Transient analysis of underground power-transmission

systems
System-model and wave-propagation characteristics

Prof. L. M. Wedepohl, B.Sc.(Eng.), Ph.D., C. Eng., F.I.E.E., and D. J. Wilcox, B.Sc. Tech. M.Sc., Ph.D.

Indexing terms: Modelling, Power transmission, Skin effect, Transients, Transmission-line theory, Underground
cables

ABSTRACT
A mathematical model suitable for the analysis of travelling-wave phenomena in underground power-transmission
systems is presented. The model is developed in terms of a generalised angular frequency, and may therefore be
applied to the solution of steady-state problems or, by means of Fourier-transform techniques, to the solution of
transient problems. The model takes into account skin effect in the conductors and in the soil. It is then shown
how the system model may be analysed using multiconductor-transmission-line theory to give the transient res-
ponse of the cable system. The wave-propagation characteristics are given for the natural modes of a certain
cable system. These characteristics are examined with a view to their implications on transient phenomena.

LIST OF PRINCIPAL SYMBOLS The method of analysis may be interpreted in terms of

V = conductor-voltage vector natural modes of propagation. The characteristics of these
I = conductor-current vector modes, as a function of frequency, are given for a represen-
Z = series-impedance matrix tative cable system. The essential nature of these charac-
Y = shunt-admittance matrix teristics is explained, as they have an important bearing on
J = current density the understanding of transient phenomena.
H = magnetic-field intensity Briefly, the paper forms a basis for further work on the
E = electric-field intensity transient performance of underground-cable systems.
r = radius
= radius of central conductor
= radius over main insulation
= radius over conducting sheath MATHEMATICAL REPRESENTATION OF THE
= outer radius of cable TRANSMISSION SYSTEM
A = r
3~r2
U) = angular frequency The analysis is based on the two matrix equations
P = resistivity or charge density
a = conductivity (la)
e = permittivity dx
= permeability
= Euler's constant dl. (lb)
Y = -Y .V
m dx
s = cable separation
n = number of effective metallic conductors in system where V and I are vectors of dimension n representing, res-
pectively, the voltages and currents at a distance x along the
cable system containing n metallic conductors. Z and Y are
1 INTRODUCTION square matrices of dimension n x n.
The spreading of urban centres, and the ever increasing de-
mand for electrical power within them, is leading to the use 2.1 Principal assumptions
of relatively long cable circuits operating at high voltage. Eqns.2 constitute a mathematical model of the transmission
Under these conditions, we can expect to be faced with a system, and may be formulated on the basis of the following
problem of transient overvoltages being induced in the principal assumptions:
conductors of the underground system whenever it is sub-
jected to a sudden disturbance—a switching operation or per- (i) The cable system consists of n effective metallic con-
haps an insulation failure. ductors whose axes are mutually parallel and are paral-
lel to the surface of the Earth.
The transient stress across the main insulation owing to
(ii) The system is longitudinally homogeneous.
such a disturbance may be estimated with enough accuracy
for most practical purposes by ignoring the presence of the (iii) The anticipated attenuation of conductor voltages and
soil and treating the problem as one of classical wave prop- currents is negligible along a length of system com-
agation along a coaxial transmission line. However, as this parable with its lateral dimensions.
method ignores the presence of the soil and neighbouring It is convenient to assume that, the system consists of N
cables, it precludes the calculation of both sheath transient cables each with a cross-section of the type shown in Fig. 1.
voltages and transient induction effects into other cables. Each cable then has two metallic conductors, of which one is
The method of transient analysis proposed here is based on the central conductor and the other is a conducting sheath or
the theory of wave propagation in multiconductor systems, 1 armour. This is the basic type of design normally employed
and therefore takes into account all the metallic conductors for high-voltage cables. The same basic construction can be
in the system as well as the ground itself. In the theory, the used to represent a screened communication cable where the
transmission system is defined in terms of a series-
impedance matrix Z and a shunt-admittance matrix Y. It is
shown in the paper how these matrices may be derived in
conducting sheath
terms of a generalised angular frequency to include an allow-
ance for the frequency-dependent nature of the parameters.

