366 IEEE Transactions on Power Apparatus and Systems, Vol. PAS-104, No.

2, February 1985
OVERHEAD LINE PARAMETERS FROM HANDBOOK FORMULAS AND COMPUTER PROGRAMS

H.W. Dommel, Fellow, IEEE

The University of British Columbia
2356 Main Mall
Vancouver, B.C., V6T 1W5 Canada

Abstract - Overhead line parameters can be 71 8

calculated from handbook formulas, or with more general
computer-oriented methods. At power frequency, the
differences between the two approaches are usually
negligible, but they can become large at higher frequen-
cies. This paper discusses the causes of these differ-
ences for the engineer who wants to compare results from
computer programs with those obtained from handbook
formulas. It contains no new theories, but simply sum-
marizes the experience gained in analyzing such
differences over many years.

1. INTRODUCTION
Nowadays, overhead line parameters are usually
obtained with computer programs. Some of these programs Fig. 1. Tower Configuration
may still be based on handbook formulas, but most of -[ dVTx ] = [Z] [Ij] (1)
then use more general computer-oriented methods which -

are valid for any number of phases and ground wires at

any frequency. A prudent engineer may not want to trust for the longitu-dinal voltage drops along the line, and
the output of such general-purpose programs blindly, but the system of phasor equations
may want to compare at least some results with those
obtained from handbook formulas. There are differences -[ ] = jw [C] [Vl (2)
in the results from the two approaches, which this paper
tries to explain. While the differences are sometimes for the current changes along the line (shunt conduc-
of little practical importance, they must be understood tances, as usual, are ignored here).
if one wants to gain confidence in the results of a
general-purpose program. The elements of the 8 x 8 impedance matrix [Z] in
Eq. (1) are usually calculated from Carson's formula
After a brief discussion of computer-oriented [21. The diagonal element Zii is the series impedance
methods, the positive and zero sequence parameters per unit length of the loop formed by conductor i and
obtained with them are compared with those obtained from ground return, and the off-diagonal element Zik = Zki is
handbook formulas. the series mutual impedance per unit length between the
two loops conductor i/ground return and conductor
k/ground return. Carson's formula contains integrals
2. COMPUTER-ORIENTED METHOD which cannot be solved in closed form. They have been
developed into reasonably well converging infinite
A general method well suited for the calculation of series for small arguments of the parameter a,
overhead line parameters with computers was described by
M.H. Hesse more than 20 years ago [1i. To explain this a = 2.81 * 10i3 D /1F (3)
method, a single-circuit three-phase line with twin
bundle conductors and two ground wires, as shown in where f = frequency in Hz,
Fig. 1, will be used as an example. It must be empha- p = earth resistivity in Qm,
sized, however, that the method is- completely general, D = distance in m between conductor i and image
and could as well be used for a double-circuit line, or below earth surface of conductor k for mutual
for a single-circuit three-phase line in parallel with a impedance (or twice conductor height in m for
bipolar dc line, or for any other configuration which self impedance).
one might think of.
For large arguments a, asymptotic expansions are usually
For the case of Fig. 1, there are 8 parallel used. Most handbook formulas were derived from these
conductors. Two systems of equations describe the series, with only the first one or two terms retained.
steady-state behaviour of these 8 conductors, namely the In computer-oriented methods, it is best to add as many
system of phasor equations terms as are necessary for obtaining a specified degree
of acculracy. Fig. 2 shows that the errors with trunca-
-tion of the series af ter the first or second term would
84 SM 545-0 A paper recommended and approved be unacceptable for the mutual impedance between two
by the IEEE Power System Engineering Committee of conductors in cases of wide separation, or alternat-
the IEEE Power Engineering Society for presentation ively, in cases of less wide separation but higher
at the IEEE/PES 1984 Summer Meeting, Seattle, frequiency or lower earth resistivity [3].
Washington, July 15 - 20, 1984. Manuscript submit-
ted January 31, 1984, made available for printing Much simpler imnpedance formulas with closed-form
May 10, 1984. solutions have recently been developed by Gary, Deri,
Tevan, Semlyen and Castanheira [4,5]. They give results
close to those obtained from Carsoni's formula (largest
differences approximnately 10% in the range 100 Hz to
10 kHz, and smaller elsewhere). These new formulas may
0018-9510/85/0002-0366$01.00© 1985 IEEE

