Electric Power Systems Research 181 (2020) 106173

Contents lists available at ScienceDirect

Electric Power Systems Research

journal homepage: www.elsevier.com/locate/epsr

Transient analysis of compact distribution lines with dielectric-coated phase T

cables
Alberto De Contia,*, Alex C. Silvab, Osis E.S. Lealb,c
a
LRC – Lightning Research Center, Department of Electrical Engineering, UFMG – Federal University of Minas Gerais, Av. Antônio Carlos, 6627, Pampulha, Belo
Horizonte, MG, 31.270-901, Brazil
b
PPGEE – Graduate Program of Electrical Engineering, UFMG – Federal University of Minas Gerais, Av. Antônio Carlos, 6627, Pampulha, Belo Horizonte, MG, 31.270-
901, Brazil
c
UTFPR - Federal University of Technology – Paraná, Via do Conhecimento, km 1, Pato Branco, PR, 85.503-390, Brazil

ARTICLE INFO ABSTRACT

Keywords: This paper studies wave propagation characteristics and the transient response of single-phase and three-phase
Compact distribution lines compact distribution lines. Although compact line configurations present bare and dielectric-coated cables in
Dielectric-coated cables relative proximity, it is shown that their external per-unit-length parameters can be calculated using simplified
Proximity effect expressions valid for widely-spaced cables. An attempt is made to identify whether the insulating layer sur-
Propagation characteristics
rounding the phase cables must be included for an accurate simulation of switching and lightning transients on
Transient analysis
compact distribution lines. The results show that the dielectric coating reduces the propagation velocity of
certain modes to values slightly below the speed of light. However, this reduction does not have a significant
influence on the initial peak values of the calculated switching and lightning overvoltages. Only the wave tails of
the resulting voltage waveforms are more strongly modified depending on the terminal conditions of the line.

1. Introduction line parameters [8,9], which needs validation. In Ref. [10], a first at-
tempt was made to identify the influence of the insulating layer sur-
In recent years there has been an increasing use of dielectric-coated rounding the phase cables on the propagation characteristics of a
phase cables on overhead distribution lines. This practice, which has compact distribution line. However, the analysis was restricted to a
been used in countries like Japan for decades [1,2], has the advantage three-phase configuration and to the calculation of overvoltages caused
of reducing the number of faults due to unintended contact with trees by a direct lightning strike.
and extraneous objects, increasing the insulation level of the line, and This paper extends the work presented in Ref. [10] by investigating
allowing clearances to be reduced [2–6]. This has led to the proposition switching and lightning transients on both single-phase and three-phase
of compact distribution line configurations that have been used as an compact distribution lines. The adopted methodology consists in first
alternative to conventional lines consisting of bare cables [3]. investigating the influence of the insulating layer surrounding the phase
Despite the growing interest on the use of dielectric-coated phase cables on the per-unit-length parameters of these lines. This includes
cables in distribution lines, the literature dedicated to study the tran- determining if simplified formulas that are valid for widely-spaced
sient performance of compact distribution lines is relatively scarce cables can be used in such calculation. The particular case of a single
[7–11]. In fact, no comprehensive analysis exists on the transient re- overhead cable is also discussed for reference. Then, the influence of the
sponse of these lines when excited by switching impulses and lightning insulating layer surrounding the phase conductors on the propagation
currents. Compact distribution lines form an inhomogeneous system velocity associated with each transmission line mode is investigated.
with bare and insulated cables in relative proximity. However, in Finally, it is discussed to what extent neglecting the insulating layer
transient studies it is common practice to neglect the insulating layer covering the phase cables can be considered acceptable when simu-
surrounding the phase conductors for simplicity (e.g., [7–9]). This lating electromagnetic transients on compact distribution lines. This is
means that compact distribution lines are often simulated as a system of performed by simulating lightning and switching transients on compact
bare conductors, even though the validity of this assumption is unclear. lines using an FDTD code to solve telegrapher’s equations for different
Also, the proximity effect is usually neglected in the calculation of the terminal conditions. The presented analysis provides elements for a

⁎
Corresponding author.
E-mail address: [email protected] (A. De Conti).

https://doi.org/10.1016/j.epsr.2019.106173
Received 2 April 2019; Received in revised form 15 July 2019; Accepted 16 December 2019
0378-7796/ © 2019 Elsevier B.V. All rights reserved.
A. De Conti, et al. Electric Power Systems Research 181 (2020) 106173

Table 2
Details of the systems of conductors shown in Fig. 1.
Cable Core Insulation εr (insulating DC resistance (Ω/
radius thickness layer) km)
(mm) (mm)

A, B, C 4.10 3.05 2.3 0.822

M 4.75 – – 4.5239
N 4.50 – – 0.7482

Fig. 1. Simulated line configurations.

