374

IEEE Transactions on Power Apparatus and Systems, Vol. PAS-103, No. 2, February 1984

EARTH RESISTIVITY MEASUREMENT INTERPRETATION TECHNIQUES

F. Dawalibi, Senior Member* Safe Engineering Services Ltd. 12201 letellier, Montreal, Canada. C. J. Blattner, Senior Member* Niagara iMohawk Power Corp. 300 Erie Blvd., Syracuse, New York.

ABSTRACT

Earth resistivity measurement interpretation techniques developed as part of a major EPRI research project on transmission line grounding are described and discussed. The interpretation techniques include graphical curve matching and an advanced computer program (RESIST). The curve matching technique requires a set of theoretical Master Curves with which a field curve can be compared directly. Program RESIST is based on the analytical methods used in a more elaborate computer program which has been in operation for several years. For an electrode spacing of 2.5 meters or greater, satisfactory agreement is obtained between measured and computed results using both interpretation techniques.

In both cases, the overhead network and the earth path, including buried metallic conductors such as counterpoises and ground electrodes, are part of the circuit in which the surge or fault current circulates. Generally, an analysis of these abnormal conditions is based on a reasonably accurate representation of the overhead circuit. The earth path, however, is usually modelled as a perfect conductor or in a very simplified form. This seldom leads to realistic results. The apparent inconsistency of these engineering approaches can be explained by the mathematical difficulties involved in the solution of three dimensional current flow in earth. Often, the wide variations observed in the characteristics of earth, generally described as a semi-infinite nonhomogeneous media, are used as justification for not pursuing detailed modelling of the earth path for fault currents. Thirty years ago, the lack of suitable high-speed digital computers was a serious obstacle to accurate modelling of the earth. Now, there are no computational limitations to the development of an accurate model of the structure of the earth. Recently published analytical works on power system grounding describe accurate, computer based, computational techniques for the design of grounding systems.

1.0

INTRODUCTION

This paper describes earth resistivity measurement interpretation techniques developed as part of a major research project on transmission line grounding. The Final Report [1l includes a comprehensive description of advanced theories and techniques pertaining to the analysis, design and measurement of transmission line grounding systems with a particular emphasis on safety and mitigation techniques to improve safety around exposed structures. Several analytical methods described in the report are new or were not available previously in the open literature. This is the case of the resistivity interpretation method described in the appendix of this paper. The research project was sponsored by the Electric Power Research Institute (EPRI).
The design of a power system requires that normal and abnormal conditions be considered in order to correctly determine the design requirements and characteristics of the installed power equipment. Of the abnormal conditions which can occur on a power network the two most frequent are:

Large variations in earth resistivity need not be an obstacle to the development of detailed earth structure models. Relatively simple equivalent earth models can effectively be used to accurately predict transmission line grounding performance as evidenced by the field measurements described in reference [1]. Finally, the earth structure at any particular site can be accurately determined by a suitable selection of the method and test equipment.
1.1 DISCUSSION OF MODELLING PROBLEM

D Lightning strokes
o Phase to

The development of a mathematical model to represent the electrical properties of earth can be a formidable task because of the widely nonuniform characteristics of earth. Fortunately for transmission line grounding purposes, the earth can be reasonably approximated by a two-layered soil structure. This soil structure is characterized by the layer resistivities P1, P2 and the upper layer thickness h. The lower layer is considered infinite. In some cases the thickness of the upper layer is large enough so as the earth model may be considered fairly uniform.
The variables P1, P2 and h are generally determined by interpreting the apparent resistivity values measured using the Wenner (or four probe) array.

ground faults

* F. Dawalibi is the principal author of the EPRI report C. Blattner served as an referenced in this paper industry advisor on the EPRI Task Force.
.

A paper recommended and approved 83 SM 456-1 by the IEEE Transmission and Distribution Committee of the IEEE Power Engineering Society for presentation at the IEEE/PES 1983 Summer Meeting, Los Angeles, California, July 17-22, 1983. Manuscript submitted January 10, 1983; made available lor printing May 13, 1983.

