You are on page 1of 11

IEEE Transactions on Dielectrics and Electrical Insulation Vol. 14, No.

3; June 2007 751

External Impedance and Admittance of Buried Horizontal


Wires for Transient Studies Using Transmission
Line Analysis
Nelson Theethayi, Rajeev Thottappillil
Division for Electricity and Lightning Research
Uppsala University
Box 534, S – 75121, Uppsala, Sweden

Mario Paolone, Carlo Alberto Nucci


Department of Electrical Engineering
University of Bologna
Viale Risorgimento, 2 - 40136
Bologna, Italy

and Farhad Rachidi


EMC Group, Swiss Federal Institute of Technology
EPFL-STI-LRE, ELL-138, Station 11, CH-1015
Lausanne, Switzerland

ABSTRACT
The paper investigates the applicability of some closed form expressions for the ground
impedance and ground admittance of buried horizontal wires (bare and insulated) for
lightning or switching transient analyses based on transmission line (TL) theory. In
view of the frequency contents that typically characterize such transients, the behavior
of the ground impedance and admittance is studied for a wide frequency range up to 10
MHz. Low frequency approximation of the ground impedance is not always
appropriate for transient analysis. Sensitivity analyses show that, unlike overhead
wires, the ground impedance for buried wires is little sensitive to the ground
conductivity. On the other hand, the ground admittance varies strongly with the
ground conductivity. The paper also discusses the results of transient analysis of buried
cables performed by means of electromagnetic transient programs (EMTP) that
neglect the ground admittance. The limits of such an approximation are discussed in
order to evaluate the applicability of EMTP-like programs to the transient analysis of
buried conductors. Transient pulse propagation in time domain based on finite
difference time domain (FDTD) method of solution of TL equations is also discussed
for a future inclusion of non-linear phenomena, like soil ionization and
arcing/breakdown mechanisms, in the soil. The analysis presented could be useful in
estimating surge propagation characteristics of buried wires for appropriate insulation
coordination and transient protection.
Index Terms - Transmission line theory, lightning, power system transients,
impedance, admittance, cables.

1 INTRODUCTION stroke is, or switching transients are, characterized by


frequency components ranging from a few tens of kHz to a
UNDERGROUND transmission systems are common few hundreds of kHz. However, for lightning subsequent
whether it be power, telecommunication or railway systems. return strokes the frequency components could be up to a few
Any normally operating systems are prone to system faults MHz [1,2]. In this study, double exponential impulse current
caused either due to lightning, switching or other similar sources, characterized by 10 μs and 0.1 μs, rise time (assumed
electromagnetic interference (EMI). Lightning first return to be upper and lower limits for lightning return stroke rise
time [1,2]) with a normalized peak of 1 A, are used for
Manuscript received on 6 November 2006, in final form 27 March 2007. analyzing pulse propagation aspects.

1070-9878/07/$25.00 © 2007 IEEE


752 N. Theethayi et al.: External Impedance and Admittance of Buried Horizontal Wires for Transient Studies

For transmission line analysis, influences of a finitely linear effects, a time domain solution of TL equations is
conducting ground are described by ground impedance and desirable. In this paper, we propose an efficient way of
ground admittance terms [3-14]. Closed form expressions, solving lossy transmission line equations in time domain by
simple and reasonably accurate, are needed for the ground using the FDTD method with recursive convolutions.
impedance and admittance for a later application in transient The paper is organized as follows: First, the system under
analysis of complex outdoor systems involving multiple study is discussed based on the TL theory. Secondly,
conductors (e.g. transmission lines - MTL - with wires above discussion is directed on various expressions for the ground
ground, on the ground and below ground like in the railway impedance and admittance and their applicability for transient
systems). Under such circumstances, researchers and analysis based on TL solution. Thirdly, the importance of
engineers resort to simple TL type models, which are ground admittance is discussed and illustrated. Fourthly, the
computationally efficient and sufficiently accurate compared time domain FDTD method of solving TL equations using
with more rigorous and complex models based on the solution recursive convolution is presented. Finally, conclusions are
of Hertz potentials [6,8,9]. Thus, the present work focuses on given.
simple expressions for the ground impedance and admittance
of buried wires.
2 SYSTEM UNDER STUDY
Existing ground impedance expressions (low frequency
approximations) obtained by different researchers [3,5,11,12] The geometrical configuration of the buried bare wire and
are accurate at low frequencies. However, they diverge at the insulated wire is shown in Figure 1 left and right window,
higher frequencies compared with wide frequency range respectively. The soil is characterized by its conductivity
expressions proposed in [6,8,9]. Limitations of low frequency ( σ g ), permittivity ( ε g = ε rg ε 0 ) and permeability ( μ 0 ). For
approximations of the ground impedance and their insulated wires, we have, additionally, the insulation
applicability for transient analysis are studied in this paper. permittivity ( ε in = ε rin ε 0 ). The wires are buried at a depth of d
Computationally efficient and simple expressions for ground
meters and have radius of a meters. For insulated wires the
impedance without involving integrals and special functions
thickness of insulation is defined in terms of the external
(like Bessel or Henkel), will be compared with more rigorous
radius b, as shown in Figure 1.
Sunde’s [4] and Wait’s [6] expressions.
As mentioned above, another important quantity that model
the ground effects on surge propagation is the ground
admittance. Usually, ground impedance and admittance are
related to each other through the ground propagation constant
[7]. Transient analysis programs like EMTP [12,15] ignore the
ground admittance and use only admittance due to the
insulation of buried cables. This is perhaps because, for
overhead conductor systems, the ground admittance has an Figure 1. Bare (left) and insulated (right) wire in the soil under study.
overall negligible influence on the surge propagation and only
the capacitive admittance needs to be taken into account [16-
18]. However, we will show, as recently pointed out in For a differential length dx, TL representations for the bare
[13,14], that ground admittance plays an important role for the wire and the insulated wire are shown in Figure 2 [7], where V
buried conductor system and cannot be omitted in the and I are the line voltage and current respectively. It is worth
analysis. noting that in a general TL representation, series impedance
and shunt admittance parameters control the wave propagation
Further, in EMTP, a low frequency approximation based on on the wire, assuming transverse electromagnetic (TEM)
Pollaczek’s theory [3] for the ground impedance is adopted. It propagation [10]. The total series impedance is the sum of the
is worth mentioning here that the EMTP cable model is based internal impedance (due to EM fields within the conductor
on a fitting procedure, where in the appropriate modal line leading to skin effect phenomena [10]) and the external
impedance and admittance parameters are fitted in frequency impedance (due to the magnetic fields in the medium outside
domain. Detailed information on such procedures is discussed the conductor). The external impedance is considered in this
in works by Marti [19] and Ametani [20,21]. It will be shown paper owing to its much larger contribution compared with the
that all those approximations could lead to inaccuracies in internal impedance. The external impedance itself is the sum
transient studies because frequency dependent impedance and of two impedances for buried insulated cables, namely, the
admittance parameters seriously affect propagation inductive impedance (due to the magnetic field in the
characteristics. insulation) and the ground impedance (which represents the
When large transient currents or voltages propagate in bare penetration of the magnetic field in the finitely-conducting
wires or insulated cables, there will be non-linear arcing or soil). For bare wires, the external impedance reduces to the
breakdown mechanisms occurring within the soil or insulation ground impedance. The total shunt admittance is composed of
medium. The mechanisms of soil ionization in grounding the capacitive admittance of the insulation in series and the
systems and insulation breakdown in cables are common ground admittance. The latter accounts for the penetration of
during power system transients/faults. To include such non- the electric field in the soil. For bare wires, the transverse
IEEE Transactions on Dielectrics and Electrical Insulation Vol. 14, No. 3; June 2007 753

