You are on page 1of 11

IEEE Transactions on Dielectrics and Electrical Insulation Vol. 18, No.

1; February 2011 221

Occurrence Probability of Lightning Failure Rates


at Substations in Consideration
of Lightning Stroke Current Waveforms
Shigemitsu Okabe, and Jun Takami
Tokyo Electric Power Company
4-1, Egasaki-cho, Tsurumi-ku, Yokohama, Kanagawa, 230-8510, Japan

ABSTRACT
To estimate lightning stroke overvoltages exactly, the occurrence probability of lightning
stroke current waveforms must be accurately evaluated. This paper firstly formulated the
occurrence probability distribution of lightning stroke current waveforms, taking into
account the correlation between the current amplitude and the front duration. Next,
lightning overvoltages were calculated, with the current amplitude and the front duration
of lightning current as statistical parameters for Gas Insulated Switchgears (GIS’s) and
transformers in UHV substations. While overvoltages caused by back-flashovers at GIS’s
are affected by the front duration of lightning current, overvoltages at transformers are
relatively less dependent on the front duration. Finally the failure rate was evaluated by
considering not only the current amplitude but also the front duration. These values were
smaller than those evaluated from the frequency of occurrence of current amplitude
alone. Further, the proposed front duration of 1.7 ȝs was examined.
Index Terms = Lightning stroke current, current waveform, front duration,
overvoltage, failure rate, substation equipment, lightning impulse withstand voltage.

1 INTRODUCTION stroke current and the capacitance of transformers as parameters,


show that the relationship between each parameter and generated
ACCORDING to the latest evaluation studies [1, 2] of overvoltages are not necessarily uncomplicated. The parameters
assumed lightning stroke current waveforms based on actual have complicated effects on the overvoltages [3]. However,
measurements - aimed at rationalizing the lightning protection evaluation of the failure rates of power stations and substations
design of transmission line and substation equipment - an has until now only been based on the frequency of occurrence of
appropriate front duration for each voltage class is proposed, with the assumed lightning stroke current amplitude; the influence of
the correlation between the current amplitude and front duration of the front duration was not taken into account [4-9].
lightning stroke currents taken into account [2]. In addition, the
possibility of reducing the lightning impulse withstand voltage This paper formulates the probability distribution of the
(hereinafter referred to as LIWV) for UHV substation equipment is occurrence of lightning stroke current waveforms with a
suggested, by changing front duration to be longer, at least on a correlation between the current amplitude and the front duration
trial calculation basis [3]. On the other hand, a decrease in the of measured data taken into account. This is essential for the
reliability (or an increase in the failure rate) of equipment becomes statistical evaluation of lightning stroke overvoltages. Lightning
concern when the LIWV is reduced, and the accurate evaluation of stroke overvoltages are then calculated a vast number of times,
reliability, or of lightning failure rates, is an increasingly important with the current amplitude and the front duration of lightning
subject for the rationalization of insulation. stroke current waveforms as statistical parameters for Gas
Insulated Switchgears (GIS’s) and transformers of UHV
In order to appropriately evaluate failure rates and rationalize substations. The statistical distributions of generated
the lightning protection design of substation equipment, it is overvoltages are obtained and the lightning failure rates are
necessary to clarify the probability distribution of the evaluated, on a trial calculation basis. The authors have
occurrence of lightning stroke overvoltages. To do this, the proposed the front duration 1.7 ȝs of the assumed lightning
probability of the occurrence of lightning stroke current current for the UHV system in Japan [2], whose duration was
waveforms must be accurately estimated. For example, results derived in consideration of a correlation of the front duration
of a detailed analysis of how lightning stroke overvoltages occur and current amplitude as a substitute of 1.0 ȝs of the
at substations, with the front duration of the assumed lightning conventional value. The newly assumed lightning stroke current
with the front durations of 1.7 ȝs is also evaluated from the
Manuscript received on 25 March 2010, in final form 18 June 2010. perspective of reliability.

1070-9878/11/$25.00 © 2011 IEEE


222 S. Okabe, and J. Takami: Occurrence Probability of Lightning Failure Rates at Substations in Consideration of Lightning Stroke

2 OCCURRENCE PROBABILITY OF The cumulative frequency of current waveforms can be


LIGHTNING STROKE CURRENT obtained by integrating the probability density of the
occurrence P(i,tf) with respect to log (i) and log (tf).
WAVEFORMS
With regard to equation (4), the constants obtained from the
The occurrence probability of lightning stroke current observed lightning stroke waveforms to transmission towers
waveforms is derived from observed lightning stroke current (the number of data N: 120 [1]) are summarized in Table 1.
to transmission tower data [1], with the correlation between Figure 1 shows the frequency distribution of the occurrence of
the current amplitude and the front duration taken into account. lightning stroke current waveforms obtained from equation (4)
Here, lightning stroke current waveforms are simulated using
ramp waveforms. Table 1. Constants related to the probability of the occurrence of lightning stroke
current waveforms [2].
First, the probability of the occurrence pi(i) of the lightning
stroke current amplitude “i” is conventionally expressed Statistical
Parameter Unit value
according to the following equation, assuming the logarithmic
m1: mean value of the lightning stroke
normal distribution [10-12] etc. current amplitude kA 29.3

1 ­° (log(i )  log(m1 )) 2 °½ (1) Vi : standard deviation of log iI log value, kA 0.28


pi (i ) exp® 2 ¾
2S V i °̄ 2V i °¿ Log(m2): mean value of the difference
of the front duration from the log value, Ps - 0.0127
where i : the lightning stroke current amplitude [kA], m1: the average regression curve
mean value of lightning stroke current amplitude [kA], and
ıi : the standard deviation of log (i). Vf : standard deviation of “the
difference in log value between the
Then, [2] showed that the front duration (log value) in the front duration of lightning stroke
log value, Ps 0.135
fixed current amplitude section was normally distributed current and the regression curve of
centered on the average regression curve - the relationship the distribution of the front
duration”
between the current amplitude and the front duration.
Accordingly, the probability of the occurrence pfi(tf) of the N: number of data event 120
front duration of lightning stroke current tf [ȝs] with the
lightning stroke current amplitude i [kA] is expressed by the 0.007-0.008
0.006-0.007
following equation. 0.005-0.006
0.004-0.005

