Sie sind auf Seite 1von 10

UDC 621.3.015.3:551.594.2:621.311.

Electrical Engineering in Japan, Vol. 106, No. 4, 1986


Translated from Denki Gakkai Ronbunshi, Vol. 106B, No. 5, May 1986, pp. 403-410

A Two-Conductor Model for Lightning Surge Calculation


NAOTO NAGAOKA and AKIHIRO AMETANI
Doshisha University

1.

2.

Introduction

Model Circuit

Lightning surge analysis plays a very


important role in the insulation design of
substations and transmission lines [ll].
In conventional lightning surge analysis,
only one power line conductor and an equivalent overhead ground wire are taken into account and the mutual inductance between them
has been neglected [ 2
41. In consequence,
calculation accuracy has been poor and this
has a particularly bad effect on the economy
of 1000 kV or higher voltage transmission
line.

Figure l(a) shows a model circuit used


for the analysis in this paper. The first
tower near the substation is assumed to be
hit by a stroke of lightning with impedance
Z z = 400 S2.
The lightning current is also

To solve the difficulty, lightning


surge analysis has been conducted which takes
into account all the conductors [5] and it
gives a considerably different result from
the conventional analysis utilizing a twocircuit model. In particular, conventional
analysis gives very high arcing horn voltages
and entry voltages of substation. This would
be a considerably serious problem from viewpoints of economy of UHV substation. For
this reason, the exact lightning surge analysis is indispensable for rational design of
UHV transmission line and substation. However, it requires a long computing time and
a large memory because, for instance, a
double circuit transmission line has to be
represented by eight conductors. To reduce
computation difficulty, this paper proposes
a new method in which the polyphase transmission line is represented by one power line
conductor and one ground wire, the two conductors being magnetically coupled with each
other. The proposed method makes it possible
to represent a double-circuit line by a twoconductor circuit and it reduces the dimension of matrix used for the analysis from 8
x 8 to 2 x 2 .
It greatly reduces the computing time and memory requirement.

where I = 200 kA, K = 1.0024, a


0
-1
and b = 2.8188 us-'.
ps

assumed to be represented by a 2/70 ps impulse current such that

I= I0 * K O
(exp(-

a t )- exp (- b t ) }

(1)

KO-1. 0024, u=O. 01025p~-~.


bs2.8188 p-'

0.01025

Five towers are used; they are erected


at intervals of 450 m and the distance between the first tower and the dead end tower
of substation is assumed to be 100 m. The
power line conductor and the ground wire are
grounded at the fifth tower through resistances H
350 S2 and R = 360 n respectively.

The conductor arrangement is shown in


Fig. l(b).
Each tower is divided into four
sections, each being represented by a lossless distributed constant circuit and a resistance-inductance parallel-connected circuit as shown in Fig. l(c).
Each section
represents that portion of the steel tower
which is between neighboring two arms. The
dead end tower is also represented by a twosection model as shown in Fig. l(d) [7].
Constants of the model tower are summarized
in Table 1. A one-phase model of UHV substation is shown in Fig. l(e), where Cb and

C are capacitances representing bushing and


8
shunt reactor respectively. The gas insulated bus is represented by a lossless distributed constant circuit with characteristic impedance of 70 S2 and propagation speed
of 270 m/ps. The ground resistivity is assumed to be 100 h.

The proposed method is compared with


the exact method and the effect of line
simulation i s discussed. Numerical calculation is conducted using EMTP [6] and the
numerically calculated results are compared
with the analytical results.
66

1SSN0424-7760/86/0004-0066$7.50/0
Q 19886 Scripta Technica, Inc.

Ill
RL

211
R2

zt.3

211

R5

R1
214

2t2
R6

Rp

No.5

No.2

R4

No.1

( a ) L i g h t n i n g s t r o k e model

Cb'3MpF

(C)

S t e e l tower
model

( d ) Dead end
tower
mode 1

c s * ~ F

(e) Gas i n s u l a t e d bus model

Fig. 1.

Table 1.

A model circuit of lightning surge calculations.

Tower and structure model data

11. a8 o

Rt

17. 59 R

wires
( b ) An e q u i v a l e n t
ground wire model

125 0

210 Q
300 m/cr

Zll
VII
RI

(a) M u l t i p l e ground

Dead end tower

S t e e l tower

300 m/rs
26.01 0

(.I

18.66 pH
100 0
3M) m/rs
22.41 0

(a)

Fig. 2 .

b)

An equivalent ground wire model.