3. Per-unit-length parameters

more rigorous simulation of electromagnetic transients on compact 3.1. Formulation

distribution lines. It also contributes to a more clear definition of si-
tuations in which the insulating layer surrounding the phase conductors The per-unit-length series impedance Z and shunt admittance Y of a
can be possibly neglected in the modeling of such lines. transmission line with n overhead conductors above a lossy ground can
This paper is organized as follows. Section 2 presents the simulated be written as [12]
line configurations. The calculation of per-unit-length parameters con- Z = Zi (s ) + Zg (s ) + sLe (1)
sidering an inhomogeneous system of bare and dielectric-coated cables
in relative proximity is discussed in Section 3. The influence of cable Y = G + sC (2)
insulation on the modal velocities associated with compact distribution
where s is the Laplace variable, Zi (s ) is a diagonal matrix containing
lines is investigated in Section 4. Section 5 presents transient simulation
the internal impedance of the conductors, Zg (s ) is the ground-return
results, followed by conclusions in Section 6.
impedance, Le is the external inductance, G is a diagonal matrix con-
taining the line conductance, and C is the line capacitance, all of order
n × n. In this paper, the elements of Zi (s ) are calculated using the
2. Line configurations
approximate expression given in Ref. [12] for the particular case of a
solid circular conductor, Zg (s ) is calculated with Sunde’s expressions
Three line configurations are considered. The first one is the over-
[13], and the elements of G are assumed equal to 18.64 nS/km, which is
head cable shown in Fig. 1(a), which consists of a solid aluminum core
in the range of default conductance values used in electromagnetic
covered with a layer of cross-linked polyethylene (XLPE). The second
transients simulators [14]. For widely-spaced insulated cables, the ca-
one is the single-phase compact distribution line illustrated in Fig. 1(b).
pacitance matrix can be written as [15]
In this configuration, the topmost conductor is a steel cable known as
messenger (M), which provides mechanical support to an XLPE-covered C = (P0 + P ) 1
(3)
cable (phase A) through a set of regularly-distributed polymeric
where the entries in P0 and P are given by [15]
spacers. In practice, the messenger is connected to the neutral wire (N)
at every grounding point. The final configuration is the three-phase 1 2h i 1 Dij
compact distribution line shown in Fig. 1(c). The messenger cable now P0 (i, i) = ln , P0 (i, j) = ln
2 0 ri + i 2 0 dij (4)
provides mechanical support to three XLPE-covered phase cables (la-
beled A, B, and C in the figure), being again connected to the neutral at 1 ri + i
every grounding point. The vertical and horizontal coordinates of the P (i , i ) = ln , P (i , j ) = 0
2 ri 0 ri (5)
cables shown in Fig. 1 are listed in Table 1. Their electrical parameters
are given in Table 2. Further details about compact distribution lines In (4) and (5), the subscripts i and j refer to the i-th and j-th cables,
can be found in Refs. [3–6]. respectively, hi is the cable height, ri is the external conductor radius, i
The compact line configurations shown in Fig. 1(b) and (c) corre- is the insulation thickness, ri is the relative permittivity of the in-
spond to inhomogeneous systems of conductors consisting of bare and sulating material, 0 is the vacuum permittivity, dij is the center-to-
dielectric-coated cables. This paper clarifies whether neglecting the center separation between cables, and Dij is the center-to-center se-
insulating layer covering the phase cables is a safe assumption for the paration between the i-th cable and the image of the j-th cable. If i = 0 ,
calculation of lightning and switching transients on such lines. It is also the insulating layer vanishes, leading to P = 0 in (3). In this case, C
determined whether the per-unit-length parameters associated with reduces to the capacitance of a bare cable, that is, C = C0 = (P0 | i= 0 ) 1
both single-phase and three-phase compact distribution lines can be [15]. Since the insulating material presents no magnetic properties, the
calculated without considering the proximity effect. external inductance Le can be calculated as Le = µ 0 0 C0 1 [16], where
µ 0 is the vacuum permeability. The approximate expressions (3)–(5) are
available, for example, in electromagnetic transient simulators such as
the Alternative Transients Program (ATP) [17].

Table 1
3.2. Influence of insulating layer and proximity effect on C
Coordinates of the systems of conductors shown in Fig. 1 (z is the conductor
height, y is the horizontal displacement).
The calculation of C with the approximate expressions (4) and (5) is
Cable Single-cable line Single-phase compact Three-phase compact line
only accurate for a system of widely-spaced cables because they are
line
derived assuming uniform charge distribution around the conductor
y(m) z(m) y(m) z(m) y(m) z(m) periphery. This does not pose any problem for the case illustrated in
A 0 9.00 0 8.70 −0.095 8.83 Fig. 1(a), which consists of a single overhead cable with hi (ri + i ) .
B – – – – 0 8.67 However, extra care must be taken in the calculation of the per-unit-
C – – – – 0.095 8.83
length parameters of the remaining configurations because the cables
M – – 0 9.00 0 9.00
N – – −0.20 7.00 −0.354 7.00 are relatively close.
To investigate if the proximity effect is significant enough to modify

2
A. De Conti, et al. Electric Power Systems Research 181 (2020) 106173

Table 3 Table 5
Capacitance of the single-cable line of Fig. 1(a). Capacitance matrix elements of the three-phase compact line of Fig. 1(c).
Element Approximate Expressions MoM Elements Approximate Expressions MoM