1 ­° ( z (tf )  log(m )) 2 ½° 0.003-0.004 0.008


pfi (tf ) exp® 2
¾
(2) 0.002-0.003 0.007
2
2S V f °̄ 2V f °¿ 0.001-0.002 0.006
0-0.001 0.005 Normalized
0.004
where tf : the front duration of lightning stroke current [ȝs], 0.003
probability [%]

z(tf): the difference in log value between the front duration of 180
0.002
lightning stroke current and the regression curve of the 150
120
0.001
90 0
distribution of the front duration, ıf : the standard deviation Current 60 1
0
amplitude [kA] 30 2
of log z(tf), and log (m2) : the mean value of the difference of 0
5
4
3
Front duration [ȝs]
the front duration from the average regression curve (the
difference in log value). z(tf) is obtained from the following (a) Three-dimensional display
equation.
0.0
­ i ½ (3) Front duration of lightning current; y=0.717e
x/230
0.5
z (tf ) log tf  log ®1.31 exp( )¾
¯ 230 ¿ 1.0
1.5
The probability density of the occurrence P(i,tf) of the 200kA,1.7ȝs Front
lightning stroke current waveform with the current amplitude i 2.0
duration [ȝs@
and the front duration tf is derived from equations (1) to (3), 2.5

and the following equation is derived. Average regression curve; y=1.31e


x/230 3.0
3.5
P (i, tf ) pi (i ) ˜ pfi (tf )
4.0

1 ­° (log(i )  log(m1 )) 2 4.5


exp® 2
2SV iV f °̄ 2V i 180 150 120 90 60 30 0
5.0

2
ª ­ i ½ º ½ Current amplitude [kA]
«log t f  log®1.31exp( 230 )¾  log(m2 )» °
(4) (b) Two-dimensional display (relationship between the current amplitude and
¯ ¿ °
¬ 2
¼
¾ the regression curve of the average front duration )
2V f °
° Figure 1. Occurrence probability of lightning stroke current waveforms as a
¿ function of the amplitude and the front duration.
IEEE Transactions on Dielectrics and Electrical Insulation Vol. 18, No. 1; February 2011 223

and Table 1 both in three-dimensional and two-dimensional Setting conditions


displays. Script control
- Setting lightning stroke conditions
- Setting the analysis circuit - Inputting the current
3 CALCULATION METHOD FOR FAILURE - Setting observation nodes waveform
RATE OF SUBSTATION - Instructing the execution of
EMTP analysis EMTP
To predict the failure rate with a high degree of accuracy,
- Extracting the maximum
the probability of the occurrence of overvoltages as well as the overvoltages
- Overvoltage analysis
probability of insulation breakdown of equipment (insulators) - Maximum overvoltage output -Calculating the frequency of
due to overvoltages is to be obtained, and then the failure rate occurrence
calculated by combining these two probabilities. However, -Creating Database of
Evaluation of the
since it is difficult to evaluate the probability of insulation failure rate overvoltages/frequency
breakdown of equipment at present, this study defines the - Calculating the frequency
failure rate as the frequency of occurrence of overvoltages of occurrence of
Probability of lightning strokes overvoltages exceeding
exceeding the LIWV. to transmission lines taken LIWV.
In the preceding examinations, assumptions with less validity into account
had to be made when calculating overvoltages, such as - Absolute evaluation
- Relative evaluation
assuming the current amplitude and either the front duration or
the front steepness to be independent of each other, or the front Figure 2. Calculation procedures for substations’ failure rates based on
duration constant for the lightning stroke waveform - the basic probability of occurrence of lightning stroke current waveforms and EMTP
conditions. In the previous section, the frequency distribution of overvoltage analyses.
lightning stroke current waveforms was obtained based on
measured data. So the lightning stroke overvoltage is now
calculated using the distribution of the lightning stroke current including inputting lightning stroke current waveforms to the
waveform parameters such as current amplitude and front EMTP data and instructing the execution of the EMTP
duration, expressed by equation (4). While the frequency calculation, extracting the maximum overvoltages and
distribution of overvoltages should be calculated by the analysis calculating the frequency of occurrence, creating a database of
of overvoltages using not only lightning stroke current overvoltages and their frequency, and calculating the
waveforms but also the combination with transmission towers frequency of occurrence of overvoltages exceeding the LIWV.
struck by lightning, the ac phases of phase conductors at the
time of lightning strokes and so on, as other types of parameter, 3.2 ANALYSIS CONDITIONS FOR LIGHTNING
an overvoltage analysis in present study was made with only OVERVOLTAGES
lightning stroke current waveforms used as parameters because The overvoltage analysis model for UHV transmission line
present study aims to be a basic study focusing on the effects of and substation equipment is the same as that in [3]. The
lightning stroke current waveforms. analysis conditions are listed in Table 2. The location of
The present study is carried out for UHV transmission and lightning strokes and back-flashovers were determined on the
substation equipment, and as criteria for estimating the failure assumption that lightning strokes hit the first transmission
rate of substations, overvoltages generated at GIS terminals tower nearest the substation and flashover occurred at the
and at transformers are evaluated by comparing them with the arcing horn of the same tower. As shown in Figure 3,
LIWV. Configuration I was used as the circuit configuration of a
substation in the GIS overvoltage analysis, where a circuit
3.1 CALCULATION FLOWCHART FOR LIGHTNING breaker at the service line entrance opens and imposes a high
FAILURE RATE surge voltage on the GIS. Configuration II was used for the
The calculation flow for the lightning failure rate is shown transformer overvoltage analysis. This allows only a small
in Figure 2. First, the conditions are established and an surge current diversion in the substation to impose a high
overvoltage analysis of a substation made using the ATP- surge voltage on transformer terminal. Transformers were
EMTP [13] with the amplitude and the front duration of simulated by capacitance, and a shell-type transformer with
lightning stroke current as parameters. Next, the maximum 5800 pF and a core-type with 16600 pF were used. With
voltages generated are extracted from the overvoltage analysis regard to the flashover model, a non-linear inductance leader
results in order to obtain the overvoltage distribution. The development model was used, where the pre-discharge
frequency of occurrence of overvoltages here corresponds to phenomenon was simulated by non-linear inductance [14, 15].
the probability of the occurrence of lightning stroke current Refer to [3] for details of the analysis model.
waveforms – conditions for the generation of overvoltages. The lightning stroke current waveform was simulated by a
The lightning failure rate is then calculated using the ramp waveform. Equation (4) obtained in the previous section
frequency of occurrence of overvoltages exceeding the was used for the probability of occurrence of the current
assumed LIWV. amplitude and the front duration in Figure 1. For the analysis,
Scripts are created to control the subsequent series the range of the current amplitude was divided into
of numerous operations and finally estimate the failure rate, 100 cases, from 3 kA to the max. 300 kA with 3 kA increment,
224 S. Okabe, and J. Takami: Occurrence Probability of Lightning Failure Rates at Substations in Consideration of Lightning Stroke

Table 2. Analysis conditions. Japan [9]. 5000 cases of patterns for lightning stroke current
waveform analysis were prepared. The analysis used ATP-
Location
Upper phase back-flashover due to a lightning EMTP [13].
stroke to the first transmission tower
Lightning stroke

