Está en la página 1de 11

Dynamic Simulation of Lightning Attachment

to Earthed Overhead Transmission Line


Structures
Mohammad Reza Bank Tavakoli
Behrooz Vahidi
Department of Electrical Engineering,
Amirkabir University of Technology,
Hafez Avenue, No. 424, Tehran, Iran
vahidi@aut.ac.ir
In this paper a novel method is presented to investigate lightning stroke attachment to earthed over-
head transmission line structures. This method is based on the dynamic simulation of lightning
downward leader movement toward the ground in three dimensions. To perform the simulation, a
three-dimensional model for clouds, towers, wires, and a perfectly conducting ground is created. A
stepping approach is also introduced in accordance with the field observations of negative downward
lightning leaders to model the downward leader movement toward the ground. In each step of the
leader movement, the electric field is computed numerically in all points of the simulation space and
in three dimensions. Meanwhile, stable positive upward leader inception from some test points on
earthed structures is tracked by means of an adopted precise model of the leaderstreamer system
to find the ultimate striking point. The main computational volume is occupied by an electric field cal-
culation, which is performed using the charge simulation method. A direct movement model for the
downward leader is also simulated and compared with the stepping approach. The percentage of
strokes that attach to the phase wires, ground wires, and towers is extracted using this method. The
maximum return stroke current for the strokes that attach to the phase wires is also computed by this
dynamic simulation and is compared with the traditional methods that are available for transmission
line lightning performance investigation.
Keywords: lightning, upward leader, downward leader, charge simulation method, transmission line
1. Introduction
During recent decades, the lightning phenomenon has
been investigated by many researchers in different cate-
gories, such as the nature of lightning, the effects of light-
ning, the methods for protection against possible damages,
etc. Tall structures of electrical power transmission lines
are one of the major targets of lightning strokes. During
recent years, engineering efforts have been made to design
transmission lines with the lowest possible outage due to
lightning. One of the major measures for this purpose is
to install the ground wires on top of the tower to shield
c
the phase wires underneath. The task of these shield or
overhead ground wires is to act as collectors of the flashes
and insofar as possible to prevent flashes from terminat-
ing on the phase conductors and causing a flashover. How-
ever, flashes cannot be totally prevented from reaching the
phase wires, unless the phase wires are completely sur-
rounded by shield wires, which is not practically possible
[1]. Consequently, engineering methods have been devel-
oped to study transmission line performance against light-
ning. By using these methods, the maximum stroke cur-
rent that can pass the shield wire and attach to the phase
age. Therefore, it is desired that the high current strokes
strike the shield wires not the phase wires. The electro-
neering method to investigate the transmission line light-
SIMULATION
SIMULATION, Vol. 86, Issue 7, 2010 417427 July
wire is calculated [212]. High stroke currents attaching to
DOI: 10.1177/0037549709352258
geometrical model (EGM) has been used as the main engi-
the phase wires may lead to insulation failure and line out-
2010 The Society for Modeling and Simulation International
417 Volume 86, Number 7
Tavakoli and Vahidi
ning performance so far [29]. In essence, the EGM uses
the striking distance equation (which is extracted from the
final step of lightning attachment to the target point) and
simple geometrical equations in two dimensions to ana-
lyze the lightning performance of the transmission line.
There are different striking distance equations that have
been proposed by investigators up to now [29]. Although
the EGM is quite straightforward, simple, and quick, it
essentially encounters the following drawbacks: (1) the
EGM does not model the real configuration of overhead
transmission lines, such as overhead line declination (sag
of wires), the shape of the towers, and electric field en-
hancement by grounded structures, which may affect the
target point1 (2) the EGM does not take into account the
movement of the downward lightning leader from clouds
toward the ground and only considers the last step or
final jump (which is not physically correct, while on each
step of downward leader movement from the cloud to the
ground the electric field changes due to the charge carried
down by the leader and the target point may be affected).
In this paper, to overcome the above-mentioned prob-
lems, a novel method is proposed to analyze the per-
formance of transmission lines against lightning. This
method, which models the overhead line and its compo-
nents and is similar to the real configuration in three di-
mensions, tracks the movement of the negative downward
lightning leader from the cloud to the ground in order to
locate the final striking point. For this dynamic simula-
tion, clouds, two towers of a span, phase and ground wires,
and the ground are modeled in the simulation space based
on their physical shapes. The downward leader movement
is also modeled in a novel approach to be as similar as pos-
sible to field observations of downward lightning leaders,
which is essentially stepping for negative downward lead-
ers [13, 14] (near 90% of downward leaders are negative
[1]). However, a direct downward model of the lightning
downward leader (which is quicker but less accurate) is
also simulated and compared with the stepping approach.