C0 C ΔC C0 C ΔC C0 C ΔC C0 C ΔC
(pF/m) (pF/m) (%) (pF/m) (pF/m) (%) (pF/m) (pF/m) (%) (pF/m) (pF/m) (%)

AA 6.6239 6.8818 3.89 6.6239 6.8818 3.89 MM 11.2823 11.4225 1.24 11.2858 11.4288 1.27
AA 11.7788 12.7395 8.16 11.7809 12.7436 8.17
BB 11.1779 12.0720 8.00 11.1806 12.0778 8.02
CC 11.7719 12.7318 8.15 11.7749 12.7376 8.18
C (and consequently Le ), a methodology based on the method of mo-
NN 7.7229 7.7339 0.14 7.7230 7.7341 0.14
ments (MoM) that is able to calculate the external parameters of a MA −3.2490 −3.3936 4.45 −3.2492 −3.3933 4.43
system of bare and insulated cables with arbitrary location above a MB −1.6514 −1.6317 −1.19 −1.6538 −1.6365 −1.05
perfectly conducting ground was employed by the authors [18]. This MC −3.2526 −3.3973 4.45 −3.2541 −3.3998 4.48
methodology is an extension of the formulation presented in Ref. [19], MN −0.7119 −0.6764 −4.99 −0.7119 −0.6764 −4.99
AB −3.1981 −3.6013 12.61 −3.1985 −3.6023 12.62
which was originally formulated for the particular case of a linear array
AC −2.7075 −3.0399 12.28 −2.7078 −3.0413 12.32
of insulated cables in free space. It was validated in Ref. [18] through AN −0.7133 −0.7337 2.86 −0.7127 −0.7343 3.03
comparisons with the finite element method (FEM). In MoM, the charge BC −3.2029 −3.6067 12.61 −3.2031 −3.6073 12.62
densities on the conductor and dielectric surfaces are expressed as a BN −0.9558 −0.9928 3.87 −0.9557 −0.9925 3.85
CN −0.6750 −0.6918 2.49 −0.6752 −0.6922 2.52
Fourier series that depends on the peripheral angle. Each element of
charge contributes to the electric field and to the electric potential at
arbitrary points inside and outside the charge distribution. By super-
Since Zm and Ym are diagonal, the modal propagation constant
imposing the effect of the charge distributions, closed-form expressions
m = Ym Zm is also a diagonal matrix. The real and imaginary parts of
can be written for the potentials and electric fields [19]. The solution is
m correspond to the attenuation and phase constants associated with
obtained by taking into account the appropriate boundary conditions
each mode [12,16]. The modal velocity m can then be calculated as
and considering the point-matching method [18,19].
[12,16]
The elements of the capacitance matrices associated with the con-
figurations shown in Fig. 1 calculated with either the approximate ex- m = [Im{ Ym Zm }] 1
(6)
pressions (3)–(5) or the method of moments are listed in Tables 3–5. In
order to indicate the relative influence of the insulating layer on the where is the angular frequency and the operator Im{⋅} extracts the
capacitance, the percentage difference ΔC associated with the inclusion imaginary part of its argument. For the single-cable line of Fig. 1(a), the
of the cable insulation was also calculated. By comparing the results transformation from phase to modal quantities was not necessary be-
obtained with the method of moments and the approximate solution, it cause only one mode exists. In this case, Eq. (6) was used directly, except
is seen that the proximity effect has a negligible influence on the cal- that Zm and Ym are now scalars respectively corresponding to Z and Y .
culation of C. This means that (3)–(5) can be safely considered in the Fig. 2 illustrates m associated with the line configurations of Fig. 1
determination of the external parameters of the line configurations assuming either an inhomogeneous system with dielectric-coated phase
shown in Fig. 1. It is also seen that the consideration of dielectric-coated cables, or a homogeneous system with bare cables only. In the calcu-
phase cables introduce variations up to 3, 5 and 8% in the self-elements lations, a ground resistivity g = 100 m and a ground relative permit-
of C for the line configurations shown in Figs. 1(a)–(c), respectively. tivity rg = 10 were considered. The results show that for a system of
The variations in the off-diagonal elements can reach up to about 12% bare cables the phase velocity approaches the speed of light c as the
in the case of the three-phase compact line configuration. frequency increases. In all cases, the presence of the insulating layer
leads to a reduction in the phase velocities associated with some modes
to values ranging from 0.95c to 0.97c. The most critical case corre-
4. Influence of cable insulation on phase velocity
sponds to the three-phase compact line, in which only two out of five
modes approach c. For this particular line, two of the three remaining
To investigate the influence of the insulating layer covering the
modes are coincident in the considered frequency range, so only four
phase cables on the propagation characteristics of the line configura-
curves are seen for each simulated condition.
tions shown in Fig. 1, it is convenient to write Z and Y in their modal
The relatively small reduction of the modal phase velocity is related
domain equivalents Zm and Ym . This requires performing the matrix
to the minor variations observed in the capacitances shown in Tables
products Zm = T tZT and Ym = T 1Y (T 1)t , where T , T t and T 1 are the
3–5 due to cable proximity and insulation. Indeed, cable separation and
transformation matrix, its transpose, and its inverse, respectively. The
height are both ultimately much larger than the external radius of the
columns of T contain the eigenvectors of YZ [12,16]. For calculating T
cables including the insulation. In addition, the thickness of the in-
avoiding element switchovers, the Newton-Raphson technique pro-
sulating layer has the same order of magnitude of the conductor radius
posed in Ref. [20] was used.
and its relative permittivity is only about twice as large as that of the
air. As a consequence, most of the energy associated with the propa-
Table 4
gating waves is likely to travel in the air.
Capacitance matrix elements of the single-phase compact line of Fig. 1(b).
To further discuss this idea, a simple analysis can be performed by
Elements Approximate Expressions MoM assuming a lossless line consisting of a single cable with height h and
core radius r covered with insulating material with thickness and
C0 C ΔC C0 C ΔC
(pF/m) (pF/m) (%) (pF/m) (pF/m) (%)
relative permittivity r . In this case, m in (6) reduces to = (Le C ) 1/2 ,
which is now invariant with frequency [16]. By manipulating (3) and
MM 9.0816 9.1824 1.11 9.0820 9.1831 1.11 Le = µ0 0 C0 1 for the scalar case, it is shown in the Appendix A that
AA 9.0374 9.5244 5.39 9.0378 9.5249 5.39
NN 7.6134 7.6256 0.16 7.6135 7.6257 0.16 c
MA −4.1122 −4.3338 5.39 −4.1125 −4.3343 5.39
=
MN −1.2094 −1.1743 −2.90 −1.2094 −1.1744 −2.89 r (7)
AN −1.4296 −1.5067 5.39 −1.4296 −1.5066 5.39
where