14.93 pH
50m

3.

Two-Phase Circuit Model

Suppose that the first tower is hit at


0 by lightning. If each ground wire
carries a current I then the total ground

(3)

where ( V . ) = conductor potential at the first


3
tower, (I.)= conductor current at the first
3
tower, Zjk = Zo * line characteristic imped-

9'

wire current i s given by 4 1 because two


9

ground wires exist and each wire is extended


in the left and right directions. Accordingly, the remaining current I - 4 1 flows
9
through impedance Zz, which consists of

ance.
Assuming that the back flashover occurs on phase a , Eq. (3) is reduced to

parallel connected steel tower impedance


and lightning stroke impedance ZI. Thus the
tower top potential V is given by
9
= z1-( I - 4 ZO)

vg

(2)

Since the two ground wires behave in the


same way, Eq. ( 4 ) is further reduced to

where

(5)

The voltage and current relation at the first


tower is expressed as

where

67

From the foregoing discussion, it is


clear that the eight-conductor circuit under
discussion can be represented by a two-conductor circuit with characteristic impedances
as given by Eq. (6). The two ground wires
are replaced by an equivalent ground wire as
shown in Fig. 2. Equivalent impedances of
two conductor circuit are calculated from
the characteristic impedances of eight-conductor circuit but it requires a considerably complicated calculation to compute the
latter characteristic impedances. To reduce the computation complexity, it would be
convenient to use the geometric mean distance
(GMD) as shown in Fig. 2(b) or to use the
line constant calculation algorithm stored in
EMTP

Fig. 3.

where h

r. are the height above ground and


j' J

4.

A multiconductor model.

radius of the jth conductor respectively,


Deik, deik are the distances between the jth

and kth conductors and between the jth conductor and the image of the kth conductor
respectively.

Analytical Method for Calculating


the Lightning Surge

Referring to Fig. 3 , Eq. (3) gives

Numerical calculation by digital computer is not so convenient when it is desired to learn only the rough values of
overvoltages due to lightning stroke. Analytical methods are more appropriate for such
purposes. They also provide an effective
means for confirming validity of the numerically calculated results. A method to calculate the peak value of overvoltage analytically follows.

v,(i) =

(211

-2 1 2 )

zg

(t)

(8)

VIl ( t )= VC' ( t )= (21, +2 2 s ) *Z&)


vb (t)= vb' ( t )= ( 2 1 4

Ve(t)= VC' (2) = ( 2 1 s

+
+

224)

226)

(9)

I&) (10)

*Z&)

(11)

421)

(12)

Equations ( 2 ) and (8) give

4.1 Multiconductor model

21 ~ ( Z ) / ( ~ I I

I g ( t )=

The equivalent circuit for the multiconductor circuit just after a stroke of
lightning in Fig. l(a) is shown in Fig. 3.
The equivalent circuit is valid until the
wave reflected at node 2 (entry tower of
substation) arrives at node 1 (the first
tower). All of the towers except the first
tower are neglected in this equivalent circuit because the second and subsequent
towers are more remotely distant from the
first tower than the entry tower of substation in view of the traveling time of lightning surge. Line conductors-lying on the
left hand side of the first tower are represented by a resistance matrix corresponding
to the characteristic impedance 2,u including

which is substituted into Eqs. (8) - (11) to


give the voltages of power line conductors
and ground wires. These voltages constitute
a forward traveling wave E
such that
If

where superscript t represents the transpose


of matrix.
Forward traveling wave E,,

arrives at

LJ

node 2 (substation entry) 0 . 3 3 us (=

0 is
the traveling time over the distance of 100
m between the first tower and the dead end
tower of substation) after it has left node
1 (the first tower). As soon as it arrives
at node 2, it produces forward traveling
wave E
as shown in Fig. 3 . Neglecting the
2f
distortion of traveling wave to simplify analysis, E
is given by
2f

the mutual inductances. The line conductors


between the first tower and substation are
represented by a distributed constant circuit with characteristic impedance 2
The
0'
power line and ground wire inside the substation are represented by resistance RG and
R

2
1
: -k

which are equal to characteristic imped-

( E d t ) )= (Elf ( t-T o ) )

ance of gas insulated bus and surge impedance of dead end tower respectively.

