Professional Documents
Culture Documents
I. Introduction
It is not necessary to explain the significance of a lightning surge analysis in a power system. For the last 30 years,
the electromagnetic transients program (EMTP) originally developed by the Bonnevill Power Administration, US
Department o Energy [1, 2], has been the most powerful tool to carry out a predictive calculation of lightning overvoltages
for insulation design and coordination of a substation and transmission lines. However, it is well-known that the EMTP is
based on a circuit theory assuming TEM mode wave propagation, and the parameters of a circuit are to be given for an
EMTP simulation [3].
Recently, digital control circuits have become quite common in power stations / substations and in intelligent
buildings, and a number of electromagnetic disturbances due to lightning and switching operation have been reported [4].
Also, traveling wave propagation of a partial discharge on a gas-insulated bus [5] and a detailed analysis of archon flashover
on a transmission tower [6] become significant for a more reliable operation and insulation coordination. These
phenomena involves non-TEM mode propagation in a very high frequency range, and can not be analyzed by a conventional
circuit- theory based approach such as the EMTP. Because of the above explained situation, numerical electromagnetic
analysis methods are becoming an alternative but a very effective approach to analyze a lightning surge [7, 8].
This paper summarizes basic assumptions and related problems of the EMTP and thus explains the application limit
of the EMTP for a lightning surge analysis. Then, a brief summary of a numerical electromagnetic analysis (NEA)
methods is presented together with application examples. Finally, a difference between the EMTP and the FDTD
simulations is investigated.
II. Recommended Modeling Method by EMTP [3]
The EMTP is based on a circuit theory which is derived from TEM mode wave propagation. Therefore, it should be
noted that the EMTP is not applicable to a phenomenon associated with non-TEM mode propagation. Also, all the
Zl=400
lightning current
tower
No. 5
matching impedance
eA eC eB eB eC eA AC source
Fig. 1
xt
tower
No. 2
tower
No. 1
8-phase
line
model
8-phase
line
model
Rf
xt
I(t)
Rf
archorn
xs
gantry
8-phase
line
model
Rf
substation
Rfg
Simplified illustration of the recommended model system for a lightning surge simulation
L1-1
parameters of a circuit are to be prepared. Fig. 1 illustrates a basic configuration of a model system recommended in Japan
for a lightning surge simulation by EMTP.
A.
Number of towers
Five towers from a substation gantry are to be considered, because traveling waves reflected from the towers affect a
lightning surge voltage in a substation in Fig.1. The first reflection, when lightning strikes the first tower at t =0, from the
last tower arrives at the substation entrance after two travel times T between the first and the last towers.
For example, xt =300m with light velocity c0 =300m/s gives the time T 10 s. Thus, a simulation result is
accurate up to 10s, which is enough in general to observe the maximum overvoltage and the time to the peak.
B.
h=h 1
SW1
GW
L1
Zt1 , c0
R 1, L1
h2
L3/2
SW3
Zt1 , c0
L3/2
R 2, L2
middle
Zt1 , c0
x3
h4
R 3, L3
lower
Zt4 , c 0
x4
R 4, L4
Fig. 2
Ln
L0
SW2
upper
x2
h3
SW1
SW2
L2
x1
Table 1
(b)
system
voltage
lightning
current
surge imp.
[]
footing
res. []
[kV]
[kA]
x1
x2
x3
Zt1
Zt4
Rf
1100
200
107
12.5
18.5
18.5
130
90
10
10
500
150
79.5
7.5
14.5
14.5
220
150
275
100
52.0
9.0
7.6
7.6
220
150
10
154(110)
60
45.8
6.2
4.3
4.3
220
150
10
77(66)
30, 40
28.0
3.5
4.0
3.5
220
150
10-20
V-I characteristic of an archorn flashover. Then the first EMTP simulation with no archorn flashover is carried out in Fig. 1,
and the first flashover phase and the initial time to are determined from the simulation results of the voltage waveforms across all the
archorns. By adopting the above parameters, the second EMTP simulation only with the first phase flashover is carried out to determine
the second flashover phase. By repeating the above procedure until no flashover occurs, a lightning surge simulation by the piecewise
linear model is completed. Thus, a number of pre-calculations are necessary in the case of multiphase flashovers, while the nonlinear
inductance model needs no pre-calculation and is easily applied to multiphase flashovers. The detail of the leader progression model is
explained in Reference [18], and that of the nonlinear
Arrester
An inductance of a lead wire and the arrester itself is connected in series to the arrester model, because it affects a
transient voltage and current of the arrester. Also, a capacitance of the arrester is considered if necessary. To represent a
very fast impulse characteristic of the arrester, an IEEE model [20] or a model of a nonlinear resistance with a nonlinear
inductance [21] is occasionally adopted. The latter model takes into account the hysterisis of an arrester V-I characteristic
by adding the nonlinear inductance to the nonlinear resistance in series.