3
A. De Conti, et al. Electric Power Systems Research 181 (2020) 106173

5. Time-domain analysis

5.1. Modeling

In order to study the influence of dielectric-coated cables on the

propagation of transient voltages on multiconductor transmission lines,
a rigorous phase-domain solution of telegraphers’ Eq. (9) was per-
formed using the 1st order FDTD method discussed in Ref. [16]. In
these equations, v (x , t ) and i (x , t ) are n × 1 vectors containing voltages
and currents at coordinate x along the line at time t, (t ) is the transient
impedance given by (10), ‘*’ is the convolution integral, and L 1 { } is
the inverse Laplace transform operator. The convolution in (9) is solved
recursively by representing each element of (t ) as a sum of exponential
functions [16]. For this, [Zi (s ) + Zg (s )] s 1 was fitted in frequency do-
main using the vector fitting technique assuming a single set of stable
poles [21,22]. No passivity enforcement procedure was required to
assure a strictly passive behavior for (t ) .

v (x , t ) i (x , t ) i (x , t )
= Le + (t )*
x t t

i (x , t ) v (x , t )
= Gv (x , t ) + C
x t (9)

1
Zi (s ) + Zg (s )
(t ) = L
s (10)

As shown in Ref. [11], the use of a full phase-domain transmission

line model such as the one considered in this paper is necessary for a
Fig. 2. Phase velocity associated with the line configurations of Fig. 1. (a) rigorous calculation of electromagnetic transients in compact distribu-
Single-cable line; (b) single-phase compact line; (c) three-phase compact line. tion lines. Alternatively, the universal line model ULM [23] could have
been used instead, leading to equivalent results. However, the FDTD
1 method was preferred because only the fitting of (t ) is necessary for
2h 2h solving telegrapher’s equations, whereas the ULM requires the fitting of
r = ln ln 1 1
r r r (r + )1 r
(8) both the propagation function and the characteristic admittance of the
line [23].
is the effective relative permittivity (1 r r ) that would be necessary
to reduce the propagation speed of an overhead bare cable immersed in
5.2. Results
an homogeneous medium to a value equivalent to that of a dielectric-
coated cable immersed in air. Fig. 3 illustrates the ratio k = / c = 1/ r 5.2.1. Single-cable line
as a function of r and calculated from (7) and (8) assuming h = 9 m To investigate the transient response of a dielectric-coated overhead
and r = 4.1 mm. As expected, k reduces with increasing the thickness cable considering a wideband excitation, a 1-V step voltage was applied
and the relative permittivity of the insulating layer. However, a satura- at the sending end of the cable of Fig. 1(a) by an ideal voltage source.
tion trend is observed for k as both and r increases. This means that the The line was either left open-ended or connected to a 500-Ω resistance
propagation speed reduction resulting from covering an overhead cable that approaches its characteristic impedance if losses are neglected.
with an insulating layer is likely to have a practical limit. Similar trend is Three conditions were tested, namely a bare cable, a dielectric-coated
expected in an overhead multiconductor line containing multiple di- cable with r = 2.3 and = 3.05mm (as per Table 2), and a dielectric-
electric-coated cables, except that the reduction in the propagation speed coated cable with hypothetical values of r and equal to 10 and 6.1
will also be affected by the cable separation. mm, respectively. In the simulations, g = 100 m and rg = 10 were
considered. The voltages calculated at the receiving end of the line are
shown in Fig. 4.
It is observed in Fig. 4 that the inclusion of an insulating layer sur-
rounding the cable does not have a significant influence on the resulting
waveforms for the line terminated at a 500-Ω resistance, except for a
minor variation in the initial peak voltage. For the open-ended line, the
initial peak voltage remains nearly unchanged after the inclusion of the
insulating layer, regardless of the values of r and . However, the os-
cillations caused by the successive reflections at the line ends are strongly
affected by the propagation velocity reduction caused by the dielectric
coating. This can be justified by analyzing the period of the oscillations
seen in Fig. 4(b), which can be calculated as T = 4l / , where l is the line
length and is given by (7). For a bare cable, approaches c and as a
consequence T reaches the minimum value Tmin . For a dielectric-coated
Fig. 3. Reduction of propagation speed (compared to the speed of light) asso- cable, the oscillation period rises to T = r Tmin . This explains the time
ciated with the line configuration of Fig. 1(a) due to the increase of the relative delay observed on the tail of the waveforms shown in Fig. 4(b), which
permittivity of the insulating material for different insulation thicknesses. increases with increasing the values of r and .