This dynamic simulation of lightning attachment to trans-
mission lines also requires checking the upward connect-
ing leader inception from a target point. A quite precise
self-consistent model of the positive upward connecting
leader is adopted in this paper in order to find the target
point where a stable upward connecting leader is incepted
[1517]. All of the above models need the electric field
to be calculated in the simulation space on each step of
downward lightning leader movement toward the ground.
This most time-consuming part of the simulation is imple-
mented using the charge simulation method [18, 19].
Using the dynamic simulation approach, statistical data
for lightning attachment to the ground, wires, and tow-
ers can be extracted. Moreover, the maximum lightning
stroke peak current that can pass the shield and attach to
the phase wires is also calculated and is compared with
the values calculated using the EGM.
Generally, by using the dynamic simulation approach,
it is possible to construct a virtual model of lightning at-
tachment to transmission line structures on computer that
is close to the behavior of the lightning strokes that are
observed in nature. The results of this method can be used
for transmission line lightning performance analysis and
design. They can also act as a measure for evaluating the
precision of different striking distance equations used by
the EGM, because the dynamic simulation approach is
closer to what is happening in reality when a lightning
stroke attaches to the transmission line structures. Never-
theless, this method has its own cost. By increasing the
details of modeling, the required analysis time increases
as well. However, in the design stage and by our nowa-
days every-day faster computers, the long simulation time
is not a limiting factor as far as simulation cases are lim-
ited to practical situations.
The rest of this paper is organized as follows. In Sec-
tion 2 the configuration setup for modeling the clouds,
towers, and wires is described. Stepping and direct move-
ment models of negative downward leaders, as well as
the self-consistent upward lightning leader model, are pre-
sented in Section 3. Section 4 introduces the simulation
procedure and results. Finally, Section 5 concludes this
paper.
2. Models of Earthed Equipment
Any simulation of lightning attachment to ground struc-
tures requires electric field calculation in the environment
(which is composed of clouds, towers, ground, and wires)
to find other related parameters, such as space charges,
stable leader inception, decision on downward leader
direction, etc. Here we use charge simulation method
[18, 19] to calculate the electric field. When using the
charge simulation method, the ring, point, and line charges
are placed anywhere in the simulation space to enforce
boundary conditions to the values specified by the prob-
lem definition. The following virtual charges are used for
electric field calculation in the configuration of Figure 1:
(1) The model of clouds represents the charges in thun-
derstorm and their distribution. Simple unipolar
negative ring charges are used in this paper to model
the cloud charges. According to Figure 1, at the
height of 2000 m and with an extension of 5000 m,
concentric ring charges are placed with their centers
above mid-span axes. According to the field tests,
the storms are said to produce the electrical field in
the range of 120 kV/m at the ground level [20].
Here, in each calculation step, the cloud charges are
calculated so that the electric field near the ground
becomes 10 kV/m.
(2) Ring charges are also used to model the towers and
to set the boundary condition of zero voltage at the
tower structures. According to Figure 1, the exact
shape of the tower (including the base, side arms,
cross arm, and shield wire arms) is modeled so that
SIMULATION 418 Volume 86, Number 7
DYNAMIC SIMULATION OF LIGHTNING ATTACHMENT TO EARTHED OVERHEAD TRANSMISSION LINE STRUCTURES
Figure 1. Charge locations for field calculation by using the charge
simulation method
the set of ring charges modeling the tower creates
a shape similar to the tower itself. By assuming the
contour points around the tower on the middle of the
consecutive ring charges, the boundary condition of
zero voltage in three dimensions around the tower
is ensured.
(3) Short horizontal and vertical line charges are em-
ployed to model the wires and their declination (sag
of wires) using the well-known hyperbolic function
[12], see Figure 1:
z = h
1
cosh
1
k
2
S
2
x
34
(1)
k =
2
S
cosh
1
2
H
1
h
1
3
(2)
where h
1
is the height of the wire at mid-span, H
1
is the height of the wire at the connection point to
the towers and S is the span length, which are shown
in Figure 1. The boundary condition of zero voltage
for the ground wires and the peak positive nomi-
nal voltage for the phase wires are assumed (see Ta-
ble 1). For a balanced three-phase system, the al-
ternating voltage on each phase wire is changing
permanently with system nominal frequency (i.e.
50 Hz in our case). In essence, it is not possible to
know exactly the voltages on each phase wire when
the lightning approaches the wire in reality (it is an
important instance for the lightning to decide where
to attach). To take the worst case into account, an
instant is assumed when the power frequency volt-
age on one of phase wires is at its peak (this as-
sumption immediately indicates that the other two
Table 1. Simulation parameters and their values
Parameter Symbol Value Unit
Span length D 430 m
Sag of Wires Sag 9 m
Mesh length in X direction dx 43 m
Mesh length in Y direction dy 20 m
Ground (shield) wire radius 0.0049 m
Phase wire radius 0.0158 m
Phase wire peak voltage 400
_
22
_
3 kV
(boundary condition)
phases should have half-negative values because, in
a balanced three-phase system, in any instance, the
summation of voltages on all three phases is equal
to zero). However, choosing the middle phase at
its peak and the outer phases at half-negative value
keeps the symmetry in the model. Therefore, the
simulation, which is quite time consuming, can be
done only for part of the span, while a possible
worst case for lightning attachment is still simu-
lated. Any other assumption will void the symme-
try and eventually will lead to a dramatic increase
in the time required for simulation.