3.3 EVALUATION CONDITIONS FOR LIGHTNING


phenomena

Lightning stroke Ramp waveform, Current amplitude: 3 - 300 kA,


current Front duration: 0.1 - 5.0 ȝs㧘Stroke duration : 70 ȝs FAILURE RATE
Lightning stroke 3.3.1 PROBABILITY OF LIGHTNING STROKES TO
400 :
impedance
TRANSMISSION LINES
ac phase ac voltage superimposed
The probability of lightning strokes to transmission lines
Flashover
Non-linear inductance model based on the leader (Nt) could be worked out in detail using electrogeometric
method
modeling with the help of computers. However, the following
Transmission line

Transmission
line
8 phase Semlyen model simplified method is used in this study because it has been
4-story transmission tower model (constant adopted by the conventional design guidelines for back-
Transmission
tower
determined by actual measurement), Grounding flashovers [9]. Regarding the relationship between the
resistance: 10 :
transmission tower height “h” and the frequency of lightning
Gantry 2- story model, Grounding resistance: 4 : strokes, the frequency of lightning strokes is proportional to h
Corona effect Not considered. if the tower is assumed to be a stand alone-structure. If,
however, in the case of horizontally long structures such as
All GIS Double bus - 4 bus tie system - 4 circuits - 4 banks transmission lines, this is proportional to, as in studies based
Configurat
Open circuit breaker at the service line entrance on electrogeometric models. This relationship holds good as
Circuit

ion I
Configurat
long as the height is up to about 100 m.
1 circuit㧙1/4 bus㧙1 transformer
ion II The probability of lightning strokes to transmission lines
GIS
Single phase distributed constant circuit, Lossless (Nt) is based on the information above and the actual
line, Surge impedance: 91 : occurrence of 43 lightning strokes per 100 km of transmission
Substation

In closing: same as the GIS line when the height of transmission towers was 25 m, with an
Circuit breaker
In opening: S-type capacitance simulation IKL of 30 to 35 (which more or less corresponds to the
Capacitance simulation
calculation results using the electrogeometric model.)
Transformer - Core-type 16600 pF h IKL [times / 100 km / year]
- Shell-type 5800 pF Nt 43 u u (5)
25 30ᨺ35
Bushing Capacitance simulation: 300 pF
Here, the transmission tower height h is 110 m, the average
Model taking fast transient current characteristics height of UHV transmission towers. As for IKL (Mesh
Surge arrester
into account㧔V10kA= 1550 kV㧕 covering 15 minutes of latitude and longitude: about 25
km×27 km), the lightning stroke density (1 km×1 km) where
UHV designed transmission lines pass through was obtained
Service line
entrance Circuit configuration I : using a lightning positioning and tracking system (LPATS)
Overvoltage at a GIS analyzed with the
circuit breaker at the service line entrance and converted into IKL. In consequence, IKL = 40 was used.
opened

Circuit configuration II :
3.3.2 RANGE OF ASSUMED LIGHTNING STROKE
Overvoltage at a transformer analyzed with LOCATIONS
one transformer in one circuit
Circuit breaker
(Solid line in the figure) With regard to the location of lightning strokes, as referred
at the service line
(CB is 18m away SA) to in Section 3.2, only the top of the first transmission tower is
Open
considered as a lightning stroke location for the overvoltage
analysis since the present study is a basic one focusing mainly
Close (Tr. is 12m away SA)
Transformer
on the effects of lightning stroke waveforms. Severe lightning
stroke overvoltages are generally generated when back-
flashover conditions occur close to the substation, and
therefore the spans including one of the first towers are
assumed as the range of the location of lightning strokes for a
rigorous evaluation, taking the attenuation of surge due to
Figure 3. Circuit configuration of UHV substation subject to lightning surge corona and other factors into account. Also since the surge
overvoltage analysis. current is shunted two ways and back-flashovers will be less
likely in the case of lightning strokes in the middle part of the
and the range of the length of the front duration was divided span, it is appropriate to assume a range within about a quarter
into 50 cases, from 0.1 ȝs to the max. 5.0 ȝs with 0.1 ȝs of the span length on both sides of the transmission tower.
increment. Since the stroke duration has only a small effect on Consequently, the range of the location of assumed lightning
lightning surge overvoltages, it was constant for 70 ȝs. This strokes was set to a total length of 200 m on the both sides of
has been conventionally used in lightning protection design in the first tower for transmission lines designed for UHV.
IEEE Transactions on Dielectrics and Electrical Insulation Vol. 18, No. 1; February 2011 225

3.3.3 FREQUENCY OF OCCURRENCE OF 3500-4000


LIGHTNING STROKE CURRENT IN RANGE OF 3000-3500
ASSUMED LIGHTNING STROKE LOCATIONS 2500-3000
2000-2500
With the frequency of occurrence Pi of lightning stroke 4000
1500-2000
current waveforms IL(ip,tf), with current amplitudes and front 3500
1000-1500
3000
durations mentioned in the previous section as parameters, 2500 500-1000
2000
given by equation (4), the frequency of occurrence of Voltage [kV] 1500 0-500
or below
1000
lightning strokes in the range of assumed lightning stroke 500 -500-0
0
location P(i, tf) can be obtained by the following equation (6). -500
-1000--500
-1000
In the equation, L is the range of assumed lightning stroke 270
240
locations, L = 200 m for UHV transmission lines. 210
180
0
L [times / route / year ] (6)
150
P (i, tf ) Pi u N t u 120
90
2
100000 60 3
Front
Current amplitude 30 duration
3.3.4 ESTIMATION OF THE LIGHTNING FAILURE [kA] 0 5 [ȝs]