Unlike most engineering problems, interpretation of earth resistivity measurements is an "inverse" problem; i.e., from the electrical response to impressed current at specific locations on the earth surface, the electrical properties of the conducting media (earth) are to be determined. In contrast, conventional electrostatic problems determine the electrical response or the excitation current sources, based on the known properties of the conducting material. These are known

0018-9510/84/0002-0374$01.00 ( 1984 IEEE

375
as the Laplace and Diriclet problems. Obviously, the "inverse" problem, where the physical constants of the material are unknown, presents more difficulties than those problems where the physical constants of the material are known functions of position.
FOUR TERMINAL TEST SET

model of the earth structure is usually so great that it is difficult to choose initial values to these parameters and have a computer algorithm converge to an acceptable solution within a practical time frame. Consequently, the selection of initial values becomes a fundamental task in the interpretation process.
Success or failure in generally dependent on the knowledge of earth engineer responsible
measurements. this important initial assessment is the experience of the engineer and electrical properties available to the for the interpretation of the

Moreover, the number of parameters required to represent a

POTENT AL PROBES

CURRENT PROBE

Figure 1.1

Wenner Arrangement

1.3 POINT SOURCE ELECTRODE IN A TWO-LAYER EARTH

There are several mathematical approaches which have been used to calculate potentials in a layered soil structure [3-10]. The method which is adopted here is to search for a particular solution to Laplace's equation which satisfies the boundary conditions of the problem. One of the simplest and most important eprth structure models is the two-layer earth model. It can be shown that the potential in earth can be always expressed as the sum of a normal potential (uniform soil), and a disturbing potential, which accounts for the deep layers of soil. Therefore, the two-layer model can be used as the simplest equivalent earth structure for interpreting practical resistivity measurements, as shown in Figure 1.2.

There is one further problem with the "inverse" solution of resistivity measurements. It is not always possible to obtain a unique solution to a data interpretation problem. Because of inaccuracies in the measurements (usually 5% with classical geoelectric instruments), several models of earth structure can be found to give satisfactory agreement with the measured results. These models will usually differ in the characteristics of the deep soil layers.
The above discussions are not presented to discourage the power system engineer from performing a scientific interpretation of resistivity measurements, but rather to make him aware that this task requires careful preparation, investigation, and engineering judgement. The difficulties mentioned previously, while imposing a considerable challenge to the geologist, have significantly smaller impact on the electrical engineer. Firstly, the existence of multiple solutions to the substratum structure is of little consequence in determining the response of ground electrodes, particularly those of the transmission line tower grounds. Secondly, a two-layer earth model, is generally sufficient for modelling substation and transmission line tower grounding systems. Finally, there are numerous charts, algorithms, and simple engineering visual estimation techniques, which can be used to determine an equivalent two-layer earth-model with reasonable accuracy.

P1

M(x,y Z)

Figure 1.2

The two-Layer Earth Model

1.2 RESISTIVITY MEASUREMENT

Because of the wide variations in the structure and properties of earth materials, there are numerous methods and techniques for determining the structure of earth. A complete description of all of these methods is beyond the scope of this paper. However, information on a variety of these methods may be found within [1]. In power system engineering applications, the Wenner method L2] is used almost exclusively for resistivity measurements. It is our opinion that this method, when used with suitable test equipment, will provide sufficient data for an accurate earth structure model for the analysis of power system grounding installations. Because of the relative simplicity of results interpretation and widespread availability of test equipment supporting the use of this method, it is recommended as an effective and suitable standard test

The apparent resistivity, Pa, as measured using the Wenner method is derived for the two-layer earth condition in Chapter 4 of LlJ and is expressed as:
p

Pi

1 +

4kn

)2

(i 1)
-

1 [4+(2nh/a)21]2 JJ

where,

Pa = the apparent resistivity as measured using the Wenner Method.

a
= separation distance between current and potential probes.

procedure.

The Wenner four-electrode arrangement is shown in Figure 1.1. Four electrodes are driven in the earth alonq a straight line. The electrodes are uniformly spaced and the burial depth of the electrodes is usually less than 10 per cent of the spacing between two adjacent electrodes. Thus, each electrode will appear as a point with respect to the distances involved in the measurement.