J. Substation
(1) Gas insulated bus and cable
A cable and a gas-insulated bus are represented either as three single-phase distributed lines with its coaxial mode
surge impedance and propagation velocity or as a three-phase distributed line system. For a gas-insulated substation
involves quite a number of gas-insulated buses/lines, the pipes are, in most cases, eliminated by assuming the voltage being
zero.
(2)
(3)
Grounding mesh
A grounding mesh is in general not considered in a lightning surge simulation, and is regarded as a zero-potential
surface. When dealing with an incoming surge to a low-voltage control circuit, the transient voltage of the grounding mesh
should not be assumed zero, and its representation becomes an important but difficult subject [22].
L1-3
AC source voltage
An ac source voltage is often neglected in a lightning surge simulation. It, however, has been found that the ac
source voltage affects a flashover phase of an archorn especially in the case of a rather small lightning current. Fig. 4 is a
measured result of archorn flashover phases as a function of the ac source voltage on a 77kV transmission line in Japan for a
summer [26]. The measurements were carried out in two 77 kV substations by surge recorders installed in the substations.
From the recorded voltages and currents, Fig. 4 was obtained. The figure clearly shows that the archorn flashover phase is
quite dependent on the ac source voltage, i.e. a flashover occurs at a phase of which the ac voltage is in the opposite polarity
of a lightning current. Table 2 shows a simulation result of archorn peak voltages (archorn not operating) on (a) the
77kV line and (b) a 500kV line [27]. The simulation was carried out in a similar circuit to Fig. 1, but another five towers
were added instead of the gantry and the substation.
The parameters are the same as those in Table 1 for a 77 kV system except the lightning current of 40 kA based on the
field measurement[26]. The lower phase archorn voltage is relatively smaller than the other phase archorn voltages on the
500kV line compared with those on the 77kV line. Thus, an archorn flashover phase on an EHV line is rather independent
from the ac source voltage, and the lower phase flashover is less probable than the other phase flashover. On the contrary,
flashover probability is rather same on each phase and a flashover is dependent on the ac source voltage on a low voltage
line.
C. Tower model
(1) Problem of recommended tower model
Fig. 5 shows simulation results of archorn flashover phases by a simple distributed line tower model i.e. neglecting
the RL circuit in Fig. 2 with the parameters in Table1, and by the recommended model illustrated in Fig. 2. The simulation
circuit is the same as that described for Table 2 in the previous section. This figure should be compared with the field test
result shown in Fig. 4. It is clear that the recommended model can not duplicate the field test result, while the simple
80
lower
middle
upper
60
voltage [kV]
40
20
angle [deg.]
-20
30
60
90
-40
-60
-80
Fig. 4
transmission voltage
upper
middle
lower
500kV
4732 / 1.025
4334 / 1.073
3423 / 1.122
L1-4
Lower
Middle
Upper
Phase [deg.]
30 60 90 120 150 180 210 240 270 300 330 360
Voltage[kV]
Voltage[kV]
80
60
40
20
0
0
-20
-40
-60
-80
80
60
40
20
0
0
-20
-40
-60
-80
Lower
Middle
Upper
Phase [deg.]