4
A. De Conti, et al. Electric Power Systems Research 181 (2020) 106173

Fig. 4. Voltages at the receiving end of a single-cable line for the application of Fig. 6. Voltages between phase and neutral at the receiving end of a 1-km long
a unit-step voltage at the sending end. (a) 500-Ω load at receiving end; (b) open single-phase compact line for the condition shown in Fig. 5(a) assuming
circuit at the receiving end. vs (t ) = u (t ) , is (t ) = 0 , RS = 0 and RG = 50 Ω: (a) RL = 500 Ω; (b) RL = .

5.2.2. Single-phase compact line tail when successive reflections take place at the line ends.
The analysis is now performed for the single-phase compact line Another relevant analysis that can be performed to investigate the
shown in Fig. 1(b), which is assumed to have a length of 1 km. A unit- transient response of the line considering or not the presence of a di-
step voltage is applied by a voltage source vs (t ) with internal resistance electric-coated phase cable consists in studying overvoltages due to a
RS = 0 at the sending end of the phase cable, which is either left open or lightning strike at the messenger cable. This analysis is of practical
connected to a resistor RL = 500 at the receiving end. To reflect the interest because the messenger is a bare cable at the topmost position of
standard practice of bonding neutral and messenger at every grounding the structure shown in Fig. 1(b). This could put it as a preferential point
point, cables M and N in Fig. 1(b) were shorted and grounded through a for the attachment of direct lightning strikes [2]. For this particular
pair of resistors RG = 50 at the line terminations. The simulated analysis, the current source is (t ) in Fig. 5(a) was assumed to inject a
configuration is shown in Fig. 5(a), where is (t ) = 0 . Fig. 6 illustrates the waveform with time characteristics that reproduce median data of
phase-to-neutral voltages calculated at the receiving end of the line subsequent return-stroke currents measured at Mount San Salvatore
assuming two different conditions: (i) all cables are bare; (ii) the phase [24]. The current waveform was modeled as a sum of two Heidler
cable is dielectric coated. The simulations considered the cable data functions [25] as
presented in Tables 1 and 2. 2
The results shown in Fig. 6 are similar to those obtained for a single- Im t (t / 1m )nm
i (t ) = exp
phase cable, except that the voltage waveforms are now influenced by m=1 m 2m 1 + (t / 1m )nm (11)
the presence of two grounded cables running parallel to phase A. If
RL = 500 , an oscillatory behavior is introduced at early times due to 1m nm 2m
1/ nm
= exp
successive reflections taking place on messenger and neutral [see m
2m 1m (12)
Fig. 6(a)]. If RL = , a reduction is observed in the phase-to-neutral
voltages compared to the case where messenger and neutral are absent where I1= 10.7 kA, n1= 2, 11= 0.25 μs, 21= 2.5 μs, I2= 6.5 kA, n2= 2, 12=
[see Fig. 6(b)]. The presence of dielectric coating around the phase 2.1 μs, and 22= 230 μs [26,27]. For comparison purposes, its amplitude
cable does not significantly change the initial phase-to-neutral voltage, was normalized to −1 kA. A direct lightning strike with realistic peak
but leads to an increase in the period of the oscillations observed on its value would inevitably lead to overvoltages exceeding typical break-
down values associated with compact distribution lines, which vary
between 200 kV and 500 kV depending on structure type, polarity of
the applied impulse voltage, and insulating material covering the cable
[4–6].
Phase-to-neutral voltages calculated at the receiving end of the line
for the injection of the considered lightning current waveform into the
messenger cable at the sending end are shown in Fig. 7. In the simu-
lations, vs (t ) = 0 was considered in the equivalent circuit shown in
Fig. 5(a). Two different line termination conditions were tested. One
considered the hypothetical case of a floating phase cable
(RL = RS = ), while the other considered a nearly matched sending
end (RS = 500 ) with the receiving end open (RL = ). In all cases,
RG = 50 was assumed. Once again, the simulations were performed
either considering or neglecting an insulating layer covering the phase
cable. The results exhibit the same behavior identified in the previous
analysis: considering a dielectric-coated phase cable does not change
the initial phase-to-neutral voltage significantly, but its effect is re-
markable whenever successive reflections of significant amplitude take
place due to impedance discontinuities at the line ends. This is the case
Fig. 5. Simulated cases. of the hypothetical condition RL = RS = illustrated in Fig. 7(b).