Paper 6872 P,first received 4lh September and in revised central

form 20th November 1972
Prof. Wedepohl is with the Department of Electrical Engineer-
ing & Electronics, University of Manchester Institute of
Science & Technology,PO Box 88,Sackville Street, Manchester Fig.l
M60 1QD, England, and Dr. Wilcox is with Mons Polytechnic,
Mons, Belgium Basic cable construction
PROC. IEE, Vol. 120, No. 2, FEBR UAR Y 1973 253

P6
group of central conductors may be reasonably represented (iii) z 3 , the inner sheath internal impedance. This impe-
by a single equivalent conductor for calculating longitudinal- dance is calculated from the voltage drop on the inner
induction effects. surface of the sheath per unit current which returns
via the inner conductor
There are n = 2N metallic conductors, and the soil in which
the cables are buried constitutes an (n + l)th conductor (iv) z 4 , the sheath mutual impedance. This impedance is
which is chosen as the voltage reference for the system. given from the voltage drop along the outer (inner)
surface of the sheath per unit current returning through
the inner (outer) conductor. In this case, the outer
2.2 Formulation of shunt-admittance matrix conductor is the soil
Maxwell's field equations show that, in any electromagnetic
field, (v) z 5 ,the outer sheath internal impedance. This is given
from the voltage drop along the outer surface of the
sheath when the current returns through the outer
conductor
3t ~ ep
(vi) z 6 , the impedance due to the time-varying flux in the
where p is the volume charge density at any point. This outer insulation
equation has the solution p = p0 exp (—a/et), which shows (vii) z 7 , the self impedance of the earth-return path.
that, if a charge is introduced into a conducting medium (a *
0), it will dilute itself with a time constant of e/o seconds. In At low frequencies, zx is equal to the d.c. resistance of the
a good conductor such as copper, this time constant will be inner conductor. At higher frequencies, the classical formula
of the order of only 10~19 s, and therefore a charge intro-
duced into the body of a good conductor will be very quickly
displaced onto its surface. The process is not so rapid in the pm
soil, which is a poor conductor, but, even for a relatively high ohms per metre (2)
soil resistivity, say 1000 fim,the time constant is still quite 27rr 1 I 1 (mr 1 )
small, being of the order of 10~8 s. Thus, if the smallest
unit of time of interest is not less than, say, 10~6 s,the which takes account of skin effect. An approximate formula,
electric charges in the cable system may be assumed to be which is generally more suitable for digital computation, is
surface charges. deduced in Appendix 8.1, and quoted here:
In the present model, it is, indeed, assumed that the smallest _ pm coth
time interval of interest in the transient response is large 1
+ - ohms per metre (3)
enough to justify the assumption that all charges are surface 27rr1
charges. This greatly simplifies the calculations, as the The maximum error in the resistive (real) part of this im-
instantaneous value of the time-varying lateral electric field pedance is 4% and occurs when | rm^ | = 5. The maximum
has the same form as the electrostatic field. This latter is error in the reactive component is 5%, and occurs when
readily deducible from elementary theory. Irm^ | = 3*5. Away from these maxima, the formula is very
accurate. This formula is justified by the fact that it general-
2.2.1 Admittance submatrices ly requires rather less digital computation than that of eqn.
2, to which it is an approximation.
The shunt admittance submatrix of the ith c?ble has the di-
mension 2 x 2 and is given by The impedances z2 and z 6 present no problem, and are
given by
v
i -Y

Y2
where
+
In(r 4 /r 3 )
Yl = gi jw2ire 1 /ln(r 2 /r 1 )

Y2 = S2 +jw277€ 2 /ln(r 3 /r 4 ) where /i x and /i 2 are,respectively,the magnetic perme-

ability of the inner insulation and the permeability of the
outer insulation.
in which gx and g2 represent the leakage conductances across
the inner and outer insulations, and e x and e 2 the correspond- 2.3.2 Sheath impedances
ing permittivities.
Classical formulas for the three sheath impedances are
given in Appendix 8.1. These formulas involve Bessel func-
2.2.2 Assembly of admittance Matrix tions, and are seen to be rather complicated. The following
The admittance matrix Y is assembled from the N sub- three approximate formulas give good accuracy for
matrices Yj by placing them along the leading diagonal. The (r3-r2)/(r3 + r2)<l/8:
soil acts as an electrostatic shield between cables, and
hence offdiagonal submatrices are null. Thus
Zo = coth (mi) ^ ohms per metre (4)
27rr2(r2 + r,)

z 4 = ——° 7 cosech (mA) ohms per metre (5)

7r(r2 + r 3 )

ohms per metre (6)

