You are on page 1of 7

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

net/publication/251930823

Grounding of overhead transmission lines for improved lightning protection

Article · January 2010


DOI: 10.1109/TDC.2010.5484321

CITATIONS READS
6 3,490

3 authors, including:

William Chisholm
Kinectrics
117 PUBLICATIONS   2,307 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Transient Grounding; Graduate Course Lecturer View project

CIGRE B2.007 View project

All content following this page was uploaded by William Chisholm on 31 March 2015.

The user has requested enhancement of the downloaded file.


1

Grounding of Overhead Transmission Lines


for Improved Lightning Protection
William A. Chisholm, Fellow, IEEE, Emanuel Petrache, Member, IEEE
and Fabio Bologna

Abstract—Improved grounding can be a cost-effective method However, Fig. 1 also suggests that OHGW lose 40% of their
to improve power quality by reducing the number of lightning effectiveness for typical 69-kV lines unless a low frequency,
flashovers on shielded overhead transmission lines. low-current footing resistance Rf < 20 Ω can be achieved.
Improvements can be valued against benchmark values of cost
per avoided customer momentary dip. Effectiveness of
improvements can vary widely because the local soil resistivity If the number of customers affected by a lightning tripout is
varies with a wide statistical distribution. A portable impulse factored in with the effectiveness, we can estimate the number
Zed-Meter® can be used to measure the local transient response of annual power quality disturbances from lightning as a
of transmission tower ground electrodes. function of footing resistance. Figure 2 shows that there is
roughly equal benefit to treating 138-kV and 345-kV lines.
Index Terms—conductivity, dipole antennas, grounding,
impulse testing, lightning, power quality, power transmission
reliability,

I. INTRODUCTION

T HE lightning flash normally terminates on the overhead


groundwire (OHGW) of a shielded transmission line and
is conducted to nearby ground electrodes, such as the tower
foundations and supplementary buried wires. In most cases,
the product of the ground impedance and the impressed
current gives a potential difference that does not exceed the
insulation impulse overvoltage strength. Fig. 1 shows that
OHGW protection gives a high degree of reliability, with
typical failure rates in the range of 1 to 8% for 345-kV lines.

Fig. 2. Number of Typical Customer Disturbances as Function of Low-


Frequency, Low-Current Footing Resistance Rf [1]

II. LOW-FREQUENCY, LOW-CURRENT FOOTING RESISTANCE


The low-frequency, low-current resistance of an electrically
conducting solid can be expressed as a geometric resistance,
based on the soil resistivity ρ, the electrode surface area A and
the maximum extent s1 [2] s2 [3] in one or two dimensions or
the three-dimensional geometric radius s3 [3]. This
geometric resistance is given by:
ρ ⎛ 11.8 ⋅ s32 ⎞ (1)
R = ln⎜⎜ ⎟⎟
geometric 2π s A
3 ⎝ ⎠

Fig. 1. Reliability of Overhead Groundwire Lightning Protection as Function Where Rgeometric is the resistance of the solid electrode to
of Low-Frequency, Low-Current Footing Resistance Rf [1] remote earth (Ω),
ρ is the uniform soil resistivity (Ωm),
W. A. Chisholm is with Kinectrics and the Université du Québec à
Chicoutimi, Toronto, Ontario, Canada (e-mail: W.A.Chisholm@ieee.org). s3 is a geometric radius of the electrode, rx2 + ry2 + rz2 (m),
E. Petrache is with Kinectrics, Toronto, Ontario, Canada (e-mail:
Emanuel.Petrache@kinectrics.com) 11.8 is (2πe√3)/3, and
F. Bologna is with EPRI, Charlotte, NC, USA (email: fbologna@epri.com) A is the surface area of overall electrode volume (m2).
2