PI
k

= surface layer

resistivity of thickness h
an

P2 = second layer resistivity which extends to depth.

= reflection coefficient
=

infinite

(P2 - P1) / (P2

P1)

376
1.4 POINT SOURCE ELECTRODE IN EARTH WITH INCLINED LAYERS

If the boundary separating two regions of earth with different resistivities is not horizontal, but inclined at an angle to the surface, an exact mathematical solution for the potential function is arduous to obtain. The potential function is obtained through a double integration process on two dummy variables, of a complicated function containing hyperbolic sines and Bessel functions of the second type [3,9,10].
When the angle of dip is 90ga simple solution can be obtained. Figure 1.3 illustrates this case. The case where a vertical interface between two dissirnilar earth resistivity areas exists will be referred to as a vertical fault structure. In practice, this effect can be the result of a number of different mechanisms, included but not limited to geological faulting.

P2

/10"- 2 C
h

Line of Fault

t
m

.A
I0

rL

d2
-

-I-

dI

PI

Figure 1.4

Vertical Fault (Top View)

V2

1.5 INTERPRETATION OF THE

I

MEASUJREMENTS

The simplest interpretation problem is the case where the measured apparent resistivities, Pa, vary minimally around an average value P. This indicates that earth at the measurement site is reasonably uniform and has a resistivity P.
Observed resistivity variations can be attributed to small local discontinuities, which may be neglected, or to inaccuracies in the measurements due a number of factors such as stray currents in earth or inadequate sensitivity of the measuring equipment.

a - SIDE-VIEW

- TOP- V IEW

Figure 1.3

Vertical Fault

The apparent resistivity, Pa, as measured usingIthe Wenner method, for the vertical fault condition, is deDrived within [1] and is expressed as:

Unfortunatley, such cases rarely occur in practice. In most cases, apparent resistivity, plotted as a function of the electrode spacing, shows large variations with probe spacing. This indicates that the earth is nonuniform.
In general, apparent resistivity curves change smoothly and do not exhibit sudden changes. When the latter occurs, it is a clear indication that the array has just crossed a vertical fault or a local discontinuity close to earth surface. The magnitude of the jump is an indication of the difference between the resistivities of the two adjacent earth materials. The presence of buried pipes or other metallic structures close to the surface is also a typical cause of sudden changes in apparent earth resistivity, as shown in Figure 1.5.

Pi
Pa =

(1-k)

11+ k2

K( 1-K)

[4(sinw+h/a)2+COS2J]2

K(1+K )
[4 (s nub-h/a) 2cos2~]21 [4(s inw-h/a )2+cos2]2

(1 -2)

where,
Pa = the apparent resistivity as measured by the Wenner Method.
a
= separation distance between current and potential probes.
L/)

P1 = resistivity of region on one side of vertical fault line.

cL a-

fault line.

P2

resistivity of region

on

opposite side of vertical

P I PE

h = distance from center of array direction and the line of fault (See Figure 1.4).

SPAC NG

ci = angle of array direction to line of fault (See Figure 1.4).

Figure 1.5

Presence of Buried Metallic Structures

The method used for interpreting the measurements can be grouped into two simplified categories:
o Empirical interpretation
o Analytical interpretation

377 taken by an experienced crew under the best measurements of conditions, will never give a perfect match with analytical results computed from the optimum earth model derived from the measurement data.

2.0 LOGARITHMIC CURVE MATCHING

The apparent resistivity functions (equations 1-2 and 1-3) may be written in terms of the dimensionless ratios K, Pa'P1 and h/a:
Horizontal Two-Layer Earth
00

Analytical interpretation is, in theory, independent of the person conducting the interpretation. In contrast, the results of an empirical interpretation are significantly influenced by the background and experience of the interpreter.
It is preferable to use a combination of both approaches for maximum accuracy and a minimum of uncertainty. For example, when analytical methods indicate that two or more earth models are reasonable, the most realistic choice can be determined from empirical considerations or visual inspection of the curves. In any case, it should be emphasized that experience is of paramount importance in the interpretation process.

pa /p

Kn[ [

4n2(h/a)

2]

