Farid Dawalibi, Member IEEE Dinkar Mukhedkar, SeniorMember IEEE

The Shawinigan Engineering Co. Ltd. Ecole Polytechnique

Montreal, Quebec, Canada Montreal, Quebec, Canada

Abstract-Exact determination of current distribution along a has t o be demonstrated for unsymmetrical or morecomplexshaped
grounding electrode is, in all practical purpose, only solvable by succes- electrodes. The main problem being to find an electrical center for the
sive approximations. In this paper various methods which solve accu- electrode. Also, doubts were raised aboutthe adequacyof such a
rately the most general problemof interconnected electrodes are methodfor overlapping conductors which saturate immediate sur-
presented and discussed. One of these methods called herein the Multi rounding earth [ 7 ] . In such situations different current distribution
Step Analysis seems to have definitive advantages with respect to assumptions might have to be used.
accuracy, computing time and simplicity. It is hoped that the following Suchempirical currentdistributionfunctions assume thatthe
will be helpful for those who are concerned with a reasonable accurate current enters physically the electrode at its electrical centerand
design ofgrounding networks whichtakes intoaccountthe overall radiates subsequently in all directions. Is the above method still valid
parameters of the system. when current enters the electrode at other location(s)? It is obvious
fromthe previous, that definitive anwers can not be made for the
LIST OF SYMBOLS moment. Extensive investigations should have to be carried out before
n : Number of straight conductors in an electrode any conclusion can be made. Reliable answers can be obtained from
experimental measurements. This is one of the Authors futureobjective.
L : Total length of an electrode(Meters)
There are however other attractivesolutions.Oneofthem is
I : Injected current (Amperes) based on the so called Matrix Method [8],[ 9 ] . This method how-
ever requires large computing time. Alternatives to this method, which
l j : Length of conductor j (Meters)
require less computing time are described in this paper.
i : Linear current density (Amperes/Meter)
1. Calculation of the current distribution
s j : Current density distribution factorof conductor j
v..1J : Potential caused by conductor i at point j or Mean value of the In this section the potential due to a straight horizontal wire in a
potential caused byconductor i along conductor j (Volts) two layer earth is only designated by the letter V. The expression of V
is given in detail in reference [3] and in an equivalent form in a
R : Earth resistance of the electrode (Ohms) companion paper [ 10] .
x k : Designates a parameter of electrode k. The study is carried out in the following sequence:

x(h) : Designates iteration Number h - Case of an isolated electrode

- Case of interconnected electrodes
z : Impedance (Ohms)
- Case of additional return electrodes
INTRODUCTION
1.1 Case of an isolated grounding electrode
Exact solution in grounding problems requires determination of
current distribution in the grounding electrode. Unfortunately, even for
This case is illustraded in Figure 1
very simple problems such as a straight horizontal conductor located in
a uniform homogenous medium of infmite extent, the current distri-
bution along the conductor must satisfy an integral equation which has
to be solved by successive approximations [ 1] . Conductor j
;I1
Therefore, for complexelectrodes and non uniform soils, other
techniques must be used to avoid analytical complexities and excessive
study cost.
Various computer-aided techniques may be used for this purpose.
The selection of any technique being strongly dependant on the accu-
racy desired and economical aspects. However, Authors opinion is that
whatever themethod used, it should be adequateto determinethe
current distribution, not only along an isolated electrode but also for
interconnected complexelectrodesas in real life power systems. It
should also take into account effects of any close-in faults (Proximity
effects of return electrode). Conductor I
In previous publications [21 to [61 the authors demonstrated that (Xj, sjs Zj)
in the case of symetrical grounding grids a first approach would be to
assume a linear current distribution with respect to the centers of the Fig. 1 Isolated Electrode
main and return electrodes. The analytical results fitted satisfactorily
with the experimental ones. However the adequacy of such a method Lets consider an isolated groundingelectrode composed of n
conductors j (j=l, n) andburied in a two layer earth.
Paper F 75 522-3, recommendedandapproved by the IEEE Substations For the purpose of the study theelements j are straight conductors
Committee of the IEEE Power Engineering Society for presentation at the IEEE
PES SummerMeeting, San Francisco, Calif., July 20-25, 1975, Manuscript sub- and their length are small compared to the total length of the electrode
mitted January 31,1975; made available for printing May 5,1975. (lj << L).