Part of forward wave E

(14)

is reflected at node
2f
2 and the remaining part is transmitted to
the substation. Transmission coefficient a2

Characteristic impedance 2,U is calculated from approximate formulas such that


(see the Appendix).
68

at node 2 is given by [8]

complicated calculation. In the preceding


discussion, it is assumed that no back flashover occurs. Its effect can be taken into
account by utilizing the principle of superposition while overly complicated calculations are required. However, the effect of
back flashover is calculated relatively
easily, assuming that the back flashover
occurs at t = 0 and using a simple circuit
in which the tower top is short circuited
with the power line conductor.

where Zs is the equivalent impedance of substation.


Accordingly, voltage V 2 at node 2 is
given by

4.2

Two-conductor model

Voltage E2b reflected at node 2 is given by


The two-conductor model corresponding
to the circuit in Fig. 3 is shown in Fig. 5.
Impedance matrices Zo and ZS of the twoBackward traveling wave E2 b arrives at node 1
r 0 seconds after and therefore backward trav-

conductor model are given by

eling wave Elb at node 1 is given by

(Eis(1)) = (Ezs ( t - To))

where

(18)

601n( 2 hg/rg), Z o o =60h ( 2 ha/ru)


Zgu= 601n(Dgu/dgu)
and meanings of h and r are shown in Fig.
Zgp=

Equivalent characteristic impedance Zs of


substation is smaller than line characteristic impedance Zo and therefore backward

polarity waves. The voltage at node 1 up to


time 2 T is given by Eqs. (8) to ( 1 2 ) . After
0
time 2 T 0 , traveling wave E l b of negative

I g ( t )= (Zl/(Zgg 2 21))Z(t>

(22)

vg(4=Zgg4(t)

(23)

Vu (1)=Z g u *&(t)

(24)

polarity is superposed reducing the voltage


at node 1. Hence, the voltage at node 1
takes a maximum value at time 2r0. The re-

Transmission coefficient a 2 at node 2 is de-

flected wave of negative polarity is also


generated at the foot of the first tower.
Denoting the traveling time of steel tower
by r I , the wave reflected at the foot arrives at tower top at time t

= 2r0.

rived from Eqs. (15) and (20) as follows:


la21 =-

This

means that the tower top voltage takes a


maximum value at 2 r 0 or 2 r 1 or more precisely, at
t

smaller value of 2r0 and 2 r l

2(b).
Application of Eqs. (8) - (12) to the
two-conductor model gives

traveling waves E2b and E l b become negative

2Rs(Rc+ Z u a ) - 2 Rs*Zga
- 2 R G * Z ~ U2 R ~ ( R s f Z g g ) ( 2 5 )

where

(19)

The preceding equation is valid as far as


the amplitudes of waves reflected at node 2
and tower foot are very small.
I

The maximum voltage impressed to tower


arm i a calculated by taking into account the
waves traveling from the tower top and tower
foot,

Fig. 4 .

The above procedures determine the voltages a t various points of the first tower.
The maximum voltage at node 2 can also be
calculated by taking into account the traveling waves reflected at various points inside the substation but it requires a very

Voltage waveshape a t node 1 .

RI

-Elf

--Mb

Ex-

E2bc

Fig. 5. A two-conductor model.


69

at node 1 is calculated in the same manner


as was done in Sect. 4.1.

'1

5.

(a)

Numerical Examples and Discussions

EMTP is used for numerical calculations. Semlyen model is used when it is


desired to take into account the frequency
dependence effect of line constants and the
distributed constant model is used when the
frequency dependence effect is neglected.

Power line voltage

5.1

Case of no back flashover

(1) Frequency dependence effect of


line constants

Fig. 6.

The calculated results for eight conductor model in Fig. 1 with frequency dependence effect of line constants taken into
account are shown in Fig. 6 and those with
frequency effect neglected are given in Fig.
7. In Fig. 7 are used line constants at
transient fundamental frequency f,, which is

Calculated results by Semlyen model.

determined from the distance between the


first tower and the dead end tower of substation. More precisely, frequency f, is
given by

ft=1/4ro=750kHz
where
(a)

Fig. 7.

Comparing the response curves in Figs.


6 and 7 with each other, it is learned that
the frequency dependence effect of line constants is very slight. Maximum voltages calculated are compared with the analytically
obtained maximum voltages in Table 3, which
also indicates that the frequency dependence
effect of line constants is very slight.
Neglect of frequency dependence effect of
line constants is very effective for reducing the computing time and memory requirement.

Calculated results by the eight-conductor model.

