See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/224167781

Flashover model of arcing horn in transient simulation

Conference Paper  in  Electrical Insulation, 1988., Conference Record of the 1988 IEEE International Symposium on · July 2010
DOI: 10.1109/ELINSL.2010.5549756 · Source: IEEE Xplore

CITATIONS READS
2 612

2 authors, including:

Thinh Pham
G&W Electric Co.
15 PUBLICATIONS   45 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Thinh Pham on 12 October 2016.

The user has requested enhancement of the downloaded file.

 Flashover Model of Arcing Horn in Transient
Simulation
Thinh Pham and Steven Boggs
Institute of Materials Science
University of Connecticut
97 North Eagleville Road, Storrs, CT 06269-3136, USA

Abstract—The flashover performance of an arcing horn in a standard 1x70 μs waveform and in an actual twin-circuit
transmission line is important in lightning surge analysis. Many 220kV transmission line struck by various lightning
flashover models have been proposed for transient simulations, waveforms. The simulations were performed using the
but they are only validated over a limited range of waveforms, ATP/EMPT transient program with the integrated simulation
which is very limited relative to what may be encountered in the language MODELS.
field. This paper compares four flashover models of the arcing
horn of a 220kV transmission line insulator under standard II. FLASHOVER MODELS
lightning impulses and as installed in a twin-circuit, 220kV
transmission line. The results provide insight into appropriate A. Volt-time curve
models for simulating flashover behavior of the arcing horn in The voltage-time curve has been determined experimentally
transmission line. for the specific gap of arcing horn using in 220kV transmission
line insulator (Figure 1). This curve is determined using the
Keywords-Flashover, arcing horn, EMTP simulation, lightning
standard lightning impulse waveform applied across the arcing
surge, insulation co-ordination
horn. During the simulation, the voltage across the insulator is
I. INTRODUCTION compared with the volt-time curve. If simulated voltage
exceeds volt-time curve, flashover occurs in the simulation (an
Service interruption of an overhead transmission line from ideal switch is closed through the inductance of the arc. As
lightning is one of the primary concerns of electric utilities. A recommended by IEC [7], the inductance of the arc is 1µH/m.
direct stroke to the top of a a tower or to a phase conductor The initial time of the V-t curve must be synchronized to the
(shielding failure) causes overvoltage on insulator string, and instant that the voltage reaches the position of the insulator in
flashover occurs if the overvoltage exceeds the voltage-time question, which is differs for each tower in a transmission line.
withstand characteristics of the insulator. An arcing horn across When the simulation is performed, this “time lag” must be
the insulator protects the insulator by flashing over before the included in the model.
insulator string.
Flashover behavior of an arcing horn is one of the impor- 4000

tant parameters in lightning insulation coordination calculations

for a transmission line. For a specific probability of distribution 3500
of lightning current (and waveform), flashover behavior of the
arcing horn is combined with other protective methods such as
line arresters and controlled footing resistance to mitigate the
Voltage (kV)

3000
effects of lightning current without a line outage. Many
flashover models for air gaps [1-6] have been proposed for use
2500
in the study lightning-induced flashover. None of the models is
universal in the context of an arcing horn, as each model is
validated over a very limited range of voltage waveforms. In an 2000
actual transmission line, the lighting overvoltage across the
insulator is far from a standard waveform, and even the
lightning current source varied greatly. 1500
0.0 0.5 1.0 1.5 2.0 2.5
The voltage waveform across an insulator depends on time (μs)
various factors such as the geometry of the transmission line Figure 1. Voltage-time characteristic of the arcing horn equipped
(length of span, height of towers, footing impedance), the with 220kV insulator.
distance between the lightning stroke point and the insulator in
question, the flashover behavior of adjacent insulators, the B. Equal area criterion
position of lightning current (stroke to tower top or phase
conductor), etc. In this paper, four common flashover models The method for determining flashover developed by Kind
of the arcing horn were analyzed for two cases, under the and widely recommend by IEC [7] and CIGRE [8] is based on
a voltage-time area above a threshold voltage U0. Breakdown