(2-1 )
-

4n2(h/a)2

2]j

Empirical methods are based on experience gained through numerous measurement and interpretation exercises. Thus, such methods can be described as statistical in nature. Essentially, it is observed that the shape of an apparent resistivity curve is closely related to the earth structure and its characteristics at the site. Therefore, certain properties of the measured curve are used to deduce the resistivity and thickness of the earth layers. Although there may be inherent inaccuracies in some of these methods, they are of less consequence to the design engineer than they are to the geologist. Empirical methods may be useful for on site interpretations and serve as a good starting point for more rigorous methods. A complete discussion of empirical methods is included in the referenced EPRI report, and will not be repeated in this paper.

Vertical Fault

Pa"/
+

(1

K)

1 + K2 +
-

K(1-K) [4(sinw

+ h/a) 2
n

to -,).

cos 2w ]

K(1+K) [(2sinw-h/a)2+cos2j1

If one of the above dimensionless apparent resistivity functions Po = Pa/ Pl is plotted in logarithmic coordinates, i.e., ln(Pa/ PI F(ln(a/h)) the coordinates, of a point on the resistivity curve will be:
y
=

lnpa

lnp

on the y axis

(2-3)

Analytical methods, as termed herein, follow a constant methodology, i.e:

Step 1 The measured results are examined and preliminary interpretation is performed, typically based on the empirical methods described previously. Step 2 One or several possible earth models are proposed.
Step 3 The measured results are compared with those calculated from the proposed imodels.
Step 4 The most suitable model is retained. If more that one model is suitable, these are considered to be equivalent.

x = Ina - lnh on the x axis (2-4) If we now assume that a number of apparent resistivity reference curves, designated as Master curves, are plotted, for various reflection coefficients K, then: (2-5) y = lnp

x = lna Assuming that Pl = 1 Q-m and h = 1 m.

(2-6)

Resistivity curves derived for two-layer earth structures where P1 and/or h are not equal to unity, will be shifted by -ln(P1) vertically and -ln(h) horizontally with respect to the corresponding Master curves. The shapes of the curves are thus preserved. This property of the apparent resistivity curve when plotted in logarithmic coordinates, is the basis for the logarithmic curve matching method.

Step 5 The selected model is optimized. Often the opimization process is based on engineering judgment. Sometimes one may choose to conduct additional surveys in order to check the validity of some assumptions or to eliminate uncertainties. However, in power system design this step is generally omitted.
Two analytical methods of earth resistivity measurement interpretation were developed for the EPRI project and are described in the following sections. First, the use of precalculated curves is described. Next, a computer program (RESIST) based on the method of steepest-descent is described.

Thus, a field curve can be compared directly with a set of theoretical Master curves, through a series of appropriate translations of the field curve plotted on a transparent paper. If a satisfactory match is found between the field and a theoretical curve, then the real earth reflection factor K, is equal to that of the computed curve.
This method requires that a set of precalculated Master reference curves be available to the interpreter. Such curves can be easily determined based on Equation 2-1 for horizontal layers and Equation 2-2 for a vertical fault. Figure 2.1 is a set of Master curves for the case of a two-layer earth. When a field curve falls between two curves, the corect values can be interpolated. If more precision is required, additional curves for closer K values should be constructed.

The word 'analytical" can be misleading as it is often interpreted as meaning "accurate" or "rigorous". Earth resistivity measurements are rarely accurate to within 1%, even when sophisticated equipment is used. Usually, careful measurements with conventional equipment are accurate to within about 5%. Careless measurements, inexperience or poor equipment can lead to measured results significantly different from the real values. Even

Figure 2.2 is a set of Master curves applicable to vertical fault when the direction of the traverse is at 00 angle with the line of fault. It should be noted that additional charts (different traverse angles) are required for a complete set of master charts. The construction of such charts is straigthforward with a programmable calculator. Reference charts for 30, 60 and 900 angles are provided in Volume 2 if El].

