Professional Documents
Culture Documents
A Method of A Lightning Surge Analysis Recommended in Japan Using EMTP
A Method of A Lightning Surge Analysis Recommended in Japan Using EMTP
Abstract—EMTP has been widely used in Japan to analyze that should be done based on discussions of the recommended
switching and lightning overvoltages to design power stations models.
and substations from the viewpoint of insulation coordination.
The Japanese standard of high voltage testing, JEC-0102-1994,
and the insulation design and the coordination of an 1100-kV II. RECOMMENDED MODELING METHOD [5]–[10], [15]
line of Tokyo Electric Power Company were mostly based on Fig. 1 illustrates a basic configuration of a model system rec-
EMTP simulations. This paper presents a method of a lightning
surge analysis using the EMTP recommended in the Japanese ommended for a lightning surge simulation.
standard and related guidelines, and suggests further work to be
done based on discussions of applicable limits and problems of the A. Number of Towers
recommended models. Five towers from a substation gantry are to be considered,
Index Terms—EMTP, insulation coordination, lightning surge, because traveling waves reflected from the towers affect a light-
substation. ning surge voltage in the substation. The first reflection, when
lightning strikes the first tower at , from the 6th tower (next
I. INTRODUCTION to the last = 5th tower in Fig. 1) arrives at the substation entrance
after two travel times T between the first and the sixth towers.
A LIGHTNING surge overvoltage is a dominant factor to
determine the insulation level of a substation. It is very
hard to observe the lightning overvoltage experimentally, and
For example,
gives the time
with light velocity
. Thus, a simulation result is accurate
up to 10 , which is enough in general to observe the maximum
thus a numerical simulation has been adopted to investigate
overvoltage and the time to the peak.
it [1]–[10]. The EMTP [11]–[13] has been widely used since
the 1970s in conjunction with increasing usage of digital com- B. AC Sources and Matching Termination
puters. In Japan, the EMTP was introduced in 1976, and has
become a powerful tool to investigate a transient in a power The left side of the last tower is represented by a multiphase
system since 1980. In 1982, a Working Group (WG) of a Light- matching impedance (resistance matrix), which is given as a
ning Surge Analysis1 was established in the IEE of Japan to characteristic impedance matrix including a mutual impedance
develop and propose a sophisticated but commonly recognized of the transmission line at a dominant transient frequency
method of a lightning surge analysis using the EMTP [5]–[10], defined by [16]:
and was dissolved in 2000. Based on the method developed by
the WG, switching, fault and lightning surge analyses had been
carried out on an 1100-kV transmission system by the EMTP
for its optimized and rationalized insulation design and coordi-
nation [14]. A successful completion of the 1100-kV project had
(1)
led to rationalization of the Japanese standard of high voltage
testing, JEC-193-1973 in that time, and a special committee for where , : total line length.
renewing the standard was established in 1991 in cooperation An ac voltage source is connected to the other side of the
with the WG. A new standard, JEC-0102-1994, was completed matching impedance as illustrated in Fig. 1 to take into account
and issued in 1994 [15]. the effect of the ac steady state voltage on a lightning surge.
This paper summarizes the method of a lightning surge anal-
ysis developed by the WG and applied to the 1100-kV system C. Lightning Current and Impedance
design and to the JEC-0102-1994, and suggests further work
A lightning current is represented by a ramp waveform with
the wavefront duration and the wavetail .
Manuscript received March 17, 2003; revised January 31, 2004. Paper no. Occasionally, a concave current of which the waveform is de-
TPWRD-00106-2003.
A. Ametani is with Doshisha University, Kyo-Tanabe, Kyoto, Japan (e-mail: fined in the following equation is adopted in Japan to represent
aametani@mail.doshisha.ac.jp). a real lightning current more accurately [5], [17]:
T. Kawamura is with the Shibaura Institute of Technology, Minato-ku, Tokyo,
Japan.
Digital Object Identifier 10.1109/TPWRD.2004.839183
(2)
1Chairman: T. Kawamura from 1982 to 1994 and A. Ametani from 1994 to
2000. where is the peak current , i.e., .
0885-8977/$20.00 © 2005 IEEE
868 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 20, NO. 2, APRIL 2005
Fig. 1. Simplified illustration of the recommended model system for a lightning surge simulation.