The single driven rod and the reinforced-concrete cylindrical resistance tends to be a small fraction of the geometric
foundation of a steel pole are examples of electrically resistance and is relatively insensitive to the value of F, so that
conducting, solid electrodes with rx=ry, the radius, and rz the a constant F=2 for any electrode shape is suitable.
length or depth below the surface.
The surface area of compact electrodes, for example
The resistance of a solid rectangular conducting box of length consisting of four poured-concrete caissons or grillage below
l, width w and depth below surface rz will have the following directly buried steel legs, can be substituted for Awire. The
dimensions of rx and ry: resulting “fill factor” correction, ln(A/(F⋅Awire)), is likely to be
low, so errors in estimating L for these types of electrode can
P l 2 + w2 be tolerated.
P = 2(l + w); D = ; A = lw + P ⋅ rz ; rx =
π 2 (2) Fig. 3 implements (3) to calculate the number of vertical
(6 D − 8rx ) + (6D − 8rx )2
− 24( D + 6r − 6 Drx )
2 2
10 mm diameter rods needed to achieve the 10-Ω resistance
ry = x
that is needed for typical 90% effectiveness of OHGW on a
12
69-kV line in Fig.1. With a 1x1 m foundation footprint, the
The dimensions rx and ry in (2) give an ellipse of perimeter P geometric resistance alone will be greater than 10 Ω, and no
that matches the box perimeter 2(l+w) in the horizontal plane. number of rods will suffice, unless each rod is at least 4.6 m
long.
Equations (1) and (2) are very suitable for calculation of
rectangular ground grid resistance, with an accuracy of better
than 2% compared to numerical reference calculations [3].
However, an additional term is needed for the overall
resistance. Practical electrodes are not usually solid: they are
usually wire-frame approximations to solid shapes [3,4]. For
example, the resistance of a wire grid of overall length L
approximating a rectangular box would consist of two terms:
the geometric resistance of (1) and a so-called “contact
resistance”, suggested to be Rcontact=ρ/L in a simple model of
substation grounding [5]. Another illustration, more relevant
to transmission tower grounding than substation grids, is that a
buried horizontal wire approximates a vertical rectangular
plate of the same length and depth, with a thickness equal to
the buried wire radius. These electrodes have the same
geometric resistance (1) as the first term in (3), along with an Fig. 3. Number of Vertical Rods to achieve Rf=10 Ω in soil with ρ=100 Ωm
added contact resistance term that varies inversely with L:
The separation of wire frame resistance into two terms,
⎡ geometric and contact, allows more convenient modeling of
ρ ⎢ 1 ⎜ 11.8 ⋅ s3
⎛ 2⎞
⎟ 1 ⎛⎜ A ⎞⎤ ionization effects. These tend to increase the effective radius
R = ⎟⎥
f 2π ⎢ s
ln
⎜⎜ A ⎟⎟ + L ln⎜ F ⋅ A ⎟⎥
(3) of the individual wires, with a corona that expands radially
⎣ 3 ⎝ ⎠ ⎝ Wire ⎠
⎦ until the gradient falls to levels of 300-400 kV/m. With the
Where Rf is the resistance of a wire frame to remote earth (Ω), separation in (3), it becomes clear that ionization increases the
consisting of Rgeometric + Rcontact, wire radius rWire and the wire frame surface area Awire, but does
ρ is the uniform soil resistivity (Ωm), not affect the geometric resistance.
s3 is the geometric radius of the electrode rx2 + ry2 + rz2 (m),
III. TOWER-TO-TOWER VARIATION OF RESISTIVITY
A is the overall surface area of the electrode in contact with
Resistivity ρ (Ωm), or its reciprocal conductivity σ (typically
soil (m2),
expressed in mS/m), varies considerably over the typical
L is the total length of the wire frame approximating A (m),
footing-to-footing span length of about 300 m. A statistical
AWire is the wire frame surface area, 2πrWireL (m2), and study established that the tower-to-tower variation of
F is a constant, 8/n2 where n is the number of radial wires resistivity could be modeled with a lognormal distribution,
meeting at a point or otherwise 1<F<2. described by a median value and the standard deviation of the
logarithm of the resistivity. Lognormal distributions are often
The contact resistance decreases with increasing wire length used in describing other lightning parameters such as the peak
L, with an adjustment for fill factor based on the ratio of magnitude of the first-stroke current, with a median of 31 kA
overall area A to wire surface area AWire. When the ratio and a standard deviation of the natural logarithm of current
A/(F⋅AWire)<1, the contact resistance is zero. The contact
3