2.3 Formulation of the series-impedance matrix
The series-impedance matrix Z is built up from N2 sub- where m = V(jaj/i/p),p is the sheath resistivity and A =
matrices. The N submatrices which fall on the leading di- r 3 — r 2 (the thickness of the sheath). For cable sheaths
agonal of Z are obtained as described in the following Section. normally encountered in practice,the above condition is well
satisfied, and perfectly adequate accuracy may be expected
2. 3.1 Leading-diagonal impedance submatrices from the above simple formulas. The derivation of these
formulas is briefly indicated in Appendix 8.1.
The impedance submatrix of each individual cable is con-
structed from seven component impedances. These are 2.3.3 Self impedance of the earth-return path
(i) z1} the internal impedance of the inner conductor The self impedance of the earth-return path of a buried
(ii) z 2 , the impedance due to the time-varying magnetic cable has been worked out by Pollaczek,5 and, as with Car-
field in the inner insulation son's formulas for the overhead line, leads to a rather
254 Pi?OC. IEE, Vol.120, No. 2, FEBRUAR Y 1973
involved infinite series. Fortunately, up to quite high fre- plicated, and the impedance may be obtained directly from
quencies, only the following terms need be taken into account: the impedance formula given by eqn. 34; i.e.

•' • z 7 = ^ - {— ln(ymr4/2) + 1/2 — 4mh/3} ohms per metre

2n \a\ + V(a2 + m 2)
(7)
where y is Euler 's constant and h is the depth at which the -^jiVCa 2 + m 2 ) - e-'jiVfo 2 + m 2 )
cable is buried. This expression is very accurate for fre-
quencies for which | mr 4 | < 0*25. In fact, eqn. 7 should give
high accuracy in all practical cases of interest.
where £•» and 6U are,the moduli of the sum and difference,
2.3.4 Assembly of the submatrix respectively, of the depths of the ith and jth cables and XJJ
is the horizontal distance between them. The integral con-
From the viewpoint of series impedance, the cable may be verges rapidly, and is therefore suitable for numerical
represented by the equivalent circuit shown in Fig. 2. This evaluation. However, in view of the simplicity of eqn. 8, the
•numerical integration is best reserved for the case where
|mjj| > 025. In this case,the integral is particularly well
central behaved.
conductor
Finally,by virtue of the reciprocality of mutual impedance,
conducting
sheath
Z 5 +Z 6 *Z 7 -Z 4
earth o
3 ANALYSIS
Fig. 2
Impedance equivalent circuit The voltages and currents represented by V and I and given
in eqns. 1 are implicitly functions of the generalised angular
representation may be deduced by considering the effects of frequency u>. For harmonic problems,i.e.problems involving
two circulating currents: one flowing down the central con- only one frequency, u> is simply the angular frequency in
ductor and returning through the sheath, and the other flow- question, and V and I represent phasors. For transient prob-
ing down the sheath and returning through the earth. This lems, V and I represent the Fourier transforms of the volt-
equivalent circuit corresponds to the following impedance ages and currents at values of w lying along the path of the
submatrix: Fourier integral necessary to convert the solution back into
the time domain. In this latter case, u> need not be real, and,
indeed, it is advantageous to choose a path of integration dis-
z6 +z 7 -z, placed below the real axis 2 ' 3 in the u> plane which requires
Zr, + Z c + Z , — Z, a complex value of w.
The elimination of the current vector I from eqns. 1 gives
the multiconductor Telegrapher's equation.
2.3. 5 Assembly of the impedance matrix
The submatrices Zj are assembled along the leading diagonal d2V
of the Z matrix in the same order as for the shunt-admit- = ZYV
dx2
tance matrix. The matrix Z then takes the form

Z
which has the solution
13
Z
21 Z
2 Z
23
V = exp (-xV(ZY))V+ + exp (+xV(ZY))V" (9)
Z =
Z Z
31 32 for the conductor voltages at a distance of x along the sys-
tem. V+ and V~ are vectors containing the 2n constants of
integration which are deduced from a knowledge of the sys-
The offdiagonal submatrices Zjj are not null,but take account tem boundary conditions. The matrix functions may be
of the mutual inductance between cables. efficiently evaluated as a consequence of the eigenvalue
definition.6
The net current in the ith cable returning through the soil,
i.e.the algebraic sum of the conductor and sheath currents, The solution for the system currents follows from the above
gives rise to a voltage drop in the soil adjacent to the jth result by using eqn. la to give
cable. This voltage drop gives the e.m.f.induced into each
conductor of this cable; i.e.the same e.m.f.is induced in both I = Z" 1 V(ZY) [exp {-xV(ZY)}v+ - exp {+xV(ZY)}V"] (10)
conductors. It follows from these remarks that all four
elements of the submatrix Z« must be equal;i.e. The distribution vectors for the n natural modes of propa-
gation are given by the n columns of the eigenvector matrix
of ZY and the distribution vectors for the n natural modes
Z = of current propagation are given by the n rows of the in-
verse eigenvector matrix. These vectors permit a physical
indentification of the natural modes.
The elemental impedances ZJJ may be calculated from Pol- The propagation coefficients of the n natural modes are
laczek's formulas. Formulas equivalent to those of Pol- given by the eigenvalues of the matrix V(ZY).
laczek have been deduced in Appendix 8.1 by means of inte-
gral-transform techniques. For |ms«| < 0*25,the mutual For transient problems,the above solutions are transformed
impedance zx between the ith and jth cables is given by back into the time domain by means of the Fourier integ-
ral. 2 ; 3 Thus,for example,