TABLE I
RECOMMENDED VALUES OF LIGHTNING PARAMETERS
TABLE II
MAXIMUM ARCHORN VOLTAGES AND THE TIME OF APPEARANCE
TABLE III
MEASURED AND CALCULATED SURGE IMPEDANCES OF VERTICAL CONDUCTORS
flashover was less probable as explained in Section III-B. Thus, 3) Frequency-Dependent Effect of a Tower: The frequency-
the recommended tower model tends to result in lower flashover dependent effect of a tower is readily taken into account in
probability of the lower phase archorn. An – parallel circuit a transient simulation by combining the frequency-dependent
between two distributed lines in Fig. 2 represents traveling-wave tower impedance [28] with Semlyen’s or Marti’s line model
attenuation and distortion along a tower. The and values [34], [35] in the EMTP [36]. That is:
were determined originally based on a field measurement [ in 1) obtain frequency responses of the propagation constant
(4)], and thus those are correct only for the tower on which the and the characteristic impedance either
measurement was carried out. Sometimes, the R-L circuit gen- from a measured result or from the frequency-dependent
erates unreal high frequency oscillation. This indicated a neces- impedance and the admittance formulas in references;
sity of further investigation of the – circuit. 2) obtain the parameters of Semlyen’s or Marti’s line model
2) Impedance and Admittance Formulas: A number of for a transient simulation from the above result.
tower models have been proposed, but most of them are not 4) Influence of Surge Impedance and Frequency-Depen-
general, i.e., a tower model shows a good agreement with a dent Effect: It should be pointed out that the influence of
specific case explained in the paper where the model is pro- the surge impedance and the frequency-dependent effect of a
posed. The following IEEE/CIGRE formula of the tower surge tower is heavily dependent on the modeling of a tower footing
impedance is well known and is widely adopted in a lightning impedance, which will be discussed in the following section.
surge simulation [2], [4]: When the footing impedance is represented by a resistive model
as recommended in Japan or by a capacitance model, then the
(5) influence of the tower surge impedance and the frequency-de-
pendent effect of a traveling wave along the tower becomes
rather noticeable. On the contrary, those cause only a minor
where is the equivalent radius of effect when the footing impedance is represented either by an
the tower represented by a truncated cone, , and inductive model or by a nonlinear resistance. Fig. 7 shows an
, , tower top, midsection and base radii [m]; example [25], [36]. The measurement was carried out on a
height from midsection to top [m]; 500-kV tower by applying a current in Fig. 7(a-1) to the top of
height from base to midsection [m]. the tower. The tower top voltage predicted by a distributed-line
When the tower is not a cone but a cylinder, then the above model with a constant tower surge impedance and no R-L
equation is rewritten by circuit Fig. 7(c-1), differs from that by the frequency-dependent
tower model Fig. 7(b), which agrees with the measured result,
in the case of the footing impedance being a resistance. On the
(6)
contrary, in the case of an inductive footing model, the tower top
voltage obtained by the distributed-line model shows a rather
where is the radius of a cylinder representing a tower good agreement with the measured result. It should be also
The recommended value of a tower surge impedance for each noted that some 10% variation of the tower surge impedance
voltage class in Table I was determined by field measurements in does not affect the result in the inductive footing impedance
Japan. Although the surge impedance is a representative value, case.
it cannot be applied to every tower (Table III). Thus, it is concluded that the frequency-dependent effect of
Wave deformation on tower structures (L- or T-shaped iron wave propagation along a tower can be neglected and the value
conductor) can be included in a lightning surge analysis, if re- of the surge impedance is not significant unless a tower footing
quired, based on the approach in [33]. impedance is represented by a resistive or a capacitive model.
872 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 20, NO. 2, APRIL 2005
Fig. 7. Influence of a tower on the tower top voltage. (a) Measured result. (b) Frequency-dependent tower model with a resistive footing impedance.
(c) Distributed-line tower model with various footing impedances.