Authorized licensed use limited to: ULAKBIM UASL - Yuzuncu Yil Universitesi. Downloaded on March 19,2021 at 08:09:23 UTC from IEEE Xplore. Restrictions apply.
 367
matrices for symmetrical components. The zero, positive
i nf/km and negative sequence parameters are simply the diagonal
elements of these matrices, with Zneg = Zpos, while the
off-diagonal elements are normally ignored. For
a _. untransposed lines,' the off-diagonal elements do contain
-l f-60 Hz
earth reoistivity=10 nm useful information about coupling effects between
12 sequences quantities, however, and are used in [7, pp.
93-103] to derive unbalance factors.
A tezu

_ _ _ __^ ~ ~ ~ ~ ~~~.
4. COMPARISON BETWEEN BUNDLING PROCEDURES

The bundling procedure by matrix reduction, as

described in Section 3, differs from the procedure used
in most handbook formulas and in some computer programs,
where the bundle of subconductors is replaced by a sin-
ter" gle equivalent conductor from the beginning [7, pp.
term 111-114]. Formulas are usually only given for the more
important case of symmetrical bundles, even though they
Fig. 2. Mutual reactance between two could be derived for asymmetrically bundled conductors
parallel conductors as well. In Eq. (4), replacing the bundle of subconduc-
tors by one equivalent conductor located at the centre
replace Carson's formula one day, but they are not dis- of the bundle is achieved by using requiv in place of
cussed here because they have been adequately described
in [4,51.
r equiv * r * AN-1 (6)
The elements of the 8 x 8 capacitance matrix [C] in
Eq. (2) are easier to calculate, and are real rather where N = number of subconductors in bundle,
than complex. They are obtained indirectly, by first r = radius of subconductor,
building a "potential coefficient" matrix [P], with A = radius of bundle.
hi I Dik For the impedance calculation, the geometric mean radius
Pii 2 c ri iPik2% i T (4)
GMR of one conductor is replaced by GMRequiv, with the
same formula as Eq. (6) (except GMR in place of r).
where hi - height above ground of conductor i,
average
ri radius of conductor i,
=
In practice, the bundling procedures with matrix
Dikm distance between conductor i and image below
reduction and with equivalent conductors produce almost
earth surface of conductor k, identical results. For the case of a 500 kV three-phase
dik- direct distance between conductors i and k, line with the data of Table 1, the results from both
E: = permittivity of free space.

Once (PI is known, [C] is found by matrix inversion, Table 1. Data for 500 kV three-phase line
[C] =
[pi-1 (5)
Phase arrangement: Horizontal tower configuration
MATRIX REDUCTION AND TRANSFORMATION Spacing between phases = 40 feet
3. Average height above ground = 50 feet
Bundle with 4 subconductors, r eauiv = 7.80524 inches
Usually one is not interested in the details
contained in the 8 x 8 matrices of 'Eq. (1) and (2). A GMRequiv 7.41838 inches
=

simpler ''description is obtained 'by reducing' them to spacing between sub-

conductors = 18 inches
3 x 3 matrices for the phase quantities, which still Subconductors r = 0.45 inches
contains more detail than 'most handbook formulas would GMR = 0.3672 inches
allow.. The reduction is accomplished by first introduc- dc resistance = 0.1686 S/mile
iRg the bundling conditions into the equations. For
No ground wires
example, if conductors 1 and 2 form phase A, then V1= V2 Earth resistivity = 100 S3n
VA and Ii + I2 IAI in Eq. (1). For continuous ground
wires which are grounded at every tower, e.g., for cond-
uctors 7 and 8, one simply sets V7' = V8 = 0 in Eq. (1).
The reduction to smaller matrices is then achieved, in Table 2. Comparison between bundling procedures
.the example, by introducing IA as a new variable, and by
eliminating It, I2, I7 and I8. The reduction procedure
for ground wires is correct as long as the ground wire Positive and zero Bundling by Equivalent
potential is continuously zero. For typical tower spans sequence paramet- matrix reduction conductors
of 250 to '350 m, this assumption is reasonable up to ers at 60 Hz
approximately 250 kHz [6]. For bundling, the reduction
procedure is correct as long as the potentials on the R os( Q/mile) 0.042223 0.042205
subconductors are continuously equal, which is a XP (Q/mile) 0.53394 0.53399
reasonable assumption up to approximately 500 kHz with
spacers typically 100 m apart. CPOS (pF/mile) 0.021399 0.021397
R (aQ/mile) 0.31740 0.31738
Even the 3 x 3 matrices are often too detailed. Xzero( Q/mile) 2.0065 2.0065
For example, only positive sequence parameters are Czero(IF/mile) 0.013456 0.013455
needed in power flow studies, or positive and zero
sequence parameters in short-circuit studies, Sequence
parameters are easily obtained from the 3 x 3 matrices procedures are shown in Table 2. At least in this case,
for phase quantities by transforming them to 3 x 3 they are practically identical at 60 Hz, though they