admittance reduces to the ground admittance. In the next form approximations. Since only final equations are
Section, we shall devote particular attention to the ground presented, interested readers may refer to associated
impedance and admittance and their effect on the surge references for details. We shall see the behavior of ground
propagation along buried cables. impedances as the frequency is increased within the validity of
TL approximation. Let us assume that the external radius of
the wire is Rab (if the bare wire is in analysis then Rab = a
otherwise (insulated wire) Rab = b). Note that in all the
expressions to be discussed below, for mutual impedance
between two parallel wires, Rab needs to be replaced by the
horizontal distance between the buried wires and d by the
average depth [4].
Figure 2. Differential length transmission line representation for bare (left) The ground impedance for buried wires was first derived by
and insulated (right) wire in the soil under study. Pollaczek [3], as a low frequency approximation, i.e., with
frequency satisfying ω << σ g ⋅ ε g−1 or using propagation
For bare wires, TL equations with the ground impedance constant γ 'g = jωμ 0σ g . Saad et al. [11] have shown that
and admittance as Zgb and Ygb, respectively, are given by (1). equation (4) is an excellent and valid approximation for
dV ( x, jω) Pollaczek’s expression. Note that in (4), K0(.) is Bessel’s
= Z gb ⋅ I ( x, jω) (1a) function of second kind and order zero. It is also shown in
dx [11] that equation (4) is identical or a better approximation
dI ( x, jω)
= Ygb ⋅ V ( x, jω)
compared with other low frequency approximations proposed
(1b) by Wedephol and Wilcox [5], Ametani [20,21] and Semlyen
dx
[12]. So those expressions are not presented here.
For insulated wires, TL equations with inductance and
jωμ0 ⎛⎜ ⎞
capacitance, due to insulation thickness and its material, as L − 2 dγ '
and C, respectively, and with the ground impedance and Z Saad etal
= ( '
)
K 0 γ g Rab +
2e g ⎟ (4)
admittance as Zgi and Ygi, respectively are given by equation
g
2π ⎜
⎝ ( )
4 + Rab2 γ 'g
2


(2). The per-unit length inductance and capacitance due to
insulation thickness and its material are calculated using For wide frequency range (within the validity of TL
equations (3) [5,10]. theory), Sunde’s [4] ground impedance expression without
dV (x, jω) low frequency approximation is given by equation (5). Note
= ( jωL + Z gi ) ⋅ I (x, jω) (2a) that the Pollaczek’s expression is similar to equation (5) but
dx with the low frequency approximation for the propagation
dI ( x, jω) ⎛ CYgi ⎞
⎟ ⋅ V ( x, jω)
constant.
= jω⎜
( )
(2b)
dx ⎜ jωC + Y ⎟ ⎛ K (γ R ) − K γ R 2 + 4d 2 + ⎞
⎝ gi ⎠
⎜ 0 g ab 0 g ab ⎟
μ ⎛b⎞ jωμ0 ⎜ ∞ − 2 d λ2 + γ 2 ⎟ (5)
L = 0 ln⎜ ⎟ (3a) Z gSunde = ⎜ e g
cos (λ R ) ⎟
2π 2
⎜⎜ ∫ dλ
2π ⎝ a ⎠ ab

λ + λ +γg 2 2 ⎟⎟
2πε in ⎝ 0 ⎠
C= (3b)
⎛b⎞ One of the difficulties with equation (5) is that, as the
ln⎜ ⎟ frequency is increased, the integral term converges slowly
⎝a⎠
leading to longer computation time and possible truncation
errors. Further, it was found that the first two Bessel terms in
3 REVIEW OF GROUND IMPEDANCE AND (5) are oscillatory when frequencies approach 1 MHz. The
GROUND ADMITTANCE EXPRESSIONS propagation constant in equation (5) is
FOR BURIED HORIZONTAL WIRES γ g = jωμ0 (σ g + jωε g ) .
In any TL problem involving ground return, ground Wait [6], Bridges [8] and Chen [9] have independently
impedance and ground admittance play an important role [4- proposed more complex ground impedance expressions
10]. In buried wires, waves propagate and return mostly derived from rigorous electromagnetic theory (exact
within the soil medium, unlike in overhead wires, where expressions). The exact expression for ground impedance has
waves have only return path through the soil. Several two modes, namely, TL or TEM mode (contributing to
researchers have developed expressions for the ground differential mode currents on the line) and radiation/antenna
impedance of buried wires starting from fundamentals of mode (contributing to common mode currents on the line) [8].
electromagnetic theory or modification of expressions Considering only the TL mode, it is shown later by an
developed earlier [3-11]. Only those applicable to the present example that Wait’s ground impedance expression is
work for TL analysis are presented, with a view on closed equivalent to Sunde’s expression (5). Wait also shows [6] that
754 N. Theethayi et al.: External Impedance and Admittance of Buried Horizontal Wires for Transient Studies