3. Simulation of Lightning Leaders
3.1 Downward Leader Model
Generally, the major portion (90% average) of down-
ward lightning leaders brings negative charges toward the
ground [1]. Therefore, the negative downward lightning
leader and consequently the positive upward connecting
leaders are almost always considered to analyze the light-
ning performance of overhead lines [1]. The same as-
sumption is also made here for the dynamic simulation
of lightning stroke attachment to overhead line structures.
There are two possibilities to model the downward
lightning leader, see Figure 2. The first is to have a down-
ward leader coming toward the earth directly and with-
out stepping, parallel to the Z-axis in Figure 1. This ap-
proach is taken by some researchers in their own research
[1, 17]. Although this method is quite far from the pho-
tographs and field observations that are available for neg-
ative downward leaders [21], it has the advantage of con-
suming a lower amount of time for the simulation and
is quite easy to implement. The observations of negative
downward leaders revealed that these leaders approach the
earth from the cloud in consecutive steps [2123]. There-
fore, the second approach, which is mainly implemented
in this paper, is to simulate the stepping nature of the neg-
ative downward leader similar to the field observations.
Nevertheless, modeling the stepping nature is hard to im-
plement and requires the model to find the next jump-
ing point (which needs the electric field calculation in all
SIMULATION Volume 86, Number 7 419
Tavakoli and Vahidi
Figure 2. Downward lightning leader, line charge segments and
contour points for finding the next jumping point: (a) stepping
downward model1 (b) direct downward model
points of the simulation space). It is also quite time con-
suming.
All of the efforts in this paper are made to bring the
model of the negative downward leader closer to the phys-
ical observations. The propagation of the negative down-
ward leader is modeled by steps with a length of 1/20
of the distance of the tip of the downward leader from
the ground. This assumption ensures the step length in-
creases gradually when the downward leader approaches
the ground and limits the step length to an upper bound
of 100 m in accordance with the field observations [23].
Similarly a lower bound of 10 m is applied in the model
to confine the step length of the downward leader near
to the ground (based on the same observations). In the
stepping model, it is important to know the direction of
leader movement toward the ground. It is assumed that the
next leader jump will be a point where the maximum volt-
age gradient exists between the current leader tip and the
jumping point [12], see Figure 2(a). In the direct model,
the direction is always vertically down and thus only one
degree of freedom exists in the Z axes direction. The steps
of the downward leader are modeled by horizontal and
vertical line charges. The charge density of the downward
leader is computed by the following equation, based on
electrostatic considerations of measured waveforms of the
return stroke current [9]:
34z5 = I
p
5
m
0
2
1
z z
0
H
c
z
0
32
1
z
0
H
c
3

m
1
m
2
4z z
0
5
1 m
3
4z z
0
5 m
4
4z z
0
5
2

1
063e
10z
0
75
067
2
1
z
0
H
c
346
(3)
where m
0
= 16476 10
5
, m
1
= 46857 10
5
, m
2
=
369097 10
6
, m
3
= 06522, m
4
= 3673 10
3
, z
0
is the
height of the downward leader tip in m, H
c
is the cloud
height in m, I
p
is the return stroke peak current in kA, 3
is the downward leader charge density in C/m and z is a
variable representing the height of the point on the leader
where charge density is to be calculated, see Figure 2. The
charge simulation method that is utilized to calculate the
electric field requires us to compute free-space charges of
the downward leader. To do so, the downward leader in
Figure 2 is divided into sections with unequal lengths. In
each section, the absolute charge should be calculated by
multiplying the charge density at the middle of that section
(see Equation (3)) to the section length. However, this is
true only if the change in charge density in the points of
the section is negligible. The length of the line charges
that are used to model the downward leader is intention-
ally increased from the tip of the downward leader to the
base of the cloud. The reason behind this type of modeling
is the shape of Equation (3). According to this equation,
the change in the charge density of the downward leader
near its tip is quite high. By dividing the leader near its
tip to small lines, the change in the charge density in these
small lines would be small and the charge of the section is
correctly calculated.
3.2 Upward Leader
When the downward lightning leader approaches the
earth, electric field enhancement by the charges of this
downward leader cause a stable upward connecting leader
to incept from a point on the ground structures. This is
an important stage of lightning stroke attachment to any
structure on the earth and is called the discrimination in-
stance for stroke to attach the target point. The precision
of the model of the upward connecting leader directly
influences the reliability of target point identification.