Authorized licensed use limited to: ULAKBIM UASL - Yuzuncu Yil Universitesi. Downloaded on March 19,2021 at 08:09:23 UTC from IEEE Xplore. Restrictions apply.
368

would probably differ somewhat more at higher obtained with

frequencies.
pos dm *- zm (9)
5. INFLUENCE OF GROUND WIRES ON POSITIVE SEQUENCE .equiv Dm
RESISTANCE 3
where hm= 1hit-E, (geometric mean height),
While it is well known that ground wires have an
influence on zero sequence parameters, it is less well 3
known that they can influence positive sequence parame- dm= /dAd-ACdBC (geometric mean distance),
ters, too. Of practical importance is the increase in 3
the positive sequence resistance Rp,, if the line has Dm = VD/DACDBc (geometric mean distance to
ground wires which are grounded at every tower. Since images).
the mutual impedances from the three phase conductors
A,B,C to the ground wire G are never exactly equal, For lines of lower voltage ratings, the differences
there is always a small longitudinal voltage induced in would be less because the phases would be closer
the ground wire, even for 2ymmetrical posif ie sequence together.
currents with IB = IA e IC = IA e+J. The formula for zero sequence capacitance in
-dVG [8,9],
= (ZAG + ZBG e i + ZCG e+l20O) ' (7)
C = 2 wc
With the ground wire grounded at every tower, this (Siemens) (10)
induced voltage produces a circulating current which '2hm dm
flows through the ground wire, towers and ground
(Fig. 3). This circulating current produces additional requiv d
m
can be derived by averaging the diagonal elements of
Eq. (4) among themselves, as well as averaging the of f-
IG diagonal elements among themselves, to account for tran-
sposition. Computer-oriented methods do the averaging
in the elements of the [C]-matrix. Both give practi-
cally the same answer. For the line of Table 1, Eq.
(.10) produces a value for Czero which is 0.23% lower
than the value obtained from computer-oriented methods.
IG In [10], Eq. (10) is further simplified by assuming
Dm 2hm, or
Fig. 3. Circulating current in ground wire
Czero = zeo2nihm0 (Westinghouse), (11)
losses, which show up as an increase in the value of the (2hm)
positive sequence resistance in computer-oriented meth-
ods. Handbook formulas would not show this increase.
In one particular case of a single-circuit 500 kV line,
this increase was 6.5% at 60 Hz. which produces a value which is 4% higher than the value
from computer-oriented methods for the line of Table 1.
To avoid the losses associated with these circula- While Eq. (11) is theoretically less accurate, the value
ting currents, some utility companies use "segmented" obtained from it may actually be closer to measured
ground wires in an arrangement which has the form of a values because the influence of towers on the zero
" T": The ground wire is grounded in the middle, and sequence capacitance, which is neglected in all form-
insulated at the adjacent towers to the left and right. ulas, increases the calculated zero sequence capaci-
At both ends of the segmentation section, the ground tance. This increase is typically 8 to 9% on 110 kV
wire is interrupted as well, to prevent circulating lines, 6% on 220 and 380 kV lines, and 4% on 700 kV
currents from flowing. In computer-oriented methods, lines [11, p. 218].
segmented ground wires are handled by ignoring them in
the series impedance calculation (or by setting the
mutual impedances to the other conductors to zero) but 7. COMPARISON FOR SEQUENCE IMPEDANCES
by taking them into account in the capacitance calcula-
tion. The formulas for zero and positive sequence imped-
ances in most handbooks are based on the assumption that
parameter a in Eq. (3) is so small that only the first
6. COMPARISON FOR SEQUENCE CAPACITANCES term in Carson's infinite series need be retained. For
normal phase spacings this is a reasonable assumption at
For positive sequence capacitance, most handbooks power frequency (50 or 60 Hz). Then, after all diagonal
give the formula and off-diagonal elements in [Z] of Eq. (1) have been
averaged out among themselves, respectively, to account
2 Tr for transposition, the correction terms for the influ-
C = 0 (8)
M ence of the finite earth resistivity become
Q.n
requiv
Rself = Rmutual = ur 04 km (12)
with dm = geometric mean distance among the three
phases. For the 500 kV line of Table 1, this produces a and
value which is approximately 4% lower than that obtained
from computer-oriented methods. The difference is caused
by ignoring the influence of height above ground in Eq.
(8), or more specifically, by assuming that the geomet-
AXself = 2X10 4[0.6159315-fn(2.81410¾3.2hm /- ) ] S/km