378

3.0 COMPUTER PROGRAM RESIST

The resistivity measurement interpretation process is considerably simplified when program RESIST is used. Detailed information on the use of this program is included in Chapters 3, 4 and 10 and in Appendix B of [1]. Because RESIST requires a minimum of data from the user, it can be used by engineers with very little exposure to power system grounding. The input data consists of the probe spacings and apparent resistances (or resistivities) measured using the Wenner method. RESIST automatically selects and calculates all the other data needed to proceed with the final computations.

For example, RESIST must select initial values for the upper layer thickness and layer resistivities before starting its iterative search algorithm. The initial values retained are appropriate for relatively small grounding systems such as those employed in transmission line structures.
This program determines an equivalent two-layer earth model from the measured apparent resistivity data. The resistivity values must have been measured using the equally-spaced four probe or Wenner method. The equivalent earth model is characterized by the thickness of the first layer and by the resistivity values of the upper and lower layers of soil.

RESIST was developed for use in an interactive mode. In this mode, the user is prompted to provide the required input data and answers in free-format.
The program can also be readily operated in batch mode. In this mode, the user must organize the data in a file prior to running the program. The batch mode is particularly advantageous when running several cases with small differences in the input data.

Figure 2.1

Horizontal Two-Layer Earth Master Chart

)o

-to

PA

/10'
VERTICAL FAULT/

-.1

-.5
-.7

_2

_5

_1 .5 _1

... . . . . . -. . . . .~~~~~~~~~~~~1.0 ... -.9 .

-.99

Figure 2.2

Vertical Fault Master Chart

379

The following method was used as the basis for the computer program RESIST, developed by SES. This method was selected for the following reasons:
* A more elaborate version of the method has been operational for several years and has been extensively tested in practical cases. It has proved to be a very stable, reliable method which requires a minimum amount of input data. Initial values for the presumed earth model are not required from the user. Therefore, it can be used by engineers with limited experience in resistivity interpretation. * The method is easy to implement.
* The program is particularly suited to transmission line tower grounding design where very large spacings between the electrodes of an array are not necessary.

Table 4.1 clearly shows that the soil at the site is not uniform. Therefore, an equivalent two-layer earth model must be determined to improve the accuracy of the grounding performance calculation. The resistivity values of Table 4.1 were used as input data to computer program RESIST as shown in Figure 4.2.
......... etc.

YOU MAY ENTER 1- THE APPARENT MEASURED RESISTANCE (V/I), OR 2- THE APPARENT RESISTIVITY VALUES APPARENT RESISTANCE ?
ENTER YOUR FIELD DATA RESULTS IN SEQUENCE. ONE LINE FOR EACH MEASUREMENT AT A GIVEN SPACING. TERMINATE WITH END AT THE BEGINNING OF A NEW LINE. EXAMPLE SPACING (M OR FT), MEASURED VALUE (OHMS OR OHMS-METER) 5.0,433.0 ?2.5y320 ?5.0,245 X7.5,18 ? 10.0,162 X 12.5,168 ? 15.0,152 ? end
......... etc.

This program is based on the method of steepest-descent.

The method of steepest-descent is most readily vizualized by an analysis of the two-variable function p(x,y), illustrated in Figure 3.1. The analytical development of this method is shown in the Appendix. The gradient of this function is calculated at an initial point Mo defined by xo, yo. The values of x and y are then selected so that the function decreases along the direction defined by the gradient vector. The process is repeated until the function along the initial direction starts to increase. The process will stop when all possible directions of the gradient indicate that the present (x,y), coordinates corresponds to a minimum of the function (zero gradient).
This process will normally converge to a minimum of the function. However, there is no guarantee that the rninimum obtained will be the only one nor that it is the minimum of the minima. Experience shows that when a secondary minimum is obtained, the initial starting point was most likely within the zone of influence of this minimum. In this case another pair of initial x and y values should be selected and the process started again.

