CALCULATION OF LIGHTNING FLASHOVERS AND BACKFLASH LEVEL

ON 230kV TRANSMISSION LINES

Bander J. Al-Qahtani * M. H. Shwehdi

SAOO-NGPD-TSU Electrical Engineering Department

Saudi Aramco King Fahd University of Petroleum & Minerals
Abqaiq, Saudi Arabia Dhahran, Saudi Arabia
Bander.qahtani@aramco.com mshwehdi@kfupm.edu.sa

transient voltages generated across power apparatus either

by a direct stroke to the power line/apparatus, or by an

ABSTRACT
Lightning has been one of the important problems for
insulation design of power systems and it is still the indirect stroke. However, no two lightning strokes are the
main cause of outages of transmission and distribution same. Therefore, the statistical variations of the lightning-
lines. Lightning caused outages can be reduced by stroke parameters must be taken into account in assessing
lightning protection devices such as ground wires and the severity of lightning strokes on the specific design of a
lightning arresters. power line or apparatus.

This paper presents a comparative studies used to The lightning return-stroke current and the charge
determine the lightning backflashovers level on 230kV delivered by the stroke are the most important parameters
transmission lines utilized by Saudi Electric Company to assess the severity of lightning strokes to power lines
(SEC) in Saudi Arabia, using two well known and apparatus. The return-stroke current is characterized
approaches CIGRE, and the simplified method. The by a rapid rise to the peak, Ip, within a few microseconds
studies include lightning flashovers, backflash and then a relatively slow decay, reaching half of the peak
analysis, as dependent on the tower design parameters value in tens of microseconds. The return-stroke current is
which is considered the main parameters that reduce specified by its peak value and its waveshape. The
the rate of lightning bachflashovers in the waveshape, in turn, is specified by the time from zero to
transmission lines. The study results can be applied to the peak value (tf, front time) and by the time to its
reduce the number lightning flashovers and therefore subsequent decay to its half value (th, tail time). The tail
reduce the transmission lines outages. time being several orders of magnitude longer than the
front time, its statistical variation is of lesser importance
KEY WORDS in the computation of the generated voltage. The
Lightning flashovers, backflashovers, simulation and generated voltage is a function of the peak current for
ground wires both the direct and indirect strokes.

For backflashes in direct strokes and for indirect strokes

1. Introduction the generated voltage is higher the shorter the front time
of the return-stroke current [1]. The front time (and the
A complete awareness of the parameters of lightning tail time, to a lesser extent), influence the withstand
strokes is essential for the prediction of the severity of the capability (volt-time characteristics) of the power

P.O.Box 66467, Dammam 31576, Kingdom of Saudi Arabia

 2

apparatus. The charge in a stroke signifies the energy When lightning hits the ground several hundred meters
transferred to the struck object. The ancillary equipment away from the line (indirect stroke), the electric and
(e.g., surge protectors) connected near the struck point magnetic fields of the lightning channel can induce high
will be damaged if the charge content of the stroke voltage on the line for the insulators of the low-voltage
exceeds the withstand capability of the equipment. The distribution lines to spark over causing a short circuit of
return-stroke velocity will affect the component of the the system. Thus, assuming the lightning channel to be a
voltage which is generated by the induction field of the current source, the transient voltages across the insulator
lightning stroke [1]. Field tests have shown that the of a phase conductor are generated in three ways: (i)
parameters of the first stroke are different from that of the lightning striking the phase conductor (shielding failure),
subsequent strokes. (ii) lightning striking the tower or the shield wire
(backflash), and (iii) lightning striking the nearby ground
(indirect stroke). The severity of these three types of
2. Lightning Flashes transient voltages is influenced by different lightning
parameters [2, 3].
Lightning damages a power apparatus in two ways: (i) it
raises the voltage across an apparatus such that the The significance of lightning parameters on power
terminals across the struck apparatus spark over causing a systems is gauged by the severity of the transient
short circuit of the system or the voltage punctures overvoltages they create and the consequent damages to
through the apparatus electrical insulation, causing the power system. As mentioned before, these
permanent damage. (ii) The energy of the lightning stroke overvoltages are generated by three different ways.
may exceed the energy handling capability of the
apparatus, causing meltdown or fracture.
3. Computation of Insulator Voltage
A lightning flash generally consists of several strokes
which lower charges, negative or positive, from the cloud The lightning return-stroke current is the most significant
to the ground. The first stroke is most often more severe parameter in the estimation of the response of electrical
than the subsequent strokes. Low current continues to apparatus and systems to lightning strikes. The return
flow between two strokes, thus increasing the total energy stroke current rises to its peak in a few microseconds and
injected to the struck object. The transient voltage from then decays to the half value in a few tens of
the lightning strike is generated by: (i) direct stroke and microseconds [4]. The return-stroke current is identified
(ii) indirect stroke. For direct strike, it can strike an by three parameters: peak value Ip, front time tf and time
apparatus. In that case, the apparatus will be permanently to half value th. The difficulty with the exponential
damaged. Most often, lightning strikes the phase function representing a return-stroke current is that it is
conductor of the power line. In that case, a traveling not easy to select the parameters of these analytical
voltage wave is generated on the line; it travels along the expressions to fit the three parameters (Ip, tf and th).
line and is impressed across the terminals of an apparatus However, this problem does not arise if the return-stroke
or most often the insulator between the phase conductor current is represented as linearly rising and linearly falling
and the cross-arm of the tower at the end of the span. If functions [4]:
the voltage is high enough, the insulator flashes over
causing a short circuit of the system. I (t ) = α1tu (t ) −α2 (t −t f )u (t − t f ) (1)