ric mean distance Dm from one phase to the image of

another phase is approximately equal to twice the geo-
AXmutual= 2u1 4[0.6159315-/n(2.81*10 3*Dm ,/E )1 {km
(13)
metric mean height. Almost identical results would be

Authorized licensed use limited to: ULAKBIM UASL - Yuzuncu Yil Universitesi. Downloaded on March 19,2021 at 08:09:23 UTC from IEEE Xplore. Restrictions apply.
 369

where hm, Dm in m, f in Hz, and p in S`b. With these three-phase lines, but not as easy to illustrate as for
correction terms, the zero and positive sequence a two-conductor line). Handbook formulas ignore this
impedances can easily be derived from the self and eddy current effect in the ground.
mutual impedances, with
The zero sequence impedance obtained with the
Zpos Zself -

Zmutual' (14) correction terms of Eq. (12) and (13) is

and -4 659/
Zzero (Rac43)+j6O1O in(-3---D2) in Q/km
Zzero = Zself +2Zmutual* (15)
GMRequivdm
Using the correction terms of Eq. (12) and (13) (18)
leads to the simple expression for the positive sequence
impedance with f in Hz, p in Sfl, Rac in 0/km, and all distances
in m. This is the for ula found in most handbooks,
including [8,9,10]. Table 4 compares the results from
zPOS Rac + j2wlO
GMequiv
in Q/km, (16) the handbook formula (1&) with those 'obtained from
computer-oriented methods with accurate earth
which is found in all handbooks, with Rac = ac resis- resistivity correction terms. The zero-sequence
tance of the bundle. It is surprising that the influ- resistance Rzero and., inductance Lzero of the handbook
ence of ground resistivity and of conductor height, formula are reasonably accurate in the lower frequency
which is present in the self and mutual impedances,
disappears completely in Zpos of Eq. ( 16). This can Table 4. Accurate and approximate zero sequence
easily be verified, however, if one kno% that Zpos resistance and inductance.
without earth resistivity correcti germs is
pOs(AliR-AX=O)
po AX ac+ 3j2t10
ac d b ffi
n
(;HRequivDm S
/km.(17). ACCURATE APPROXIMATE
FROM EQ. (18)
Table 3 compares the results from the handbook formula
(16) and from computer-oriented methods with accurate f R Lze Rz L
earth resistivity correction terms for the 500 kV line (Hz) (Q/mi e m/ie (i;Vmi,e mH/mi )
of Table 1. In this comparisoni skin effects within the
10-6 0.04215 13.94 0.04215 13.94
10 0.08905 6.170 0.08980 6.158
Table 3. Accurate and approximate positive sequence 100 0.4960 5.084 0.5187 5.046
resistance and inductance 1000 4.169 4.052 4.807 3.934
10000 32.12 3.164 47.69r 2.823
ACCURATE APPROXIMATE 100000 184.0 2.568 476.6 1.711
FROM EQ. (16)