30 60 90 120 150 180 210 240 270 300 330 360
distributed line model shows a good agreement with the field test result. The reason for the poor accuracy of the
recommended model is that the model was developed originally for a 500kV line on which the lower phase flashover was
less probable as explained in the previous section. Thus, the recommended tower model tends to result in lower flashover
probability of the lower phase archorn. An R-L parallel circuit between two distributed lines in Fig. 2 represents traveling
wave attenuation and distortion along a tower. The R and L values were determined originally based on a field
measurement ( in eq. (4)), and thus those are correct only for the tower on which the measurement was carried out.
Sometimes, the R-L circuit generates unreal high frequency oscillation. This indicated a necessity of further investigation of
the R-L circuit.
(2)
The recommended value of a tower surge impedance for each voltage class in Table 1 was determined by field
measurements in Japan. Although the surge impedance is a representative value, it can not be applied to every tower as is
clear in Table 3.
Wave deformation on tower structures (L- or T-shape iron conductor) can be included in a lightning surge analysis, if
required, based on the approach in Reference [30].
(3)
(4)
It should be pointed out that the influence of the surge impedance and the frequency-dependent effect of a tower is
heavily dependent on the modeling of a tower footing impedance, which will be discussed in the following section. When
the footing impedance is represented by a resistive model as recommended in Japan or by a capacitance model, then the
influence of the tower surge impedance and the frequency-dependent effect of a traveling wave along the tower becomes
rather noticeable. On the contrary, those cause only a minor effect when the footing impedance is represented either by an
inductive model or by a nonlinear resistance. Fig. 6 shows an example [7], [33]. The measurement was carried out on a
500 kV tower by applying a current in Fig. 6(a-1) to the top of the tower. The tower top voltage predicted by a
distributed-line model with a constant tower surge impedance and no R-L circuit, Fig. 6(c-1), differs from that by the
frequency-dependent tower model, Fig. 6(b) which agrees with the measured result, in the case of the footing impedance
being a resistance. On the contrary, in the case of an inductive footing model, the tower top voltage obtained by the
distributed-line model shows a rather good agreement with the measured result. It should be also noted that some 10%
variation of the tower surge impedance does not affect the result in the inductive footing impedance case.
L1-5
Ref.
height
h [m]
[31]
radius
r [mm]
15.0
25.4
15.0
2.5
9.0
2.5
6.0
2.5
3.0
50.0
3.0
25.0
3.0
2.5
2.0
2.5
0.608
43.375
0.608
9.45
0.608
3.1125
average of absolute error [%]
[30]
320.0
459.0
432.0
424.0
181.0
235.0
373.0
345.0
112.0
180.0
250.0
323.0
462.0
431.3
407.0
187.2
228.0
365.5
341.2
104.7
191.2
256.9
2.5
322.9
462.0
431.3
407.0
185.7
227.2
365.4
341.1
98.4
189.8
256.5
2.7
445.2
584.4
553.7
529.4
308.0
349.6
487.8
463.5
220.8
312.2
378.9
44.8
385.2
524.4
493.7
469.4
248.0
289.6
427.8
403.5
160.8
252.2
318.9
22.6
Hara*4
Ref. [31]
325.2
464.4
433.7
409.4
188.0
229.6
367.8
343.5
100.8
192.2
258.9
2.8
Voltage [V]
300
200
100
300
300
200
200
200
Voltage [V]
300
100
100
2
Time [s]
0
-100
-100
0
2
Time [s]
100
-100
Measured result
Voltage [V]
Voltage [V]
(a)
2
Time [s]
2
Time [s]
Thus, it is concluded that the frequency-dependent effect of wave propagation along a tower can be neglected and the
value of the surge impedance is not significant unless a tower footing impedance is represented by a resistive or a capacitive
model.
(5)
L1-6
lightning surge voltage at the tower and thus decreases a lightning overvoltage at a substation. Therefore a simulation
neglecting the current dependence gives a severer overvoltage, i.e. a safer side result from the insulation design viewpoint.
By this reason, again a pure resistance model is recommended in Japan.
(3) Non-uniform and frequency-dependent characteristics
All the grounding electrodes, either horizontal (counterpoise) or vertical, show a non-uniform characteristic [35],
which corresponds a so-called critical length of the electrode [36]. The characteristic can be taken into account in an
EMTP simulation by adopting a model circuit described in Reference [35] together with the frequency-dependent effect.