5
A. De Conti, et al. Electric Power Systems Research 181 (2020) 106173

dielectric-coated phase cables present a spike at early times that is

absent in the waveforms calculated assuming bare cables. This behavior
was not observed when studying the transient response of the single-
cable and single-phase compact lines to a unit-step excitation, which is
explained as follows. For the single-cable and single-phase compact
lines, the voltage source vs (t ) was connected at the sending end of phase
A and the resulting voltages were calculated at the receiving end of this
same phase. This means that the bulk of the energy arriving at the
sending end of the cable is due to the direct cable excitation by the
source, with some interference from voltages induced by the messenger
and neutral (when present). In the cases shown in Fig. 8, however,
phase-to-neutral voltages on phase B are induced by electromagnetic
coupling. If dielectric-coated cables are used, some modes will travel
slower than the speed of light at high frequencies while others will
travel virtually at c [see Fig. 2(c)]. The induced voltages on phase B will
therefore be the composition of voltage waves that do not arrive si-
multaneously at the line termination. This will cause the spikes seen in
Fig. 8. Although not shown, similar behavior was observed in the phase-
Fig. 7. Voltages between phase and neutral at the receiving end of a 1-km long to-neutral voltages associated with phase C, but not on phase A, which
single-phase compact line for the condition shown in Fig. 5(a) assuming is the one excited by vs (t ) . In fact, phase-to-neutral voltages associated
vs (t ) = 0 and RG = 50 Ω and the injection of subsequent stroke current at the with phase A do not present an early spike, behaving similarly as the
sending end of the messenger. (a) RS = 500 Ω, RL = ; (b) RS = RL = . waveforms shown in Figs. 4 and 6.
The analysis is now repeated assuming the injection of the sub-
Otherwise, the effect of the cable insulation on the overall waveform sequent return-stroke current described in previous section at the
characteristics is nearly negligible, as shown in Fig. 7(a) for the case sending end of the messenger cable. Once again, the injected current
RS = 500 RL = . waveform has a normalized peak value of -1 kA and time characteristics
that reproduce the median parameters of subsequent return-stroke
5.2.3. Three-phase compact line currents measured at Mount San Salvatore. The simulated case is shown
The final analysis refers to the three-phase compact distribution line in Fig. 5(b), where vs (t ) = 0 and RG = 10 are considered. In the si-
configuration shown in Fig. 1(c). The simulations were performed mulations, phase A was terminated at both ends with 500 Ω resistors,
taking as reference the equivalent circuit shown in Fig. 5(b). As before, phase B was connected to RSB = 500 and RLB = , and phase C was
neutral and messenger were shorted and grounded at the line ends, but left floating with both ends open (RSC = RLC = ) . The simulations
now RG is reduced to 10 Ω to increase the amplitude of reflected vol- were performed including or neglecting the presence of an insulating
tages at the line terminations. A unit-step voltage was applied at the layer surrounding the phase cables. The phase-to-neutral voltages cal-
sending end of phase A by voltage source vs (t ) , which has internal re- culated at the receiving end of phases A, B and C are shown in Fig. 9.
sistance RSA = 0 . Phases B and C were grounded at the sending end with It is seen in Fig. 9 that the phase-to-neutral voltages associated with
RSB = RSC = 500 to emulate a matched condition. The receiving end phases A and B are nearly insensitive to the consideration of the di-
of phases A, B, and C was either left open or grounded through resistors electric coating. Also, the waveforms do not present the spikes observed
RLA = RLB = RLC = 500 . Fig. 8 illustrates the phase-to-neutral voltages for the unit-step excitation of the line at early times. This is related to the
calculated at the receiving end of phase B assuming either that all fact that the rising part of the injected lightning current is not as steep as
cables are bare or that the phase cables are dielectric-coated. The si- that of the step voltage. As a consequence, the non-simultaneous arrival
mulations considered data listed in Tables 1 and 2. of different mode voltages at the line termination is less critical for the
It is observed in Fig. 8 that the waveforms calculated considering composition of the resulting waveform than for the unit-step excitation.
The effect of considering dielectric-coated phase cables is more re-
markable on the tail of the phase-to-neutral voltages associated with phase
C. The successive reflections taking place at the open line ends present
amplitudes that are much larger than those observed if bare cables are
assumed. This result confirms that whenever reflections of significant
magnitude are expected due to impedance discontinuities at the line ends,
the simulation of the transient response of compact distribution lines as-
suming bare cables might lead to greater deviations on the tail of the re-
sulting voltages. Nevertheless, the initial peak voltages are not sig-
nificantly affected by such consideration, at least for the injection of
typical lightning currents. This result is in line with the analysis presented
in Ref. [10], which considered different line terminations, different current
injection modes, and a different lightning current waveform.