2TT
— ml ohms per metre (8) V (t) = j - /_+°° V(w) exp (jwt)dw

where S-H is the distance between the ith and jth cables,/! is
the magnetic permeability of the soil,y is Euler's constant 4 NATURAL MODES OF PROPAGATION
and I represents the sum of the depths of the ith and jth
cables. This formula reduces to that given by eqn. 7 by re- The nature of the natural modes of wave propagation in cable
placing Sji by the cable radius r 4 . The formula will,in fact, systems will be illustrated by considering a cable circuit
be valid up to quite high frequencies in practice—of the consisting of three identical power cables layed in a flat con-
order of 100 kHz for cables buried in the same trench. figuration at a depth of 0 76 m and with a cable separation of
At frequencies for which | ms-jjl > 0-25, the full form given 015 m,as shown in Fig. 3. The cable-design specification is
by eqn. 40 may be used. However, this form is rather com- given in Appendix 8.4 (Fig. 3).
PROC.IEE, Vol.120, No.2, FEBRUARY 1973 255
4.1 Characteristics Modes b and c are seen from Table 1 to be intersheath
The attenuation characteristics of the six natural modes of modes. Mode b is energised by injecting unit current into
wave propagation of the present system are shown in Fig. 4 the sheath of the central cable (sheath 2) and extracting half
and the velocity characteristics are shown in Fig. 5. The of it from sheath 1 and the other half from sheath 3,while
mode 3 is energized by injecting unit current into sheath 1
and extracting it from sheath 3. These two modes are seen
from Fig. 4 to have a quasiconstant attenuation over quite a
air wide frequency band (of the order of two decades). This band
is very important, as it tends to dominate the frequency
soil spectra of many of the transient responses of interest.
15cm I 15cm
20 n m 75cm 160i—

120 —

Fig. 3
Cable configuration

10-Or—

10 100 1000 10000 100000

frequency, Hz
Fig. 5
O l - Velocity characteristics of natural modes of propagation

Fig. 5 shows that,for frequencies in excess of about 200 Hz,

the velocities of these two modes are constant. A surprising
0 01 feature is that these velocities are only of the order of one-
quarter of the natural velocity (165 km/ms) associated with
the cable insulation. A further point of interest is that, at
the higher frequencies,these modes are the least lossy.
0001 These modal characteristics are evidently rather different
100 1000 10000 100000 in nature to those of overhead transmission lines, where,in
frequency, Hz the latter case,all modal velocities,even the ground mode,
are a high fraction of the velocity of light and where low
Fig. 4 velocity implies high attenuation.
Attenuation characteristics of natural modes of propagation The nature of the sheath characteristics may be explained as
follows. Consider,for example,the case of the intersheath
mode between the two outer conductors (1 and 3). If the
modal-current vectors at 1000 Hz are given in Table 1. The presence of all the system conductors except these two
modes have been designated a to f in the Figures, and may sheaths is ignored,this 2-conductor path will have the
be identified with the help of Table 1. impedance
Mode a is a zero-sequence sheath mode, and is seen to have
the characteristics of relatively high attenuation and very
low velocity. The velocity is 13 5 km/ms at 1000 Hz and z p = 2RS + 2 z i + 2(ze - z m ) (11)
increases only slowly with frequency. The low velocity is
due to the high inductive impedance of the soil path, which is where R s is the sheath resistance, z^ is the impedance due
of the order of 100 greater than the inductive impedance of to the inductance associated with the cable outer insulation,
the cable outer insulation. z e is the self impedance of each cable due to the earth path

TABLE 1
MODE-DISTRIBUTION VECTORS FOR CURRENT

Modal vectors for current

Mode a b c d e f

Velocity, km/ms 13-5 49-0 389 144-8 139-6 141-7

Attenuation, dB/km 0'2637 0-1096 0-0866 00991 0-0945 0-0972

Conductor
1 0001 94-3 0018 279 0-016 94 0-337 0 0343 180 0503 0
Sheath 1 0337 0 0343 180 0503 0 -0-337180-0-34413 -0505187
Conductor
2 0001 94-2 0-034 99 0350 0 -0-6620 0
Sheath 2 0-350 0 -0-662 0 -0-350180 0-664 193 0
Conductor
3 0001 943 0018 279 0-016 274 0337 0 0343 180 -0503 180
Sheath 3 0-337 0 0-343 180-0-503180 -0-337180 -0-34413 0505 7