under the condition d ⋅ ε 0 ⋅ μ0 ⋅ ω 2 << 1 , for all the frequencies and Wait’s expressions will be used as reference values, as
they are identical in the frequency range studied here (see
below this limit, quasi-static or TL approximations are valid.
corresponding curves in Figures 3a and 3b). From Figure 3a
This limiting condition is about 5 MHz for wires at depths
between 0.5 m – 1.0 m and for different ground conditions. for 1 mS/m ground conductivity, it can be seen that
Wait’s quasi-static ground impedance is equation (6) [6]. expressions of Sunde (5), Wait (6), Saad et al. (4) and LogExp
(9) have comparable amplitude. The amplitude response of
jωμ0 ⎛ ⎞⎫
(1 + Δ )ln⎜⎜ − j ⋅ 1.12 ⎟⎟⎪
Log expression (8) or the Vance expression (7) is
Z gWait = systematically below all the other responses due to missing
2π ⎝ κRab ⎠⎪ wire depth term. This observation is applicable for other
⎪ ground conditions too.
κ = ε g μ0ω 2 − jωμ0σ g ⎪
ς = 2 jκd ⎪
⎬ (6)
⎡ 2 ⎤⎪
1 ⎢ K 0 (ς ) + ς K1 (ς )⎥ ⎪
Δ= ⎢ ⎥⎪
K 0 ( jκRab ) ⎢ 2 −ς ⎥


⎢ ς 2
(1 + ς )e ⎥⎪
⎣ ⎦⎭
A closed form approximation for equation (5), neglecting
the air- earth interface (infinite earth model), was proposed by
Vance [7] as shown in equation (7). Observe that equation (7)
is dependant only on the wire radius, not the burial depth,
unlike Sunde’s equation (5) and Wait’s equation (6). Note in
equation (7) H1,0(.) are Henkel functions.
ωμ0 H 0 ( jγ g Rab )
1

Z gVance =
2πγ g Rab H11 ( jγ g Rab )
(7) Figure 3a. Amplitude of Z g ( jω )−1 for comparing (4), (5), (6), (8) and (9):
ground conductivity is 1 mS/m and εrg=10.
Similarly, Petrache et al. [13] propose a simple logarithmic
approximation for the ground impedance as given by equation
(8).
jωμ 0 ⎛⎜ 1 + γ g ⋅ Rab ⎞

Z gLOG = ln (8)
2π ⎜ γ ⋅R ⎟
⎝ g ab ⎠
It was found that Vance’s equation (7) and logarithmic (8)
approximations are nearly identical, so, for simplicity, (8) is
considered here. Wait [6] mentions that infinite earth models
are applicable if 2 jd ε g μ0ω 2 − jωμ0σ g >> 1 . Hence,

neglecting the wire depth may result in inaccuracies,


especially for low frequencies. Theethayi [14] proposed a
modified empirical logarithmic-exponential approximation
(9), which is similar to equation (8) with an extra term
accounting for wire depth.

jωμ0 ⎧⎪ ⎛⎜ 1 + γ g Rab ⎞⎟ ⎡ 2e ⎤ ⎫⎪
−2 d |γ g |
Figure 3b. Argument (radians) of Z g ( jω )−1 for comparing (4), (5), (6), (8)
Z gLOGEXP = ln
⎨ ⎜ + ⎢ ⎥ ⎬ (9)
2π ⎪⎩ ⎝ γ g Rab ⎟⎠ ⎢⎣ 4 + γ g Rab ⎥⎦ ⎪⎭
2 2
and (9): ground conductivities are 10 mS/m (top) and 0.1 mS/m (bottom) and
εrg=10.
For better identification of the deviation points between
equations (4), (5), (6), (8) and (9), the amplitude and phase
response for the ratio Z g ⋅ ( jω )−1 investigated as a function of The deviation of Saad et al. formula from the other
expressions is due to the low frequency approximation
frequency for a buried wire of 2 cm radius and 0.5 m depth. adopted in [11]. As it can be seen from the figures, the low
Figures 3a and 3b show the amplitude and argument plots for frequency limit is about 0.5 MHz for 10 mS/m, 0.1 MHz for 1
various ground conductivities up to 10 MHz. The ground mS/m and 0.02 MHz for 0.1 mS/m as shown in Figure 3b.
relative permittivity and permeability are 10 and unity Note the velocity of propagation is connected to the phase
respectively. We would like to mention here that the Sunde’s quantity [7].
IEEE Transactions on Dielectrics and Electrical Insulation Vol. 14, No. 3; June 2007 755

From these observations, the use of low frequency have considered before. From Figures 4a and 4b, it can be
approximation, such as equation (4), for the ground observed that, although there is not much of a significant
impedance in conjunction with high resistivity soils may not difference in the magnitude of the ground impedance for
be valid. This has also been observed for overhead wires various ground conductivities, the magnitude of the ground
[17,22]. admittance is indeed very sensitive to the ground conductivity.
For bare and insulated wires, the ground admittance The same observations apply for phase responses as well. This
corresponding to Figs. 1 and 2 is given by equations (10) [7] is unlike overhead wires, wherein the ground impedance is
more sensitive to ground conductivity and the ground
γ g2 admittance has negligible influence [16-18].
Ygb = (10a)
Z gb
4 CONSEQUENCES OF NEGLECTING
γ g2 GROUND ADMITTANCE
Ygi = (10b)
Z gi Transient simulation packages like EMTP [12,15] ignore
the ground admittance and use low frequency approximation
of the ground impedance, i.e., they neglect displacement
currents (ground permittivity) for buried cables. The model
adopted in EMTP is similar to that proposed in [5]. For buried
bare wires, the ground admittance [13] cannot be ignored.
There aren’t any available models for counterpoises etc.,
unlike line and cable models in EMTP [12,15].
To clearly demonstrate the importance of the ground
admittance for buried cables, two cases are considered: Case-1
taking into account the ground admittance and Case-2
ignoring the ground admittance, i.e., tending YgiÆ∞ in
equation (2). The system being simulated is an insulated cable
as in Figure 1 buried at a depth of 0.5 m having a radius of 2
cm and 2 mm thickness of insulation. The ground conductivity
is 1 mS/m, the ground relative permittivity is 10, and the
insulation relative permittivity is either 2 or 5. The length of
the line is 1 km, and the current source is at one end of the line
with the other end open. We consider two types of double
exponential current source of the form I = I 0 (e − χ ⋅t − e −ϕ ⋅t ) that
Figure 4a. Ground impedance and admittance magnitude response for various
ground conductivities for bare wire buried at 0.5 m and εrg=10. has a peak value of 1 A. The exponential parameters for the
fast impulse (0.1 μs rise time) are χ=1x104 and ϕ=5x107 and
for the slow impulse (10 μs rise time) are χ=1x104 and
ϕ=4x105. The currents at distances 0 m, 150 m and 300 m
from the point of injection are plotted in Figures 5a and 5b for
fast and slow impulses, respectively.