D. Tower Footing Impedance perimental result [21]. A - characteristic model and a piece-
wise linear inductance model show a reasonable accuracy ex-
1) Linear Footing Impedance: It has been known in gen- cept that the maximum voltage of the former is greater and that
eral that the footing impedance tends to be capacitive in the of the latter is lower than that calculated by the nonlinear induc-
case of a high-resistivity earth, and inductive in the low resis- tance model. It might be noteworthy that a current and energy
tivity earth case. A problem of the representation is: the footing consumed by an arrester in a substation are dependent on an ar-
impedance can be resistive, inductive and capacitive depending chorn model. The switch and the flashover switch models result
on the season and the weather when a measurement is made, in much higher energy consumed by an arrester.
i.e., the impedance is temperature- and soil moisture-depen-
dent. Therefore, it is not easy to select a model of the footing
impedance and this is the reason why a resistance model is F. Transmission Line, Feeder, Gas-Insulated Bus
adopted in Japan. 1) Frequency-Dependent Transmission Line Impedance:
2) Current-Dependent Nonlinearity: A number of papers Although a frequency-dependent line model, i.e., Semlyen’s or
have discussed the current-dependence of a tower footing Marti’s model [34], [35], is recommended, the maximum error
impedance, and have proposed various models of the nonlinear of Marti’s model is observed to be about 15% at the wavefront
footing impedance. It has been a common understanding that of an impulse voltage on an 1100-kV untransposed vertical
the current-dependence decreases a lightning surge voltage twin-circuit line in comparison with a field test result [38]. The
at the tower and thus decreases a lightning overvoltage at estimation of possible errors incurred by using these models
a substation. Therefore a simulation neglecting the current in a lightning overvoltage simulation is not straightforward,
dependence gives a severer overvoltage, i.e., a safer side result because it involves a nonlinearity due to an archorn flashover
from the insulation design viewpoint. For this reason, again a dependent on a lightning current, an ac source voltage, a
pure resistance model is recommended in Japan. flashover phase and so on. This is an important subject to be
investigated in future.
E. Archorn Flashover Model 2) Finite Length of a Line and a Gas-Insulated Bus:
Carson’s and Pollaczek’s earth return impedance of an over-
There exist a number of archorn flashover models. To in- head line and an underground cable were derived based on
vestigate the accuracy, phase-wire voltages at the first tower in the assumption of an infinitely long line. A real line is not
Fig. 1 are calculated by various archorn models and are com- infinitely long at all. The separation distance of a UHV/EHV
pared [37]. A switch (time-controlled) model and a flashover transmission line between adjacent towers and the length of a
switch model show not satisfactory agreement especially in the gas-insulated bus are in the same order of their height . If the
wavefront with the nonlinear inductance model of which the ac- condition that is far greater than and is far greater than the
curacy has been confirmed to be high in comparison with an ex- radius , is not satisfied, Carson’s and Pollaczek’s impedances
AMETANI AND KAWAMURA: A METHOD OF A LIGHTNING SURGE ANALYSIS RECOMMENDED IN JAPAN USING EMTP 873
(7)
where
;
;
horizontal separation between conductors and ;
, height of conductors i and j;
complex penetration depth [40].
It should be clear in Fig. 8 that Carson’s impedance assuming
an infinitely long line is far greater than that of a real finite line,
when is not greater enough than 1. The reason for this is
readily understood from the following equation [41]:
(8)
(9)
Fig. 8. Finite line impedance in comparison with Carson’s impedance. fin =
finite line, (7). (a) Resistance. (b) Inductance.
As is clear from the above equations, involves mutual
coupling from the infinitely long conductor “ ”, while in-
volves mutual coupling from the conductor “ ” with the finite 3) Feeding Line From a Transmission Line to a Substation—
length . In fact, becomes infinite because of the infi- Inclined Conductor: A feeding line from the first tower to the
nite length, and thus “per unit length” impedance is necessarily substation via the gantry in Fig. 1 is inclined, i.e., the height is
defined. It should be noted that the per-unit length impedance gradually decreased. As a result, its surge impedance is also de-
of (8) has included the mutual coupling from the infin- creased gradually and thus no significant reflection of traveling
itely long conductor “ ”. On the contrary, , if we define waves occurs along the feeding line until the substation entrance
the per unit length impedance in (9), includes the mutual cou- in physical reality. Because the surge impedance (about 70 )
pling only for the finite length . Thus of a gas-insulated bus or a bushing is much smaller than that of
an overhead line (300 to 500 ), noticeable reflection appears
(10) at the substation entrance, if the inclined configuration of the
overhead feeding line is not considered. It is better to consider
On the contrary, the per-unit length admittance of an infinitely the inclined configuration of a feeding line if an accurate simu-
long line is smaller than that of a finite-length line. lation is required. A maximum difference of 7% in a substation
From the above discussion, it should now be clear that entrance voltage is observed when the inclined configuration is
Carson’s and Pollaczek’s earth return impedances may not be considered [42].