(2) Analytical result obtained from


eight-conductor model

The maximum voltages calculated by analytical means are shown in case 3 of Table 3.
Characteristic impedances used for the analysis are derived from Eq. (7) and other constants are set at

Therefore, forward traveling wave E at


2f
node 2 is given by Eq. (14) and voltage V2
is derived from Eqs. (16), (25) and (26).
is obtained from Eq. (17)
Reflected wave E
2b
using voltage V 2 and forward traveling wave
2f*

denotes the traveling time (equal

to 0.33 ps) over a distance of 100 m. Characteristic impedances at this frequency are
given in Table 2.

Power line voltage

Equations (23) and (25) give the forward


traveling wave at node 1 such that

'to

(27)

Table 3 indicates that the analytically obtained maximum voltages at node 1 agree well
with the numerically obtained ones. This

As forementioned, the maximum voltage

70

Characteristic impedance (2,)

Table 2 .

at f

750 kHz

104.6 147.4 95.7 63.0 99.2 77.0 5 4 2


lOa 6 560.4 99.2 77.0 5 4 2 147.4 95.7 63.0
147.4 99.2 349.7 126.8 77.6 106.4 89.3 63.6
95.7 77.0 126.8 mi 110.3 89.3 Sa 6 72.4
63.0 54.2 77.6 110.3 316.0 63.6 7 2 4 71.1
99.2 147.4 106.4 89.3 63.6 349.6 126.8 77.7
77.0 95.7 89. 3 90.6 7 2 4 126.6 335.1 110.3
63.0 63.6 72.4 71.1 77.7 110.3 316.0
\ 542

Ow1 '560.4
Ow1

b
C

t'

a'

Table 3.

Maximum voltage and times by eight-conductor model without


back-flashover
3

Case No.
Distributed

Line model

&dyeenc o n s t a n t
(")

Tower top
Upper arm
Middle arm
Lower arm
Upper arm horn
Middle arm horn
Lower arm horn
Upper power l i n e conductor
Middle power l i n e conductor
Lower power l i n e conductor
Upper power l i n e conductor
* f i r s t peak middle power
l i n e conductor

Lower power l i n e conductor

(")

13,217
12,260
10,348
6.110
7,180
6,796
5,692
5.081
3.682
2,652

mode 1
(kV) (pa)

0.68

(kV) )s'(
12,947
11,880
9,910
7,603
7,069
6,671
5.495
4,811
9,370
2,287

0.67
0.61
0.55
0.66
0.60
0.55
0.67
0.67
0.67

433 242
241 0.70
332 15.5
95 0.70
261 15.5
48 0.70

Analytic
solution

4 3 3.42
241 0.70
335 15.4
94 0.70
253 15.4
45 0.70

0.67
0.67
0.61
0.55

0.67
0.61
0.51
0.67
0.67
0.67

...............
256 0.70
...............
101 0.70

...............
51 0.70

demonstrates that the analytical solution is


sufficiently accurate if it is only desired
to learn the maximum voltages of the tower
hit by the lightning stroke. Computational
difficulty arising in deriving the analytical solution is caused by the necessity of
transmission coefficient matrix calculation
which requires inverse matrix calculation as
indicated in Eq. (15). This makes it necessary to use computers if the number of conductors is very large. Numerically calculated voltages at node 2 which are produced
by the first traveling wave also agree well
with the analytically derived voltages.

-_7

( 3 ) Calculated results by the two-conductor model

Fig. 8.

71

Calculated results by the two-conductor model.

Characteristic impedance ( 2 , ) at f = 750 kHz

Table 4 .

for the two-conductor model

[
a

Table 5.

332.6

123.4

123.4

347.2

Maximum voltages by the two-conductor model without backflashover

Line model
~~

I:

.A

c*

~~~

Tower top
Upper arm
Middle arm
Lower arm

Upper arm horn


Upper power l i n e conductor
Upper power
Energy of subline cons t a t ion
ductor

Table 6.

Case No.

Analytic
solution

Distributed
cons tan t
(kVydel (pa)

13,037
12,138
10.267
8,065
7,317

0.67

4*826

o.66

593

2.41

0.67
0.61
0.55
0.67

(kV)

(Ir4

12,951
11,866
9,898
7,594
6,882
4,984

0.67

0.67
0.61
0.55
0.67
0.67

....,.... ......

Analytical parameters for the two-conductor


model
(a) Characteristic impedance
gr
J

331.8
127.7

349.3

(b) Transmission coefficient


OW

-0.17
0.36

0'59
-0.10

The voltage waveforms at node 1 obtained by the proposed method are shown in
Fig. 8. Two ground wires and phase-a power
line conductor only are taken into account
in calculating line constants. The two
ground wires are assumed to constitute a
double conductor and its line constants are
calculated by EMTP. Therefore, the characteristic impedances are expressed as a 2 x 2
matrix as shown in Table 4 and they are used
for calculating the maximum voltages as
shown in Table 5.

Fig. 9.

A back-flashover model.

Computing time with HITAC-M200H is 35


seconds for eight-conductor model and 17 seconds for two-conductor model. From the preceding discussion, it is learned that the
proposed method is satisfactorily effective.
Disagreement of substation entry voltage is
caused by the effect of mutual coupling between phase-a power line conductor and other
power line conductors.

Voltage waveforms in Fig. 8 agree well


with those calculated from the eight-conductor
model in Figs. 6 and 7. The maximum voltages
calculated by the two-conductor model as
shown in case 4 of Table 5 also agree well
with those calculated by the eight-conductor
model as shown in cases 1 and 2 of Table 3
except for the voltages at Substation entry.

(4) Analytical results by two-conductor model


72

9J

Fig. 10.

97

$1

(a) Tower line voltage

Calculated results of a back-flashover surge by the eightconductor model.

a
11

$1

$?

:j

( b ) Arcing horn voltage

'J

(a) Phase-a power line voltage

Fig. 11.

( b ) Phase-a arcing horn voltage

Calculated results of a back-flashover surge by the twoconductor model.

Equations (16) and ( 2 5 ) and transmission


coefficients in Table 6 give

The analytical solutions for two-conductor model are shown in case 5 of Table 5 .
characteristic impedances calculated fromEqs.
( 2 0 ) and ( 2 1 ) are summarized in Table 6 together with the transmission coefficient calculated from Eq. ( 2 5 ) . Values of RG and Rs

v
2
0

Ez/o ( t )

*(Ezfa(f))

- 0.59 -0.17
-[-O. 10 0.36

in Eq. ( 2 8 ) are used without change in Eq.


(25)

I*(

V ~ ( ~ - T O )(30)
V#(t-n))

which indicates that about 36% of forward


traveling wave E
at node 2 is transmitted

Comparison between Tables 3 and 5 indicates good agreement between the analytical
results obtained from two- and eight-conductor models. The two-conductor model makes it
possible to calculate the maximum voltages
very easily using only a desk calculator and
therefore it would be very useful for practical purposes.

2fa

to the substation to determine the power line


voltage V2a at node 2 . The remaining 64% is
reflected toward node 2 with negative polarity. About -10% of E
traveling on the
2f9
ground wires is induced in the power line
conductor. Since V = 2.6 V a ,
9

Parameters in Table 6 give much information regarding lightning surges. For instance, Eqs. ( 2 2 ) - ( 2 4 ) and characteristic
impedances in Table 6 give

Va=128kI, Vq=332 kI=2.6Va

(4

( Vza(tJ=(a21

Vza=-O. 1~2.6Va+0.36Va=O.lOVa (31)


The result agrees well with the calculated
results shown in Tables 3 and 5 . If the
mutual coupling between the power line conductors and ground wires is neglected, Z is

(29)

ga

where

set equal to zero in Eq. ( 2 5 ) to obtain


0.55 0
(az]=I 0 0.341

The foregoing result indicates that the


ground wire voltage at node 1 is about 2 . 6
times as high as the power line voltage.

which shows that VPa is 34% as large as Va'


73

< .7. 7
.:
r
...

u ) r : . - ; I J .

I:%

~~~,~a

Is,

(b) Phase-a a r c i n g horn voltage

( a ) Phase-a power l i n e v o l t a g e

Fig. 12.

Calculated results of a back-flashover surge by the conventional


model.

This value of V2a is more than three times


as large as the value given by Eq. (31) and
therefore it is impossible to neglect the
mutual inductance between the power line conductors and the ground wires.
The mutual coupling between the power
line conductors and steel tower is not negligible if the steel tower is as high as UHV
towers. Since no theoretical expressions
are available which deal with this effect,
it is neglected in this paper.
5.2

Fig. 13.

Calculated results of maximum voltages by various methods.

Case of back flashover


given by the conventional method is about
twice as high as that given by the two- or
eight-conductor model.

The back flashover model presented in


[9, 101 are used in the subsequent discussion. It is represented in Fig. 9 and it
simulates the lightning leader, final jump
and arc by respective branches. The model is
regarded as simulating the back flashover
phenomenon much better than the conventional
single inductance model.