During recent years and based on different laboratory
tests and field observations, different criteria for stable
upward connecting leader inception have been proposed
[10, 17, 24]. Recently, a precise self-consistent model is
introduced based on the space-charge development in the
corona zone of the leaderstreamer system [16, 17]. This
model tracks the movement of positive leaderstreamer
system and checks for some physical criteria to deter-
mine whether a leaderstreamer system is stable or not.
The lightning will attach to a point where the leader
streamer system initiating from that point proves to be
stable, based on the physical criteria [17]. The model of
the upward connecting leader first calculates the initial
streamer length and the charge produced in the local ion-
ization region of the streamer zone for each test point on
the ground structures as follows:
L
s
405 =
U
0
E
str
E
1
(4)
SIMULATION Volume 86, Number 7 420
DYNAMIC SIMULATION OF LIGHTNING ATTACHMENT TO EARTHED OVERHEAD TRANSMISSION LINE STRUCTURES
7Q 405 =
1
2
K L
2
s
405 4E
str
E
1
5 (5)
where U
0
and E
1
are the voltage and field on the fitted
line of the background electric field (which is calculated
by the charge simulation method) [17]. E
str
is the con-
stant electric field in the streamer zone, which is taken to
be 450 kV/m for positive upward connecting leaders, and
K is a geometrical factor depending on the configuration
and the shape of the streamer and is taken to be 3.5
10
11
C/Vm. L
s
405 and 7Q405 are the initial length and
charge of the streamer zone, respectively. The criterion
used to determine when the initial streamer charges tran-
sit to a leaderstreamer system is given by the inequal-
ity 7Q405 8 19C. If it is fulfilled, the mechanism of
leaderstreamer system advancement is then simulated by
the following equations [17]:
U
T
4i 5 = E
o
L
l
4i 5 x
0
E
o
ln
1
E
0
E
o

E
0
E
o
E
o
e
L
l
4i 52x
0
4
(6)
L
s
4i 5 =
U
0
E
str
L
l
4i 5 U
T
4i 5
E
str
E
1
(7)
7Q 4i 5 = K [L
s
4i 15 L
l
4i 5]
{E
str
[L
l
4i 5 L
l
4i 15]
U
T
4i 15 U
T
4i 5} (8)
7L
l
4i 5 =
7Q 4i 5
q
(9)
L
l
4i 15 = L
l
4i 5 7L
l
4i 5 (10)
where U
T
4i 5 is the leader potential in step i, E
o
and E
0
are the initial value and final quasi-stationary values of
the leader gradient and are taken to be 400 kV/m and 30
kV/m, respectively [25], x
0
(=10 m) is a constant that de-
pends on the speed of the upward leader and the constant
of the conductance equation of the leader channel. q is
a constant that represents the charge per unit length nec-
essary to achieve the thermal transition from the diffuse
glow to the leader channel [26], which is assumed to be
50 9C/m.
The procedure is continued until L
l
4i 5 reaches a
specified length (taken to be 5m). If during this advance-
ment for any test point, 7Q 4i 5 stays always positive, then
it is approved that a stable upward leader is incepted from
that point, i.e. the target point is found. If 7Q(i) always
stays positive, the total charge Q(i) in the streamer re-
gion of the leaderstreamer system is always increasing,
which is an indication of the breakdown initiated from
the test point. Also based on the self-consistent model,
if the leaderstreamer system reaches a length such as 5
m, the speed and charge development in the active re-
gion of the streamer zone would high enough to ensure
that the leaderstreamer system is stable and electrical
breakdown is imminent from the test point. These criteria
are verified with laboratory experiments and also rocket-
triggered lightning experiments, see [14, 16, 17, 27]. At
this stage, it just remains to recognize whether the stable
leaderstreamer system is initiated from the phase wires
(shielding failure), shield wires, or towers. The direction
of all upward connecting leaders is taken to be toward the
tip of the downward leader.
The ground strikes are assumed to occur if no stable
leader is incepted from any test point on the structures
(towers, phase wires, and ground wires) until the down-
ward leader reaches a height of 5 m above ground. This
approach is taken because the goal of simulation is to find
the location where the stroke strikes. Obviously, if the tar-
get point is not the phase wires, ground wires, or towers,
the stroke will attach to the ground in our model. More-
over, based on the downward model, each step is 1/20 of
the current leader height. Therefore, near to the ground the
step length will decrease considerably. If the simulation is
not interrupted at a specified height, the simulation time
required for the downward leader to reach the earth will
increase dramatically. It is quite wise to stop the simula-
tion at a specified height of the downward leader tip from
ground and assume a ground strike in order to save time
[12].