113
 Let xj, yj, zj be the coordinates of the center of each conductor
j and V1 the potential induced by conductor i at point xj, yj, zj (this
point is at the surface of the conductor). Let also ii be the linear current
density of conductor i and I the total current injected by the electrode
into earth. Finally Si is the current density distribution factor and is and the following for Si
defined as follows:
I
si -= ii (1 1
L
1.1.1TheMatrixMethod

Neglecting the voltage drop in the electrode conductor is permis-

sible and therefore the following equations can be written:

h being the iteration number. The initial values of the s k should

satisfy equation (4).
The convergence speed being dependantonthe initial values
selected, it is evident that one should take advantage of.well known
physical rules which state that:
Where V is the electrode potential rise. - Outer conductors carry the heaviest current densities
This system of n simultaneous equations has n+l unknowns (V is
- Saturation occurs where great number of conductors share a
not known). It can be written as a matrix equation:
narrow surrounding media:
The voltage Vc is calculated from the formula given by the inte-
gration method of reference [ 3 ] (or as indicated in [ 101, Vi is then a
mean value)
The s o l u t i o n of which i s
1.1.3 The Multi Step Method

Themethods described above have the disadvantage of dealing

It is possible to solve thisequation by assuming V=l and then with numerous unknowns at the same time. In brder to reducethis
adjusting the 9 according to knownboundaryconditions (Whenever number of unknowns one can take advantage of the following:
the injected current is known;multiply by an adequate proportionality Experience and previous analytical work [ 1 1, [9] have shown that
factor). the resistance of an electrode is practically independant of the current
Unfortunately this method requires large computer time in order distribution in the electrode. Also the potential created by a conductor
to inverse the large size of the Vij matrix which is obtained from practi- at a remote point is independant of the currentdistribution inthe
cal electrodes used in the industry. Therefore, regardless of the accuracy conductor.
and simplicity of the method, other techniques should beconsidered Instead of subdividing the electrode into n elements in one unique
when economical studies are to be completed. step, let's divide the electrode in a smaller number of elements at each
step of a n step process. For example based on the previous remarks, it
1.1.2IterativeMethod is always possible to subdivide, at each step the grounding electrode in
two sections:
From equation (1) and the following:
- A conductor close to the point where thepotential is calculated
11 + 12 +...+11 +... 1, = L (point 1)
ill1 + 1212 +... iili +... i,l, = I - The remaining of the electrode (r)

It is simple to establish that: Consequently we can write the following:

Sill + s212 +....+S i l i +...+s,1, = L Sill + (L - 11) s,1 = L (7)

Multiplying then, the two sides of (2) by liVs leads to: 1 designates the conductor center (at thesurface of the electrode),
where the potential is calculated ( x i , y1, zl), and r1 the remaining of
the electrode subdivided with respect to point 1.
Also we can write that:
S l V l I + S,1V,1 = V RI (8)
=

Summing up, and afterreductionthe following expression is Where R is the resistance of the conductor.
obtained for V: From (7) and (8) the following is obtained:

114
 Repeating the above for different points i=l, n at the centers of
I
the electrodes, leads to: I

Fig. 2 InterconnectedElectrodes

It is possible to determine thereforethecurrent distribution

factors Si by assuming R=l and adjusting afterwards the Si according
to equation (4):
In order to solve the problem the study is carried out in successive
steps. At each step increased details of the electrode are introduced in
order to obtain theadequate accuracy.
A frst preliminary approach tothe problem has already been
However (9) and (4) provide an attractive way to determine con- presented by theAuthors in reference [41. Briefly, themethod
currently the $ and the resistance R of the electrode. Replacing the Si consists in the calculation of a linear current density factor Sa on the
by their values from (9) in (4) leads after reduction to: overall interconnected grounding area basis and a linear current density
factor Se on each individual electrode basis. This method however
yields to two major difficulties:
- Determination of electrical centers
R = - Selection of anadequaterelation forthe linear current distri-
bution.
The adequacy of such an approach has therefore to be justified by
its extensive application to various practical grounding configurations
and : followed by a detailed analysis of the results.