To examine the difference using the


two-conductor model by analytical means, the
ground wire is short-circuited with the
power line conductor at node 1 in Fig. 5 as
stated in Sect. 4.1. According to Sect. 4
and Eqs. (30) and ( 3 2 ) , the two-conductor
model and conventional model give the power
line conductor voltages at nodes 1 and 2 such
that

Voltage waveforms at node 1 calculated


from eight-conductor model, two-conductor
model and conventional model (two-conductor
model with mutual coupling neglected) are
given in Figs. 10 - 12. While the voltage
waveform in Fig. 11 agrees well with the
phase-a voltage waveform in Fig. 10, the
conventional model gives considerably different waveforms from those given in Figs. 10
and 11. For instance, a voltage appears before the flashover switch is closed in Figs.
10 and 11 while the power line voltage is
kept equal to zero until the flashover switch
is closed. This is because the mutual coupling between the power line conductor and
ground wire is neglected in Fig. 12.

V4=63.31, Vza=O. 26Vaz16.51


Vot=52.71, Vza'=O. 34Va'=17. 91
The foregoing equation indicates that the
conventional model gives a 15%smaller power
line voltage at node 1 and a 10% higher
power line voltage at node 2. The results
agree well with the calculated results as
shown in Fig. 13. The preceding discussion
explains qualitatively the difference caused
by the conventional method.

Maximum voltages calculated by the


three methods are compared in Fig. 13. It
indicates that the eight-conductor model
(case 2) gives very similar results to the
two-conductor model (case 4) while the conventional model (case 6 ) gives considerably
different arcing horn and power line voltages. In particular, the arcing horn voltage

6. Conclusions
Main conclusions drawn from this study
are as follows:

(1) The proposed two-conductor model


is as accurate as the exact eight-conductor
74

model and it reduces both the computing time


and required memory capacity by half. Therefore, the proposed method is very useful if
it is only desired to analyze the overvoltages on the power line conductor under back
flashover.

where

(2) The analytical method discussed


in this paper is as accurate as the numerical calculation method if no back flashover
occurs. It is also possible to explain
qualitatively the results of numerical calculation even if back flashover occurs. The
analytical method would be useful to learn
the maximum voltages on the steel tower hit
by lightning or to confirm validity of the
numerically calculated results by simple
means. It would also be useful to clarify
the physical aspects of lightning strokes.

Surge impedance

zo is derived from Eq. (Al)

as follows:

REFERENCES

1.

(3) It has been made clear that the


frequency dependence effect of line constants is negligible in lightning surge
analysis. This means that the distributed
constant equivalent circuit with frequency
dependence effect neglected can be used for
lightning surge analysis.

2.
3.

( 4 ) The conventional two-conductor


model is inaccurate because the mutual coupling between the two conductors is neglected.

4.

5.
Acknowledgement
The authors wish to thank members of
Surge Analysis Committee of the Institute of
Electrical Engineers of Japan for their valuable comments on this work.

6.

APPENDIX

8.

7.

Derivation of Eq. (7)

9.

The series impedance and parallel admittance of lossless transmission line are
given by

10.

75

AIEE Committee. A method of estimating


the lightning performance of transmission lines, AIEE Trans. Power Apparatus
Syst., Vol. 69, 1187, 1950.
Inoue. Trans. I.E.E., Japan, Vol. 93-B,
p. 618, 1973.
CRIEPI. Lightning Protection Design of
Generating Plant and Substation, Research Report of CRIEPI, 175034, March
1976.
R.E. Clayton et al. Surge arrestor
protection and very fast surges, 1.E.E.E
Trans. Power Apparatus Syst., Vol. PAS102, 2400, 1983.
J. Ozawa et al. Lightning surge analysis in a multiconductor system for substation insulation design, ibid., Vol.
PAS-104, 2244, 1985.
W. Scott-Meyer. EMTP Rule Book, B.P.A.
April 1982.
Kawamura et al. Paper of Technical
Meeting on High Voltage Engineering,
I.E.E., Japan, HV-84-51, 1984.
L.V. Bewley. Traveling Waves on Transmission Systems, Dover, 1963.
Suzuki. Technical Report of CRIEPI,
181021, Nov. 1981.
Shindo et al. Paper of Technical Meeting on High Voltage Engineering, I.E.E.,
Japan, HV-84-52, 1984.

Das könnte Ihnen auch gefallen