Many overhead power lines are equipped with shield Where α1 = Ip/tf, and α2 = (2th–If)Ip/2tf (th–If). For short tf in
wires to shield the phase conductors. Even then, shielding the order of a few microseconds, eqn. 1 seems to work
failures occur when lightning bypasses the shield wires very well. With eqn. 1, the three parameters of the return-
and strikes a phase conductor. When lightning strikes a stroke current can be varied very easily. Starting with the
tower, a traveling voltage is generated which travels back return-stroke current, the various voltage components
and forth along the tower, being reflected at the tower across the insulator were computed.
footing and at the tower top, thus raising the voltages at
the cross-arms and stressing the insulators. The insulator
will flash over if this transient voltage exceeds its 3.1 First and Second Voltage Components
withstand level (backflash). Even if lightning strikes a
shield wire, the generated traveling voltage wave will To compute the first voltage component, i.e. the cross-
travel to the nearest tower, produce multiple reflections arm voltage Vca, the tower was assumed to be a vertical
along the tower, causing backflash across an insulator. transmission line of a fixed surge impedance Zt. The
voltage and current waves were assumed to travel along
 3

the tower with a constant p.u. velocity of βt. The first r − r′

reflections from the adjacent towers for the shield-wire z′ I ( r ′, t − )
µ0 c (5)
4π ∫
voltages were also included in the computation. The A(r , t ) = dr ′
tower footing resistance was assumed to be constant, Rtf. 0
r − r′
The tower-top was terminated by shield wire(s) and the where r and r ‫ ׳‬are field and source points, respectively: I
lightning channel of constant surge impedance Zch. The is the return-stroke current: q0 is the constant linear charge
cross-arm voltage due to the multiply reflected voltage density of the leader stroke: hc is the cloud height: and is
waves along the struck tower was computed by following the instantaneous height of the upward-moving head of
a previous method as shown in [4]. Although Zt, βt, Rtf the return stroke above ground. For a stroke to ground, z ‫׳‬
and Zch were assumed as constant, they were used as input increases as a function of time and the return-stroke
variables which could be changed for parametric analysis. velocity, with its lower and upper limits 0 and hc. For a
stroke to tower of height hc, the lower and the upper limits
The second voltage component is the voltage induced on of z ‫ ׳‬are ht and hc. Thus, for a stroke to tower, the
the phase conductor due to electromagnetic coupling with voltages induced on the phase conductor were computed
the shield wire. This voltage is equal to kcf Vt, where Vt is for two different cloud heights (hc and ht), and then the
the tower-top/shield-wire voltage and the coupling factor second induced voltage (for ht) was subtracted from the
kcf is equal to Zps/Zsh. Zps is the mutual surge impedance first induced voltage (for hc).
between the phase conductor and the shield wire; Zsh is
the shield-wire surge impedance [4]. The tower-top
voltage was computed following the same procedure as 4. Computation of Backflash Rate
for Vca. The insulator-string voltage due to the first and
second voltage components is: The overhead ground wires or shield wires have been
located so as to minimize the number of lightning strokes
Vins =Vca −kcf Vt (2) that terminate on the phase conductor. The remaining and
vast majority of strokes and flashes now terminate on the
overhead ground wires. A stroke that so terminate forces
3.2 Third Voltage Component current to flow down the tower and out on the ground
wires. Thus voltage are built up across the line insulation.
The third voltage component is the voltage induced on the If these voltages equal or exceed the line CFO, flashover
phase conductor due to the electromagnetic fields of the occurs. This event is called a backflash. By referring to
lightning channel. The computation of the phase- figure 1, equations for the crest voltage, the voltage at the
conductor voltage followed previous analysis [4], with the tower top prior to any reflections from the footing
difference that, in the present case, the stroke hits the resistance, and the final voltage can be derived as follows
tower top instead of the ground. This difference is
manifested in the inducing voltage Vi, which is the
voltage in space (in the absence of the phase conductor)
caused by the residual charge in the upper part and the
return-stroke current in the lower part of the lightning
channel. Vi is:
hp
∂A
Vi = ∫ (∇φ + ) dz (3)
0
∂t
Where Φ is the scalar potential due to the residual charge
in the upper part of the lightning channel, and A is the
vector potential due to the return-stroke current in the
lower part of the channel. For stroke to ground, Φ and A
are