RATE
(a) Three-dimensional display
To evaluate the failure rate, the frequency per year of
occurrence of overvoltages generated in the substation 0
equipment (GIS’s and transformers) exceeding the set LIWV is 1
200kA,1.0ȝs : 2622kV
calculated. The frequency of occurrence of individual
overvoltages can be expressed by the frequency of occurrence P 200kA,1.7ȝs : 2137kV 2 Front duration
in equation (6) of lightning stroke current within the range of [ȝs]
3
assumed lightning stroke locations - the condition for generating
overvoltages. The lightning failure rate can be calculated by 4
integrating the frequency of occurrence of overvoltages
exceeding the LIWV assumed in advance, taking the probability 270 240 210 180 150 120 90 60 30 0
of lightning strokes to transmission lines into account.
Current amplitude [kA]
(b) Two-dimensional display
4 OVERVOLTAGE ANALYSIS RESULTS Figure 5. Distribution of overvoltages generated at the terminal of GIS’s
with changes in current amplitude and front duration of lightning stroke
4.1 DISTRIBUTION OF OVERVOLTAGES current waveforms.
GENERATED AT GIS’S
In the Circuit Configuration I, Figures 4 and 5 show
examples of overvoltage waveforms and the distribution of The part where overvoltages generated are negative values
overvoltages in three-dimensional and two-dimensional in Figure 5 is the area where back-flashovers do not occur.
displays, respectively, generated at the GIS terminals and with The boundary area with positive values indicates the boundary
changes in lightning stroke current waveform parameters. As between the presence and absence of back-flashovers. In the
shown in Figure 4, in the case of GIS’s, the maximum value figure, the critical points of back-flashover occurrence are
appears at the spike waveform of wavefront, the peak value almost linear. These critical characteristics can be almost
varying with changes in the front durations of lightning stroke linearly approximated with respect to the current amplitude
current waveforms. and the front duration, suggesting a condition in which the
3000 front steepness is almost constant. However, when the current
tf : 1.0 ȝs amplitude is around 50 kA, back-flashovers do not occur
tf : 2.0 ȝs
tf : 3.0 ȝs because of lack of time to generate back-flashovers resulting
2000 from shorter front duration less than approx. 1 ȝs. While
Voltage [kV]

back-flashovers partly occur in the range 500 kV to 1000 kV,


1000 late in the wave tail of the lightning current, there is little
lightning stroke current flow to phase conductors and the
voltage increase is small.
0
According to the distribution of generated voltages and the
contour lines of overvoltages, in the area where the front
-1000 duration is about 0.5 ȝs or longer, the higher the current
0 1 2 3 4 5
amplitude, and the shorter the front duration, the higher the
Time [ȝs] voltage tends to be. While the area of occurrence of
Figure 4. Examples of overvoltage waveforms generated at GIS’s with overvoltages between 1500 kV and 2000 kV is wide, the area
changes in front duration of lightning stroke current with amplitude of between 1500 kV and 1600 kV was found to be the widest of
200kA. The time 0 is defined as the time of the lightning to a tower and ac all after detailed checking. This was due to the suppression of
phase voltage is superimposed.
226 S. Okabe, and J. Takami: Occurrence Probability of Lightning Failure Rates at Substations in Consideration of Lightning Stroke

generated overvoltages, by surge arresters, to the residual 3000


voltage level of the surge arresters. Conversely, when the
front duration is shorter than about 0.5 ȝs, overvoltages 2000
decrease. This is because back-flashovers occur at the wave

Voltage [kV]
tail, after lightning stroke current peaks, as a result of front 1000
durations shorter than the time required for back-flashovers to
tf : 1.0 ȝs
occur. It is also thought that the voltage increase due to pre- tf : 2.0 ȝs
0 tf : 3.0 ȝs
discharge has an impact on decreasing the surge suppression
effect of surge arresters [3].
-1000
Under the analysis conditions established in the present 0 2 4 6 8 10
study, the highest overvoltage was 3760 kV, with a current Time [ȝs]
amplitude of 300 kA and a front duration of 0.1 ȝs; a
combination of the highest current amplitude and the shortest Figure 6. Examples of overvoltage waveforms generated at shell-type
transformers with changes in front duration of lightning stroke current with
front duration. amplitude of 200kA. The time 0 is defined as the time of the lightning to a
tower and ac phase voltage is superimposed.
4.2 DISTRIBUTION OF OVERVOLTAGES
GENERATED AT TRANSFORMERS
2000-2500
In the circuit configuration II, overvoltages generated at the 1500-2000
primary terminal of shell-type transformers and core-type 1000-1500
transformers were analyzed and calculated. Figures 6 and 7 500-1000
show examples of lightning stroke overvoltage waveforms 0-500
or below
and the distribution of overvoltages, respectively, generated at 2500 -500-0
the primary terminal of shell-type transformers. As shown in 2000 -1000--500
Figure 6, the maximum value is generated during the 1500
wavefront of overvoltages; however, there is no steep spike as 270 1000
240
500 Voltage [kV]
observed in the case of GIS’s and changes in overvoltages 210
180 0
with changes in the front duration of lightning stroke current 150 -500
waveforms seem small. 120 -1000
90 0
Current amplitude 1
The part where generated overvoltages are negative values in 60
2
[kA] 30 3 Front duration
Figure 7 is the area where back-flashover does not occur and the 0 4 [ȝs]
boundary area to positive values indicates the boundary between
the presence and absence of back-flashovers. These critical (a) Three-dimensional display
characteristics of back-flashover generation are almost identical 0
to those of GIS’s. The characteristics of generated overvoltages 200kA,1.0ȝs : 1741kV
are also the same as those at GIS’s in that the voltage area 1
corresponding to the residual voltage of surge arresters around 2 Front duration
1550 kV is the widest. 200kA,1.7ȝs : 1800kV
[ȝs]
3
The way back-flashovers occur is strongly affected by the
front duration in the case of both transformer and GIS. 4
However, the overvoltage distribution after back-flashovers
occur is different to that of GIS’s. While it would be natural 270 240 210 180 150 120 90 60 30 0
to conclude that the contour line of overvoltages after back-
Current amplitude [kA]
flashovers occur shows a positive correlation between the
current amplitude and the front duration (a straight line (b) Two-dimensional display
from the right bottom to the top left in the figure) - similar Figure 7. Distribution of overvoltages generated at primary terminal of shell-
to critical characteristics of flashover generation - the type transformers with changes in current amplitude and front duration of
contour line of overvoltages at transformers is closer to lightning stroke current waveforms.
being in parallel with the front duration axis than in the
case of GIS’s. Overvoltage waveforms are steep during the the overvoltage is regarded as being determined by the size of
wavefront due to the effect of the front duration of the charge from a lightning surge.
lightning stroke current and the maximum voltage Under the analysis conditions set out in this study, the
appearing at the peak during the wavefront of waveforms. highest overvoltage at shell-type transformers was 2219 kV,
However, if a large capacitance, such as a transformer, is with a current amplitude of 300 kA, and a front duration of
connected to the line, the impedance to steep surge voltage 0.3 ȝs. Compared with voltages generated at GIS’s, changes
is reduced and the influence of the front duration of in voltage are smaller on the whole, and the maximum voltage
lightning stroke current decreases. The generated voltage at is about 59 % of the maximum voltage at GIS’s. Overvoltages
transformers increases as if charging the entire substation, so generated at core-type transformers are shown in Figure 8.
IEEE Transactions on Dielectrics and Electrical Insulation Vol. 18, No. 1; February 2011 227