978-1-4244-6301-5/10/$26.00 @2010 IEEE

is assumed to occur at a given value of area. This can be provides a reasonable correlation with impulse voltage waves
expressed mathematically as: of 1/50µs, 2/50µs, and 3/50µs.
F, where u(t) is the applied voltage across
the insulator and T0 is the time from which u(t)>U0. U0 and F
are unknown parameters evaluated from volt-time curve of the
arcing horn [8]. For our model, U0=1737kV and F=0.2664V-s.
In an ATP/EMTP simulation, a switch element closes though
the inductance of the arc when this criterion is fulfilled. Figure 2. Flashover model of arcing horn proposed by Ueda [6]
C. Leader propagation method In ATP/EMTP simulation, the first computation is per-
Based on the method developed by Motoyama [4], who formed with SW1 and SW2 open. The close times of these
explained physical discharge of breakdown process from switches are determined by measuring the current through the
experiments for long air gaps (from 1 to 3m) and short tail arcing horn. By repeating this procedure from the insulator
lightning impulse (from 1.2 to 1.4µs of rise time, and from 1 to which is the most prone to breakdown to those which are less
3µs time to half value), the breakdown process is divided into so until no flashover occurs, the simulation can be completed.
two steps:
III. SIMULATION RESULTS
Step 1: Leader onset process. When average applied voltage
on the gap exceeds a constant value, the leader onset condition A. Standard wave form aplies on the arcing horn
is fulfilled and defined by the following equations: 12
[kA]
400 50 for positive 10
polarity
8
460 150 for negative
6
polarity
4
where Ts is leader onset time, D is the gap length (m)
Step 2: Leader development process. The average value of 2

leader developing velocity (νLAVE) is defined by the following

0
equation: 0 4
(f ile linear_ramp.pl4; x-v ar t) c:XX0001-
8 12 16 [us] 20
Figure 3. Linear ramp waveform of lightning current.

A lightning strike is modeled as a current source in parallel

where K1 is a constant, 2.5m2/(Vs) for 0≤xLAVE<D/4 and with a lightning path impedance of 400Ω [9]. The current
0.42m2/(Vs) for D/4≤xLAVE< D/2, and E0=750kV/m. xLAVE is source is a linear 1/70µs ramp waveform of variable amplitude
the average value of leader developing length which is defined (Figure 3). The time to flashover (in second) of each model is
as presented in Table 1.

TABLE I. TIME TO FLASHOVER FOR THE FOUR MODELS (s)

I(kA) V-t Kind Motoyama Ueda
Breakdown occurs when the leader length xLAVE>D/2 and ends 4 No flashover No flashover No flashover No flashover
when u(t)<E0(D-2xLAVE). In an ATP/EMTP simulation, a 4.2 No flashover No flashover No flashover No flashover
switch element closes though the inductance of the arc when 4.3 No flashover No flashover 2.9E-6 3E-6
the breakdown condition is met. 4.4 No flashover No flashover 2.7E-6 2.48E-6
4.5 No flashover No flashover 2.5E-6 2.41E-6
D. Non-linear inductance model 4.8 3.2E-6 6.3E-6 2.16E-6 2.27E-6
This model originated from leader development method of 5 2.9E-6 4E-6 2E-6 2.2E-6
Shindo et al. [2], and was developed by Ueda et al. [6] in order 10 9.4E-7 8.7E-7 1.03E-6 1.35E-6
to develop a model which could be validated for both long and 20 5.5E-7 4.4E-7 6.7E-7 9.25E-7
30 3.9E-7 2.95E-7 5.3E-7 7.55E-7
short air gaps. The flashover of arcing horn is modeled by a 40 3.1E-7 2.2E-7 4.3E-7 6.5E-7
non-linear resistance (Rn) and non-linear inductance (Ln) in
series (Figure 2). Rn is used to hold the critical flashover
The only difference among the models is in the time to
voltage irrespective of the current. SW1 closes when the leader
breakdown. The four models can be classified in two groups in
current starts, i.e., when it exceeds 200A. Ln is the non-linear
term of breakdown behavior. The first group consists of the V-t
inductance of the arcing horn. SW2 closes to complete
model and Kind model, while the second group is formed by
flashover across the gap when the current exceeds 1000A. Lf is
the Motoyama model and Ueda model. Below 4.8kA of
inductance of the arc as previously mentioned. This model
lightning current, breakdown cannot occur in the first group of
model but it is still observed in the second one. The latter group
stops to breakdown when lightning current below 4.3kA. When
breakdown occurs in the wave front (t<10-6s) or the current
magnitude is greater than 10kA, the order of the models in
terms of increasing time to breakdown is Kind, V-t, Motoyama
and Ueda. In this case, the difference in time to breakdown is
small so that the model predictions are similar. When
breakdown occurs in the tail of the wave, the time to
breakdown of Motoyama model and Ueda model is nearly the
same.
B. Flashover performance in an actual transmission line