Frequency = 1000 Hz
256 PROC.IEE, Vol.120, No. 2, FEBRUARY 1973
and z m is the mutual impedance between them. The impe- 39*2 km/ms and the modal attenuation to be constant at a D =
dances z e and z m are, effectively, given by eqns.7 and 8; 9-98 x 10- 6 (0-0866 dB/km). These estimated values are
seen to be in good agreement with those obtained from the
1 4 computer results given in Table 1. However, according to the
— ln ~ — , - mh (12) simple theory presented above, the attenuation should remain
constant as the frequency increases from 250 Hz, whereas it
Zrv, = -In 1 4
-ms (13) is seen to remain constant only up to about 10 kHz. The ex-
2 " - y mh planation is that the simple theory neglects skin effect which
where s is the cable separation. Now causes the resistance to increase as the square root of fre-
quency; this is confirmed in Fig. 4, as the attenuations of all
but the ground mode increase half a decade for one decade
(14) increase in frequency at the higher frequencies.
and hence the series impedance Zp of the intersheath path1 The factor k increases with cable separation, and hence both
becomes the velocity and the attenuation of the intersheath modes de-
crease with increasing cable separation.
(15) Modes d,e and f are seen from Table 1 to have essentially
coaxial natures, with velocities which approach the natural
which is exactly the same series impedance as would be velocity for the dielectric medium,i.e. 165 km/ms. Mode d
obtained if the soil in which the cables are buried were re- is a zero-sequence-type mode which is energised by inject-
placed by air. This will remain valid so long as | m s | < 0'25, ing unit current into each cable conductor and extracting it
i.e.up to a frequency of 1'7 MHz in the present instance, from the corresponding sheath. Mode e attempts to be an
which should be quite high enough for most practical transient interconductor mode energised by injecting two units of
studies. current into the conductor of cable 2 and extracting half of
this from the conductor of cable 1 and the other half from
Consider now the shunt admittance yp of the same path; the
the conductor of cable 3. However,much of the flux linking
earth acts as an electrostatic shield oetween the cables, and with the interconductor path also links with the correspond-
hence ing intersheath path, and therefore induces an intersheath
circulating current which opposes this flux linkage, tending
_ 1^ jae27T (16) to convert the interconductor mode into a coaxial mode and
y
P " 2 In(r 4 /r 3 ) an intersheath mode. At rather low frequencies,the circulat-
ing sheath current is restricted by the relatively high sheath
The propagation coefficient yp of this pure intersheath mode resistance,and mode d will,indeed,behave as a true inter-
is then given from eqns. 15 and 16 as conductor mode. As the frequency increases, the sheath
resistance becomes progressively less and less effective in
(17) supressing the sheath current and, in fact,the Figures show
that the characteristics of mode e merge with those of the
Now, if the inductance L of the cable outer insulation is pure coaxial mode e at about 100 Hz. A similar transforma-
given by tion into a coaxial mode occurs for mode f. However, at the
lower frequencies such that modes e and f retain an essential
interconductor-mode nature,the attenuation (and, incidentally,
(18) the velocity) tends to be lower than is perhaps expected,for
much the same reasons as those given for the intersheath
and the sheath-earth capacitance C is given by modes;this is just as well,as these are the modes of normal
power transmission.
C = 27T€/ln(r4/r3) (19)