Since the capacitance of transformers is larger, the influence 5 EVALUATION OF LIGHTNING FAILURE
of the front duration becomes smaller than on shell-type RATE OF SUBSTATION
transformers. Accordingly, the maximum overvoltage is
comparatively lower and the overvoltage characteristics with 5.1 FAILURE RATE OF GIS’S
changes in current waveforms are flatter. The highest
Based on the overvoltages at GIS’s with back-flashovers
overvoltage at core-type transformers was 2070 kV, with a
analyzed in the previous section, overvoltages exceeding the
current amplitude of 300 kA, and a front duration of 0.6 μs.
LIWV are defined as insulation failures here. The frequency
While the highest overvoltage at GIS’s appeared with the
of occurrence of overvoltages is obtained by calculating
shortest front duration of lightning stroke current, the highest
equation (6). The failure rates of GIS’s are obtained by
overvoltage at transformers of both types appeared in
calculating the frequency of occurrence of overvoltages
conditions other than the shortest lightning stroke current
exceeding the specified LIWV and integrating them, which
front duration.
are expressed in the per year per route (line). The results of
As described above, the characteristics of overvoltages the calculation of the failure rate of GIS’s are shown in Table
shows that neither the increased current amplitude nor the 3. Figure 9 shows the relationship between the failure rate of
shorter front duration of lightning stroke current necessarily GIS’s and the LIWV. In the UHV field-test equipment, the
lead to a simple increase in overvoltages. So as far as test voltage for GIS’s is set to 2250 kV [16] and the failure
lightning stroke current conditions in studies of lightning rate is very low at 8.41×10-6 [times/route /year]. (Hence the
failure rates are concerned, it will be necessary to carry out frequency of failure (MTBF: Mean Time Between Failure) is
statistical evaluations of front durations of lightning stroke about once in 119000 years per route.) If the LIWV is reduced
current, taking into account their correlation with current to 2100 kV, 1950 kV, and 1800 kV, with the lightning
amplitudes. impulse withstand voltage level specified by IEC [4] as a
guide, the failure rate is increased to 2.5, 6.5 and 18.9 times,
respectively, from the baseline of the failure rate with the LIWV
2000-2500 at 2250 kV; however, the failure rate remains at 1.59×10-4
1500-2000 [times/route/year] even if the LIWV is reduced to 1800 kV.
1000-1500
While this is limited to the failure due to back-flashover, the
frequency of occurrence is about once in 6290 years per route.
500-1000
Evaluating the failure rate conventionally, based only on the
0-500
or below
frequency of occurrence of the assumed lightning stroke current
-500-0
2500 amplitude of 200 kA (the cumulative frequency of occurrence of
-1000--500
2000 0.3 % of the current amplitude exceeding 200 kA), would mean
1500 that failures occur once in about 700 to 1000 years [9]. The
270 1000 failure rate mentioned above is considerably lower than this. In
240 500
210 Voltage [kV] the case of an LIWV set to 2250 kV, the evaluation based on the
180 0
150 -500 frequency of occurrence of generated overvoltages resulting in
120 -1000 MTBF of 119000 [year, route/time] is significantly different
90
Current amplitude
60 0 from the conventional evaluation based only on the frequency of
[kA] 1
30 2 occurrence with the current amplitude resulting in MTBF of 700
0 3
4 Front duration to 1000 [year, route/time]. This is an significant difference from
[μs]
the point of view of reliability to promote electrical insulation
(a) Three-dimensional display rationalization.
A high assumed failure rate with the conventional method is
0
200kA,1.0μs: 1592kV mainly due to not taking into consideration the frequency of
1 occurrence of the front duration of the assumed lightning
stroke current. In the present analysis, the overvoltage of 2622
200kA,1.7μs: 1616kV 2 Front duration kV is generated by the conventionally assumed lightning
[μs]
stroke current waveforms with a current amplitude of 200 kA
3
and a front duration of 1.0 μs. Figure 10 shows the
4
overvoltage distribution overlaid with the density distribution
of occurrence of lightning stroke current. The failure rate
conventionally used was evaluated using only the current
270 240 210 180 150 120 90 60 30 0
amplitude, regardless of the front duration. Consequently,
Current amplitude [kA]
overvoltages generated under the condition of 200 kA or
(b) Two-dimensional display higher are assumed to result in failure in all cases as shown by
the hatched area surrounded by a light red line in Figure 10a.
Figure 8. Distribution of overvoltages generated at primary terminal of core-
type transformers with changes in current amplitude and front duration of
On the other hand, in the case of evaluation using overvoltages,
lightning stroke current waveforms. these overvoltages depend on the influence of the current
228 S. Okabe, and J. Takami: Occurrence Probability of Lightning Failure Rates at Substations in Consideration of Lightning Stroke

Table 3. Calculation of the lightning failure rate of GIS’s. amplitude and the front duration. Therefore, the evaluation of
overvoltages exceeding 2622 kV with a front duration of 1.0 ȝs
Failure rate Relative failure rate using the conventional assumed lightning stroke current
LIWV [times/route/year] compared with LIWV of waveform results in the hatched area surrounded by a light red
(MTBF [year]) 2250kV
line in Figure 10b. The area surrounded by a light red line in
1550 1.48×10-3 (676) 176 Figure 10b is determined by overvoltages with a front duration
1675 -4
5.14×10 (1950) 61.1 of 1.0 ȝs. The light red line here is the 2622 kV contour line.
This area is small and, further, is far from the area where the
1800 1.59×10-4 (6290) 18.9
-5
probability of occurrence of lightning stroke current is high and
1950 5.44×10 (18400) 6.47
therefore the frequency of occurrence is low. The higher the
2100 2.13×10-5 (46900) 2.53 current amplitude, the longer the front duration becomes.
2250 -6
8.41×10 (119000) 1.00 Therefore, in the area of large currents such as the one with a
2400 3.13×10-6 (319000) 0.30 current amplitude of 200 kA, there is a large gap between an
evaluation with one parameter shown in Figure 10a, and an
[u10-3] evaluation with two parameters shown in Figure 10b.
The evaluation of the failure rate with overvoltages is more
1.5
essential, and the front duration has a large impact on
[circuits/route/year]