However, this approach requires a measured result of the grounding electrode to be simulated.
A general solution for a transient response of a grounding electrode is rather easily obtained by a numerical
electromagnetic analysis method such as a finite-difference time-domain (FDTD) method [8], [37] including non-TEM
mode propagation.
E.
R1
Z t 2 Vt 2
0 .7 km
No.2
2. 0km
No.3
5 .8km
No.4
10. 4km
No .5
21 .7 km
No.6
40 .0 km 44.8 km
No.7
No.8
Z t3 Vt 3
L3
l 3 = 20.0m
Z t1 3 []
Z t 4 []
Vt1 4 [m / s]
120
10.19
l 4 = 68.0m
R1[]
L1[H ]
R2 3 []
L2 3 [H ]
R4 []
L4 [H ]
42.80
R3
Z t 4 Vt 4
L4
R4
R f = 10.0
120
300
8.186
16.31
13.10
34.38
1
Voltage [pu]
1
Voltage [pu]
0 km
No.1
l 2 = 20. 0m
R2
L2
0.5
0.5
500m
1000m
2000m
4000m
10
20[sec] 30
Time
500m
1000m
2000m
4000m
40
10
20
30
Time [sec]
40
L1-7
Carson
fin
Carson
10kHz
fin
50Hz
Carson
-2
-4
fin
101
102
103
length x [m]
2.5
f=1MHz
50Hz
Carson
Li [mH/km]
log10Ri [/km]
fin
Carson
fin
Carson
1.5
10kHz
f=1MHz
fin
104
101
103
length x [m]
(a) Resistance
(b) Inductance
Fig. 8 Finite line impedance in comparison with Carsons impedance
fin = finite line, eq. (7)
1 + 1 + (dij / x ) 2
j 0
Sij
{x ln
+ x ln x 2 + dij 2 + x 2 + Sij 2 + dij Sij} []
2
2
dij
1 + 1 + (Sij / x )
(7)
It should be clear in Fig. 8 that Carsons impedance assuming an infinitely long line is far greater than that of a real
finite line, when x / h is not greater enough than 1. The reason for this is readily understood from the following equation
[43].
ZCar = 0 0 A( xi, xj)dxidxj = 0 B( xi )dxi
(8)
Zfin = 0X1 0X 2 A( xi, xj)dxidxj = 0X1 B' ( xi )dxi
(9)
As is clear from the above equations, ZCar involves mutual coupling from the infinitely long conductor j, while
Zfin involves mutual coupling from the conductor j with the finite length X 2 . In fact, ZCar becomes infinite because
of the infinite length, and thus per unit length impedance is necessarily defined. It should be noted that the per unit
length impedance ZCar of eq. (8) has included the mutual coupling from the infinitely long conductor j. On the
contrary, Zfin , if we define the per unit length impedance in eq. (9), includes the mutual coupling only for the finite length
X 2 . Thus,
(10)
ZCarson > Zfinite
On the contrary, the per unit length admittance of an infinitely long line is smaller than that of a finite length line.
From the above discussion, it should now be clear that Carsons and Pollaczeks earth return impedances may not be
applied to a lightning surge analysis, because the separation distance x between adjacent towers is the same order as the
line height. The same is true for a gas-insulated bus, because its length, height and radius are in the same order. This
requires further work which is interesting and significant.
It is noteworthy that the propagation constants of a finite line is nearly the same as that of an infinitely long line, but
the characteristic (surge) impedance is smaller, because of a smaller series impedance and a greater shunt admittance of the
finite line. Furthermore the ratio of the surge impedances of two finite lines is nearly the same as that of two infinitely
long lines. Finally, traveling wave reflection, refraction and deformation on the finite line is not much different from those
on the infinitely long line.
(3) Feeding line from a transmission line to a substation Inclined conductor
A feeding line from the first tower to the substation via the gantry in Fig. 1 is inclined, i.e. the height is gradually
decreased. As a result, its surge impedance is also decreased gradually and thus no significant reflection of traveling waves
occurs along the feeding line until the substation entrance in physical reality. Because the surge impedance (about 70) of
a gas-insulated bus or a bushing is much smaller than that of an overhead line (300 to 500), noticeable reflection appears
at the substation entrance, if the inclined configuration of the overhead feeding line is not considered. It is better to
consider the inclined configuration of a feeding line if an accurate simulation is required. A maximum difference of 7% in
a substation entrance voltage is observed when the inclined configuration is considered [43], [45].