r − r′ Figure 1. Surge voltages at the tower and across the insulation [5]
hc q0 ( r ′, t − )
VTT = ( K sp K TT ) I
1 c
φ( r , t ) =
4π ε0 ∫ r − r′
dr ′
z′
VTA = ( K sp K TA ) I (4.1)
(4)
VF = Re I
 4

Also, note that the crest voltage VIF across the insulation
And the current through the footing resistance is caused by the footing resistance is
Re VIF = (1 − C ) Re I (4.8)
IR = I (4.2)
Ri For a flashover to occur, the voltage across the insulator
V1, must be equal to or greater than the CFO of the
Where
insulation. Replacing V1 of Eq. (7) with CFO, the current
TT obtained is the critical current IC at and above which
K TT = Re + αT Z T
tf flashover occur, i.e.,
TA CFO
K TA = Re + αT Z T IC = (4.9)
tf ( KTA − CK TT ) K SP
Since KTT is in many cases approximately equal to K TA,
 T     
then approximately, 
K SP = 1 − α R (1 − αT ) 1 − 2 S  + α RαT 1 − 4 TS  + (α RαT ) 2 1 − 6 TS  + ..... 
    CFO
 tf   tf  I C =  tf  
(4.3)
(1 − C ) KTT K SP  (4.10)

For these equations: The probability of a flashover is the probability that the
stroke current I equals or exceeds the critical current IC, or
∞
Z g Ri Z − Ri Z g − 2 Ri
Re = αT = T ≈ Pr ob ( I ≥ I C ) = P( I C ) = ∫ f ( I )dI (4.11)
Z g + 2 Ri ZT + Ri Z g + 2 Ri IC

Zg The backflash rate BFR is this probability times the

αR = (4.4) number of strokes, NL, that terminate on the ground wires,
Z g + 2 Ri or
Also, the tail of the voltages can be conservatively BFR= N L P ( I C ) (4.12)
approximated by a time constant τ:
Where
τ=
Zg
Ri
TS (4.5)
NL = Ng
( 28h 0.6
+ Sg )
(4.13)
That is, the equation for the tail of the surge is 10
−(t −t f ) / τ Where h is the tower height (meters), Sg is the horizontal
eTT = VF e (4.6) distance between the ground wires (meters), and Ng is the
To be complete the definition of the variables are: ground flash density (flashes/km2-year), thus the BFR is
tf = time to crest of the stroke current, μs in terms of flashovers per 100 km-years.
C = coupling factor
ZT = surge impedance of the tower, ohms The equations for KTT and KI show that the voltage across
Zg = surge impedance of the ground wires, ohms the insulation increases as the time to crest of the stroke
TT = tower travel time, μs current decreases. This is caused by the tower component
TA = tower travel time to any location on the tower A, of voltage. Thus the critical current increases as the time
μs to crest increases. Therefore, theoretically, all fronts
TS = travel time of a span, μs should be considered. To do this, the equation for BFR
I = stroke current, KA should be changed to the following:
IR = current through footing of struck tower, KA
Ro = measured or low-current footing resistance, ohms BFR=0.6 NL P(IC) (4.14)
Ri = impulse or high-current footing resistance, ohms
τ = time constant of tail, μs
Now, to provide first estmate of the backflash rate, the
BFR, examine figure 6. The surge voltage on the ground
wires produces a surge voltage on the phase conductor 5. Simulation & Results
equal to the coupling factor C times the voltage on the
ground wires, or CVTT. Also note that the voltage VTA is The 230 kV HV line of figure 2 whose characteristics are
located on the tower opposite the phase conductor. given in table I, are used to calculate the backflash rate
Therefore, the crest voltage across the insulation V1 is using different methods. Also, this case study will include
the following
V1 = I [ KTA − CK TT ] K SP (4.7)
 5