the propagation coefficient y p can be expressed by

5 CONCLUSIONS
y p = jwk[LC{l + Rs/(k2jwL)}]i/2 (20) A model of a buried-cable system has been developed in
terms of a generalised angular frequency (which may be
where real or complex) which may be used for the solution of
steady-state problems, e.g. interference into nearby com-
ln(s/r 3 ) munication cables due to power-circuit harmonics,or,by
k = In(r 4 /r 3 ) (21)
applying Fourier-transform techniques, to the solution of a
wide range of transient problems.
At the higher frequencies,the square-root term in eqn.20
may be expanded as a convergent power series. In particular, The attenuation and velocity characteristics of a represen-
consider the case where the frequency is sufficiently high tative cable system have been presented. The remarkably
that only the first two terms of this expansion need be con- low velocities and attenuations of the intersheath modes
sidered; then have been satisfactorily explained, to the extent that they
may be estimated over quite a wide range of frequencies by
simple 'slide-rule' calculations. The nature of these charac-
(22) teristics, notably those of the intersheath modes, have an
important bearing on the transient behaviour of underground-
cable systems. For example,the velocity of the intersheath
The modal velocity is then given by and earth modes will determine the time taken for a system
earth to communicate its presence along the system, and
(23)
U hence the period of time during which sheath overvoltages
P ~ k 7(LC) may develop before being checked by the system earths.
Also, the intersheath modes are an important vehicle by
and the attenuation factor is given by which a disturbance in one cable of the system can be com-
municated to other cables. On the other hand, the attenuation
characteristics will determine the decay of natural system
resonances resulting from an abrupt disturbance.
In these equations, l/V(LC) is the natural velocity for the The study of transient sheath overvoltages and other transi-
cable insulation and V(L/C) is the sheath-earth surge ent induction effects form the subject of further work to be
impedance. carried out by the authors.
At frequencies above Rs/7rk2L,both the modal attenuation and
the modal velocity should be constant and are given by eqns.
24 and 23, respectively. The frequency Rs/7rk2L may be 6 ACKNOWLEDGMENTS
deduced for the case in hand from the data specified in Ap- The authors wish to acknowledge the facilities granted in the
pendix 8.4, and is found to be 250 Hz;the factor k is found to power-systems laboratory of the Department of Electrical
be 4 2 1 . Thus, at frequencies in excess of 250 Hz, eqns. 23 Engineering & Electronics, University of Manchester Insti-
and 24 predict the modal velocity to be constant at u p = tute of Science & Technology, and the encouragement and
PROC. IEE, Vol. 120, No. 2, FEBRUAR Y 1973 257
helpful advice of Prof.Ch.Gregoire of the Faculte Poly- At low frequencies,the higher-order terms of the expansion
technic de Mons. The authors wish to thank CEGB head- may be neglected. Now,mr1 is a complex quantity with a
quarters for financial help and for helpful discussions with phase angle of 45°, and hence the second term in the expan-
A.Stalewski and A.Rosen of that organisation. sion gives an inductive component while the third term gives
They also wish to thank the University of Manchester com- a resistive component. Terms are compared, and k is chosen
puting service for running the programs used to provide the to give the correct resistive component; thus k3 = 90/192 or
results. k = 0-777.
It may be noted that an error in the inductive component is
more tolerable than an error in the resistive component, as
REFERENCES the internal inductance will,in practice,tend to be swamped
by the external inductance.
WEDEPOHL, L.M.: 'Application of matrix methods to
the solution of travelling-wave phenomena in polyphase
systems', Proc. IEE, 1963,110, (12), pp. 2200-2212 8.2 Internal impedance of a solid sheath with annular
MULLINEUX,N.,DAY, J.,and REED,J.R.: 'Develop- cross-section
ments in obtaining transient response using Fourier
transforms: use of the modified Fourier transform', 8.2.1 Classical formulas 4
Int. J. Elec. Eng. Educ, 1966,4, pp. 31-40
MULLINEUX,N.,DAY,S. J.,and REED, J.R.: 'Develop- K 0 (mr 2 )I 1 (mr 3 )}/D
ments in obtaining transient response using Fourier
transforms: Gibbs phenomena and Fourier integrals',
ibid., 1965,3, pp. 501-506 Z
4 = ^7T"/D (27)
SHELKUNOFF,S. A.: 'The electromagnetic theory of co-
axial transmission lines and cylindrical shields',Bell
Syst. Tech. J., 1934,13, pp.532-579
POLLACZEK,F.: 'Sur le champ produit par un conduc-
teur simple infiniment long parcouru par un courant
alternatif, Revue Ge"n.Elec, 1931,29, pp. 851-867 where D = I 1 (mr 3 )K 1 (mr 2 ) — I 1 (mr 2 )K 1 (mr 3 )
PIPES, L. A.: 'Matrix methods for engineering'
(Prentice-Hall, 1963)
MULLINEUX,N.,and REED, J.R.: 'Calculation of elec- 8.2.2 Approximate formulas
trical parameters for short and long polyphase trans-
mission lines', Proc.IEE, 1965,112, (4), pp. 741-742 coth (mA)
IRVING, J.,and MULLINEUX.N.: 'Mathematics in 77 "
physics and engineering' (Academic Press, 1959)
z 4 = —,—-, r cosech (mA)
"\T9. + T
l>
8 APPENDIX