Figure 4b. Ground impedance and admittance phase response in radians for
various ground conductivities for bare wire buried at 0.5 m depth and εrg=10.

In Figures 4a and 4b, we present both the ground impedance


and ground admittance amplitude and phase responses for Figure 5a. Importance of ground admittance, simulations for fast pulse (0.1 μs
various ground conductivities, for the same example as we rise time) ground conductivity is 1 mS/m and εrg=10.
756 N. Theethayi et al.: External Impedance and Admittance of Buried Horizontal Wires for Transient Studies

Note TL equations are solved here using equations for By comparing the results of Figures 6a and 6b with those of
general solution of the terminated line in frequency domain Figures 5a and 5b, it can be seen that the low frequency
from Tesche et al (for details see chapter 6 in [10]). The approximation of the ground impedance adopted in the
inverse Fourier transforms of the line responses were EMTP-RV is not influencing the results when the ground
evaluated in Matlab environment [23]. Authors have also admittance is neglected. Moreover, the results of Figures 6a
solved similar problems using the FDTD method with and 6b confirm that a more detailed model is needed in
recursive convolutions [24,25], in which the ground EMTP-like programs in order to improve the transient
impedance and admittance were Vector Fitted [26] as sums of analysis of buried cable.
exponentials. Both methods give similar results with
negligible error. This will be discussed in the next section.

Figure 6b. Simulations for fast pulse (10 μs rise time) ground conductivity is
Figure 5b. Importance of ground admittance simulations for slow pulse (10 μs 1 mS/m performed using the EMTP-RV program.
rise time) ground conductivity is 1 mS/m and εrg=10.

Figure 7a. Attenuation ratio for Case-1 (Yg included) and Case-2 (Yg
neglected) for insulation relative permittivity 2 & 5, ground conductivity is 1
Figure 6a. Simulations for fast pulse (0.1 μs rise time) ground conductivity is mS/m and εrg=10.
1 mS/m performed using the EMTP-RV program.
It seems that neglecting the ground admittance for
The same examples are simulated using the EMTP-RV underground cables can predict incorrect attenuation and
program [27,28], where the FD-cable model is adopted [19] velocity of wave propagation. Observe the different rise times
with the geometry as described earlier. The parameters used in and magnitudes of reflections in Figures 5a and 5b. The
the FD-cable model are the following: frequency calculation reason for this difference can be observed considering the
interval 0.01 Hz ÷ 10 MHz with 10 points per decade, the propagation aspects in frequency domain. The line
number of poles for the fitting is equal to 40. The results are propagation constant for Case-1 and Case-2 are given by
shown in Figures 6a and 6b for fast and slow impulses equations (12) and (13), respectively, in Laplace domain, i.e.,
respectively. s ⇔ jω .
IEEE Transactions on Dielectrics and Electrical Insulation Vol. 14, No. 3; June 2007 757
1 The solution of TL equations applicable to the present work
⎡ ⎛ sC ⋅ Ygi ⎞⎤ 2

γ 1 = ⎢(s ⋅ L + Z gi )⋅ ⎜⎜
by the FDTD method is very similar to the Yee approach [31]
⎟⎥ (12)
⎢⎣ sC + Y ⎟⎥ of solving the Maxwell’s equations in one dimension [32],
⎝ gi ⎠⎦ wherein solutions for electric and magnetic fields are obtained
γ 2 = [(s ⋅ L + Z gi )⋅ sC ] 2
1
(13) instead of corresponding voltages and currents. Note that
solved electromagnetic fields have to be integrated
Note that neglecting the ground admittance gives a appropriately to obtain corresponding voltages and currents.
propagation constant given by equation (13). Observe that The path dependant integration is overcome in TL solutions
when Ygi → ∞ , γ 1 → γ 2 . For either Case-1 or Case-2, the ratio since the propagation is assumed to be TEM along the
of attenuation factors α1 α 2 [‘ α ’ is the real part of ‘ γ ’] and conductor length and all the coupling terms are included in the
the velocity ratio of propagation ν 1 ν 2 , [obtained from the per unit length transient impedances and admittances, thereby
the voltages and currents are obtained directly. Researchers
ratio ω , where ‘ β ’ (phase constant), the imaginary part of
β like Noda et al. [33] and Baba and Rakov [34] have
‘ γ ’] are shown in Figures 7a and 7b, respectively, with the numerically solved full wave Maxwell’s equations, using the
medium properties same as before for examples used in FDTD method with the three dimensional Yee algorithm, for
simulating Figures 5a, 5b, 6a and 6b. different practical thin wire problems.
It is seen that above a certain frequency, the attenuation In view of the above, the FDTD method of solving TL
constant and propagation velocity with ground admittance equations, in conjunction with a recursive convolution
included are very different from the case with the ground technique, is described in detail in the Appendix.
admittance neglected. This frequency is about 1 kHz for the In order to validate the developed numerical model (see
attenuation ratio and 10 kHz for velocity ratio for 1 mS/m. Appendix A for details), differences between the application
The deviations will be even larger for poorer ground of the proposed FDTD method and the solutions of TL
conductivities. equations in frequency domain performed by means of the
method described by Tesche et al. in [10] (later transformed
back into the time domain using inverse Fourier transforms)
are illustrated. This comparison is obtained by making
reference to a 1-km long bare wire having a radius of 7.5 mm
and buried at a depth of 0.5 m. The ground electrical
parameters are σg=1 mS/m and εrg=10. The current source,
considered with a normalized amplitude of 1 A and a
( )
waveshape given by I S (t ) = 1.1274 e − (1×10 )⋅t − e − (4×10 )⋅t , is
4 5

injected in one cable termination with the other termination


open. The results obtained using the two methods relevant to
current at several observations along the cable, namely, 100
m, 200 m, 300 m, 400 m and 500 m, are shown in Figure 8.