1. The effect of decrease of resistance from Ro versus Ri factor to the lower phases, those phases that are most
2. One versus two shield wires likely to flashover. For example, for the 230-kV double-
3. The effect of underbuilt shield or ground wire circuit, two-ground-wire line with a shield wire height of
35 meters and coupling factor to the top, middle, and
As shown in the figure 3 & 4 the backflash rate for the bottom phase of 0.350, 0.248, and 0.183, respectively,
above mentioned high voltage lines with span length of installing a ground wire at 12 meters above ground at the
300 meters and CFO of 1200kV has been calculated by center of the tower increases these coupling factors to
using CIGRE method software and simplified method. 0.441, 0.347, and 0.307, respectively. Thus all coupling
The comparison appears acceptable for the line with factors are increased and are more uniform. Figure 7
tower height of 35 meters, but for tower height of 70 shows the dramatic decrease in BFR for this case.

BFR, Flashovers/100 km-yrs

meters the simplified method is inadequate. So, the
CIGRE method is always the proper tool.
10
9
Using the CIGRE method, the BFR of the single circuit 8
7
230 kV is shown in Fig. 5 as a function of RO with the 6 CIG
M
RE
ethod
ratio ρ/RO as a parameter. To illustrate the effect of the 5
4
Sim
M
plified
ethod
decrease of resistance with current, a curve labeled Ri=RO 3
2
for which the footing resistance is not decreased is also 1
0
presented. 1 2 3 4 5 6
X10Ro, ohms

Figure 3. Comparasion of BFRs for CIGRE method and simplified

method, 230kV double circuit towers with two ground wires and height
BFR, Flashovers/100km-yrs

of 35 meters

12

10

8
Sim plified
M ethod
6
CIG RE
4 M ethod

0
1 2 3 4 5 6
X10Ro, ohms

Figure 4. Comparasion of BFRs for CIGRE method and simplified

Figure 2. 230 kV Tower Dimensions method, 230kV double circuit towers with two ground wires and height
of 70 meters
BFR, Flashovers/100km-yrs

For some applications, where the cost of two shield wires 12

is not economically and technically justified, or where 10

there is low ground flash density, a single shield wire can 8

p/Ro=40
be used. The single wire increases the value of Re, 6 p/Ro=20

decreases the coupling factor, and thus increase the BFR. 4

p/Ro=10

To illustrate, the curves of Fig. 6 have been constructed to 2

compare one and two shield wires for a 230 kV double- 0

circuit line and two shield wires for a single-circuit 230 1 2 3 4

X10Ro, ohms
5 6 7 8

kV line. Using one shield wire on the double-circuit line

essentially doubles the BFR as compared to the two-
shield-wire case. Figure 5. Effect of decrease to high-current footing resistance

A ground wire located below the phase conductors cannot

truthfully be called a shield wire, since it has no shielding
function. Rather, its function is to increase the coupling
 6

BFR, Flashovers/100km-yrs
9

7. Acknowledgment
8
7
6
5 2GrdWire
4
3
1GrdWire
The authors express appreciation to Saudi Electric
2
1
Company engineers for thier time and support also their
0 gratitude to KFUPM for educational, studies facilities and
1 2 3 4 5 6
X10, ohms support.

Figure 6. Tow shield wires for the 230kV double circuit line with height
References:
BFR, Flashovers/100km-yrs

of 35 m decrease the BFR, p/Ro=20

10
[1] P. Chowdhuri, J. G. Anderson, W. A. Chisholm,
9 T. E. Field, M. Ishii, J. A. Martinez, M. B. Marz, J.
8
7 2GrdWires McDaniel, T. R. McDermott, A. M. Mousa,T. Narita,
6
5
2G rd
D. K. Nichols, & T. A. Short, Parameters of
4
3
W ires+under
built grdwire
Lightning Strokes: A Review, IEEE Transactions
2
1
and Power delivery, March 28, 2003.
0
1 2 3 4 5 6 7 8
X10Ro, ohms
[2] P. Chowdhuri, A.K. Mishra & B.W. McConnell,
Volt-time characteristics of short air gaps under
nonstandard lightning voltage waves, ibid., Vol. 12,
Figure 7. An underbuilt ground wire decreases the BFR, 230kV double
circuit line with height of 35 m, p/Ro=20. No. 1, pp. 470-476, 1997.