applied to a lightning surge analysis, because the separation
distance between adjacent towers is the same order as the G. Corona Wave Deformation
line height. The same is true for a gas-insulated bus, because The reason for corona wave deformation being not considered
its length, height and radius are in the same order. This requires in a lightning surge simulation in Japanese guideline is that a
further work which is interesting and significant. simulation result neglecting the corona is expected to be higher
It is noteworthy that the propagation constants of a finite line than that considering corona and thus the result is on a safer side
is nearly the same as that of an infinitely long line, but the char- from the insulation viewpoint. The possible errors incurred by
acteristic (surge) impedance is smaller, because of a smaller se- ignoring corona are observed to be less than 10% when light-
ries impedance and a greater shunt admittance of the finite line. ning strikes the first tower in Fig. 1 [8]. Although sophisticated
Furthermore, the ratio of the surge impedances of two finite lines corona models have been proposed [43], [44], the reliability and
is nearly the same as that of two infinitely long lines. Finally, stability in a lightning overvoltage simulation is not confirmed.
traveling-wave reflection, refraction and deformation on the fi- It is noteworthy that the corona wave deformation can result
nite line is not much different from those on the infinitely long in a higher overvoltage at a substation entrance under a specific
line. condition. Fig. 9 shows a field test result of a normalized voltage
874 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 20, NO. 2, APRIL 2005
(11)
where
—normalized by applied voltage ;
maximum phase-wire (PW) voltage at the receiving
end (substation entrance);
normalized voltage with no corona discharge.
The experiment was carried out on a 6.6-kV line with one
phase wire and one ground wire which were terminated by re-
sistances and at the remote end. An impulse voltage up
to 800 kV was applied to the sending end of the ground wire.
The back flashover in Fig. 9 was represented by a short circuit
of the ground and phase wires. For corona wave deformation
decreases a traveling-wave voltage on a line, it is a common un-
derstanding that the line voltage is decreased by the corona wave
deformation and thus the ratio is less than 1. In the case of no
corona discharge, is nearly equal to 1 on a short distance line.
It was observed that a measured result of on a single-phase
line was less than 1.
Fig. 9(a) shows that in the case of no back-flashover be-
comes greater than 1 as the applied voltage is increased, i.e., K
Fig. 9. Measured result of normalized voltage ratio —Effect of corona wave
corona discharge occurs. On the contrary in Fig. 9(b), is less deformation. (a) Without back-flashover. (b) With back-flashover.
than 1. The reason for the phenomena is readily explained as a
result of different attenuation on a phase wire and a ground wire
due to corona discharge, and negative reflection of a heavily at-
tenuated traveling wave on the ground wire. The phenomena
are less noticeable in the positive polarity case. Details are ex-
plained in [45] and [46]. The phenomena have been also real-
ized qualitatively by an EMTP simulation. The increase of a
phase-wire voltage at a substation entrance is expected to be
more pronounced on an EHV/UHV transmission line on which
a corona discharge hardly occurs on a phase wire because of a
multiple bundled conductor, while a heavy corona discharge is
expected on a ground wire.
overhead line between towers and a gas-insulated bus, be- [25] H. Motoyama, Y. Okumura, and A. Ametani, “The effects of a tower
cause the line/bus length is in the same order of the height, footing resistance and a lightning current waveform on a lightning
surge,” in IEE Japan, Research Meeting, Jul. 1987, Paper PE-87-19.
and the bus height is also in the same order of its radius. [26] W. Diesendorf, Insulation Co-Ordination in High Voltage Electric
Thus, an earth-return impedance of a finite-length con- Power Systems. London, U.K.: Butterworths, 1974.
ductor has to be taken into account. [27] A. Ametani et al., “Investigation of flashover phases in a lightning surge
by new archorn and tower models,” in Proc. IEEE PES T&D Conf. 2002,
3) Similarly, an inclined configuration of an overhead line Yokohama, Japan, 2002, pp. 1241–1246.
from the first tower to a substation entrance is to be con- [28] , “A frequency-dependent impedance of vertical conductors on
sidered, if an accurate simulation is required. a multiconductor tower model,” IEE Proc.-GTD, vol. 141, no. 4, pp.