σln I = 0.48 [8]. What was surprising in the study of observed The distributions of resistance were well characterized by
resistivity ρ from helicopter electromagnetic surveys was that lognormal forms having unique median values and a common
the median standard deviation σln ρ was 1.21. This suggests shared log standard deviation of σln R=0.92. Spot-checks of
that the statistical distribution of resistivity and thus resistance the F- and t-statistics confirmed that the lognormal
Rf from tower to tower is considerably wider than that found distributions shared the same variance but had different
for the peak current. means. In use, this means that there is a ±1σ (68%) chance
that the resistance of a particular tower falls in a range
IV. TOWER-TO-TOWER VARIATION OF RESISTANCE RF between exp(0.92) and exp(-0.92) times the median for the
Transmission lines constructed with relatively uniform line, a 16% chance that it will be a factor of 2.5× higher and
foundation details, with similar number of legs, spacing and 16% chance that it will be less than 40% of the median.
depth of burial, will have relatively constant values of s3 and A
in (3). This means that most of the observed tower-to-tower V. VARIATION OF SOIL RESISTIVITY WITH FREQUENCY
variation in low-frequency resistance will be a result of tower- Lightning currents are impulsive with important frequency
to-tower variation in resistivity ρ. A utility database of more content in the range of 100 kHz. This frequency is obtained
than 10,000 individual low-frequency resistance by equating the sine-wave peak and derivative with that of
measurements on untreated 500-kV single circuit lattice typical lightning surges (31 kA median, 25 kA/μs peak rate of
towers was used to evaluate whether the lognormal current rise) with those of a sine wave of the same peak
distribution was appropriate, using Pearson-Hartley amplitude.
classification [9]. Footing resistance values for 48 different
lines were used to construct the statistics β1,β2 and k: Many researchers have established that the frequency
dependence of resistivity ρ can be exploited in geological
1 N
1 N surveys to characterize surface and below-grade materials.
μ1 =
N
∑ yi
i =1
μ2 =
N
∑(y
i =1
i − μ1 ) 2 For example, scans of apparent resistivity with logarithmically
1 N
1 N spaced frequencies in the range of 0.4 to 100 kHz form the
μ3 =
N
∑(y
i =1
i − μ1 ) 3
μ4 =
N
∑(y
i =1
i − μ1 ) 4
(4)
basis of airborne electromagnetic resistivity surveys.

β1 = μ β2 = μ4
2
3 In general, soils have electrical properties that can be defined
μ 23 μ 22 using complex resistivity ρ* as a function of frequency f:
β 1 ( β 2 + 3) 2
k=
4(4 β 2 − 3β 1 )(2 β 2 − 3β 1 − 6) ⎧⎪ ⎛ 1 ⎞ ⎫⎪
ρ * ( f ) = ρ ′ − jρ ′′ ≈ ρ dc ⎨1 − m⎜⎜1 − ⎟⎟ ⎬ (5)
⎪⎩ ⎝ (1 + j 2πfτ ) c ⎠ ⎭⎪
The distributions of resistance were un-classifiable, but the
distributions of the natural logarithm of resistance (ln R) gave Where: ρdc is the low-frequency resistivity (Ωm),
the useful results in Fig. 4. The statistic k was in the range of m varies from 0.1 to 1 depending on soil mineral content,
-0.3 to 0 for most lines. A value of k=0 indicates a perfect τ is a time constant in the range of 10-4<τ<104,
(log) normal distribution, as does β1=0 and β2=3. c varies from 0.2 to 0.6 depending on soil grain size,
f is the frequency (Hz), εo is 8.854⋅10-12 F/m, j is √-1

A factor-of-ten change in resistivity from 100 Hz to 100 kHz


is observed for some materials, such as clay. Others such as
sand or shale show only a 20% change in ρ from 100 Hz to
100 kHz. For sandy soil, τ = 8s has been reported [11], in
contrast to τ =1.3 ms for kaolinite (clay). There will be
corresponding differences between the low-frequency
measurements of Rf (typically carried out at about 100 Hz) and
the impulse response under lightning surge conditions (with
its characteristic frequency in the range of 100 kHz).