L1-8
Rg= , Rp=
Rg=7 , Rp=
Rg=7 , Rp=500
Rg=7 , Rp=70
8
6
normalized K [pu]
normalized K [pu]
10
4
2
0.8
0.6
Rg= , Rp=
Rg=7 , Rp=
Rg=7 , Rp=500
Rg=7 , Rp=70
0.4
0.2
0
0
0
200
400
600
800
applied voltage E0 [kV]
1000
200
400
600
800
applied voltage E0 [kV]
1000
800
600
Voltage [kV]
a-b
a-c
b-c
phase a
b
c
400
200
0
0
60
120
180
240
300
360
G.
space
discretization/do
main
fnite difference
time-domain
FDTD
TD-FI
boundary
boundary length
3D circuit
TLM
TD-FEM
FEM
frequency
FI
base equationn
Maxwell
diffrential
Maxwell
integral
Maxwell
characteristic
DAlembert
solution
feature
easy
programing
multi media
finte
element
easy program.
MOM
(TWTDA)
MOM
field integral
wide
application
Small CPU
nonlinear in
time domain
s R I(s, t)
s R
0 s s I(s, t)
c 2 3 q (s, t)ds
+c 2
C
4 R
R
R
s
t
where q( s ,t ) =
(12)
I ( s , )
d
s
C is an integration path along the wire axis, Einc denotes the incident electric field that induces current I, R=r -r, r and
t denote the observation location (a point on the wire surface) and time, respectively, r and t denote the source location (a
point on the wire axis) and time, respectively, s and s denote the distance along the wire surface at r and that along the wire
axis at r, s and s ' denote unit vectors tangent to path C in (12) at r and r, 0 is the permeability of vacuum, and c is the
speed of light. Through numerically solving (12), which is based on Maxwells equations, the time-dependent current
distribution along the wire structure (lightning channel), excited by a lumped source, is obtained.
The thin-wire time-domain (TWTD) code [52] (available from the Lawrence Livermore National Laboratory) is based
on the MoM in the time domain. One of the advantages of the use of the time-domain MoM is that it can incorporate
nonlinear effects such as the lightning attachment process [54], although it does not allow lossy ground and wires buried in
lossy ground to be incorporated.
b)
L1-10
I (s )
^
s
^
s
r
r
C(r)
Origin
Fig. 11 Thin-wire segment for MoM-based calculations. Current is confined to the wire axis,
and the tangential electric field on the surface of the wire is set to zero.
s E inc ( r ) =
2
2
j
g ( r , r )ds
C I ( s ) k s s
ss
4k
jk r r
where g( r , r ) = exp
r r
, k = 0 0 , =
(13)
0
0
is the angular frequency, 0 is the permeability of vacuum, and 0 is the permittivity of vacuum. Other quantities in
eq.(13) are the same as those in eq.(12). Current distribution along the lightning channel can be obtained numerically
solving eq.(13).
This method allows lossy ground and wires in lossy ground (for example, grounding of a tall strike object) to be
incorporated into the model. The commercially available numerical electromagnetic codes [56], [57], are based on the MoM
in the frequency domain.
(3) Finite-Difference Time-Domain (FDTD) Method
The FDTD method [58] employs a simple way to discretize Maxwells equations in differential form. In the Cartesian
coordinate system, it requires discretization of the entire space of interest into small cubic or rectangular-parallelepiped cells.
Cells for specifying or computing electric field (electric field cells) and magnetic field cells are placed relative to each other
as shown in Fig. 12. Electric and magnetic fields of the cells are calculated using the discretized Maxwells equations given
below.