339–345, 1994.
4) Corona wave deformation does not always reduce a light- [29] C. A. Jordan, “Lightning computation for transmission line with over-
ning overvoltage. head ground wires,” G. E. Rev., vol. 37, no. 4, pp. 180–186, 1931.
5) A phase-to-phase overvoltage can exceed a phase-to-earth [30] C. F. Wagner, “A new approach to calculation of lightning performance
of transmission lines,” AIEE Trans., vol. 75, p. 1233, 1956.
overvoltage especially in the case of lightning far from a [31] M. A. Sargent and M. Darveniza, “Tower surge impedance,” IEEE Trans.
substation and multiphase flashovers. Power App. Syst., vol. PAS-88, no. 5, p. 680, 1969.
[32] T. Hara et al., “Empirical formulas of surge impedance for single and
multiple vertical cylinder,” Trans. IEE Japan, vol. 110-B, p. 129, 1990.
ACKNOWLEDGMENT [33] A. Ametani et al., “Wave propagation characteristics of iron conductors
in an intelligent building,” Trans. IEE Japan, vol. 120-B, no. 1, p. 31,
The authors are grateful to all the WG members. 2000.
[34] A. Semlyen and A. Dabuleanu, “Fast and accurate switching transient
calculations on transmission lines with ground return using recursive
REFERENCES convolutions,” IEEE Trans. Power App. Syst., vol. PAS-95, no. 5, p. 561,
[1] “A simplified method for estimating lightning performance of transmis- 1975.
sion lines,” IEEE Trans. Power App. Syst., vol. PAS-104, p. 919, 1985. [35] J. R. Marti, “Accurate modeling of frequency-dependent transmission
[2] “Estimating lightning performance of transmission lines, II—update to lines in electromagnetic transient simulations,” IEEE Trans. Power App.
analytical models,” IEEE Trans. Power Delivery, vol. PWRD-8, p. 1254, Syst., vol. PAS-101, no. 1, p. 147, 1982.
1993. [36] N. Nagaoka, “Development of frequency-dependent tower model,”
[3] IEEE Guide for Improving the Lightning Performance of Transmission Trans. IEE Japan, vol. 111-B, p. 51, 1991.
Lines, IEEE Standard 1243-1997, 1997. [37] N. Nagaoka and A. Ametani, “A lightning surge analysis considering
[4] Guide to Procedures for Estimating Lightning Performance of Transmis- multiphase flashovers,” in IEE Japan, Research Meeting, 1995, Paper
sion Lines, CIGRE SC33-WG01, Tech. Brochure, Oct. 1991. HV-95-50.
[5] “A New Method of a Lightning Surge Analysis in a Power System,” IEE [38] A. Ametani, K. Adachi, and T. Narita, “An investigation of surge prop-
Japan WG Report, Tech. Rep. 244, Mar. 1987. agation characteristics on an 1100 kV transmission line,” IEEJ Trans.
[6] “Various Parameters of a Lightning Surge Analysis in Power Stations PE, vol. 123, no. 4, p. 513, 2003.
and Substations and the Influences Thereof,” IEE Japan WG Report, [39] A. Ametani and A. Ishihara, “Investigation of impedance and line pa-
Tech. Rep. 301, Jun. 1989. rameters of a finite-length multiconductor system,” Trans. IEE Japan,
[7] “A New Method for Estimating Lightning Surge in Substations,” IEE vol. 113-B, no. 8, p. 905, 1993.
Japan WG Report, Tech. Rep. 446, Nov. 1992. [40] A. Deri et al., “The complex ground return plane: a simplified model for
[8] “An Estimating Method of a Lightning Surge for Statistical Insulation homogeneous and multi-layer earth return,” IEEE Trans. Power App.