Impedance formulas z s = ^ T coth (mA) + £

27rr3(r2 + r 3 )
8.1 Internal impedance of a solid conductor with
circular cross-section
8.2.3 Derivation
8.1.1 Approximate formula
The magnetic intensity H and the electric current density J
The approximate formula proposed here is within the solid tube representing the conducting sheath are
related by the equations4
z, = —— coth (0*777 m r , ) + 0'356 - ^ r ohms per metre
27rr1 7rrf dr r
(28a)

(25)
This formula is deduced by considering first the formula (28b)

z
= 2^7 c o t h (mr )
i where r is the radial distance from the sheath axis.
This simple formula is known to exhibit similar properties Elimination of first H and then J between these two equations
to the exact expression given by eqn. 2. For example, at high result in two 2nd-order differential equations, one in H and
frequencies,this impedance tends to pm/27Trl5which is the the other in J. These equations are, in fact,Bessel's equa-
well known skin-effect formula, while at low frequencies it tions and lead to the classical result given in eqn. 27.
represents a pure resistance, although not,in fact, equal to
The approximate formulas are derived on the assumption
the required value of p/irrf.
that the sheath is thin compared with its mean radius. In
This formula may be improved by introducing a degree of this case, eqn. 28a may be written as
freedom so that
dH 2_
H= J (29)
dr (r, + r 3 )
z = coth (kmr,) + - £ - (1 - l/2k) (26)
2irr Trrf
Elimination of first H and then J between eqns. 29b and 30
where k is an arbitrary constant. The second term on the results in two 2nd-order differential equations with constant
right-hand side corrects the impedance at direct current. At coefficients. The solution of these equations with suitable
high frequencies,this formula tends towards the required boundary conditions applied leads to the approximate
skin-effect impendance. The constant k is chosen to optimise formulas.
the formula at the lower frequencies.
Expanding eqn. 26 in series form, 8.3 Earth-return impedances of buried-cable system
The ground is assumed to be a homogeneous medium whose
z = — { 1 +k(mr1)2/6-k3(mr1)4/90 + . . . ] flat surface divides space into two semi-infinite regions:
•nrf soil and air. It is further assumed that the field in the soil
whereas the corresponding expansion of the classical for- will be insignificantly different from the field that would be
mula is present if the cables were replaced by infinitely thin insu-
lated conductors and their volumes were replaced by soil.
zx = - £ - {1 + ( m r ^ / 8 - (mr,)V192 + . . . } The required results may be deduced from the electric-
7rrf field intensity E in the soil when current i flows in an
258 PROC.IEE, Vol.120, No.2, FEBRUARY 1973
insulated conductor and returns through the soil. Let this whose co-ordinates are (x,y):
conductor be buried at a depth h below the surface of the
ground, as shown in Fig. 6. This Figure shows the conductor "exp{-(h + y)V(a2 + m 2)}
in a plane perpendicular to its axis and defines the reference P2m2i
E, = - 2 77
coordinates x and y. m2)
2

region 1 air
e x p { - | y - h | V ( a + m2,)} - exp{-|y + h|V(a 2 + m2,)}
m 2)

exp(jax)dx (33)

If the co-ordinates (x,y) coincide with a second cable,the

mutual impedance z m between the two cables will be given
by

insulated conductor exp{-(h + y)V(a2 + m")}

2TT

Fig. 6 exp{-|y-h|V(a 2 + m 2 )}-exp{-|y + h|V(a2 + m 2 )}

Position of insulated filament with respect to the 2V(a2 + m 2 )
co-ordinate axes
exp(jax)dxa (34)

The general equation for electromagnetic-wave propagation where subscripts have been dropped, as all quantities relate
is only to the soil medium. It has also been assumed that ju2
= nx. The formula then corresponds to the integral form
given by Pollaczek.
V2E - V(VE) = (30)
This integral may be evaluated from a convergent series,
starting with the standard result
where V2 is the Laplacean operator and E is the electric-
field intensity.
(35)
For the present problem, it is reasonable to assume that
VE = 0 in both air and soil, and that displacement currents
may be neglected. These are the usual assumptions, and are where T = V(g2 + s 2 )
justifiable up to quite high frequencies. Eqn. 30 then simpli-
fies to
Eqn. 34 immediately simplifies to
V2E = (j (31)
% = ^ ~ {K0(mR) - K0(mR') + i j
On the basis of assumptions (i)-(iii) in Section 2.1,the elec-
tric force may be taken to be locally invariant in the direc- where R = V{x2 + (h -y) 2 },R' = V{x2 + (h + y) 2 } and
tion of propagation, and everywhere to be parallel to the
conductor. Eqn. 31 then simplifies to the following equations exp{—(h + y)V(a 2 + m 2 )}
in the two media: I, = —^—^ y— — exp(iax)da
1
2
3 E,
2
y>0 = —2 1 Sr ^(a 2 + m2) exp{-^V(a 2 + m2)} exp(jax)da
3x m I °°