overvoltages. Therefore, evaluations of failure rates with the


Failure rate

overvoltage level using two parameters, the current amplitude


1.0
and the front duration of lightning stroke current, are more
rigorous than evaluations of them using the conventional
0.5 method, where only the frequency of occurrence of the current
amplitude is used.
Meanwhile, in the present study, the evaluation is made
0 using the maximum value of generated overvoltages.
1600 1800 2000 2200 2400 Waveforms are not evaluated by being converted into standard
Lightning impulse withstand voltage [kV] waveforms [17]. Under the conditions in the present study, if
Lightning impuls withstand voltage [kV]
lightning overvoltages generated at GIS’s are converted into
Figure 9. Relationship between lightning impulse withstand voltage (LIWV) standard lightning impulse waveforms, they will be lower.
for GIS and lightning failure rate.
Thus it is anticipated that the evaluation of waveforms will
0 lead to a lower failure rate.
200kA,1.0ȝs
1 5.2 FAILURE RATE OF TRANSFORMERS
For shell-type transformers, as with GIS’s, the calculation
2 Front duration
of the lightning failure rate is shown in Table 4, and the
[ȝs]
3 relationship between the failure rate and the LIWV is shown
by the solid line in Figure 11. The higher the LIWV, the lower
4 the failure rate, and in the present calculation if the LIWV is
set to 2250 kV, no overvoltages in excess occur, resulting in a
270 240 210 180 150 120 90 60 30 0 failure rate of zero. The failure rate at the test voltage of 1950
Current amplitude [kA] kV [16] for transformers in the UHV field-test equipment was
(a) Distribution of overvoltages generated at 200 kA or higher 7.94×10-5 [times/route/year]. It was 3.99×10-4
[times/route/year] for the lowest LIWV of 1675 kV on a trial
0
calculation basis studied in [3]. Expressed by MTBF, the
200kA,1.0ȝs
1 frequencies of occurrence of lightning failure are about 12600
years and about 2500 years per route, respectively, in the case
2 Front duration of LIWV of 1950 kV and 1675 kV. In comparison with the
200kA,1.7ȝs
[ȝs]
3
case of GIS’s in Figure 9, the failure rate of GIS’s is higher in
the range of 1800 kV or higher, and if a comparison is made
4 based on the test voltages of the UHV field-test equipment for
transformers (LIWV: 1950 kV) and GIS’s (LIWV: 2250 kV),
270 240 210 180 150 120 90 60 30 0 the failure rate of transformers is about 9 times that of GIS’s.
Current amplitude [kA] Qualitatively, as described in Section 4.2, transformers have
(b) Distribution of overvoltages exceeding 2622 kV large capacitance and therefore smaller increases in
Figure 10. Distribution of overvoltages at GIS with respect to the current
overvoltages caused by a shorter front duration compared with
amplitude and the front duration of lightning stroke current, and density GIS’s; the contour line is closer to being in parallel with the
distribution of occurrence of lightning stroke current waveforms. front duration axis, as is clearly evident when Figures 7 and 8
IEEE Transactions on Dielectrics and Electrical Insulation Vol. 18, No. 1; February 2011 229

Table 4. Calculation of the lightning failure rate of shell-type transformers. 6 RELIABILITY WITH ASSUMED
LIGHTNING STROKE CURRENT
Failure rate Relative failure rate
LIWV [times/route/year] compared with LIWV of WAVEFORM
(MTBF [year]) 1950kV The study has clarified that the front duration of lightning
1550 -4
5.64×10 (1770) 7.11 stroke current has a significant impact on overvoltages and
1675 3.99×10-4 (2510) 5.02
that the effect of the front duration is not simple. It is therefore
more essential, and more appropriate, for the reliability to be
1800 2.50×10-4 (4000) 3.15
evaluated by generated overvoltages rather than by the
1950 7.94×10-5 (12600) 1.00 frequency of occurrence of lightning stroke current
2100 8.43×10-6 (119000) 0.11 waveforms This section, therefore, evaluates the reliability
2250 0.00 (’) 0.00 ensured in the context of the assumed lightning stroke current
2400 0.00 (’) 0.00
based on the generated overvoltage level. With regard to the
assumed lightning stroke current waveforms, a comparison is made
between a waveform with 200 kA and 1 / 70 ȝs in the study of
[u10-4]
0
6.0
Shell-type transformers
[circuits/route/year]

Core-type transformers 200kA,1.0ȝs


1
Failure rate

4.0 2
200kA,1.7ȝs Front duration
3 [ȝs]

2.0 4

270 240 210 180 150 120 90 60 30 0


0
1600 1800 2000 2200 2400 Current amplitude [kA]

Lightning impulse withstand voltage


voltage [kV] Figure 12. Distribution of overvoltages at shell-type transformer with respect
Lightning implus withstand [kV] to the current amplitude and the front duration of lightning stroke current and
Figure 11. Relationship between lightning impulse withstand voltage density distribution of occurrence of lightning stroke current waveforms.
(LIWV) for transformer and lightning failure rate.
0
are compared with Figure 5. As shown by a light red line in 200kA,1.0ȝs
Figure 12, the overvoltage distribution for 1741 kV 1
generated in the case of a current amplitude of 200 kA and a
2 Front duration
front duration of 1.0 ȝs has a contour line distribution more 200kA,1.7ȝs
[ȝs]
like that with the single parameter - the current amplitude - 3
shown in Figure 10a than the contour line (light red line) of
overvoltages at GIS’s shown in Figure 10b. This results in a 4
higher failure rate, because the area is wider and includes
the range where the density of the probability of occurrence 270 240 210 180 150 120 90 60 30 0
is high. As for the core-type transformers described next, Current amplitude [kA]
while the counterpart of Figure 12 is Figure 13, since the Figure 13. Distribution of overvoltages at core-type transformer with respect
dependency on the front duration is even smaller, the to the current amplitude and the front duration of lightning stroke current and
contour line is more closely parallel to the axis of the front density distribution of occurrence of lightning stroke current waveforms.
duration of the lightning stroke current supposed by Figure
8, and the evaluation result is close to the that based on the Table 5. Calculation of the lightning failure rate of Core-type transformers.
conventional method using only the distribution of the
Failure rate Relative failure rate
current amplitude. However, in the case of transformers, LIWV [times/route/year] compared with LIWV of
overvoltages are suppressed to the residual voltage level of (MTBF [year]) 1950kV
surge arresters, even when lightning stroke currents are large
1550 3.57×10-4 (2800) 7.30
and the failure rate is almost zero if the LIWV is set to 2250
kV or higher. 1675 1.91×10-4 (5240) 3.91
-4
1800 1.14×10 (8770) 2.32
The calculation of the failure rate of core-type
transformers is shown in Table 5, and the relationship 1950 4.89×10-5 (20400) 1.00
between the failure rate and the LIWV is shown by the 2100 0.00 (’) 0.00
dotted line in Figure 11. Since overvoltages at core-type 2250 0.00 (’) 0.00
transformers are lower than those at shell-type transformers,
2400 0.00 (’) 0.00
the failure rate is about half.
230 S. Okabe, and J. Takami: Occurrence Probability of Lightning Failure Rates at Substations in Consideration of Lightning Stroke