Figure 7b. Propagation velocity ratio for Case-1 (Yg included) and Case-2 (Yg
neglected) for insulation relative permittivity 2 & 5, ground conductivity is 1
mS/m and εrg=10.

5 TIME DOMAIN SOLUTIONS


Most studies on lightning-induced voltages on power lines
use a direct time domain analysis because of its
straightforwardness in dealing with insulation coordination
problems, and its ability to handle non-linearities, which arise
in presence of soil ionization, arcing and breakdown
phenomena in grounding systems and cables.
One of the most popular approaches to solve the
transmission line coupling equations in time domain is the
finite difference time domain (FDTD) technique [29]). Such Figure 8. Calculated currents using the TL solution and the FDTD method for
technique was used indeed by Agrawal et al. in [30] when a bare wire buried at depth of 0.5 m, 1 km long and radius of 7.5 mm. The
presenting their field-to-transmission line coupling equations. ground medium has σg=1 mS/m and εrg=10.
758 N. Theethayi et al.: External Impedance and Admittance of Buried Horizontal Wires for Transient Studies

It can be seen that the results obtained by the two methods dI ( x, s ) ⎡ Y ⎤


are in good agreement. Slight differences between the two = ⎢ ⎥ ⋅ sV ( x, s ) (14b)
methods could be ascribed to numerical errors inherent in both dx ⎣s⎦
methods (i.e. inaccuracies in fitting, discritization in FDTD In equations (14a) and (14b), Z and Y represent the total
methods, averaging effects in inverse Fourier transforms, series impedance or shunt admittance of the line respectively.
etc.). The ground impedance and admittance were fitted using The corresponding time domain TL equations involving
Vector fitting [26] using 10 poles. convolutions are given by equations (15).
∂V ( x ,t ) ∂I ( x ,τ )
t

6 CONCLUSIONS + ∫ ζ (t − τ ) ⋅ dτ = 0 (15a)
∂x 0
∂τ
In this paper, the behavior of some closed-form expressions
∂I ( x ,t ) ∂V (x ,τ )
for the ground impedance are investigated for a wide t
frequency range (up to 10 MHz). A simple expression for the + ∫η (t − τ ) ⋅ dτ = 0 (15b)
ground impedance of buried bare and insulated wires is ∂x 0
∂τ
presented. This is both accurate and computationally efficient Note that in equation (15), the terms ζ(t), the so-called
for the transient analysis based on transmission line theory. It
transient resistance, and η(t), the transient conductance, are
is shown that the low frequency approximation of the ground
given by equations (16a) and (16b) respectively.
impedance in conjunction with high resistivity soils is not
⎛Z⎞
appropriate for transient analysis. It is also shown that the ζ (t ) = L−1 ⎜ ⎟ (16a)
ground admittance cannot be disregarded in transient analysis, ⎝s⎠
as it is done in commercial software such as EMTP. For
⎛Y ⎞
buried wires, the ground impedance does not significantly η (t ) = L−1 ⎜ ⎟ (16b)
vary with the ground conductivity. On the other hand, the ⎝s⎠
ground admittance is strongly dependent upon the value of the Let us assume that Z and Y when fitted with Vector
ground conductivity. This is unlike overhead wires for which s s
the ground impedance varies strongly with the ground fitting [26] using n poles say up to 10 MHz, will give
conductivity and the ground admittance can be disregarded. equations (17a) and (17b) respectively.
Transient pulse propagation in time domain based on a Z n
A
finite difference time domain (FDTD) method of solution of ≈ L0 + ∑ i (17a)
the transmission line equations is also discussed for a future s i =1 s + α i

inclusion of non-linear phenomena like soil ionization and Y n


B
arcing/breakdown mechanisms in the soil. Analysis presented ≈ C0 + ∑ i (17b)
could be useful in estimating surge propagation characteristics s i =1 s + β i

of buried wires for appropriate insulation coordination and In time domain equations (17a) and (17b) appear as
transient protection. equations (18a) and (18b), respectively.
n

ACKNOWLEDGMENT ζ (t ) ≈ L0 ⋅ δ (t ) + ∑ Ai ⋅ e −α ⋅t i
(18a)
i =1
The authors thank the financial support received from the n
Swedish National Rail Administration (Banverket) and η (t ) ≈ C0 ⋅ δ (t ) + ∑ Bi ⋅ e − β ⋅t i
(18b)
Swedish Research Council (VR Grant: 621-2005-5939). i =1
To demonstrate the final time domain expression let us
APPENDIX A. IMPLEMENTATION OF THE discuss considering only equations (18a) and (15a).
FDTD METHOD FOR THE NUMERICAL Substituting equation (18a) into equation (15a), we have,
∂V ( x ,t ) ∂I (x ,τ ) ⎫
t
SOLUTION OF THE COUPLING EQUATIONS
+ ∫ [L0 ⋅ δ (t − τ )] dτ +⎪
Concerning the time domain numerical integration of the ∂x 0
∂τ ⎪
coupling equations, we use the technique of recursive ⎬ (19)
⎤ ∂I ( x ,τ )
t
⎡ n ⎪
∫0 ⎢⎣∑ Ai ⋅ e −α i ⋅(t −τ ) ⎥
convolution described by Araneo and Celozzi [25], properly
dτ = 0 ⎪
implemented with some modifications to include both the i =1 ⎦ ∂τ ⎭
ground impedance and the ground admittance of buried wires.
In equations (19), the second term with a delta function can
In order to illustrate the implementation of the FDTD method,
be written as a convolution given by equation (20) which is
consider first the buried wire equations either (1) or (2)
the same as equation (21).
written in frequency domain as (14a) and (14b) respectively,
for voltage wave and current wave equation in Laplace [L0 ⋅ δ (t )] ∗ ∂I (x ,t ) = L−1[L0 s ⋅ I (x ,s )] (20)
domain. ∂t
dV (x, s ) ⎡ Z ⎤ ∂I (x ,t )
= ⎢ ⎥ ⋅ sI ( x, s ) (14a) L−1[L s ⋅ I (x , s )] = L
0 0 (21)
dx ⎣s⎦ ∂t
IEEE Transactions on Dielectrics and Electrical Insulation Vol. 14, No. 3; June 2007 759