Variations of soil properties with frequency can be measured


with network analyzers. However, moisture content, which
also affects dielectric constant [12], is more typically
Fig. 4. Pearson-Hartley Classification of Natural Logarithm of Measured characterized using a low-amplitude transient test method
Transmission Line Footing Resistance for 48 Different 500-kV Lines called time domain reflectometery. TDR instruments measure
the transient impedance of short probes inserted into drilled
core samples or driven directly into the surface soil.
4

VI. TRANSIENT VARIATION OF FOOTING IMPEDANCE overhead groundwire system. Potential rise from the injection
point to a remote potential lead, running at 180°, was
A. Test Instrumentation computed as well. The ratio of potential rise to injected
Transient injection methods such as TDR can also be current gives time-varying impedance that has an initial
efficiently used to establish the potential rise at the base of a related to the tower surge impedance, and then settles to a
particular tower in response to injection of a test current pulse. relatively constant value. The median of this V/I relation in
A portable impulse source with four-channel 100 MS/s the time scale of 450 to 650 ns was then computed. For the
digitizers and computer control, the “Zed-Meter®”, was reference case, an impedance of 5 Ω was obtained with a soil
developed to measure transient impedance. This instrument resistivity of 50 Ωm, and when the soil resistivity was
was specially adapted to transmission tower surveys by the 1000 Ωm, this impedance increased to 46 Ω. In both cases,
use of: the relative permittivity of the soil was constant at ε’=5.
• Safe impulse source, meeting local electrical codes
for impulse shock from electric fence apparatus C. Configuration with Short Potential Lead
• Compact and self-powered instrument, capable of The current reaction lead needs to be sufficiently long to
exciting tower base against surge impedance of a ensure that any reflection from its termination does not arrive
current reaction wire. before the impedance has stabilized. Fig. 5 shows a return
• Suitable sensitivity, averaging and signal processing reflection from the current reaction lead at 750 ns,
for reliable measurements with high induced corresponding to a wave propagation velocity of (2x100 m /
potentials on test leads 750 ns) or 0.89c where c is the speed of light.
The Zed-Meter test was initially designed for the use of
straight current reaction and remote potential leads of equal
length, forming a dipole on the surface of the earth centered at
the tower. Field experience showed that there was
considerable flexibility in placing the leads, for example at
90° to the line direction, or with unequal length, or with a
“meander line” or zig-zag pattern. This experience has been
supported by modeling of the entire system, using the NEC-4
computer program [6] along with Fourier transform methods
to establish the response to a 1.5-μs test pulse current.
B. Reference Configuration
Fig. 5. Effect of Shortened Potential Lead (75, 50 or 25 m) on Calculated
Most transmission line rights-of-way have sufficient space to Impedance Profile for 50 Ωm Soil
lay out a pair of leads on the surface of the ground, running in
opposite directions from a tower leg as shown in Fig. 4. There is not much difference in the impedance profiles
measured with short remote potential leads, of 25, 50 or 75 m.
For the ρ=50 Ωm case in Fig. 5, the results are the same in all
cases. For the case of ρ=1000 Ω in Fig. 6, the median
impedance decreases by about 7% for a 25-m lead, compared
to the reference value of 46 Ω with a 100-m lead.

Fig. 4. Reference Configuration for “Zed-Meter” Impulse Test using


Impedance of Reaction Leads

The computer program, NEC-4 [6], was used to model the


injection of a current pulse between a straight, 100-m lead on Fig. 6. Effect of Shortened Potential Lead (75, 50 or 25 m) on Calculated
the surface of the ground and one leg of a lattice tower. The Impedance Profile for 1000 Ωm Soil
tower had three other legs and an electrically connected
5

The conclusion from these simulations was that, if one of the


two leads needs to be short, it should be the potential lead.
D. Configuration with Zig-Zag Current Reaction Lead
There are a number of cases, for example on frozen soil or
rock, or with guyed towers, where it can be an advantage to
extend the current reaction lead, prolonging the return of the
reflection from its termination. A combination of field trials
and modeling established that a “meander line” with a zig-zag
path on the surface of the ground was nearly as effective as a
straight current reaction lead of the same 100-m length.

