409e2416.
[23] N.I. Petrov, R.T. Waters, Determination of the striking distance of lightning to
earthed structures. Proc. R. Soc. Lond. A. 450 (1995) 589e601.
[24] E.R. Whitehead, CIGRE survey of the lightning performance of EHV trans-
mission lines. Electra 33 (1974) 63e89.
[25] L.L. Grigsby, Electric Power Generation, Transmission, and Distribution. CRC
Press, New York, 2007, (Chapter 14).