the test voltage for the UHV field-test equipment (hereinafter core-type transformers, while the overvoltage generated by the
referred to as the conventional waveform), and a waveform with proposed waveform is low at 1616 kV, the lightning failure rate
200 kA and 1.7 / 70 ȝs, as proposed in [2] (hereinafter referred was 2.50×10-4 [times/route/year] (MTBF: 4000 [year,
to as the proposed waveform). It should be noted that the study route/time]), which is shown by the area surrounded by a blue
of reliability in this study was done under specific analysis line in Figure 13, accidentally just the same as that of shell-type
conditions. In addition, for the actual test voltages, LIWV, it is transformers, because generated overvoltages were generally
necessary to consider not only the back-flashover analysis lower. The overvoltage generated by the conventional waveform
results but also the impact of direct lightning strokes on phase is 1592kV and the failure rate, the area surrounded by a light red
conductors with shielding failures [18] and overvoltages in line in Figure 13, is estimated as about same as that calculated by
other lightning surge time areas, such as disconnector surges. the proposed waveform.
With regard to GIS’s, according to Figure 10b the overvoltage As for a UHV transmission system, making the front duration
generated by the conventional waveform is 2622 kV and the of the assumed lightning stroke current 1.7 ȝs instead of 1.0 ȝs
frequency of occurrence of overvoltages exceeding this is 2.27×10- is considered appropriate, based on the properties of lightning
6
%. Therefore, when LIWV is set to 2622kV generated from the current [2]. It was found from this study that lightning failures
conventional waveform, calculating the failure rate by integration continue to be basically zero, as seen in changes in the required
using equation (6), 5.45×10-7 [times/route/year] (MTBF: 1.83×106 reliability with a revised front duration.
[year, route/time]) is obtained. This failure rate is shown by the area
surrounded by a light red line in Figure 10b. On the other hand, the
overvoltage generated by the proposed waveform is 2137 kV and 7 CONCLUSIONS
the frequency of occurrence of overvoltages exceeding this is
The frequency of occurrence of lightning stroke current
6.80×10-5 % with a lightning failure rate of 1.64×10-5
waveforms was formulated from the lightning stroke current
[times/route/year] (MTBF: 61000 [year, route/time]), which is
waveform parameter characteristics based on observed
shown by the area surrounded by a blue line in Figure 10b. In
lightning data, and an evaluation of the failure rate of UHV
comparing the contour lines of both cases, in the case of the
substations was attempted in terms of the probability of
proposed waveform as shown in Figure 10b, generated
occurrence of lightning overvoltages exceeding the lightning
overvoltages are lower than with the conventional waveform, and
impulse withstand voltage (LIWV). The results of this study
the probability distribution of occurrence based on the proposed
are summarized below.
waveform extends to the range where the probability of occurrence
is higher and the frequency of occurrence of overvoltages with the 1) Based on the lightning stroke current waveform
proposed waveform is as high 30 times as that with the parameter characteristics obtained from observed lightning
conventional waveform. This means that the reliability from the data, the probability (density) distribution of occurrence of
point of view of lightning protection design decreases. However, lightning stroke current waveforms with the correlation
the absolute value of the frequency of occurrence is extremely between the current amplitude and the front duration taken
small compared to the assumed failure rate of 1.0×10-3 to 1.5×10-3 into account was formulated.
[times/route/year] (MTBF: about 700 to 1000 [year, route/time]) - 2) Using, as parameters, lightning stroke current
an evaluation value with a cumulative frequency of 0.3 % which waveforms with the correlation between the current amplitude
exceeds the conventionally assumed value of 200 kA, based only and the front duration taken into account, overvoltage analysis
on the current amplitude distribution frequency. Therefore, the was conducted in 5000 cases each for GIS’s, shell-type
lightning stroke waveform proposed in [2] maintains a high transformers and core-type transformers, respectively. With
reliability from the engineering point of view, namely, in the regard to GIS’s, overvoltages generated during back-flashover
evaluation the failure rate is basically zero. are suppressed by surge arresters. In general, however, the
Similarly, with regard to shell-type transformers, according to higher the current amplitude and the shorter the front duration,
Figure 7, the overvoltage generated by the conventional the higher the overvoltage generated. Conversely, when the
waveform under the present analysis conditions, is 1741 kV, and front duration was shorter than about 0.5 ȝs, overvoltages
the frequency of occurrence of overvoltages exceeding this is decreased. This is regarded as being caused by delayed back-
1.29×10-3 % and the failure rate is 3.10×10-4 [times/route/year] flashover and the decreased surge suppression by surge
(MTBF: 3230 [year, route/time]). The overvoltage generated by arresters of the voltage increases caused by pre-discharge.
the proposed waveform was 1800 kV and the frequency of With transformers on the other hand, overvoltages during
occurrence of overvoltages exceeding this was 1.04×10-3 % with back-flashover were less dependent on the front duration, and
a lightning failure rate of 2.50×10-4 [times/route/year] (MTBF: generated overvoltages were lower than those at GIS’s, due to
4000 [year, route/time]). These failure rates correspond to the the their large capacitance.
areas surrounded by a light red line and a blue line in Figure 12, 3) As a result of the trial calculation of the lightning failure
respectively. Unlike the case of GIS’s, the required reliability of rate of GIS’s, in the case where a test voltage of 2250 kV for
the proposed waveform is to some extent better than that of the the UHV field-test equipment is set as the LIWV, the failure
conventional waveform and the failure rate is reduced by about rate is 8.41×10-6 [times/route/year] (MTBF: 119000 [year,
20%. In either case, the reliability conventionally assumed with route/time]) under the analysis conditions of the present study.
only the current amplitude (assumed failure rate MTBF: about If the LIWV is reduced to 2100 kV, 1950 kV, and 1800 kV,
700 to 1000 [year, route/time]) is adequately ensured. As for the the failure rate is increased to 2.5, 6.5 and 18.9 times,
IEEE Transactions on Dielectrics and Electrical Insulation Vol. 18, No. 1; February 2011 231