32E2 3 2 E, — 2 /0°° aexp{— l>l(a2 + m2)} exp(ja |x|)da

2 2
h)y < 0 (32)
3x 3y
m2)} exp(ja |x|)da[ (36)
where subscripts denote the region to which a quantity re- " 2I3
lates and 6(x) and 6(y + h) are Dirac functions whose values
are zero except when their arguments are zero; in the latter
case they take the value unity. Now,from eqn. 35,
A solution for E 2 is required that satisfies the following
boundary conditions: exp-{—^V(a2 + m2)}
2V(a2+m2)
(i) continuity of E at the surface: E 1 = E 2 a t y = 0
(ii) continuity of the vertical component of B at the and therefore
interface:
2(*2- i
3E, ju, 3E ? I2 = ^ (K0(mR')} = (mR') (mR')
-Tr-i = ^ -~ at y = 0
3y M 3Y
(37)
(iii) continuity of the horizontal component of H at the
32
surface: r • , (mR') (38)
dxdl {K0(mR')| = R'2
3E, 3E2
— - = — 2- at y = 0 The integral I 3 may be evaluated by transforming the vari-
3x 3x
able according to
Eqn. 22 may be solved without difficulty by integral-trans-
form techniques to give the electric force at a general point mtR' = - j a |x| + ^V(a2 + m2)
PROC.IEE, Vol.120, No.2, FEBRUARY 1973 259
which gives where

R' 2 2! R'

+ _}W_ exp(-mtR')jdt (39)

V(t2 - 1) ) L(e- J i i i V | i (mR1)4 R 3 l x | 3 3. \e\x\s
The path of integration along the real axis from 0 to QO in the 2 \ R ' 2 / ) 3 4! L R' 6 4 I R'?
a-plane transforms into the path C in the t-plane,as shown
in Fig.7. 2 |x|3 >) | x | 3
{ 4 (
Im 5 3 R'3 3 1! R'3

t plane + 2_ (mR')3 / ^ 2 | x | 3 + 2_ J x | 3 \ + 2_ (mR')s ) J 4 | X | .

5 3! \ R'3 3 R'3/ 7 5! i R'7

Re
5 V R'5 3 R'3

, 1 \x\(3
R' /2
4 4! R'4 +—
2 \R'

1 (mR') lx| , 1 (mR')3 [ [ x l / 2 + 2_ \x[

1 1! R' 3 R'3 L R' 3 1 R'
radius tends
to infinity
1
5 5! / R'5 3 \ R' 1 R' J
Fig. 7
and 8 = v/2 — sin" 1 (£/R'). Assembling the above results
Path of integration in the complex plane gives the solution
l/R ^ 1 always
zm = tet K0(mR, - K0(mR') + M i K0(mR')
Now mtR' is positive real in the segment of 0, and hence
the line integral along the path C must be zero. The func-
tion has no singularities on or within the closed contour C
— C — Coo, and hence, by Cauchy's theorem, j c = J c/ . Thus
after elementary manipulation, eqn. 39 may be written as
m 2 R' 4
h=
At low frequencies,the impedance will be given by the lead-
ing terms of the series expansion corresponding to eqn. 40.
- 1)( exp(- mt R')d t For cables buried at usual depths (~1 m) and for separations
of the order of 1 m and less,for |mR| < 0 25 only, the follow-
ing terms of the expansion need be taken into account:
Jc/R' 2V(1 - t 2 ) - exp(-mtR')dt
R'2 V(l-t 2 )
- ^ i-ln(ymR'/2) + J - - | (41)
The first part of the integral is easily evaluated by normal
methods; the second part is recognised8 as K2(mR'). The
third part is evaluated by expansion of the exponential func-
tion and integration term by term to give S(mR', |x|, C). 8.4 Cable-design specification
Diameter over copper conductor, cm = 2-54
Hence I 3 is given by
Diameter over main insulation, cm = 4-56
Diameter over sheath, cm = 5-08
ml) K2(mR') Diameter over outer insulation, cm = 5-59
R'4 R'2 D.C. resistance of copper conductor, Si/km = 0034
D.C. resistance of lead sheath, J2/km = 0-436
S(mR', Core-sheath surge impedance, fi = 19-4
R'2 Sheath-earth surge impedance, Q = 4-6

260 PROC.IEE, Vol. 120, No. 2, FEBRUARY 1973