1.2.1 Matrix Method

Although the Matrix method yields to large matrix inversion, the

technique is described because of its pedagogic advantages.
Let i = I/L be the uniform linear currentdensity(theoretical)
over the interconnectedelectrodeand Sf, the linear current density
1.2 Case of Interconnected Electrodes factor of conductor j of electrode k. Slkis solutionofthe following
equation:
This case is illustrated by Figure 2 .
By interconnected electrodes we mean various grounding sites tied
together via overhead ground wires (or cables) as in any utility power
network. In such cases the voltage drop caused by zero sequence fault
I
t
>' 2
j-1 j=l
1s; = I

current flowing in theground wire, because of the large distances and therefore
involved is notany more neghgible. Also, in AC networks, voltage
angular shifts can not be neglected in some circumstances. t n,.
Let I be the total current injected in the group of interconnected
electrodes k (k=l, t) and Ik the total current injected in earth by each
individual electrode k. We can write:
By following the same procedureas for an individual isolated
t electrode it is possible to write the system of simultaneous equations
>:Ik=I
k-1 which hold for the determination of Sy. However it should be kept in
mind that the potential rise of the grounding electrodes k, Vk, are not
The total buried length of each electrode k being Lk and the total identical. Even more the Vk are complex numbers because of possible
buried length of the interconnected grounding network being L. Thus: angular shifts in Ac power systems.

115
 The potential caused by a straight conductor ih of electrode h at Close electrodes are those for which the voltage drop in their tie
the center of a straight conductor jk of an electrode k is such that: circuit can be neglected. In such a case the two electrodes are considered
as a unique unit. For distant and remote electrodes the above is not
t applicable.
However the remote electrodes do not interfere mutually in their
individual performance.
When this cladlcation is achieved the gterconnected system is
The nk relations above generate t other similar relations when k is composed of distant and remote electrodesonly.Consequently it is
varied from 1 to t. permissible t o say that the potential caused by an electrode h at any
t point ik of the electrode k is constant at all points ik (the value of the
Therefore we have n = kC1 nk simultaneous equations with n ~h constant is zero if the electrodes are remote).
lh
plus t Vk unknowns. The Vk,however are related to the V potential by In the first stepthe interconnectedgroundelectrodes are con-
following equations: sidered ,as a unique electrode of length L injecting in earth a current I.
Using the results of paragraph 1.1.3, on can write:
v k = V - Z kI k

Where Zk is the impedance of the tie circuit. Also, the following

equation holds for electrode k:

and
Thus

Equations (15) and (16) reduces the number of unknowns to n+l

(the %and V). It is possible to assume V = 1 and adjust afterwards the Where:
S according to definite boundary conditions.
Note that by virtue of the possible complex nature of the Zk, the
S in general are complex numbers.
R is the overall resistance of the interconnected grounding net-
I . 2.2 Iterative Method work.
$fi is the potential caused by electrode h at electrode k center
Adjusting the S by using limit conditions are avoided by noticing (assuming uniform current distribution in electrode h).
that the S has to verify also equation (14) this permits reducing the Sk is the c k e n t distribution factor of electrode k.
system to n simultaneous equations with n unknowns, which can be @fi can therefore be written as:
solved by inversion of the adequate matrix. The enormous usual size
of this matrix precludes the use of such a method.
Instead, it is possible to follow the same approach as in section
1.1.2 above.
Equatoins (15) and (16) when combined can be written asfollows:
In the second step of the analysis electrode k is considered alone.
The total buried length of electrode k being Lk and the injected current
Ik.Using equation (11) of paragraph 1.1.3 leads to:

f o r j, = 1, nk
and k = 1, t

In order'to avoid writing complicatedexpressionswe will just

state that by use of equation (14) onecan find the value of V and then
determine the iterative formula which gives the value of S&(w+l) as a However, Ik = S r L y L
function of the Wth iterative values S&(w).