Fig. 9. Effect of Zig-Zag Configurations on Calculated Impedance, ρ=1k Ωm

E. Field Test Results with Zig-Zag Current Reaction Lead


A field test was carried out on a lattice tower with single
overhead groundwire in an urban area. Three different
versions of Configuration ZZ1 were tested. The remote
potential lead in all cases was 93 m in length and terminated
in a spike driven into the low-resistivity soil. The current
reaction lead was 152 m in length. The meander patterns
were:
• A single back-and-forth loop, 80 m from the tower
• Five back-and-forth loops, separated by 3 m
• Six back-and-forth loops, separated by 1 m
Fig. 7. Zig-Zag or “Meander” configurations for Current Reaction and
Remote Potential Leads in Zed-Meter Test The test results are shown for each lead configuration in
Figs. 10 to 12. In all cases, two values of impedance versus
Fig. 7 shows four different patterns – two with straight remote time are shown, based on the measured current into the tower
potential leads of 100-m length and two with both lines and the current into the reaction lead. A moving standard
following a back-and-forth pattern. The calculations of tower deviation is also plotted, making use of a 150-ns time window
base potential rise are shown in Fig. 8 and Fig. 9 for ρ=50 Ωm and both impedance profiles. This remains below 0.1 Ω for
and ρ=1000 Ωm respectively. most of the time of interest from 750 to 1500 ns.

Fig. 8. Effect of Zig-Zag Configurations on Calculated Impedance, ρ=50 Ωm

Fig. 10. Observed Impedance of 115-kV Lattice Tower with Single Zig-Zag
The timing of the reflection from the terminated end of the
Loop and 152-m Total Length. Median 3.6 Ω.
current reaction lead is similar in all configurations. The
median impedance over the time scale from 450 to 650 ns is
substantially the same as with the reference lead configuration
shown in Fig. 3.
6

VIII. REFERENCES
[1] W.A. Chisholm and J.G. Anderson, “Guide for Transmission Line
Grounding: A Roadmap for Design, Testing, and Remediation”, EPRI,
Palo Alto, CA: 2004. 1002021. Available: mydocs.epri.com/docs/public
[2] W.A. Chisholm and W. Janischewskyj, “Lightning Surge Response of
Ground Electrodes”, IEEE Trans. PWRD, 4(2), pp.1329-37, Apr. 1989.
[3] W.A. Chisholm and A. Phillips, “Contact and Geometric Resistance of
Wire Frame Electrodes”, in Proc. IX International Symposium on
Lightning Protection, Foz do Iguaçu, Brazil, Nov. 2007.
[4] W. A. Chisholm and F. Bologna, “Validation of Simple Fill Factor Term
for Contact Resistance of Radial Electrodes”, in Proc. Ground ‘2008,
Florianopolis, Brazil, Nov. 2008.
[5] Guide for Safety in AC Substation Grounding, IEEE Standard 80-2000.
[6] G. J. Burke, “Numerical Electromagnetics Code NEC-4 Method of
Moments,” Lawrence Livermore National Laboratory Report UCRL-
MA-109338, 1992.
Fig. 11. Observed Impedance of 115-kV Lattice Tower with Five Zig-Zag [7] E. Petrache, W.A. Chisholm and A. Phillips, “Evaluating the Transient
Loops, 3 m Apart and 152-m Total Length. Median 3.4 Ω. Impedance of Transmission Line Towers”, in Proc. IX International
Symposium on Lightning Protection, Foz do Iguaçu, Brazil, Nov. 2007
[8] CIGRE Working Group 01 of Study Committee 33, “Guide to
Procedures for Estimating the Lightning Performance of Transmission
Lines”, CIGRE Brochure 63, Oct. 1991.
[9] E.S. Pearson and H.O. Hartley, Biometrika Tables for Statisticians,
Vol. 1 (Cambridge: Cambridge University Press), 1966, pp. 67-88.
[10] S.D. Logsdon, “Soil Dielectric Spectra from Vector Network Analyzer
Data”, Soil Sci. Soc. Am. J. 69, pp. 983–989, 2005.
[11] R.L. van Dam, B. Borchers and J. Hendrickx, “Methods for prediction
of soil dielectric properties: a review”, SPIE 2005.
[12] C.H. Roth, M.A. Malicki, R. Plagge, “Empirical Evaluation of the
Relationship between Soil Dielectric Constant and Volumetric Water
Content as the Basis for Calibrating Soil Moisture Measurements by
TDR”, European Journal of Soil Science 43(1), pp. 1-13.