Figure 4.2

Program RESIST Input Data

The computer printout, shown in Figure 4.3, shows that the equivalent earth resistivity structure determined by RESIST corresponds to a 2.56 m (8.4 feet) thick first layer with a resistivity of 383 Q-m underlain by a second layer with a lower resistivity value of 147.7 Q-m.
COMPUTATION RESULTS TOP LAYER RESISTIVITY 383.4982 BOTTOM LAYER RESISTIVITY 147,6571 - 4440 REFLECTION FACTOR 2.5626 TOP LAYER THICKNESS

OHMS-METER OHMS-METER (P.U.) METERS

M (x ,y)

MINIMUM

Figure 4.3

Computation Results

Figure 3.1

The Method of Steepest-Descent

4.0 A TYPICAL EXAMPLE

The apparent soil resistivity at the site of a power system grounding installation was measured using the Wenner method. The results of the measurements are given in Table 4.1.
PROBE SPACING (i) 2.5 5.0 7.5 10.0

A comparison between the actual measured resistivities and the calculated apparent resistivities based on the two layer earth structure determined by RESIST is shown in the computer printout in Figure 4.4.
SPACING CALCULATED APPARENT MEASURED APPARENT DISCREPANCY (METERS) RESISTIVITY(OHMS-M) RESISTIVITY(DHMS-M) (PERCENT)

RESISTIVITY (Q-m)
320

APPARENT

245
182 162 168 152

2.j00 5.000 7,500 10.000 12.500 15,000

3271.4372 233.8807 187.4388 168.0307 159.5344 155 3505

320.0000 245.0000 182 .0000 162,0000 168.0000 152.0000

2.32 -4,54 2.99 3. 72 -5,04 i-220

iO

12.5 15.0

Figure 4.4

Comparison of Resistivities

Table 4.1

Apparent Soil Resistivity

380 An almost identical two-layer earth model can be determined using the logarithmic curve matching technique described in Section 2. The field resistivity curve is plotted on a transparent logarithmic graph and is then compared directly to the set of theoretical Master curves provided in Section 2 (Figure 2.1). The comparison process, illustrated in Figure 4.5, consists of obtaining a satisfactory match between the measured and theoretical curves through a series of appropriate horizontal and vertical translations of the transparent graph sheet. When this is accomplished, the thickness of the upper layer is the value of the abscissa on the vertical line passing through the Master chart abscissa corresponding to a/h=1. From Figure 4.5 it is found that the upper layer thickness is approximately 2.5 m (8.2 feet). Similarly, the upper soil resistivity is the value of the ordinate on the horizontal line passing through the Master chart ordinate corresponding to Pa/P1 = 1. Fiqure 4.5 shows that the upper soil resistivity is about 390 Q-m. Finally, this figure indicates that the measured apparent resistivity curve corresponds to a reflection factor in the range of -0.4 to -0.45. This leads to a lower resistivity value between 148 and 167 Q -m.
A comparison of the two-layer earth models obtained by computer program RESIST and the curve matching technique is shown in Table 4.2. The results demonstrate that the results are almost identical.

CONCLUSIONS Advanced earth resistivity measurement interpretation techniques have been developed and demonstrated. The new techniques include:
a- Logarithmic curve matching utilizing precalculated Master curves with which a field curve can be compared directly.

b- An advanced computer program RESIST which is based on the method of steepest-descent. The program is particularly suited to transmission line tower grounding desi gn. The interpretation techniques demonstrate that large variations in earth resistivity measurements need not be an obstacle to the development of detailed earth structure models. For power system grounding purposes, the earth can be reasonably approximated by a two-layer soil structure. For an electrode spacing of 2.5 meters or greater, satisfactory agreement has been demonstrated between measured and computed results using both techniques.

ACKNOWLEDGEMENTS
The authors wish to aknowledge the financial support of the r7I--:^ h instLtute Tor k;" rot r iOh -niof-f for M-_ n electrical t-ower researcn T FF F- rInis researcn rojeCL. Mr. J. Dunlap, the EPRI Project Manager, and the Advisory
.-

RESIST
--------

CURVES
----------

Top Layer Resistivity Bottom Layer Resistivity Top Layer Thickness

Table 4.2

383

390

Task Force members; Messrs. T. E. Bethke, A. C. Pfitzer, G. B. Niles, and R. S. Baishiki, are also acknowleged for their assistance and guidance on this EPRI project.

147.7
2.56 m

148-167
2.5 m
R EERENCES
1- F. Dawalibi, "Transmission Line Grounding", EPRI Research Project 1494-1, Final Report EL 2699, October 1982.