1.2.3 The Multi Step Method With: V = RI

Any group of two electrodes of the interconnected network can Therefore:

be classified in three sorts: vk = RI -Z~S~L~I/L
- Close electrodes
- Distant
electrodes Consequently thecurrent distribution factor of a stmght con-
- Remote electrodes iuctor i of an electrode k is given by:

116
 v,k - v,k f V

I Where Vn is the driving voltage (phase to neutral voltage of the

faulted phase in case of an LC fault).
With these additional relations and remarks, the problem can be
Where R is given by (18), Sk given by (1 9) and the #& in (1 8), solved exactly as in paragraph 1.2.
(19) given by (20).
NOTES
1.3 Cnse of Additional Return electrodes
Usually, it is possible to simplify considerably the problem of
This case is illustrated by figure 3. interconnected network:
If two group of electrodes are remote the potential caused by one
on the other can be considered negligible.
I k
I ?I
v
Also if two groups of electrodes are very close, the voltage drops
and/or angular shifts can be neglected.
Ma i n
2. Application to classical simple electrodes and comparison between
Electrodes the various methods

In order t o illustrate the various methods, the latter were applied

to a number of classical electrodes buried in a uniform soil.
The electrodes used are shown in table I.
All the electrodes are assumed to be buried in a uniform soil of
100 ohm-meters resistivity at a depth e = 0.5 meter. The radius of the
conductors is 0.05 meter.
The results obtained are shown in figure 4 and 5 . Figure 4 cor-
responds to the square grids and figure 5 tothe radial wires confi-
guration.

Table I. Typical Electrodes Used

Electrod Form of Total Number

I Iden t i f i the length of

/Y/ cation electrode (meters) slvlslor:

Fig. 3 Additional Return Electrodes A 600

Lets consider two groups of interconnected electrodes. The first

group injects positive currents in earth (Main electrodes, or Emitters)
and the second groupinjects negative currents in the earth(Return
electrodes or collectors). It is assumed that the only common electrical
path between the two group of interconnected electrodes is the earth 400 16
B
media.
In order to determine the current distribution factor of any ele-
mentary conductor of the two groups, the methods used previously are
still applicable with however certain modifications and additions listed
below :
Positive values are assigned to thecurrent distribution factor, C 400 4
current, and voltage of the Emitters and negative values to the corres-
3r l 4
ponding parameters of the collectors.
The common potentialVb of theEmitters is positive. Each
,
Emitter k potential V i is related to V by the following equation:

k
Ve = Ve- Z k
e I ek ; ( k = 1, t e )
D 120 12
The common potential Vc of the collectors is negative. Each col-
lector potential V$ is related to V, by the following equation:

k k ; (k = 1, t c )
V t = Vc- Z,Ic
E 60 6
Note that If is negative.
One additional relation is however available here:

11 7
 It was foundthatforboth electrodesA and B the Matrix,
Iterative and Multi step methods give practically uniform current
distribution in the electrode. In addition, the Iterative and Multi step
method calculate theearth resistances according to (5) and (10)
respectively:
Iterative Method
R = 0.64529 ohms (16 subdivisions)
R = 0.64750 ohms ( 4 subdivisions)
Multi Step Method
R = 0.64526 ohms (16 subdivisions)
The well known Laurents [ 1 11 formulaR=& gives 0.4431 ohms
(r is the radius of the circle having an equal area to the grid)
It was also noticed (at least for the cases studied) that the iter-
ative method converges extremelyfastas one can conclude from
figures 4 and 5 . The rapid convergence applies also for the resistance
calculation using (5). For example in the case of grid A, the Iterative
method gave a resistance of 0.5584 at itsf i i t iteration. This value went
up to a maximum of 0.6027 and then decreased regularly and stabilized
at iteration 3 to 0.586048. Fromiteration 3 to 5 onlythefourth
significative digit fluctuated. For grid A, the Multi step method gave a
value of 0.568787 ohms while the classical method of the average
potential using asummationtechnique [31 gave a value of 0.5626
ohms.
Identical conclusions can be made for the radial electrodes D and
E. The calculated resistances are given in table 11.

Fig. 4 Current distribution Factor in grid A, along: A perimeter cor

ductor ( 1 , 2 , 3 , 4 ) . A radial conductor (17, 18, 19,20). Table 11. Radial Electrodes Resistances

The Matrix method is not recorded in figure 4. During compu-

tation, the inversion of the 24 x 24 matrix caused an overflow. How-
ever in figure 5 , a complete agreement was achieved by the matrix and
the Iterative methods.
In all cases the Multi step method forces more current in the outer
conductors while the inner conductors were correspondingly unloaded.