Os posOS' Lpos range (up' to approximately 1 kHz), but differ

(Hz) (I/mile) (mH/mile) (-n 1e)- (mH/mile) appreciably at higher frequencies.
io-6 0.04215 1.417 0.04215 1.417
10 0.04215 1.416 0.04215 1.417 8. CONCLUSIONS
100 0.04229 1.416 0.04215 1.417
1000 0.05003 1.416 0.04215 1.417 Differences in overhead line parameters obtained
10000 0.3528 1.413 0.04215 1.417 with computer-oriented methods and from handbook
100000 6.229 1.401 0.04215 1.417 formulas are usually small at power frequency (50 or 60
Hz). The bundling procedure based on matrix reduction
and the use of equivalent conductors for bundles produce
conductors were intentionally ignored (Rac = Rdc), to practically identical results. The- zero and positive
clearly show the influence of earth resistivity. Table sequence capacitances may differ by approximately 4% for
3 shows that the handbook formulas are quite accurate a typical 500 kV line. Of more pract'ical importance is
for the inductance Lpo0 over a, wide frequency range, the, increase in the value of the positive -sequence
whereas R DOS becomes less accurate as the -frequency resistance on lines with ground wires which are grounded
increases (0.33% difference at 100 Hz, but-different by at ever.y tower. This increase reflects the-. losses
orders of magnitude at 100 kHz). The increase in 0, caused by the circulating currents in the ground wire.
for higher frequencies is caused by eddy currents in the
ground, as indicated for a bipolar dc line or a single- Line parameters at higher frequencies are required
phase ac line in Fig. 4 (the phenomenon is similar in for switching and lightning surge studies, for power
line carrier studies, and for similar problems. Results
in the high frequency range from handbook formulas and
from computer-oriented methods can be quite different.
This is not surprising, since most handbook formulas
were not derived for frequencies much beyond. power
frequency.

..- REFERENCES
Fig. 4. Eddy currents in earth
[1] M.H. Hesse, "Electromagnetic and electrostatic
transmission-line parameters by digital computer,"
*) The self and Mutual inductances without correction IEEE Trans. Power App. Syst., vol. 82, pp. 282-291,
terms are calculated from formulas which have the June 1963.
same form as Eq. (4), except that 1/(2wco) is [2] J.R. Carson, "Wave propagation in overhead wires
replaced by o/(2ir), and r is replaced by GMR. with ground return," Bell System Techn. Journal,
vol. 5, pp. 539-554, 1926.

Authorized licensed use limited to: ULAKBIM UASL - Yuzuncu Yil Universitesi. Downloaded on March 19,2021 at 08:09:23 UTC from IEEE Xplore. Restrictions apply.
370