[3] P. Chowdhuri, A.K., Parameters of Lighting

6. Conclusion Strokes and Their effect on Power Systems, Vol. 12,
No. 1, pp. 1047-1051, 2001
The most significant parameters of the lightning return
stroke to estimate the severity on the power system are: (i) [4] P. Chowdhuri, A.K., S. Li & P. Yan Rigorous
peak current, (ii) current front time, (iii) velocity and (iv) analysis of back-flashover outages caused by direct
total charge of the flash. lightning strokes to overhead power lines, IEEE
Proceedings, 2002
The electromagnetic fields of the lightning channel and
the magnetic fields of the traveling current waves along [5] Andrew R. Hileman, Insulation Coordination for
the power-line tower will significantly affect the Power Systems, (Eastern Hemisphere Distribution,
insulator-string voltage, and hence the outage rate due to New York, 1999)
backflash. Analytical methods to estimate the backflash
outage rate have been proposed, which should result in [6] R. Thottappillil & M. A. Uman, Comparison of
better prediction of the lightning performance of overhead lightning return stroke models, J. Geophys. Res., vol.
power lines. 98, pp. 22 903–22 914, 1993.
In this report, equations were developed to estimate the
BFR that include the tower component of voltage; their
[7] V. Cooray & R. E. Orville, The effect of the
variation of current amplitude, current rise time and
use is called CIGRE method. This method is suffiently
return stroke velocity along the return stroke channel
complex so that the use of the computer program is
on the electromagnetic fields generated by the return
suggested. The effect of decrease of the concentrated
stroke, J. Geophys. Res., vol. 95, pp. 18 617–18 630,
grounds value on the BFR was addressed. Also, the effect
1990.
of the number of shield wires as well as adding underbuilt
shield or ground wire were highlighted.
[8] Dennis W. Lenk, F. Richard Stockum & David
The 230 kV line design from SEC is considered very E. Grimes,A new approach to distribution arrester
highly engineered, using two ground shield wires with 7.3 design, IEEE Transactions on power delivery, vol. 3,
meter span at each side made almost a full cover for both No. 2, April 1988.
circuits. This tower can be considered as lightning proof.
 7

[9] P. Pinceti & M. Giannettoni, A simplified model Sc. Degree from the University of Southern California and Ph.D.
for zinc oxide surge arrester, IEEE Transactions on degree from Mississippi State University in 1975 and 1985
respectively all in electrical engineering. He was a consultant to
power delivery, Vol. 14, No. 2, April 1999. A.B. Chance Company, and Flood Engineering. Dr. Shwehdi
held teaching positions with the University of Missouri-
Columbia, Texas A & I University, University of Florida and
Biographies Penn. State University from 1991-1993. At present he is
associate professor with the King Fahd University of Petroleum
Bander J. Qahtani; Born in Al-Khobar 1979. He obtained his & Minerals (KFUPM), Saudi Arabia. His research interest
B.Sc. degree in electrical engineering with honors from King includes, power system analysis, Power Quality & Harmonics,
Fahd University of Petroleum & Minerals (KFUPM) in 2002. In overvoltages analysis on Power Systems, Transmission and
the year 2000 he was selected as distinguished student for Saudi Distribution Systems. Dr. Shwehdi is active in IEEE activities.
Aramco Scholarship program. During his studies at KFUPM he He is listed as a distinguished lecturer with the DLP of the
has conducted several term projects and studies dealing with IEEE/PES DLP upon the Board selection, was named and
Industrial power systems. Upon graduation, Bandar was awarded the 2001 IEEE/PES outstanding chapter engineer,. He
employed by Aramco as instrument engineer with Southern was named and awarded the 1999 IEEE WG for standard award,
Area Producing Engineering Department (SAPED) in Abqaiq. the GCC-CIGRE 1998 best applied research award, IEEE/IAS
He is enrolled in the Msc. Program at KFUPM. Bandar has Outstanding Supervisor for Student Research 1989, 1990, and
published and presented many technical papers and reports to the IEEE outstanding student advisor in 1990.
region, and international conferences.

M. H. Shwehdi (S'74, M'85, SM 90) received the B. SC. degree

from University of Tripoli, Libya in 1972. He obtained the M.

Table I Characteristics of Lines, Distances in meters

System
h yA yB yC Sg Sa Sb Zg ZT CA CB CC
Voltage
230 35 29 24 18 5 8 11 379 190 35. 25. 18.
a
230 35 29 24 18 0 8 11 600 190 22. 16. 12.
230 70 64 59 53 5 8 11 421 210 42. 34. 28.
b
230 35 29 24 18 5 8 11 239 190 44. 35. 31.
a
Single ground wire. b Underbuilt ground wire at h=12 m at center of tower