3. Conclusions

The Multi step method for the calculation of current distribution

in agroundingelectrode proves to be apowerfultechnique with a
number of important advantages:
1) Accurate results which are comparable to the Matrix method
results are obtained.
2) Permits the simultaneouscalculationofcurrentdistribution
and earthresistances, if desired.
3) Requires relatively low computing time
4) Can easily be extended to interconnected electrodes tied via
Fig. 5 Current distribution factor in grids D and E along one branch ground wires having fmite impedances, without tremendous increase of
(1.2.3.4.5.6). computing time and/or analytical complexities.

118
 The Iterative method introduced by the Authors gives identical T. N. Ciao, M. P. Sarma,Effectof two layer earth on the
results to the Matrix method. When the number of iterations are not electric field near HVDC electrodes, ZEEE Traniactions paper,
excessively large then its computing time is intermediate between the VO~.PAS-91, NO. 6. NOV.1972>pp. 2346-65.
F. Dawalibi, D. Mukhedkar, Resistance calculation of inter-
two othermethods. connectedgroundingelectrodes - Paper presented simulta-
neously at IEEE Summer Power meeting, 1975.
ACKNOWLEDGMENTS P. G. Laurent, Les bases generales de la technique des m i s s a
la terre dans les installations electriques, Bulletin de la Societe
Francaise des Elecmciens, 7ikme drie, tome1,No. 7, July
Mr. F. Dawalibi who is a registered Doctoral Student wishes to 195 1, pp. 368402.
thank Ecole Polytechnique for the facilities offered. Also the Authors G. FT. a g , Earth
Resistances. London: George Wewnes
express their appreciation to the National Research Council of Canada Limited, 1964.
and the Department of Energy, Mines and Resources of Government of
Canada for providing the necessary fmancial support of the work. The
Authors would like to thankthe managementof the Shawinigan Discussion
Engineering CompanyLimited,Montreal,Quebec, for their coopera-
A. B. Furdy (Appalachian Power Company, Roanoke, Va.): The
authors have presented a valuable paper which makes use of the modem
computers capability to solve the complexproblemof current dis-
REFERENCES tribution in grounding electrodes.
An area in which the program can be used immediately is to
E. D. Sunde, Earth conduction effects in transmission systems, determine the ground electrode configurations which make the best use
Dover Publications,New York, 1968 (Book). of the ground conductor.Thecurrent distributionand resistance of
F. Dawalibi, D. Mukhedkar, Ground electrode resistance simple types of electrodes can be computedto show which of the
measurement in non-uniform soils, ZEEE Transactions, Vol. existing equations give the best approximations for voltage gradients
PAS-93, NO. 1, January 74, pp. 109-1 15. and resistance. Arecently developed equationforthe resistance of
.F. Dawalibi, D. Mukhedkar,Optimum design ofsubstation horizontal ground electrodes gives 2.84 ohms for Type E in Table 11.
grounding in a two layer earth structure, Part I - Analytical If the ideas presented can be adapted to apower system with
study. IEEE Transactions paper, No. T-74-191-3. multigroundedneutral wires to determine fault current distribution,
F. Dawalibi, D. Mukhedkar,Optimum design ofsubstation perhpassafety, reliability and economy of construction will be im-
grounding in a two layer earth structure, Part I1 - Comparison proved.
betweentheoreticalandexperimental results. ZEEE Trans-
actions paper, No. T-74400-2. Manuscript received August 4,1975.
F. Dawalibi, D. Mukhedkar,Optimum design ofsubstation
grounding in a two layer earth structure, Part 111 - Study of
grounding grids performance and new electrodes configurations.
ZEEE Transactions paper, No. T-74-403-3.
F. Dawalibi, D. Mukhedkar, Sur les conditions de simulation
des mises B la terre Revue GLnPrale dElectricitd, Tome 83, Farid DawaIibi and Dinkar Mukhedkar: We will to thank Mr.A. B.
No. 3, March 74, Paris. Purdy for his interest in our work. We agree with him completely. We
See Discussions of Optimum design of substation grounding in are presently working to extend the work for multi grounded neutral
a two layer earth structure Part I1 and 111. wlres.
R. F. Harrington, Matrix methods for field problems Proc
IEEE, Vol55, pp 136-149, February 1967. Manuscript received October 1,1975.

119