It appears from equation (21) that L0 has dimensions similar conditions have to be satisfied, i.e. depending upon whether
to line series inductance and similarly, the term C0 has the voltage or current source is feeding the line or whether the
dimensions similar to shunt capacitance. Thus the final set of line is terminated in a load. For the examples treated in this
TL equations in time domain for pulse propagation that needs paper (see the conditions at the ends of the line in Figure A1),
to be solved is given by equation (22). an ideal current source IS was feeding one of the ends of the
line (source end) and at the far end of the line (load end) there
∂V ( x ,t ) ∂I ( x ,t ) ⎫ is resistive termination RL. Therefore, the recursive equations
+ L0 + ⎪
∂x ∂t ⎪ at voltage nodes 1 and node NDX+1 are given by equation
⎬=0 (22a) (23b) and (23c), respectively corresponding to Figure A1.
−α i ⋅(t −τ ) ⎤ ∂I ( x ,τ )
t
⎡ n

∫0 ⎢⎣∑ Ai ⋅ e ⎥ ∂τ dτ ⎪



⎡ C η (0) − η (Δt ) ⎤
⎢⎛⎜ 0 + ⎞ n' ⎥
i =1
⎟Vk
∂I ( x ,t ) ∂V ( x ,t ) ⎫ ⎢⎝ Δt 2 ⎠ ⎥
+ C0 + −1 ⎢ ⎥
∂x ∂t ⎪
⎪ ⎛ C η (0) ⎞ ⎢ η (Δt ) n' −1 ⎥ (23a)
⎬=0 (22b) Vkn' +1 = ⎜ 0 + ⎟ + Vk −
− β i ⋅(t −τ ) ⎤ ∂V ( x ,τ )
t
⎡ n ⎝ Δt 2 ⎠ ⎢ 2 ⎥
∫0 ⎢⎣∑i =1
Bi ⋅ e ⎥ ∂τ dτ ⎪



⎢ n' − 1 n' −
⎢ I k 2 − I k −1 2
3 ⎥

⎢ − CVkn' ⎥
It is interesting to see that equation (22) appears similar to ⎣ Δx ⎦
the overhead wires and the solutions to these equations using ⎡⎛ C0 η (0) − η (Δt ) ⎞ n' ⎤
the FDTD method are straightforward as discussed below ⎢⎜ 2Δt + 4
⎟V1 + ⎥
⎝ ⎠
−1 ⎢ ⎥
using the recursive convolution. The readers can refer to
[24,25] for details. ⎛ 0 ⎞ ⎢
C 1

⎜ + ⎟ η (Δt ) I
n' −
The FDTD method uses central difference approximations 2 Δ t ⎢ n' −1
V1 −
2
+ ⎥
V1n' +1 = ⎜ ⎟ ⎢ 1
to discretize equation (22) in terms of time and space steps. Δ ⎥ (23b)
⎜ η (0) ⎟ ⎢ 4 x

The solution to the problem is achieved by means of leap-frog
scheme [29]. In the FDTD method, there are NDX+1 voltage
⎜ ⎟
⎝ 4 ⎠ ⎢ IS + IS ( n' +1 n'
) ⎥
nodes and NDX current nodes on the line separated by a ⎢ 2Δx ⎥
length of Δx [29] as shown in Figure A1. ⎢ ⎥
⎣− CV1
n'
The ends of the line are essentially voltage nodes and every ⎦
current node is at the midpoint between two successive ⎡⎛ C0 RL ⎞ ⎤
voltage nodes. The voltage nodes are solved first and then the ⎢⎜ 2Δt − 2 ⎟ ⎥
current nodes, the recursive equations for the voltage nodes ⎢⎜ ⎟VNDX
n'
+1 ⎥
are given below and are obtained by discretizing the TL ⎢⎜ + η (0) − η (Δt ) ⎟ ⎥
−1 ⎜ ⎟
equations and then satisfying the boundary condition at the ⎛ C0 ⎞ ⎢⎝ 4 ⎠ ⎥
two end nodes. Note that the stability of FDTD method ⎜ + ⎟ ⎢ ⎥
n' +1 ⎜ 2Δt ⎟ ⎢
1
⎥ (23c)
= n' −
⎜ η (0) RL ⎟ ⎢+ η (Δt ) V n' −1 + I NDX + ⎥
depends on the time and space discritization and it must VNDX +1 2

satisfy the Courant condition i.e. Δx ≥ υ where ν is the ⎜ + ⎟ NDX +1


Δx ⎥
Δt ⎝ 4 2 ⎠ ⎢ 4
maximum phase velocity of the currents or voltages (
⎢ I n' +1 + I n'
⎢ S S ) ⎥

propagating on the line in the soil (for the present case) and ⎢ 2 Δx ⎥
the solution stability is largely dependant on this condition. ⎢− CV n' ⎥
The recursive equations for voltage and current at any node ⎣ NDX +1 ⎦
are given by equations (23) and (24) respectively. ⎡⎛ L0 ζ (0) − ζ (Δt ) ⎞ n' − 12 ⎤
⎢⎜ + ⎟I k ⎥
⎢⎝ Δt 2 ⎠ ⎥
⎛ L ζ (0) ⎞ ζ (Δt ) 2
1 −1 ⎢ 3 ⎥
n' + n' −
Ik 2
=⎜ 0 + ⎟ ⎢+ Ik − ⎥ (24)
⎝ Δt 2 ⎠ ⎢ 2 ⎥
⎢Vkn+'1 − Vkn' ⎥
⎢ − CI kn' ⎥
Δx
⎣⎢ ⎦⎥
Figure A1. FDTD annotation for the solution of voltage or current pulse The recursive relation for the voltage and current at all the
propagation along a line. nodes uses the values of ζ and η at time zero and at first time
step Δt. The convolution term CV or CI in equations (23) and
Note that voltage recursive equation (23a) is for any node (24) is calculated based on the method described in [25]. We
along the line but at the line terminations the boundary discuss here by using the term CI the convolution process; the
760 N. Theethayi et al.: External Impedance and Admittance of Buried Horizontal Wires for Transient Studies

