Artigo Linhas
Artigo Linhas
Artigo Linhas
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)
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
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)
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
6. Conclusions
6
A. De Conti, et al. Electric Power Systems Research 181 (2020) 106173
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)
7
A. De Conti, et al. Electric Power Systems Research 181 (2020) 106173