Comparison of Interpretation Results

P.

2- F. Wenner, "A Method of measuring Resistivity", National Bureau of Standards, Scientific Paper 12, NO. S-258, 1916, p. 499. 3- G. F. Tagg, "Earth Resistances", George Newnes Ltd., London 1964 (book).
4E. D. Sunde, "Earth Conduction Effects in Transmission Systems", Dover Publications, New York, 1968

(book). 5on

F. Dawalibi, D. Mukhedkar, "Influence of Ground Rods CGrounding Grids", IEEE Transactions on PAS, Vol. PAS-98, No. 6, November/December 1979, pp. 2089-2098.

6- S. Stefanesco, C. & M. Schlumberger, "Sur la Distribution Electrique Potentielle Autour d'une Prise de Terre Ponctuelle dans un Terrain a Couches Horizontales Homogenes et Isotropes", Journal de Physique et Radium, Vol. 1, Serie VII, No. 4, 1930, pp. 132-140.
7on

M. Muskat, "Potential Distribution About an Electrode the Surface of the Earth", Physics, Vol. 4, NO. 4, April 1933, pp. 129-147.

8- H. M. Mooney, E. Orellana, H. Pickett, L. Tornheim, "A Resistivity Computation Method for Layered Earth Models", Geophysics, Vol. XXXI, No. 1, February 1966, pp. 192-203.
9K. Maeda, "Apparent Resistivity for Dipping Beds, Geophysics, Vol. XX, No. 1, January 1955, pp, 123-139.

Figure 4.5

Logarithmic Curve matching Technique

10- L. S. Palmer, "Examples of Geolectric Surveys", IEE Journal, Vol. 106, part A, June 1959, pp. 231-244.

381 APPENDIX
or
At
=

Let PO(aj), j = 1, n be the series of apparent resistivity values as measured at a given site by the Wenner method for n different inter-electrode spacings aj. Let P (aj), j = 1, n be the calculated apparent resistivity values, based on a two-layer earth model at the same spacings aj used during the measurements.

F-T p1
pil

2 [ a~ a) IT

K~

rLh

12]
J

(4)

The sought for minimum is obtained when AW = 0 or practically when:

IA

<

E
algorithm

(5) are

The interpretation task consists of finding the most suitable earth model for which the difference between the set of measured and calculated values, according to certain criteria, is a minimum. In theory any criterion can be used (e.g., sum of the absolute value of the differences). In practice, the classical least-square criterion is preferred.
Let + ( P1, K, h) be the square error function defined as: ')(Pj, K, h) =

where E is the desired accuracy.

The main steps in the steepest-descent therefore:

KO, h)
(1)

1- Estimate initial values of P1, K and h (i.e.,

Pl

~p(aj) _p (aj ) E[- )P(a )

n

l2 J

2- Calculate a suitable value of T 3- Determine AP1,AK and Ah

The best fit is obtained when p is minimum. The values of Pi, K, h which lead to this minimum are determined by the steepest-descent algorithm.
The gradient vector is defined as:
V=

4- Estimate a new starting point:

Pi

0)

(i-i)

+ AP
+ AK
+

K(i)
aK

K(1-i)
= h

ap1

ah

h(
5- Calculate
a- if
b- if

i-)

Ah

Ail and compare it with a:

Ai
< a , the fit is completed.

Each component of the potential vector is determined from Equation 1. Thus: n n2

apJ
aK

)P a

(2a)

(or step 3 if T is maintained constant).

IAyI

> E , continue the process at step 2

(2b)
1
n

p
0

aK
ah

ap
ah

(2c)

In order to calculate Aip from Equation 4, the partial derivatives of P must be known. These are determined from Equation 5 where the partial derivatives of the theoretical two-layer earth apparent resistivity function are obtained from Equation 4-46 of the EPRI report [1]. These calculations lead to:
00

Assume now that AP1, AK, Ah are small stepwise changes along the gradient represented as: Ap1
= -T -

Upl

= - I +

n=l

Kn(

n(I

K2)/2K) (A:
-- a ]

(6)

DP,

(3a)