same analogy however applies to CV as well. A convolution [17] C. A. Nucci and F. Rachidi, “Interaction of electromagnetic fields with
electrical networks generated by lightning”, Chapter 8 of The Lightning
term CI is given by equation (25) for n number of exponential
Flash: Physical and Engineering Aspects, IEE Press, London, 2003.
and constant terms depending on the vector fitting [26]. [18] W. S. Meyer and T. Liu, “Alternative Transient Program (ATP) rule
Usually 10 to 15 poles give an excellent fit for equations (17a) book, © Canadian/American EMTP user group, 1987 – 1992.
and (17b). The recursive relation for each term in equation [19] L. Marti, “Simulation of transients in underground cables with frequency
(25) at a given node and time instant is given by equation (26) dependant modal transformation matrices”, IEEE Trans. Power Delivery,
Vol. 3, pp. 1099–1110, 1988.
(the derivation of which is explained in [25]). [20] A. Ametani, “A general formulation of impedance and admittance of
n cables”, IEEE Trans. Power App. Syst., Vol. 99, pp. 902–910, 1980.
CI kn = ∑ CI kn,r
' ' [21] H. W. Dommel, “Overhead line parameters from handbook formulas and
(25) computer programs”, IEEE Trans. Power App. Syst., Vol. 104, pp. 366–
r =1 372, 1985.
[22] N. Theethayi, R. Thottappillil, Y. Liu and R. Montano, “Important
parameters that influence crosstalk in multiconductor transmission
⎛ n' 1 ⎡ n' + 1 n' − ⎤ ⎞
1
lines”, Electric Power Systems Research, (2006),
⎜ CI k ,r + Ar ⎢ I k 2 − I k 2 ⎥ + ⎟ doi:10.1016/j.epsr.2006.06.014. in press, availble online.
⎜ 2 ⎣ ⎦ ⎟ [23] The Math Works, Inc., USA.
= eα r ⋅Δt ⎜
'
CI kn,r+1 ⎟ (26) [24] A. Semlyen and A. Debulenau, “Fast and accurate switching transient
α r ⋅Δt ⎡
⎜1 n' − ⎤ ⎟
1 3 calculations on transmission lines with ground return recursive
n' −
⎜ 2 Ar e ⎢ I k − I k ⎥
2 2 convolution”, IEEE Trans. Power App. Syst., Vol. 94, pp. 561–571,
⎟ 1975.
⎝ ⎣ ⎦ ⎠ [25] R. Araneo and S. Celozzi, “Direct time domain analysis of transmission
lines above a lossy ground,” IEE Proc. Sci. Measurement Technology.,
Vol. 148, pp. 73–79, 2001.
[26] B. Gustavsen and A. Semlyen, “Rational approximation of frequency
REFERENCES domain responses by vector fitting,” IEEE Trans. Power Delivery, Vol.
[1] R. J. Fisher, G. H. Schnetzer, R. Thottappillil, V. A. Rakov, M. A. Uman, 14, pp. 1052–1061, 1999.
and J. Goldberg, “Parameters of Triggered Lightning Flashes in Florida and [27] J. Mahseredjian, S. Lefebvre, X.-D. Do, “A new method for time–
Alabama”, J. Geophys. Res., Vol. 98, pp. 22887-22902, 1993. domain modelling of nonlinear circuits in large linear networks”, Proc
[2] J. Schoene, M. A. Uman, V. A. Rakov, V. Kodali, K. J. Rambo, and G. 11th power systems computation conf. (PSCC), Vol. 2 pp. 915–922,
H. Schnetzer, “Statistical characteristics of the electric and magnetic 1993.
fields and their time derivatives 15 m and 30 m from triggered [28] J. Mahseredjian, L. Dube´ and L. Ge´rin-Lajoie, “New advances in the
lightning”, J. Geophys. Res., Vol. 108 (D6), 4192, simulation of transients with EMTP: Computation and visualization
doi:10.1029/2002JD002698, 2003. techniques”, Electrimacs, 2002.
[3] F. Pollaczek, “Sur le champ produit par un conducteur simple infiniment [29] C. R. Paul, Analysis of Multiconductor Transmission Lines, John Wiley
long parcouru par un courant alternatif”, Revue Gen. Elec. , Vol. 29, pp. and Sons Inc., 1994.
851–867, 1931. [30] A. K. Agrawal, H. J. Price and S. H. Gurbaxani, “Transient response of a
[4] E. D. Sunde, Earth conduction effects in transmission systems, Dover multiconductor transmission line excited by a nonuniform
publication, New York, 1968. electromagnetic field”, IEEE Trans. Electromagnetic Compatibility, Vol.
[5] L. M. Wedepohl and D. J. Wilcox, “Transient analysis of underground 22, pp.119–129, 1980.
power transmission systems”, Proc. IEE, Vol. 120, pp. 253-260, 1973. [31] K. S. Yee, “Numerical solution of initial boundary value problems
[6] J. R. Wait,“Electromagnetic wave propagation along a buried insulated involving Maxwell’s equation in isotropic media”, IEEE Trans.
wire”, Can. J. Phys., Vol. 50, pp. 2402-2409, 1972. Antennas and Propagation, Vol. 14, pp. 302–307, 1966.
[7] E. F. Vance, “Coupling to shielded cables”, John Wiley and Sons, 1978. [32] A. Tafflove, Computational electrodynamics: The finite difference time
[8] G. E. Bridges, “Fields Generated by Bare and Insulated Cables Buried in domain method, Artech House, 1995.
a Lossy Half-Space”. IEEE Trans. Geo. Remote Sensing, Vol. 30, pp. [33] T. Noda, R. Yonezawa, S. Yokoyama and Y. Takahashi, “Error in
140–146, 1992. propagation velocity due to staircase approximation of an inclined thin
[9] K. C. Chen, “Transient Response of an Infinite Wire in a Dissipative wire in FDTD surge simulation”, IEEE Trans. Power Delivery, Vol. 19,
Medium”, Interaction notes on EMP and related subjects Dr. C. E. pp. 1913–1918, 2004.
Baum, Editor, IN453, 2001. [34] Y. Baba and V. A. Rakov, “Voltages induced on an overhead wire by
[10] F. M. Tesche, M. Ianoz and T. Karlsson, EMC analysis methods and lightning strikes to a nearby tall grounded object”, IEEE Trans.
computational methods, John Wiley and Sons, 1997. Electromagnetic Compatibility, Vol. 48, pp. 212–224, 2006.
[11] O. Saad, G. Gaba and M. Giroux, “A closed-form approximation for
ground return impedance of underground cables”, IEEE Trans. Power
Delivery, Vol. 11, pp. 1536–1545, 1996.
[12] H. W. Dommel, “Electromagnetic Transients Program (EMTP theory
book), Bonneville Power Administration, 1986. Nelson Theethayi (S’04-M’06) was born in India, in 1975. He received the
[13] E. Petrache, F. Rachidi, M. Paolone, C. A. Nucci, V. A. Rakov and M. B.E. degree in Electrical and Electronics with first class distinction from the
A. Uman, “Lightning induced disturbances in buried cables - part I: University of Mysore and the M Sc (Engineering) degree in High Voltage
theory”, IEEE Trans. Electromagnetic Compatibility, Vol. 47, pp. 498– Engineering from the Indian Institute of Science and the PhD degree in
508, 2005. electricity with specialization in electrical transients and discharges from the