1
1 1 (i, j, k + 1 2) t [2 (i, j, k + 1 2) ]
n
n 1
E z i, j, k + =
E z i, j, k +
2
2 1 + (i, j, k + 1 2) t [2 (i, j, k + 1 2) ]
n 1
n
H y 2 (i + 1 2 , j, k + 1 2)y H y 2 (i 1 2 , j, k + 1 2)y
t (i, j, k + 1 2)
1
+
1
1
Hx
n+
1
2
1
1
1
1
1
1
t
n
i, j , k + = H x 2 i, j , k + +
2
2
2
2 (i, j 1 2 , k + 1 2) yz
(14)
(15)
n
n
+ E y (i, j 1 2 , k + 1)y E y (i, j 1 2 , k )y
Equation (14) , which is based on Amperes law, is an equation updating z component of electric field, Ez(i, j, k+1/2),
at point x=ix, y=jy, and z=(k+1/2)z, and at time t=nt. Eq. (15), which is based on Faradays law, is an equation
updating x component of magnetic field, Hx(i, j-1/2, k+1/2), at point x=ix, y=(j-1/2)y, and z=(k+1/2)z, and at time
t=(n+1/2)t. Equations updating x and y components of electric field, and y and z components of magnetic field can be
written in a similar manner. Note that (i, j, k+1/2) and (i, j, k+1/2) are the conductivity and permittivity at point x=ix,
y=jy, and z=(k+1/2)z, respectively, (i, j-1/2, k+1/2) is the permeability at point x=ix, y=(j-1/2)y, and z=(k+1/2)z. By
updating electric and magnetic fields at every point using eq.(14) and (15), transient fields throughout the computational
domain are obtained. Since the material constants of each cell can be specified individually, a complex inhomogeneous
medium can be analyzed easily.
In order to analyze fields in unbounded space, an absorbing boundary condition has to be set on each plane which
limits the space to be analyzed, so as to avoid reflections there. The FDTD method allows one to incorporate wires buried in
lossy ground, such as strike-object grounding electrodes [59], and nonlinear effects.
L1-11
E-field cell
y
Ez (i, j, k+1/2)
Hy (i-1/2, j, k+1/2)
Hx (i, j+1/2, k+1/2)
Hx (i, j-1/2, k+1/2)
Hy (i+1/2, j, k+1/2)
H-field cell
E-field cell
measured result
simulation result
(a)
(b)
L1-12
(b)
(a)
Fig. 14 Simulation of partial-discharge pulse propagation in a gas-insulated switchgear by MoM.
voltage reference wire
measurement point
of tower-top
voltage rise
PG
77 m
1.37
A
21.7 V
600 ns
600 ns
0
1
1.37 A
1.5 A
voltage [V]
current [A]
67.8 V
200
400
time [ns]
600
800
20
40
60
80
69.2 V
0
200
400
time [ns]
600
800
(16)
C 0 = 2 0 / An (r2 / r1 )
where r1: radius of a bare conductor
L1-13
voltage [V]
150
measured
FDTD
100
50
0
100
200
300
time [ns]
400
500
It is observed that the simulation results in Fig.19 by the EMTP and in Fig. 20 by the FDTD agree reasonably well
with the measured result in Fig. 19. A difference observed between the measured and the EMTP simulation results is
estimated due to mutual coupling between the tower, the pipeline and measuring wires. Also, the frequency-dependent
effect of the conductor affects the difference. A difference between the measured and the FDTD simulation results seems
to be caused by a perfect conductor assumption of the FDTD method.
C.