497m 560m 496m 533m 237m 740m Figure 5. Voltage across the arcing horns of tower #4 when I=100kA,
no flashover is assumed.
#1 #2 #3 #4 #5 #6 #7

Figure 4. 220kV transmission line to be simulated

A twin circuit 220kV transmission line that comprises 6

phase conductors and 2 shielding wires was analyzed. In this
typical transmission line, 7 towers with 6 spans are simulated.
The spacing between towers varies from 237m to 740m (Figure
4). The frequency dependent J-Marti model was used to
simulate each span of the transmission line. As the double-
circuit vertical configuration with 2 shielding wires is used, a J-
Marti model of 8 conductors was considered. The tower is
represented by four lossless transmission lines [9]. Footing Figure 6. Voltage across the arcing horns of phase A along the
impedance is a simple resistance measured at low frequency [9] transmission line when I=100kA, no flashover is assumed
(Rf≈10Ω) which can produce higher potential on the top of the
tower than the current-dependent nonlinear resistance recom-
mended by IEEE [1]. In order to avoid the reflection from the
ends of the transmission line, the latter were connected to AC
voltage sources via multiphase matching impedances [9].
The simulation was performed with a lightning current
source of the same form as described in section A, which is
connected to the top of tower #4. The primarily simulation
showed that below 60kA of peak lightning stroke current,
breakdown cannot occur on any phase of the transmission line.
The simulation was therefore performed with six values of
crest magnitude: 60kA, 70kA, 80kA, 100kA, 120kA and
150kA. If one assumes no flashover in the transmission line, Figure 7. Voltage across the arcing horns of tower #4 when I=100kA,
the voltage distribution on struck tower (tower #4) is far from flashover is assumed on phase A.
the standard wave form (linear ramp) due to the reflection from
adjacent towers (Figure 5).
The insulator of phase A of tower 4 always suffers from the
severest stress no matter what the lightning current (Figure 6).
When I=100kA, flashover on phase A of tower 4 occurs for all
4 models, and the time to breakdown is almost the same for all
models. Thus the voltage waveform does not differ appreci-
ably among the models.
Flashover across phase A of tower #4 diverts the lightning
current in this phase, so that the voltage across insulators of
phase B and phase C of tower #4 is decreased and distorted
(Figure 7). The potential difference between phase A of tower Figure 8. Voltage across the arcing horns of phase A along the
#4 and phase A of other towers is inverted by the sudden transmission line when I=100kA, flashover on arcing horn on phase
increase in potential of phase A (figure 8). A of tower #4 is assumed.
 Thus the waveform changed greatly from the previous case For a standard lightning waveform, the volt-time model and
in which no flashover occurs. When the lightning current is equal-area criterion of Kind are comparable, with very small
increased, the number phases which flash over also increases difference in time to breakdown of these models. The
(Table 2). Unlike the case for the standard wave, the flashover Motoyama and Ueda models, which are based on voltage and
behavior for a transmission line also depends on the span current criteria, produced conservative predictions for low
length. values of lightning current. At high lightning currents or for
cases which involve flashover during the wave front, the four
In V-t and Kind models, flashover in phase A of stricken models become similar.
tower eliminates the possibility of breakdown in phase B of
that tower. For those models, increased lightning current causes In an actual transmission line in which a standard lightning
flashover on phase A of adjacent towers. As tower #3 is located current waveform is applied to the top of a tower, the voltage
closest to tower 4, flashover on phase A of tower #3 occurs across the arcing horn of an insulator varies from tower to
when lightning current is very high (>120kA for V-t model, tower and from phase to phase as a result of wave propagation