4. The Simulation Procedure and Results
The procedure and steps of performing the dynamic sim-
ulation are as follows:
(1) The model of towers, wires, and the ground is cre-
ated for one span (two towers and five wires), in-
cluding detailed geometry in three dimensions, and
proper charges are also placed according to the
charge simulation method.
(2) The space above the span is divided into meshes.
The width of area in the Y direction, where the sim-
ulation should be performed (according to Figure 3)
is selected so that out of this area no flash to the
wires and towers occurs. Owing to the symmetry
that exists in the problem in the X and Y directions
(Figure 3), it is enough to perform the procedure of
simulation for half of the span length and half of the
width of the simulation area. This part of the area
is shown in Figure 3 and is divided into meshes to
show the numbering procedure.
(3) In a specified mesh, a downward leader is set to de-
scend toward the ground utilizing each approach of
Figure 2. This procedure is repeated for all values
of return stroke current in a reasonable range. In
each downward leader step, the electrical field in
some test points in front of the downward leader is
SIMULATION Volume 86, Number 7 421
Tavakoli and Vahidi
Figure 3. One span of overhead line and starting meshes for the downward lightning leader
computed to find the next point of the downward
leader jump. The electric field is also computed for
each test point on the ground structures in order to
check the stable upward leaderstreamer system in-
ception.
(4) After finding the stable leader inception from the
phase wires in each mesh, the maximum return
stroke current of the lightning leader striking the
phase wires is extracted. Moreover, the height of the
downward leader on the discrimination instance and
the statistical data of stroke inception are computed
for different shielding angles. The shielding angle
is the angle between the line connecting the shield
wire to the phase wire and the vertical line to the
ground, see Figure 4.
(5) To check the effectiveness of the shielding, the pre-
ceding procedure is repeated for the same structure
but different shielding angles. A reasonable range
of shielding angle is checked to find the best angle,
which results in the least possible strokes attach-
ing to the phase wire (optimum shielding angle).
The range in which the shielding angle is changed
should be selected based on experience to include
the optimum shielding angle.
The tower dimensions and simulation parameters are
shown in Figure 4 and Table 1. In Table 1 D is the span
length (the distance between the two towers in Figure 3),
dx and dy are the mesh length in the X direction and the Y
direction, respectively (see Figure 3), sag is the amount of
wire declination at mid-span and in the Z direction, mea-
sured from the wire connection point to the towers (see
Figure 1). In this table the radius of the ground and phase
wires are also tabulated.
In Figure 5 the number of strokes out of the total sim-
ulated cases that strike the phase wires, ground wires,
ground, and towers for two different shielding angles are
shown. Please note that the shielding angle (see Figure 4)
is changed by shifting the location of the shield wire in
the X direction only. The height of the shield wires re-
SIMULATION Volume 86, Number 7 422
DYNAMIC SIMULATION OF LIGHTNING ATTACHMENT TO EARTHED OVERHEAD TRANSMISSION LINE STRUCTURES
Figure 4. 400 kV tower and its dimensions and definition of the
shielding angle
mains unchanged. Fromthis figure, it is revealed that most
strokes attach to the shield wires, which are higher and in
the path of the strokes. Slight differences can be seen be-
tween two shielding angles. A shielding angle of 12

re-
sults in a lower number of strokes to the phase wires and a
higher level of lightning absorption by the shield wires. To
make it sure, the shielding angle is lowered with constant
steps up to a point where no further decrease can be seen,
see Figure 6. According to this figure, better performance
will achieve by lowering the shielding angle to 8.
For each mesh, the maximum return stroke current that
attaches to the phase wire is also derived by the pro-
posed dynamic simulation. In Figure 7 the maximum re-
turn stroke current for the strokes that attach to the phase
wires is shown for each mesh. According to this figure,
the strokes initiating from the meshes directly above the
lines and far from towers will not incept the phase wire,
and if inception does occur the return stroke current will
be rather low. This happens because the shield and phase
wires of the overhead line decline in the Y direction (i.e.
the sag of the ground and phase wires), see Figure 1.
This declination makes the phase wires come nearer to
the ground in higher mesh numbers in the Y direction
Figure 5. Strike position statistics for two different shielding angles
Figure 6. Phase wire strokes in different shielding angles
of Figure 7. As a result, the ground enhances the elec-
tric field on the phase wires and, therefore, the downward
leader approaching the ground with a high level of charges
(i.e. higher return stroke currents, see Equation (3)) causes
the points on the ground wires, which are higher than the
phase wires, to have a higher electric field. Eventually, this
electric field enhancement makes the strokes attach to the
ground wires. Therefore, for the meshes above the line and
far from the towers, the high current strokes attach to the
ground wires, not the phase wires. This can be seen in Fig-
ure 7 along with higher mesh numbers in the Y direction.
The same prediction is also made by the EGM [19].