respectively, from the baseline with the LIWV at 2250 kV; REFERENCES
however, the failure rate remains at 1.59×10-4 [1] J. Takami, and S. Okabe, “Observational Results of Lightning Current on
[times/route/year] (MTBF 6290 [year, route/time]) even if the Transmission Towers”, IEEE Trans. Power Delivery, Vol. 22, pp. 547-556,
2007.
LIWV is reduced to 1800 kV.
[2] S. Okabe, and J. Takami, “Evaluation of Improved Lightning Stroke Current
4) As a result of the trial calculation of the lightning failure Waveform Using Advanced Statistical Method”, IEEE Trans. Power Delivery,
rate of transformers under the conditions of the present study, Vol. 24, pp. 2197-2205, 2009.
[3] J. Takami, S. Okabe, and E. Zaima, “Evaluation of Lightning Surge
in the case of the calculation for shell-type transformers, if the Overvoltages at Substations Using Assumed Lightning Stroke Current
LIWV is set to 2250 kV, no overvoltages in excess occur, Waveforms Based on Observations”, IEEE Trans. Power Delivery, Vol. 25, pp.
resulting in a zero failure rate. The failure rate at the test 2958-2969, 2010.
voltage of 1950 kV of the UHV field-test equipment was [4] IEC 60071-1-2006, “Insulation co-ordination, Part 1: Definitions, principles and
rules”.
7.94×10-5 [times/route/year] (MTBF: 12600 [year,
[5] IEC 60071-2-1996, “Insulation co-ordination, Part 2: Applications guide”.
route/time]), and 3.99×10-4 [times/route/year] (MTBF: 2500 [6] IEEE Std 1313-1-1996, “IEEE Standard for Insulation Coordination-
[year, route/time]) even when the LIWV was reduced to 1675 Definitions, Principles, and Rules”.
kV. Since overvoltages generated at core-type transformers [7] IEEE Std 1313-2-1999, “IEEE Guide for Application of Insulation
are lower than those at shell-type transformers, the failure rate Coordination”.
[8] The Japanese Electrotechnical Committee, Standard for Test Voltage, JEC-
is about half that of shell-type transformers. 0102-1994, 1994 (in Japanese).
5) From the point of view of reliability, in the evaluation [9] Sub-committee for Power Stations and Substations, Lightning Protection
Design Committee, “Guide to Lightning Protection Design of Power Stations,
of the assumed lightning stroke current waveforms used for
Substations and Underground Transmission Lines”, CRIEPI Rep., No. T40,
the lightning protection design, the assumed lightning stroke 1995 (in Japanese).
current waveform with 200 kA and 1.7 / 70 ȝs proposed in [10] K. Berger, R. B. Anderson, and H. Kroeninger, “Parameters of Lightning
[2] ensures a lightning failure rate reliability for GIS’s at Flashes”, CIGRE, Electra, No. 41, pp. 23-27, 1975.
[11] R. B. Anderson, and A. J. Eriksson, “Lightning Parameters for Engineering
1.64× 10-5 [times/route/year] (MTBF: 61000 [year,
Application”, CIGRE, Electra, No. 69, pp. 65-102, 1980.
route/time]) or less, and a lightning failure rate reliability for [12] A. J. Eriksson, H. J. Geldenhuys, and G. W. Bourn, “Fifteen Years Data of
transformers - both shell-type and core-type - of 2.50×10-4 Lightning Measurements on a 60 Meter Mast”, Trans. S. African Inst. Elec.
[times/route/year] (MTBF: 4000 [year, route/time]) or less. Engrs., Vol. 80, No. 1, pp. 98-103, 1989.
The assumed lightning failure rate, based only on a current [13] Alternative Transients Program (ATP) Rule Book. OR: Canadian / American
EMTP User Group, 1987.
amplitude of 200 kA, for setting the test voltage of UHV [14] N. Nagaoka, “A Flashover Model Using a Nonlinear Inductance”, Trans. IEE
field-test equipment, is 1.0×10-3 to 1.5×10-3 [times Japan, Vol. 111-B, No. 5, pp. 529-534, 1991 (in Japanese).
/route/year] (MTBF: about 700 to 1000 [year, route/time]). [15] A. Ametani, and T. Kawamura, “A Method of a Lightning Surge Analysis
Comparing the failure rates, the assumed lightning stroke Recommended in Japan Using EMTP”, IEEE Trans. Power Delivery, Vol. 20,
pp. 867-875, 2005.
waveform proposed has adequate required reliability with a [16] T. Watanabe, Y. Yamagata, and E. Zaima, “Insulation Coordination for UHV
basically zero failure rate. System”, CIGRE Paper, No. 33-101, 1998 Session.
The probability distribution of occurrence of lightning [17] S.Okabe, S.Yuasa, M.Koto, and E.Zaima, “Evaluation of lightning surge
waveform for LIWV reduction of substation equipment”, 13th Intern. Symps.
stroke current waveforms formulated in the present study High Voltage Eng. (ISH), P.05.66, 2003.
would be useful not only for the lightning protection design of [18] J. Takami, S. Okabe, and E. Zaima, “Study of Lightning Surge Overvoltages at
substation equipment and for countermeasures against Substations Due to Direct Lightning Strokes to Phase Conductors”, IEEE Trans.
lightning damage to power transmission lines and structures, Power Delivery, Vol. 25, pp. 425-433, 2010.
[19] S. Taniguchi, T. Tsuboi, S. Okabe, Y. Nagaraki, J. Takami, and H. Ota,
but also for analyzing phenomena dependent on the current “Improved Method of Calculating Lightning Stroke Rate to Large-sized
amplitude and the front duration of lightning stroke current. Transmission Lines Based on an Electric geometry model”, IEEE Trans.
The authors would expect it to be widely used. While the Dielectr. Electr. Insul., Vol. 17, pp. 53-62, 2010.
lightning failure rate was evaluated on a trial calculation basis
using statistical methods, overvoltage analysis could be Shigemitsu Okabe (M’98) received the B.Eng., M.Eng. and
conducted in the future using a combination of all possible Dr. degrees in electrical engineering from the University of
parameters, such as transmission towers struck by lightning Tokyo in 1981, 1983 and 1986, respectively. He has been with
and the ac phases on phase conductors, and applying the Tokyo Electric Power Company since 1986, and presently is a
group manager of the High Voltage & Insulation Group at the
method of calculating lightning stroke rate to transmission R & D center. He was a visiting scientist at the Technical
lines [19]. Those should increase estimation accuracy of University of Munich in 1992. He has been a guest professor
overvoltages and reliability. at the Doshisha University since 2005, at the Nagoya
University since 2006, and a visiting lecturer at the Tokyo University. He works as a
While the relationship between GIS’s and transformers and secretary/member at several WG/MT in CIGRE and IEC. He is an Associate Editor of
the LIWV was referred to in the present study, for the actual the IEEE Transactions on Dielectrics and Electrical Insulation.
test voltages, LIWV, it is necessary to consider not only the
back-flashover analysis results but also the impact of direct Jun Takami (M’06) received the B.Eng., M.Eng. and Dr.
degrees in electrical engineering from Doshisha
lightning strokes on phase conductors with shielding failures University, Kyoto in 1995, 1997 and 2010, respectively.
[18] and overvoltages in other lightning surge time areas, such He joined Tokyo Electric Power Company in 1997 and at
as disconnector surges. In addition, it should be noted that the present is a member of the High Voltage & Insulation
study of reliability in this paper was done under specific Group at R & D center in Tokyo Electric Power
Company. His main research interest is the insulation
analysis conditions. design of power systems. He is a member of CIGRE.

You might also like