[31 H.W. Dommel, discussion of "Electromagnetic effects and the impedance matrix in phase quantities is:
of overhead lines", by IEEE Working Group, IEEE
Trans. Power App. Syst., vol. PAS-93, pp. 900-901,
May/June 1974. 2H
P12
[4] C. Gary, "Approche complete de la propagation in 1 i.n D1 3
multifilaire en haute frequence par utilisation r 12 13
des matrices complexes" ("Complete approach to
multiconductor propagation at high frequency with
complex matrices", in French), EdF Bulletin de la k x in D in
2H2
2.
2H2D2
Q.n
Direction des Etudes et Recher=hes, Serie B, no. 12 -23
3/4, pp. 5-20, 1976. Qn
[5] A. Dern, G. Tevan, A. Semlyen and A. Castanheira, 2H3
"The complex ground return plane, a simplified
model for homogeneous and multi-layer earth
inD13 k 23 in P233r
return," IEEE Trans. Power App. Syst., vol. PAS-
100, pp. 3686-3693, Aug. 1981.
[6] L.M. Wedepohl and R.G. Wasley, "Wave propagation in The positive sequence impedance is: 2 3
polyphase transmission systems; resonance effects
due to discretely bonded earth wires," Proc. IEE,
vol. 112. DD. 2113-2119. Nov. 1965. Z = k.in (
[7] General Electric Co., Transmission Line Reference
Book 345kV and Above. New York: F. Weidner & Son
vp12p13p23
Printers, 1975.
[8] E.V. Rziha, Starkstromtechnik-Taschenbuch fuer the zero sequence impedance is.
Elektrotechniker, ("Electric power handbook for
electrical engineers", in German), Berlit: Wilhelm
Ernst u. Sohn, 1960. /8HHH ~ ~ ~ 2~ 2~ 2
91 Siemens, Formel-und Tabellenbuch fuer Starkstrom- Z o = kiQn jJ1/H1 32 32. 122 23 13
Ingenieure ("Handbook of formulas and tables for 2
electric power engineers," in German), Essen:
r
12 D13 D23
Girardet 1965.
[10] Westinghouse Electric Corp., Electrical and the off-diagonal elements are given by expressions as:
Transmission Distribution Reference Book.
and
Pittsburgh: Westinghouse Electric Corp., -1964. H P223 D D H P2 ~2
[11] H. Happoldt and D. Oeding, Elektrische Kraftwerke Z k [ 1 12 13 j In 3 12 D 13
und Netze ("Electric power plants and networks", in
German), Berlin: Springer, 1978. 2H3
1 /Pp DY 321 1
~~~~
01 "2%1n 2 1
Zol = k [ Qn _
I F1213 23 j 1 2 HD p
Discussion
A. Deri and G. Tevan (Technical University of Budapest, Hungary): We
wish to commend the author for his interesting paper. Even if a com-
puter may produce more accurate calculations, simpler expressions are We mention that Z12 .Z21 in the sequence impedances concerning the
often preferable especially when the computations are based on strong off-diagonal elements.
simplifying assumptions or the underlying data are not accurately known. It can be seen from the above formulae, that
We would like to make a few remarks concerning the calculation of 1) if the frequency is small (~50 Hz), p > > h, so that
line impedances using the concept of complex depth. This provides a
2H 2H 2 = 2H =P
12 =P 23 P 13 =H
=
powerful tool for highly accurate calculations for both earth return cor- =3
rections (Carson) and conductor skin effect [A].
1. The complex image formula is more important than shown in the
paper and its accuracy is higher: in the case of transmission line Z=kn D1 13D23
parameters, 0<450 (where tan- 0 = (hi + h2)/d), since the horizontal 1 r
distance between phases never exceeds hl + h2 for practical reasons, so These are the
in 'this case'the error of the real and imaginary part of the earth correc- ("handbook
tion term calculated from the complex image formula is less than 4 per- z = k.3in 3 H 2 formulae"
cent related to their exact value. The error related to the total impedance
is even less, since the geometric term to be added has considerable value.
0

9rD9
jI
All these are valid for the whole range of frequencies.
n13-iD2 ]
2. It happens at interference problems that 0 > 45 °. In these cases the 2
error in the complex image method is higher, especially in the vicinity
of 0 -900 but in these cases the commonly used asymptotic series for
z 12 =k
[3Qn 12D13
[-i + j
2i/3
i
a > 5 is also inaccurate. An additional' series has to be added to obtain D23
adequate accuracy [B].
= k [Tin D223
3. In relation to the sequence impedances, the complex image formulae I D
make it possible to obtain closed form expressions valid for the whole + j 1 i Qn
range of frequencies for both the positive and zero sequence impedances, /D12D13 2 v/3 D13
and also for the off-diagonal elements of the sequence impedance matrix
[Z], even for untransposed lines. 2) if the line is transposed and 1) is still valid, then ZI and ZO are
With H and P complex and D real quantities, given as above, while all the off-diagonal elements become zero.
The author's comments to our remarks would be appreciated.

H1= h1 p+ D12 h = (h +
l1
+ pi2 =H1+H2 2+ REFERENCES
H
2= h +p23
H2
2 +p D h =
D3
2 3 ) + d
23
2
23 2 3 H1
[A] A. Semlyen and A. Deri, "Time Domain Modelling of Frequency
/((,H.+H3)d-T23
23= 22 Dependent Three-Phase Transmission Line Impedance," IEEE
H .h +4p D13h/ihI)IT7d1 P13 = Paper No. 84 SM 713-4, presented at the 1984 Summer Power
1 3 1 Meeting in Seattle, WA.

Authorized licensed use limited to: ULAKBIM UASL - Yuzuncu Yil Universitesi. Downloaded on March 19,2021 at 08:09:23 UTC from IEEE Xplore. Restrictions apply.
 371