[14] N. Theethayi, Electromagnetic Interference in Distributed Outdoor Uppsala University, in 1996, 2001 and 2005, respectively. Currently he is
Electrical Systems, with an Emphasis on Lightning Interaction with employed as a researcher at the EMC group of the Division for Electricity and
Electrified Railway Network, Ph.D. Thesis, ISBN 91–554–6301–0, Lightning Research of Uppsala University. His research areas are
Uppsala University, 2005. electromagnetic compatibility, high voltage engineering, electrical power
[15] K. C. Chen and K. M. Damrau, “Accuracy of approximate transmission systems, modeling and experimental investigation of lightning phenomena and
line formulas for overhead wires”, IEEE Trans. Electromagnetic lightning interaction, analysis and design of lightning protection systems for
Compatibility, Vol. 31, pp. 396–397, 1989. power, railway and communication systems. Dr. Theethayi is a member of
[16] F. Rachidi, C. A. Nucci and M. Ianoz, “Influence of a lossy ground on Subcommittee “Lightning” of the Technical Committee TC5 of IEEE EMC
lightning-induced voltages on overhead lines”, IEEE Trans., Society, IEEE Dielectrics and Electrical Insulation Society (DEIS) and IEEE
Electromagnetic Compatibility, Vol. 38, pp. 250–264, 1996. Power Engineering Society (PES).
IEEE Transactions on Dielectrics and Electrical Insulation Vol. 14, No. 3; June 2007 761
Rajeev Thottappillil (S’88-M’92-SM’06) was born Farhad Rachidi (M’93-SM’02) was born in Geneva
in India in 1958. He received the B.Sc. degree in in 1962. He received the M.S. degree in electrical
electrical engineering from the University of Calicut, engineering and the Ph.D. degree from the Swiss
India in 1981, and the M.S. and Ph.D. degrees in Federal Institute of Technology, Lausanne, in 1986
electrical engineering from the University of Florida, and 1991 respectively. He worked at the Power
Gainesville, USA in 1989, and 1992, respectively. Systems Laboratory of the same institute until 1996.
He became an Associate Professor at Uppsala In 1997, he joined the Lightning Research Laboratory
University in 1996 and was promoted to the rank of of the University of Toronto in Canada and from
full Professor in 2000 in the subject area electricity April 1998 until September 1999, he was with
with special emphasis on transients and discharges at Montena EMC in Switzerland. He is currently the
the Division for Electricity and Lightning Research, head of the EMC Group at the Swiss Federal Institute
which is affiliated to the Department of Engineering Sciences of the Uppsala of Technology, Lausanne, Switzerland. His research interests concern
University. His research interests are lightning phenomenon, electromagnetic electromagnetic compatibility, lightning electromagnetics and electromagnetic
interference, and electromagnetic field theory. He has published more than field interactions with transmission lines. He is the convener of the joint
100 scientific articles. Prof. Thottappillil is the chairman of the EU project CIGRE-CIRED Working Group “Protection of MV and LV networks against
COST action P18 ‘Physics of Lightning Flash and its Effects.’ He is also a Lightning”, Vice-Chair of the European COST Action P18 'The Physics of
member of SC 77C of SEK, IEC on High power transients and the Lightning Flash and its Effects', Chairman of the Subcommittee “Lightning”
Subcommittee “Lightning” of the Technical Committee TC5 of IEEE EMC of the Technical Committee TC5 of IEEE EMC Society. He is member of the
Society. Editorial Board of the Journal of Lightning Research and member of the
scientific committees of various International Symposia in the field of EMC
Mario Paolone (M’07) was born in Campobasso, and lightning protection. Farhad Rachidi is author or coauthor of over 200
Italy, in 1973. Received a degree with honors in scientific papers published in reviewed journals and presented at international
electrical engineering in 1998, and the Ph.D. conferences. He was the recipient of the 2005 IEEE EMC Society Technical
degree from the same University in 2002. He is Achievement Award and of the 2005 CIGRE Technical Committee Award. In
currently working within the power systems 2006, Farhad Rachidi received the Blondel Medal from the French
group of the University of Bologna. He is Association of Electrical Engineering, Electronics, Information Technology
member of the IEEE WG on Lightning and Communication (SEE).
performance of distribution lines and of the joint
CIGRE-CIRED Working Group "Protection of
MV and LV networks against Lightning". His
research interests are power system transients,
with particular reference to LEMP-interaction with electrical networks, power
systems dynamics, power system protections and distributed generation.

Carlo Alberto Nucci (M’91-SM’02-F’07) was


born in Bologna, Italy, in 1956. He received the
M.S. (hons.) and Ph.D. degrees in electrical
engineering from the University of Bologna, Italy,
in 1982 and 1986, respectively. At the same
university he was researcher at the Power Electrical
Engineering Institute since 1983, became an
associate professor in 1992 and in 2000 became full
professor and chair of power systems. He is author
or co-author of more than 200 scientific papers
published on reviewed journals or presented at
international conferences. He is member of the IEEE Working Group
'Lightning performance of Distribution lines’; in CIGRE he serves as
chairman of the Study Committee C4 ‘System Technical performance’ and is
member of CIGRE Working group C.401 'Lightning', of which he is also
convener. He is also Fellow of the IET. His research interests concern power
systems transients and dynamics, with particular reference to lightning impact
on power lines, system restoration after black-out and distributed generation.
He is the chair of International Steering Committee of the IEEE PowerTech
and of the IEEE PES Italian Chapter PE31 in Region 8. Since January 2005 he
is regional editor for Africa and Europe of the Electric Power System
Research journal, Elsevier.

You might also like