IX. BIOGRAPHIES
William A. Chisholm (M’1981, SM’ 1989, F’2007) was
Fig. 12. Observed Impedance of 115-kV Lattice Tower with Six Zig-Zag born in Plattsburgh, NY, USA in 1955. He graduated from
Loops, 1 m Apart and 152-m Total Length. Median 2.9 Ω. the University of Toronto in Engineering Science (1977),
M.Eng (1979) and PhD (1983) in Electrical Engineering
All three test results show a relatively constant impedance from the University of Waterloo. Chisholm joined the
Ontario Hydro Research Division (now Kinectrics) in 1977
profile in the time range from 750 to 1500 ns. There is a 5% and retired in 2007. He has many research projects in
reduction in impedance with loops spaced 3 m apart, and a lightning protection, overhead line ampacity and electrical
more substantial 19% drop in the measured value with the performance of insulators under icing conditions. He spent a year at UQAC in
Chicoutimi, where he co-authored an IEEE/Wiley book, Insulators for Icing
tight 1-m separation.
and Polluted Environments. He is the Secretary of the PES Transmission and
Distribution Committee, a contributor to IEEE Standards 957, 1243, 1410 and
VII. CONCLUSIONS 1783 and a Registered Professional Engineer in the Province of Ontario.
It is possible to compute the resistance of a wide variety of Emanuel Petrache (S’02–M’04) was born in Constanta,
different ground electrodes, using a single simplified formula Romania, in 1975. He received the M.S. degree in electrical
that separates the geometric and contact resistance terms. The engineering from the University Politehnica of Bucharest,
geometric resistance varies inversely with geometric radius, Romania, in 1998, and the Ph.D. degree from the Swiss
Federal Institute of Technology, Lausanne, in 2004.
scaled by a shape factor. The contact resistance varies From 2004 to 2006 he was a postdoctoral fellow at
inversely with wire length, scaled by a fill factor. University of Toronto, Canada, in the Lightning Studies
Group. In 2006, he joined Kinectrics, formerly the Ontario
Resistivity and its linear effect on low-frequency, low-current Hydro Research Division, and is now a Principal Engineer. His research
interests include numerical computation of electromagnetic fields, lightning,
resistance varies on the scale of 300 m span length, and this and electromagnetic field interactions with transmission lines.
variation can be modeled with a lognormal distribution. Each
line has a unique median, but they tend to have similar values Fabio Bologna holds an M.Sc in Electrical Engineering
of log standard deviation σln R of 0.9. from the University of the Witwatersrand in Johannesburg,
South Africa. He worked at Technology Services
International, a division of Eskom Enterprises, as a chief
The impulse impedance of a transmission tower foundation consultant on insulators. In 2006 he joined EPRI in
resistance can be practically tested using the surge impedance Charlotte, NC and is now the for Overhead Transmission
of a reaction lead, placed on the surface of the ground, rather Program Manager. His research interests include faults on
transmission lines caused by bird streamers, lightning and
than a buried connection in contact with the soil. The fires, optical fibres in high voltage environments and high voltage outdoor
impedance of the test leads and resulting ground impedance insulation.
profiles are not sensitive to lead orientation and configuration.

View publication stats

You might also like