In Figure 8, the height of the downward leader from
the ground is sketched versus the return stroke current
when the stable upward leaderstreamer system is in-
cepted from the earthed structures (final jump of the
downward leader). The higher return stroke current in
any mesh means that a discrimination instance occurs in
SIMULATION Volume 86, Number 7 423
Tavakoli and Vahidi
Figure 7. Maximum return stroke current in each mesh, stepping
model
Table 2. Comparison of results between stepping and direct
downward leader models
Parameter Direct model Stepping model
Max return stroke current 13.5 11.5
striking the phase wires (kA)
Phase stroke (%) 6.23 9.54
CPU time 8 h 9 min 13 h 18 min
higher altitudes because the electric field on the ground
structures rises more quickly when larger charges are
brought to ground by a downward leader with a higher
peak current (see Equation (3)). This can be seen in Fig-
ure 8, where higher return stroke currents make the final
jump of the downward leader occur in higher altitudes.
Two approaches to model the downward leader (see
Section 3.1) are also compared in Figure 8. In the di-
rect downward leader approach, the discrimination alti-
tude becomes higher. A comparison is also made between
the stepping and the direct downward leader models in
Table 2. It seems that the direct model slightly underes-
timates the percentage of strokes attaching to the phase
wires.
None of the negative downward leaders in nature ap-
proach the earth directly. The stepping nature is the ab-
solute physical characteristic of negative downward lead-
ers. Therefore, the stepping approach is more accurate be-
cause it models the behavior of negative downward lead-
ers according to the observations [22, 23]. Nevertheless,
the direct downward leader model is simpler and quicker
and it yields all of the information that can be obtained
with the stepping method. The predicted maximum return
stroke current by the direct method is not that far from
the stepping model. The direct method can be used in ex-
tensive (but less accurate) simulations where the stepping
model of the downward leader makes the simulation time
become unacceptably high.
In Figure 9, the maximum return stroke current attach-
ing to the phase wires is computed using traditional EGM
methods and is compared with the value calculated us-
ing the dynamic simulation method (stepping downward
leader). According to this figure, different EGM methods
result in quite different values of the maximum current of
lightning that can attach the phase wires. Coming back to
the discussion in Section 1, the EGM methods are not ca-
pable of modeling precisely the field enhancement of the
grounded structures, the shape of the ground and phase
wires, or their natural declination and downward leader
movement toward the ground. Based on the detailed mod-
eling level of dynamic simulation approach, it can be used
as a measure to compare different EGM methods. From
Figure 9, we can see that among different EGM methods,
the prediction of the Eriksson model is more similar to the
calculated value using dynamic simulation. Some EGM
methods predict quite high values compared with the dy-
namic simulation, which may be misleading in design.
Moreover, the EGMmethods could not yield the statistical
information, such as that presented in Figure 5. Such sta-
tistics can only be derived out of extensive dynamic simu-
lations of lightning attachment to earthed structures.
According to Table 2, simulation time is rather high
for the stepping model (the time is valid for a computer
with the following specifications: central processing unit
(CPU) clock 2.8 GHz, L1 cache 32 KB, L2 cache 2 MB,
front-side bus (FSB) 800 MHz and 1 GB RAM). The total
CPU time for the simulation is also shown for all meshes
in Figure 10. A higher CPU time for mid-span meshes is
seen where the simulation continues until the downward
leader approaches the ground with a lower step length, see
Section 3.1. It is also good to mention here that for EGM
methods, the computation time is negligible as shown in
the comparison of the values in Table 2. This is because
the EGMuses simple geometry (two dimensional) to com-
pute the performance parameters and does not require sig-
nificant simulation time.
5. Conclusion and Discussion
In this work a novel method is introduced, based on the
dynamic simulation of lightning attachment to transmis-
sion overhead line structures, in order to analyze the light-
ning performance of overhead lines and to evaluate the
shielding effectiveness at the design stage. This method
creates a simulation space in three dimensions to model
the clouds, towers, wires, and ground close to reality. The
dynamic simulation approach also takes into account the
stepping nature of negative downward lightning leaders
in accordance with the field observations. A precise self-
consistent model is adopted in order to model the posi-
tive upward connecting leader, which is utilized to deter-
mine the point where the stroke eventually attaches [16,
SIMULATION Volume 86, Number 7 424
DYNAMIC SIMULATION OF LIGHTNING ATTACHMENT TO EARTHED OVERHEAD TRANSMISSION LINE STRUCTURES
Figure 8. Discrimination altitude of downward leader tip for the direct and jumping models
Figure 9. Maximum return stroke current that can pass the shield and attach the phase wires, calculated by the proposed dynamic
simulation method and EGM methods
17]. The proposed method is applied on a 400 kV over-
head transmission line and the statistics of lightning at-
tachment to earthed structures, as well as the maximum
return stroke current that can pass the shield and strike
the phase wire, are calculated and presented. A compari-
son is also made between two approaches for downward
lightning leader movement and also between the tradi-
tional EGM methods and dynamic simulation for calcu-
lating the maximum return stroke current attaching to the
phase wires. Moreover, by changing the shielding angle
SIMULATION Volume 86, Number 7 425
Tavakoli and Vahidi
Figure 10. CPU time required for the simulation in each mesh
(stepping model)
of the overhead lines, the simulation is repeated to find
the optimum shielding angle where the minimum number
of strokes to the phase wires is observed.