6. Conclusions

This paper investigates the influence of dielectric-coated phase

conductors on per-unit-length parameter calculation and wave propa-
gation characteristics of three different overhead distribution line
Fig. 8. Voltages between phase B and neutral at the receiving end of a 1-km configurations. Regarding per-unit-length parameter calculation, sim-
long three-phase compact line for the condition shown in Fig. 5(b) assuming plified expressions valid for widely-spaced cables were shown to be
vs (t ) = u (t ) , is (t ) = 0 , RSA = 0 , RSB = RSC = 500 Ω, and RG = 10 Ω: (a) valid for determining the external parameters of single- and three-phase
RLA = RLB = RLC = 500 Ω; (b) RLA = RLB = RLC = . compact distribution lines. Also, the elements of the capacitance matrix

6
A. De Conti, et al. Electric Power Systems Research 181 (2020) 106173

frequencies. A greater reduction is obtained when the thickness and the

relative permittivity of the insulating layer covering the phase cables
are increased, but a saturation trend is observed. For the tested con-
figurations, the presence of the insulating layer was seen to reduce the
phase velocities to values not lower than 0.95c.
Regarding the simulation of switching and lightning transients on
the compact distribution line configurations, it was shown that the
consideration of dielectric-coated phase cables is expected to modify
the resulting voltage waveforms whenever successive reflections of
significant magnitude take place due to impedance discontinuities at
the line ends. In this case, the period of oscillation increases with the
use of dielectric-coated cables. In most of the investigated cases, the
influence of dielectric-coated cables on initial peak values was seen to
be negligible. The only exception was the calculation of induced vol-
tages on phases B and C of the three-phase compact distribution line
due to a unit-step excitation at phase A.
In summary, a rigorous simulation of electromagnetic transients on
compact distribution lines will inevitably require the consideration of
the insulating layer surrounding the phase cables. However, in most
cases it seems possible to obtain sufficiently accurate estimates for in-
itial peak voltages if the insulating layer is neglected for simplicity. This
is particularly true for the simulation of lightning overvoltages.

Declaration of competing interests

The authors declare that they have no known competing financial

interests or personal relationships that could have appeared to influ-
Fig. 9. Voltages between phases (a) A, (b) B, (c) C and neutral at the receiving
ence the work reported in this paper.
end of a 1-km long three-phase compact line for the injection of a subsequent
stroke current at the sending end of the messenger. Simulated case, taking as
reference Fig. 5(b): vs (t ) = 0 , RG = 10 Ω, RSA = RLA = 500 Ω, RSB = 500 Ω, Acknowledgments
RLB = , and RSC = RLC = .
This work was supported in part by Brazilian agencies CNPq
of the tested configurations were seen to experience variations up to (National Council for Scientific and Technological Development) (grant
12% when dielectric-coated phase cables are considered. 304117/2016-1), FAPEMIG (Fundação de Amparo à Pesquisa do Estado
The use of dielectric-coated phase cables was shown to reduce the de Minas Gerais) (grant TEC-PPM-00280-17), and CAPES (Coordenação
propagation velocity to values below the speed of light at high de Aperfeiçoamento de Pessoal de Nível Superior) (Finance Code 001).

Appendix A

For a lossless single-phase line the propagation velocity is given by = (Le C ) 1/2
[16]. Since Le = µ0 0 P0 | =0 and C = (P0 + P ) 1 [15], using (4)
and (5) one can write as

1 2 0 1 2h 1 r+
= ln + ln
µ0 0 ln
( ) 2h
r
2 0 r+ 2 r 0 r
(A.1)
Simplifying and rearranging the terms in (A.1), one obtains

( ) + ln ( )
2h r+ 1/ r
1 ln r+ r
=
µ0 ln ( )
2h
0
r (A.2)
which is equivalent to

2h (r + )1/ r
ln
1 (r + ) r 1/ r
=
µ0 0 ln ( ) 2h
r (A.3)
This expression can be written as

2h
ln 1 1
1 rr (r + )1 r
=
µ0 0 ln ( )
2h
r (A.4)

Since c = (µ 0 0) 1/2 , it is possible to write (A.4) as = c/ r, where r is given by (8).

7
A. De Conti, et al. Electric Power Systems Research 181 (2020) 106173

References Transmission, Perpignan, 2019, pp. 1–6.