[B] G. Tevan and A. Deri, "Some Remarks about the Accurate Evalua- (3) Cable impedance. The impedance of a conductor with earth return
tion of the Carson Integral for Mutual Impedances of Lines with outside of radius r can be calculated from the formula*
Earth Return," ArchivfurElektrotechnik, vol. 67, pp. 83-90, 1984.
0
Manuscript received July 10, 1984.
7- =
3
=
2-T Qn ~~~r (5)
(5)
Adam Semlyen (University of Toronto, Toronto, ON, Canada): Most where p is the complex depth given in (2c). The errors are of the order
engineering problems are so complex that even a solution by computer of a few percent points. At larger values of the Bessel argument the evalua-
is only an approximation. Some solution-as in the case of transmis- tion of the Bessel functions becomes inaccurate, even with double presci-
sion line parameters discussed in this paper-are based on simplifying sion computation, but the error of the approximate expression (5) is still
assumptions and appear in analytical form, e.g., infinite integrals or small (the asymptotic approximation for the Bessel functions must then
series. These can usually be approximated by simple expressions, obtained be used for comparison).
sometimes by truncating the infinite series. The best approximation, even- It would certainly be useful to develop an approximation to a mutual
tually become "handbook formulae." It is certainly useful to review and impedance for conductors embedded in infinite or semi-infinite earth,
evaluate these from time to time because new and better approximations analogous to (5). This and other unsolved problems still constitute im-
are occasionally developed. portant and exciting challenges for the power system engineer.
In my discussion, I would like to refer to a group of formulae which In conclusion, I wish to express my appreciation to Professor Dom-
are based on the concept of complex depth. mel for freshening up and bringing new insight with his paper into the
(1) Carson's corrections for earth return. The infinite series given by somewhat staid topic of textbook formulae.
Carson converge well if the parameter r is not much larger than unity.
For r > 5 Carson has given asymptotic series which however, for 0 > 450,
are inaccurate for the reactance if r < 15. A corrected version [A] is now REFERENCES
available. If, however, closed form formulae based on complex depth
are used, the results are always sufficiently accurate in an engineering [A] G. Tevan and A. Deri, "Some Remarks about the Accurate Evalua-
sense, since the errors due to the simple closed form formula are less tion of the Carson Integral for Mutual Impedances of Lines with
than those resulting from the inaccurate knowledge of the earth resistivity Earth Return," ArchivfurElektrotechnik, vol. 67, pp. 83-90, 1984.
and from the assumption that it is constant. In the calculation of transmis- [B] A. Semlyen and A. Deri, "Time Domain Modelling of Feqeuency
sion line impedances the corrections for earth return effects are in error Dependent Three-Phase Transmission Line Impedance," IEEE Paper
by less than 4 percent and, since the linear (geometric) reactance is always No. 84 SM 713-4, presented at the 1984 Summer Power Meeting
dominant, the global error is less than 0.5 percent. The simplified for- in Seattle, WA.
mula is valid for all frequencies and geometries.
(2) Conductor impedance. For cylindrical geometry Bessel functions * Conjectured by L. M. Wedepohl, based on the obvious correctness
can be used for the calculation of the internal conductor impedance. For for Q-0 and t- oo; private communication.
large Bessel parameters the basic series give unreliable results and have
to be replaced by asymtotic approximations. A simplified formula, [B] Manuscript received July 10, 1984.
valid for the full range of frequencies, is:
F. L. Alvarado (University of Wisconsin, Madison, WI): This discusser
Z =/R'+Z o (1) commends the author on an interesting and educational paper. At one
c c
end of the accuracy spectrum we see the exact calculation of carson ef-
where Rc is the d.c. resistance and Z OO the impedance at very high fects, which must be performed by computers to be practical, and in-
frequencies: volves an infinite series. At the other end of the spectrum we see neglec-
ting earth effects altogether. Between these two extremes, a number of
1 useful formulas have been developed over the years for the calculation
= R c2 of series impedances. We present them here in order of increasing ac-
a 7T r curacy. Only the "self impedance" formulas are described in detail. The
mutual impedance formulas are slightly more complex. Refer to the
_0 a2Trr1 p
a original papers for these.
y2rrr (1) The Rudenberg "equivalent depth of return" formulas [1]. The
Rudenberg formulas provide an additive correction term to the ideal
with ground series impedance. Expressed in more modern terminology than
1 that in the original paper, the formulas for this correction term are (in
p=.- SI units):
0JX rflf De

representing the complex depth for the conductor. where

The error of the simple expression (1) is negligible at very low and
very high frequencies and even at intermediate frequencies it is less than
6.6 percent. In fact (1) is asymptotically correct at the power frequency