As stated Section 1, the EGM methods that are tra-
ditionally utilized to analyze overhead line performance
against lightning are essentially two-dimensional ap-
proaches and take the overhead line geometry to cal-
culate the performance parameters. The EGM is based
on striking distance equations, which are developed con-
sidering the final jump of lightning leaders [29]. The
three-dimensional structures of the towers and conduc-
tors, as well as the downward leader movement toward
the ground, enhance the electric field and consequently
affect the target point of stroke. Being a two-dimensional
approach, the EGM is not capable of modeling the struc-
tures of overhead lines precisely, e.g. the declination of
conductors, towers, and clouds. Moreover, the EGM does
not consider the downward leader movement toward the
ground. The dynamic simulation approach, which is im-
plemented in a three-dimensional space in this paper, takes
care of the above-mentioned drawbacks of the EGM. It
can be used to check the results of the EGM when ana-
lyzing the lightning performance of transmission lines for
special cases whenever there is an engineering design bot-
tleneck or an unusual lightning performance of an already-
operating overhead line.
The accuracy and precision of the dynamic simulation
method of lightning attachment to overhead lines is di-
rectly dependent on the models that are implemented in
the simulation space. The models of clouds, towers, and
conductors are adjusted in this paper to fit closely to the
reality. The downward lightning leader is also modeled
based on the field observation of negative downward lead-
ers. The adopted model for positive upward connecting
leader inception is also quite precise, which is verified
by laboratory measurements and rocket-triggered light-
ning experiments [14, 17]. Therefore, the models that are
implemented in the dynamic simulation approach are in-
dependently examined to comply with the experiments
and field observations. All of these measures bring the
lightning attachment to overhead lines, which is observed
by dynamic simulation in a virtual software environment
close to the reality. Nevertheless, the random nature of
lightning phenomenon and many non-controllable factors
in dealing with lightning confine the detailed field data
of lightning attachment to transmission lines in operation.
Therefore, the common verification approach is to check
the models that are used as the core of simulation, us-
ing laboratory measurements and available field data of
lightning leaders. The dynamic simulation approach cre-
ates a good ground to utilize the already-verified models
for the particular application of the lightning performance
of transmission lines.
On the other hand, the required simulation time in-
creases considerably when models with a higher level of
detail are implemented. The simulation needs high-speed
computers to finish in a reasonable amount of time. This
is not a limiting factor with todays high-speed comput-
ers1 however, it may become a major problem when a
large number of cases is to be simulated. Accepting the
simplifications, the direct downward lightning leader ap-
proach may work well for such cases because it requires
less simulation time.
6. References
[1] Hileman, A.R. 1999. Insulation Coordination for Power Systems,
Power Engineering Series, Marcel Dekker Inc., New York.
[2] Young, F.S., J.M. Clayton and A.R. Hileman. 1963. Shielding of
transmission lines. IEEE Transactions on Power Apparatus and
Systems, S82: 132154.
[3] Brown, G.W. and E. R. Whitehead. 1969. Field and analytical stud-
ies of transmission line shielding: part II. IEEE Transactions on
Power Apparatus and Systems, PAS-88: 617626.
[4] Mousa, A.M. and K.D. Srivastava. 1988. Effect of shielding by trees
on the frequency of lightning strokes to power lines. IEEE Trans-
actions on Power Delivery, 3: 724732.
[5] Mousa, A.M. and K.D. Srivastava. 1989. The lightning performance
of unshielded steel-structure transmission lines. IEEE Transac-
tions on Power Delivery, 4: 437445.
[6] IEEE Substations Committee. 1996. IEEE Guide for Direct Stroke
Shielding of Substations, p. 998, IEEE Std.
[7] IEEE. 1997. IEEE Guide for Improving the Lightning Performance
of Transmission Lines, p. 1243, IEEE Std.
[8] Eriksson, A.J. 1987. An improved electrogeometric model for trans-
mission line shielding analysis. IEEE Transactions on Power De-
livery, 2: 871886.
[9] Cooray, V., V. Rakov and N. Theethayi. 2007. The lightning striking
distance revisited. Journal of Electrostatics, 67: 296306.
[10] Dellera, L. and E. Garbagnati. 1990. Lightning stroke simulation
by means of the leader progression model. Part I: description of
the model and evaluation of exposure of free shielding structures.