[12] A. Ametani, N. Nagaoka, Y. Baba, T. Ohno, Power System Transients – Theory and
Applications, CRC Press, 2014.
[1] M. Washino, A. Fukuyama, K. Kito, K. Kato, Development of current limiting arcing [13] E. Sunde, Earth Conduction effects in transmission systems, 2nd ed., Dover
horn for prevention of lightning faults on distribution lines, IEEE Trans. Power Publications, New York, 1968.
Deliv. 3 (1) (1988) 187–196. [14] A.B. Fernandes, W.L.A. Neves, E.G. Costa, M.N. Cavalcanti, Transmission line shunt
[2] S. Yokoyama, T. Sato, S. Sekioka, Y. Hashimoto, Lightning performance of insulated conductance from measurements, IEEE Trans. Power Deliv. 19 (2) (2004) 722–728.
wires on overhead power distribution lines, 30th International Conference on [15] A. Ametani, A general formulation of impedance and admittance of cables, IEEE
Lightning Protection, Cagliary, 2010, pp. 1–4. Trans. Power Apparatus Syst. PAS-99 (3) (1980) 902–910.
[3] R.C.C. Rocha, R.C. Berrêdo, R.A.O. Bernis, E.M. Gomes, F. Nishimura, L.D. Cicarelli, [16] C.R. Paul, Analysis of Multiconductor Transmission Lines, 2nd ed., Wiley, New
M.R. Soares, New technologies, standards, and maintenance methods in spacer York, 2008.
cable systems, IEEE Trans. Power Deliv. 17 (2) (2002) 562–568. [17] Alternative Transients Program (ATP), Rule Book, Leuven EMTP Center, 1992.
[4] G.S. Lima, R.M. Gomes, R.E.S. Filho, A. De Conti, F.H. Silveira, S. Visacro, [18] A.C. Silva, A. De Conti, Calculation of the external parameters of overhead insulated
W.A. Souza, Impulse withstand voltage of single-phase compact distribution line cables considering proximity effects, XIV International Symposium on Lightning
structures considering bare and XLPE-covered cables, Electr. Power Syst. Res. 153 Protection, Natal, 2017, pp. 1–5.
(2017) 88–93. [19] C.R. Paul, A.R. Feather, Computation of the transmission line inductance and ca-
[5] R.E. Souza, R.M. Gomes, G.S. Lima, F.H. Silveira, A. De Conti, S. Visacro, Analysis of pacitance matrices from the generalized capacitance matrix, IEEE Trans.
the impulse breakdown behavior of covered cables used in compact distribution Electromagn. Compat. EMC-18 (4) (1976) 175–183.
lines, Electr. Power Syst. Res. 159 (2018) 24–30. [20] L.M. Wedepohl, H.V. Nguyen, G.D. Irwin, Frequency-dependent transformation
[6] R.E. Souza, F.H. Silveira, R.M. Gomes, G.S. Lima, A. De Conti, S. Visacro, matrices for untransposed transmission lines using Newton-Raphson method, IEEE
Characterization of the effect of the insulating material of covered cables on the Trans. Power Syst. 11 (3) (1996) 1538–1546.
impulse breakdown behavior of single- and three-phase compact distribution lines, [21] B. Gustavsen, A. Semlyen, Rational approximation of frequency domain responses
Electr. Power Syst. Res. 172 (2019) 161–166. by vector fitting, IEEE Trans. Power Deliv. 14 (3) (1999) 1052–1061.
[7] A. De Conti, F.H. Silveira, J.V.P. Duarte, J.C.S. Ventura, Lightning-induced over- [22] D. Deschrijver, M. Mrozowski, T. Dhaene, D. De Zutter, Macromodeling of multiport
voltages in MV distribution lines: spacer-cable versus conventional line configura- systems using a fast implementation of the vector fitting method, IEEE Microw.
tions, 27th International Conference on Lightning Protection, Avignon, 2004, Wirel. Components Lett. 18 (6) (2008) 383–385.
pp. 1–6. [23] A. Morched, B. Gustavsen, M. Tartibi, A universal model for accurate calculation of
[8] F. Napolitano, A. Borghetti, D. Messori, C.A. Nucci, M.L.B. Martinez, G.P. Lopes, electromagnetic transients on overhead lines and underground cables, IEEE Trans.
J.I.L. Uchoa, Assessment of the lightning performance of compact overhead dis- Power Deliv. 14 (3) (1999) 1032–1038.
tribution lines, IEEJ Trans. Power Energy 133 (12) (2013) 1–7. [24] K. Berger, R.B. Anderson, H. Kroninger, Parameters of lightning flashes, Electra 41
[9] A. Borghetti, G.M. Ferraz, F. Napolitano, C.A. Nucci, A. Piantini, F. Tossani, (1975) 23–37.
Lightning protection of a compact MV power line sharing the same poles of a HV [25] F. Heidler, Analytische blitzstromfunktion zur LEMP-berechnung, 18th
line, 34th International Conference on Lightning Protection, Rzeszów, 2018, International Conference on Lightning Protection, Munich, 1985, pp. 63–66.
pp. 1–6. [26] C.A. Nucci, F. Rachidi, M.V. Ianoz, C. Mazzetti, Lightning-induced voltages on
[10] A.C. Silva, A. De Conti, O.E.S. Leal, Lightning wave propagation characteristics of overhead lines, IEEE Trans. Electromagn. Compat. 35 (1) (1993) 75–86.
an overhead line with insulated phase conductors, 34th International Conference [27] A. De Conti, S. Visacro, Analytical representation of single and double-peaked
Lightning Protection, Rzeszów, 2018, pp. 1–6. lightning current waveform, IEEE Trans. Electromagn. Compat. 49 (2) (2007)
[11] A. De Conti, A.C. Silva, Three-phase compact distribution line transient analysis 448–451.
considering different line models, International Conference on Power Systems,