80
Voltage[V]
Voltage[V]
60
EXP
EMTP
120
40
0
40
20
0
-40
-80
EXP
EMTP
-20
50
100
150
200
50
Time[ns]
150
200
60
40
20
40
Voltage[V]
Voltage[V]
100
Time[ns]
20
0
50
100
150
-20
-40
EXP
EMTP
-20
EXP
EMTP
-60
200
50
Time[ns]
100
150
200
Time[ns]
80
FDTD
FDTD
Voltage[V]
Voltage[V]
120
80
40
40
0
-40
-40
0
50
100
150
200
50
Time[ns]
150
200
60
40
FDTD
FDTD
20
40
Voltage[V]
Voltage[V]
100
Time[ns]
20
0
0
-20
-40
-20
-60
0
50
100
Time[ns]
150
200
50
100
150
200
Time[ns]
L1-15
29.0m
r = 20mm
11.6m
4.0m
12.0m
4.0m
r = 0.373m
44.0m
2
1
-1
Upper phase
Middle phase
Lower phase
Voltage [MV]
6
4
2
(1)
0
0
1
Time [s]
(1)
0
0
3
2
1
(2)
2
Time [s]
-1
2
Time [s]
6
4
2
(2)
0
2
Measured [6]
Calculated
4
Current [kA]
80.0m
Measured [6]
Calculated
Voltage [MV]
12.0m
4.0m
Voltage [MV]
11.2m
16.0m
8.0m
1
Time [s]
(c) Waveforms of archorn voltages computed by (1) TWTDA and by (2) EMTP, in the
case of a middle-phase back-flashover. ( 150 kA, 1.0 s ramp current injection )
Fig.21 Archorn voltages during a back-flashover
To analyze such a very-fast transient electromagnetic field around a three-dimensional conductor system,
electromagnetic modeling codes are appropriate. Among many available codes, the Thin-Wire Time-Domain Analysis
(TWTDA) code [52, 68] based on the method of moments [53] is chosen in the present work, for this code allows to
incorporate nonlinear effects into the analysis [6].
In this section, archorn voltages of a simulated 500 kV twin-circuit tower in Fig.21 (a) hit by lightning, in the case of
one-phase back-flashover, are analyzed by a modified TWTDA code that includes a recently proposed flashover model [69,
70]. A similar analysis is also carried out by EMTP [1], and the results are compared with those computed by the modified
TWTDA code.
Fig.21 (b) shows measured waveforms of the voltage of a 3 m gap representing an archorn and the current flowing
through it [70], and those computed by the TWTDA code. Fig.21 (c) are the archorn voltages computed by (1)TWTDA and
(2)EMTP. In the EMTP simulation, the multistory tower model [13] is used, and Motoyamas flashover model is represented
by a general-purpose description language MODELS [71] in EMTP. The archorn voltages computed by EMTP agree well
with those computed by TWTDA before the back-flashover on one phase. On the other hand, after the back-flashover, the
archorn voltages of the other two phases computed by EMTP decay more steeply than those computed by TWTDA, and
they deviate from the results computed by TWTDA during about 1 s after that. The deviation is noticeable particularly in
the case of the middle- or the lower-phase back-flashover although the settling values of both results are in good agreement.
One of the reasons for these discrepancies may be attributed to somewhat high lumped resistors of the multistory
tower model, which are employed to reproduce the peak values of archorn voltages for step current injection into the tower
top. A very steep wave, injected into the top of this tower model, propagates downward without reflection at nodes, but an
upward propagating wave, which may be a reflected wave at the ground or the associated with the middle- or lower-phase
back-flashover, attenuates much at these nodes. The difference of induction or coupling between the actual dynamic
electromagnetic field around a tower struck by lightning and the TEM mode, which is a basis of an EMTP multiconductor
model, must be another reason.
VI. Conclusion
This paper has presented a lightning surge analysis by the EMTP and by numerical electromagnetic analysis methods.
Because the EMTP is based on a circuit theory assuming TEM mode propagation, it can not give an accurate solution
for a high frequency transient which involves non-TEM mode propagation. Also, the EMTP can not deal with a circuit of
which the parameters are not known.
On the contrary, a numerical electromagnetic analysis method can deal with a transient associated with both TEM and
non-TEM mode propagation. Furthermore, it requires not circuit parameter but geometrical and physical parameters of a
given system. However, it other results in numerical instability if the analytical space, the boundary conditions, the cell
size etc are not appropriate. Also, it requires a large amount of computer resources, and existing codes are not general
enough to deal with various type of transients especially in a large network.
L1-16
VII. References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33]
[34]
[35]
[36]
[37]
[38]
[39]
[40]
[41]
[42]
[43]
[44]
[45]
[46]
[47]
[48]
[49]
[50]
[51]
L1-17
403, 2000
[52] M. Van Baricum, and E. K. Miller, TWTD --- A Computer Program for Time-Domain Analysis of Thin-Wire Structures UCRL-51-277, Lawrence
Livermore Laboratory, California, 1972
[53] E. K. Miller, A. J. Poggio, and G. J. Burke, An integro-differential equation technique for the time-domain analysis of thin wire structures J.