)
ee
- 1000/o P
Jj f
(60 Hz) and its lower order harmonics. Indeed, the expansion of (1) yields,
for w 0: - An interpretation of this last formula as an equivalent depth of return
is possible because of the logarithmic nature of the additive correction
z 2 term.
Z = R 1 + ( 2 Rc + =
Rc + j w
Lc (3)
(2) The Dubanton formulas [4,5]. In the Dubanton formulas, the ideal
c
2R
c ground series impedance formulas is corrected in an amount:

where taking (2) into account, Js= Iln (P- + 1)

where

K
2r°
1
L x (4)
c 4
Because of the logarithmic nature of the correction term, these formulas
also have a depth of return interpretation. However, because of the com-
plex nature of p, this depth of return is complex! The Dubanton for-
which is the well known expression of the internal inductance of the mulas are accurate within 12%o of the Carson values for all cases tested
conductor. by this discusser.

Authorized licensed use limited to: ULAKBIM UASL - Yuzuncu Yil Universitesi. Downloaded on March 19,2021 at 08:09:23 UTC from IEEE Xplore. Restrictions apply.
372

(3) The formulas developed by Alvarado and Betancourt [A21. These H. W. Dommel: The author wants to thank all the discussers for draw-
formulas involve a slightly more complicated but still closed form cor- ing attention to the new complex depth formulas. They should have been
rection term: emphasized more in the paper, because they are almost as simple to use
on electronic calculators as the older handbook formulas, and yet cover
the whole frequency range with reasonable accuracy.
1= ln (Ph + Professors Deri and Tevan are correct in emphasizing that the com-
s7 24
~~(1 + h)3 plex depth formulas are more accurate than indicated in the paper, if
where p is defined as above. P they are used for poWer line parameters only. This is clearly shown in
The Alvarado/Betancourt formulas are within 2% of the Carson values Fig. 5, which compares positive and zero sequence parameters obtained
in all cases tested. At most frequencies and resistivities the values ob- from Carson's formula and from the complex depth formulas, for the
tained using these formulas are virtually identical to those using the Car- 500 kV line defined in Table 1. The differences are hardly noticeable
son corrections. and well within the 4 percent quoted by the discussers. The differences
The Alvarado/Betancourt formulas do not have a natural "depth of do become larger in the calculation of mutual impedances, as needed
return" interpretation. They have, however, been implemented successful- in interference calculations, but even there the maximum difference of
ly in hand-held calculators (HP-41C) and in ordinary spreadsheet en- 10 to 12 percent is small if one considers the fact that the earth resistivi-
vironments (Lotus 1-2-3), something not easily accomplished with the ty is neither uniform nor well known. As mentioned in point 2 of the
full Carson formulas. discussion of Deri and Tevan, the asymptotic expansion of Carson's for-
In addition to the obvious advantages in simplicity of computation, mula becomes inaccurate as well in this case. This can either be corrected
all closed form formulas make it possible to obtain approximate analytic by the additional series of Ref. [B] in their discussion, or by extending
expressions that incorporate frequency-dependent effects. the range of the infinite series above a 5. The latter approach was taken
=

We wish to hear the author's comments on these approximate formulas. by Luis Marti at the University of British Columbia, who achieved good
accuracy by using the infinite series with more terms up to a 5 to 20. =

Similarly, he made the asympototic expansion more accurate by adding

REFERENCES more terms to it.
The author agrees with Professor Semlyen that the complex depths
[Al] R. Rudenberg, "Elektrische Schalvorgange und verwandte formulas are accurate enough for practical purposes, as clearly illustrated
Storungserscheinungen in Starkstromanlagen," Verlag Von Julius in Fig. 5. Professor Semlyen's complex depth formula for the self im-
Springer, Berlin p. 231, 1933. pedance of underground cables is very useful, and the author hopes that
[A2] F. L. Alvarado and R. Betancourt, "An Accurate Closed-Form a similar formula will some day be found for the mutual impedance.
Approximation for Ground Return Impedance Calculations," Pro- Finally, the author would like to thank Professor Alvarado for draw-
ceedings of IEEE, vol. 71, no. 2, pp. 279-280, February 1983. ing attention to the simple Alvarado-Betancourt correction term in the
complex depth formula, which reduces the error to the small amount
Manuscript received July 31, 1984. of 2 percent.
R
zero
(fl/mile) R
pos
(flmle)

12
io2

10-
10-24,
i&2
101 104 10 104
f (lIz) f (liz)
L (rom/mile) L
PoS
(mvd/nuile)
10 I 1.5 i

1.4

0 t-.-v
l.. 3
10-2 10I 104
104 io- 2 101 104 107
f (Hz) f (Hz)
(a) Zero sequence (b) Positive sequence
Fig. 5. Comparison between Carson's formulas and the complex depth
formulas for the 5oo kV line of Table 1.
Manuscript received September 14, 1984.

Authorized licensed use limited to: ULAKBIM UASL - Yuzuncu Yil Universitesi. Downloaded on March 19,2021 at 08:09:23 UTC from IEEE Xplore. Restrictions apply.