and out of the range studied in the Kind model). The number of and the flashover behavior of adjacent phases. As a result, the
towers which suffer flashover as predicted by Kind is always voltage waveform across the insulators and arcing horns is no
the less than for other models except at I=80kA (Table II). longer standard, and the question of which arcing horns flash
over is problematic. In our actual twin circuit 220kV transmis-
TABLE II. FLASHOVER POSITION IN THE TRANSMISSION LINE sion line, the Kind model seemed to overestimate the flashover
I(kA) V-t Kind Motoyama Ueda performance of the arcing horn, while the Motoyama and Ueda
60 No flashover No flashover No flashover No flashover models produce more conservative predictions.
70 No flashover No flashover 4A 4A
80 No flashover 4A 4A, 4B 4A ACKNOWLEDGMENT
100 1A, 2A, 4A, 4A 4A, 4B 4A, 4B This article was funded in part by a grant from the Vietnam
7A
120 1A, 2A, 4A, 4A 4A, 4B 4A, 4B
Education Foundation (VEF). The opinions, findings, and
5A, 6A, 7A conclusions stated herein are those of the authors and do not
150 1A, 3A, 4A, 4A 1A, 2A, 3A, 1A, 2A, 3A, necessarily reflect those of VEF. Support from the Toshiba
5A, 6A, 7A 4A, 4B, 5A, 4A, 4B, 5A, Surge Arrester Engineering Group is grateful acknowledged.
6A 6A, 7A
200 1A, 3A, 4A, 2A, 4A, 5A, 1A, 2A, 3A, 1A, 2A, 3A, REFERENCES
5A, 6A, 7A 6A 4A, 4B, 5A, 4A, 4B, 5A,
[1] Fast Front Transient Task Force of the IEEE, “Modeling guidelines for
6A 6A, 7A
fast front transients”, IEEE Trans. Power Delivery, vol. 11, No. 1,
January 1996
Motoyama and Ueda models produced nearly the same [2] T. Shindo, Y. Aoshima, I. Kishizima, T. Harada, “A study of
result. When the lightning current is sufficiently high, flashover predischarge current characteristic of long air gaps”, IEEE Trans. on
takes place at phase A and phase B of stricken tower before Power Apparatus and Systems, vol. PAS-104, No. 11, November 1985.
flashovers occur at adjacent towers. At low current (I=80kA) [3] A. Pigini, G. Rizzi, E. Garbagnati, A. Porrino, G. Baldo, G. Pessavento,
Motoyama model gives a conservative prediction of the « Performance of large air gaps under lightning overvoltages :
Experimental study and analysis of accuracy of predetermination
number of phase which will flashover. With I=80kA, the leader methods », IEEE Trans. on Power Delivery, vol. 4, no. 2, April 1999.
current through the arcing horn of phase B-tower #4 in Ueda [4] H. Motoyama, “Experimental study and analysis of breakdown
model is about 450A, which is well below the 1000A required characteristics of long air gaps with short tail lightning impulse”, IEEE
to complete flashover. However, the voltage criterion of Trans. on Power Delivery, vol. 11, no. 2, April 1996.
Motoyam is fulfilled so that flashover still occurs on phase B of [5] CIGRE WG 33-01, “Guide to procedure for estimating the lightning
tower #4. At high crest value of lightning current (I≥150kA), procedures the lightning performance of transmission lines”, October
Ueda model provided a more conservative result than predicted 1991.
by the Motoyama model. [6] O. Ueda, S. Neo, T. Funabashi, T. Hagiwara and H. Wanatabe,
“Flashover model for arcing horns and transmission line arresters”,
IV. CONCLUSION International Conference on Power Systems Transients, Lisbon, 3-7
September 1995.
Flashover behavior of the insulator in a transmission line [7] IEC Technical Report 60071-4, “Insulation co-ordination. Part 4:
depends on various factors including the number of circuits, the Computational guide to insulation co-ordination and modeling of
distance from stricken point to the insulator under study, etc. electrical networks”, 2004-06.
This study examined flashover behavior of an arcing horn [8] CIGRE WG 33-02, “Guidelines for representation of network elements
which stands alone and as used in an actual twin circuit 220kV when calculating transients”, No. 39, 1990.
transmission line using ATP/EMTP simulations. Four models [9] A. Ametani and T. Kawamura, “A method of a lightning surge analysis
recommended in Japan using EMTP”, IEEE Trans. on Power Delivery,
of flashover were used in order to compare their behavior in vol. 20, no. 2, April 2005.
various circumstances.

View publication stats