Computational Phys., vol. 12, pp. 24-48, 1973
[54] A. S. Podgorski, and J. A. Landt, Three dimensional time domain modelling of lightning IEEE Trans. Power Delivery, vol. PWRD-2, no. 3, Jul.,
pp.931-938, 1987
[55] R. F. Harrington, Field Computation by Moment Methods Macmillan Co., New York, 1968
[56] G. J. Burke, and A. J. Poggio, Numerical Electromagnetic Code (NEC) --- Method of Moments Technical Document 116, Naval Ocean Systems
Center, San Diego, 1980
[57] G. J. Burke, Numerical Electromagnetic Code (NEC-4) --- Method of Moments UCRL-MA-109338, Lawrence Livermore National Laboratory,
California, 1992
[58] K. S. Yee, Numerical solution of initial boundary value problems involving Maxwells equations in isotropic media,(IEEE Trans. Antennas
Propagat., vol. AP-14, no. 3, Mar., pp. 302-307, 1966
[59] T. Noda, A. Tetematsu, and S. Yokoyama, Improvements of an FDTD-based surge simulation code and its application to the lightning overvoltage
calculation of a transmission tower IPST05-138-24c, Montreal, Canada, Jun. 2005
[60] E. D. Sunde Earth Conduction Effects in Transmission Systems Dover Publications, New York, 1968
[61] A. K. Mishra, A. Ametani, N. Nagaoka, and S. Okabe A study on frequency-dependent parameters and Sundes formula of a counterpoise IEEJ
Trans. PE, vol.127, no.1, pp.299-305, 2007
[62] CIGRE WG36.04 Guide on EMC in power plants and substations CIGRE Pub.124, Paris, France, Dec. 1997
[63] S. Agematu et al, High-frequency switching surge in substation and its effects on operation of digital relays in Japan CIGRE 2006, General
Meeting, Paper C4-304, Sept.2006
[64] A.Ametani et al Electromagnetic disturbances of control circuits in power stations and substations experienced in Japan UPEC 2007, 12-31/12-32,
Brightion, UK, Sept.2007
[65] A. K. Mishra, N. Nagaoka and A. Ametani Frequency-dependent distributed-parameter modeling of counterpoise by time-domain fitting IEE Proc.
GTD, vol.153, no.4, July, pp.485-492, 2006
[66] A. Aemtani, T. Chikara, Y. Baba, N. Nagaoka and S. Okabe : A characteristic of a grounding electrode on the earth sargace , IWHV 2008/Kyoto,
Paper HV-08-76, 2008-10
[67] A. Ametani, K. Oshio, N. Nagaoka and Y. Baba: Lightning surge characteristics in a chemical plant, EEUG 2010 / Helsinki, 2010.8
[68] R. Moini, B. Kordi and M. Abedi, Evaluation of LEMP effects on complex wire structures located above a perfectly conducting ground using
electric field integral equation in time domain, IEEE Trans. Electromagnetic Compatibility, vol.40, no.2, May, pp.154-162, 1998
[69] H. Motoyama, K. Shinjo, Y. Matsumoto and N. Itamoto, Observation and analysis of multiphase back flashover on the Okushishiku test
transmission line caused by winter lightning IEEE Trans. Power Delivery, vol.13, no.4, October, pp.1391-1398, 1998
[70] H. Motoyama, Development of a new flashover model for lightning surge analysis IEEE Trans. Power Delivery, vol.11, no.2, April, pp.972-979,
1996
[71] L. Dube and I. Bonfanti, MODELS: A new simulation tool in the EMTP European Trans. Electrical Power Engineering, vol.2, no.1,
January/February, pp.45-50, 1992
A. Ametani, (M71-SM84-F92-LF10) received the Ph.D. degree from the University of Manchester Institute of Science and Technology, Manchester, U.
K., in 1973. Currently, he is a Professor at Doshisha University, Kyoto, Japan.
L1-18