IEEE Transactions on Power Delivery, 5: 20092022.
[11] He, J., Y. Tu, R. Zeng, J.B. Lee, S.H. Chang and Z. Guan. 2005.
Numeral analysis model for shielding failure of transmission line
SIMULATION Volume 86, Number 7 426
DYNAMIC SIMULATION OF LIGHTNING ATTACHMENT TO EARTHED OVERHEAD TRANSMISSION LINE STRUCTURES
under lightning stroke. IEEE Transactions on Power Delivery,
20: 815821.
[12] Vahidi, B., M. Yahyaabadi, M.R. Bank Tavakoli and S.M. Ahadi.
2008. Leader progression analysis model for shielding failure
computation by using charge simulation method. IEEE Trans-
actions on Power Delivery, 4: 22012206.
[13] Lalande, P., A. Bondiou, G. Bacchiega and I. Gallimberti. 2002.
Observations and modeling of lightning leaders. Comptes Rendus
Physique, 3: 13751392.
[14] Willet, J.C., D.A. Davis and P. Laroche. 1999. An experimental
study of positive leaders initiating rocket-triggered lightning. At-
mospheric Research, 51: 189219.
[15] Bondiou, A. and I. Gallimberti. 1994. Theoretical modeling of
the development of the positive spark in long gaps. Journal of
Physics D: Applied Physics, 27: 12521266.
[16] Goelian, N., P. Lalande, A. Bondiou-Clergerie, G.L. Bacchiega, A.
Gazzani and I. Gallimberti. 1997. Asimplified model for the sim-
ulation of positive-spark development in long air gaps. Journal of
Physics D: Applied Physics, 30: 24412452.
[17] Becerra, M. and V. Cooray. 2006. Asimplified physical model to de-
termine the lightning upward connecting leader inception. IEEE
Transactions on Power Delivery, 21: 897908.
[18] Singer, H., H. Steinbigler and P. Weiss. 1974. A charge simulation
method for the calculation of high voltage fields. IEEE Transac-
tions on Power Apparatus and Systems, PAS-93: 16601668.
[19] Malik, N.H. 1989. A review of the charge simulation method and
its application. IEEE Transactions on Electrical Insulation, 24:
320.
[20] Bazelyan, E.M. and Y.P. Raizer. 2000. Lightning Physics and Light-
ning Protection, Institute of Physics Pub., Bristol.
[21] Horvath, T. 1991. Computation of Lightning Protection, Research
Studies Press Ltd., London.
[22] Cooray, V. 2003. The mechanismof the lightning flash. In V. Cooray
(Ed.), The Lightning Flash, pp. 144159, Institution of Electrical
Engineers, London.
[23] Berger, K. 1967. Novel observations on lightning discharges: re-
sults of research on mount San Salvatore. Journal of the Franklin
Institute, 283: 478525.
[24] Rizk, F.A.M. 1990. Modeling of transmission line exposure to di-
rect lightning strokes. IEEE Transactions on Power Delivery, 5:
19831997.
[25] Rizk, F.A.M. 1994. Modeling of lightning incidence to tall struc-
tures part I: theory. IEEE Transactions on Power Delivery, 9:
162171.
[26] Gallimberti, I. 1972. The mechanism of long spark formation. J.
Physique Coll. 40: 193250.
[27] Les Renardires Group. 1977. Positive discharges in long air gaps
at Les Renardires 1975 Results. Electra, 53.
Mohammad Reza Bank Tavakoli was born in Kerman, Iran, in
1981. He received the B.S. in electrical engineering from Tehran
University, Tehran, Iran, in 2003 and M.S. degree in electrical
engineering from Amirkabir University of Technology, in 2005.
Currently, he is working toward Ph.D. degree at the department
of electrical engineering of Amirkabir University of Technology.
His main fields of interests are power system components mod-
eling and simulation, power system dynamics and fast transients
in power system. He is also with the Kerman Regional Electric
Company as a planning and network consulting engineer.
Behrooz Vahidi was born in Abadan, Iran in 1953. He received
the B.S. in electrical engineering from Sharif University of Tech-
nology, Tehran, Iran in 1980 and M.S. degree in electrical engi-
neering from Amirkabir University of Technology, Tehran, Iran
in 1989. He also received his Ph.D. in electrical engineering
from UMIST, Manchester, UK in 1997. From 1980 to 1986 he
worked in the field of high voltage in industry as chief engineer.
From 1989 to present he has been with the department of electri-
cal engineering of Amirkabir University of Technology where he
is now a professor. He is IEEE senior member. His main fields
of research are high voltage, electrical insulation, power sys-
tem transient, lightning protection and pulse power technology.
He has authored and co-authored 170 papers and five books on
high voltage engineering and power system.
SIMULATION Volume 86, Number 7 427

También podría gustarte