Design of Substations,” IEE Japan WG Report, Tech. Rep. 566, Oct. Syst., vol. PAS-100, no. 8, p. 3686, 1981.
1995. [41] A. Ametani, “Wave propagation on a nonuniform line and its impedance
[9] “Modeling for an Accurate Estimation of a Lightning Surge,” IEE Japan and admittance,” Sci. Eng. Review, Doshisha Univ., vol. 43, no. 3, p. 136,
WG Report, Tech. Rep. 704, Nov. 1998. 2002.
[10] “Power System Transients and EMTP Analyzes,” IEE Japan WG Report, [42] A. Ametani and A. Ishihara, “Impedance of a nonparallel conductor
Tech. Rep. 872, Mar. 2002. system and its circuit analysis,” in IEE Japan, Research Meeting, 1992,
[11] W. S. Meyer, EMTP Rule Book, 1st ed. Portland, OR: B.P.A., 1973. Paper PE-92-173.
[12] H. W. Dommel, EMTP Theory Book: B.P.A., Aug. 1986. [43] C. de Jusus and M. T. Correia de Barros, “Modeling of corona dynamics
[13] Japanese EMTP Committee, EMTP Test Cases Book, Jun. 1987. for surge propagation studies,” IEEE Trans. Power Delivery, vol.
[14] N. Yamada et al., “UHV AC transmission,” J. IEE Japan, vol. 102, pp. PWRD-9, no. 3, p. 1564, Jul. 1994.
969–1051, 1982. [44] J. F. Guiller, M. Poloujadoff, and M. Rioul, “Damping model of traveling
[15] High-Voltage Testing Method, IEE Japan, JEC-0102-1994, 1994. waves by corona effect along extra high voltage three phase line,” IEEE
[16] A. Ametani, Distributed-Parameter Circuit Theory. Tokyo, Japan: Trans. Power Delivery, vol. PWRD-10, no. 4, p. 1851, Oct. 1995.
Corona Publishing Co., Feb. 1990. [45] A. Ametani et al., “A study of phase-wire voltage variations due to
[17] R. B. Anderson and A. J. Eriksson, “Lightning parameters for engi- corona wave-deformation,” in Proc. IPST ’99, Budapest, Hungary, June
neering application,” Electra, no. 69, p. 65, 1980. 1999, pp. 433–438.
[18] L. V. Bewley, Traveling Waves on Transmission Systems. New York: [46] , “A basic investigation of substation entrance voltage variation due
Dover, 1963. to corona wave deformation,” Trans. IEE Japan, vol. 120-B, no. 3, p.
[19] M. Ishii et al., “Multistory transmission tower model for lightning surge 403, 2000.
analysis,” IEEE Trans. Power Delivery, vol. PWRD-6, no. 3, p. 1372,
Jul. 1991.
[20] T. Shindo and T. Suzuki, “A new calculation method of breakdown
voltage-time characteristics of long air gaps,” IEEE Trans. Power App. A. Ametani, (M’71–SM’84–F’92) received the Ph.D. degree from the Univer-
Syst., vol. PAS-104, p. 1556, 1985. sity of Manchester Institute of Science and Technology, Manchester, U.K., in
[21] N. Nagaoka, “An archorn flashover model by means of a nonlinear in- 1973.
ductance,” Trans. IEE Japan, vol. B-111, no. 5, p. 529, 1991. Currently, he is a Professor at Doshisha University, Kyoto, Japan.
[22] “Modeling of metal oxide surge arresters,” IEEE Trans. Power Delivery,
vol. PWRD-7, no. 1, p. 302, Jan. 1992.
[23] I. Kim et al., “Study of ZnO arrester model for steep front wave,” IEEE
Trans. Power Delivery, vol. PWRD-11, no. 2, p. 834, Apr. 1996. T. Kawamura, (M’59–SM’89–F’91) received the D.Eng. degree from Tokyo
[24] T. Ueda, M. Yoda, and I. Miyachi, “Characteristics of lightning surges University, Tokyo, Japan, in 1959.
observed at 77 kV substations,” Trans. IEE Japan, vol. 116-B, no. 11, p. Currently, he is a Professor with the Shibaura Institute of Technology, Tokyo,
1422, 1996. Japan, and is Professor Emeritus at Tokyo University.