AK = -T

ap
(3b)
-=

16plh
a2

ah

Kn

A-'

B-'

(7)

Ah

= -T

(3c)
4 nKnl n=l
where A = 1 + (2nh/a) 2
B
=

11:

Where T is a positive value expressed in p.u. of V), suitably selected to generate a smooth search for the m-ninimum. The above changes cause a small variation A in the error function

(A-:

B)

(8)

AWl= al APi

api

a* AK + a'p Ah a~~h 3K

(9)

382

Discussion Eldon J. Rogers, (Bonneville Power Adm., Vancouver, WA): It would appear the author's computer technique for resolving Wenner test earth resistivity into two-component earth would start with parameters determined from the logarithmic curve matching. I have used Roman's logarithmic curve fitting method for two-layer earth, described by the authors, to analyze earth resistivity data measured for the substation grid sit (Refer to [3], Ch. 3, pp. 72). Generally, Wenner resistivity survey data fall into several categories: The data are obviously due to twolayer earth and logarithmic comparison is easily made; the data appear to be two-layer but logarithmic fit is inaccurate; or, it is evident the data are 3- or 4-layer earth. As most of the substation data fall in the last two categories, the author's computer technique could be useful to determine their two-layer equivalent. Even for towers, earth resistivity variations may require more than one Wenner test location to adequately describe the earth volume. One important aspect of resistivity survey is determining the earth's resistivity near the surface. The earth's resistivity near the surface can be calculated from the measured resistance (Rn) of each Wenner test probe (Pn = YY L Rn/(In8L/D- 1) ). For a typical test probe length (L) of 0.3m, earth resistivity samples are obtained of an earth cylinder 0.5m in depth and 0.6m diameter. The composite earth surface layer resistivity is found by the parallel combination of all sample volumes
00

In his closure will the author include equations for the Vertical Fault case similar to Eqs. (6), (7) and (8) of the Appendix. Will his computer technicque be adaptable to the case of the rod electrode penetrating two-layer earth? The authors are to be congratulated for reviewing the logarithmic curve matching technique and developing a computer program to fit the two-layer earth model to the earth resistivity survey.

Manuscript received August 2, 1983.

F. Dawalibi and C. J. Blattner: We thank Mr. Rogers for his discussion and pertinent comments. The initial parameters required by the steepest-descent technique could be determined from the logarithmic curve matching method. Mr. Rogers points out rightfully that our technique is particularly useful to determine an equivalent two-layer model for complex soil structures. When soil is practically a two-layer configuration, then the logarithmic curve matching method provides accurate solutions. Program RESIST is designed to give the best two-layer fit based on the Wenner test values. There is no selection of an initial value of the gradient vector V. The components of this vector are defined by Eq. 2 of the paper. Guidance for the selection of the accuracy is included in [1] (see p. B-3). Eq. 4 has a correct numerator, (it was corrected after the review by the Technical Committee). We are grateful to Mr. Rogers for reporting the typo errors in Eq. 7 and 8. It seems that these mistypes appeared a few years ago when the first author was involved in the revision of IEEE Guide 81. Unfortunately, the mistypes were reintroduced in this paper and also in Ref. 1. Fortunately however, the code in program RESIST is correct. The case of vertical layers requires that the direction of the traverse and the location of each probe of the array be known relative to the vertical fault plane (interface betweeen layers). This introduces several possibilities with 3 equations similar to Eqs. 6, 7 and 8. Although we have not yet derived all these new equations., we believe that it is a relatively easy task (see p. 4-26 of [1]). The rod electrode penetrating two-layer earth requires major modifications to the computer algorithm. Manuscript received September 26, 1983.

[Pi

n/I (1/Pn) ).
I

When the composite Wenner test data for a grounding electrode site differ significantly from the two-layer earth model (for example, data from three-and four-layer earth), will the author's RESIST program determine the best two-layer fit? How are the p.u. value of V, "r", and the desired accuracy "e" Eq.(5) selected? It would appear their selection depends on how well the data fit the theoretical model. Would the author please discuss? What are the theoretical considerations which justify the assumptions used to form Eqs. 3a, 3b, 3c shown in the Appendix? There appear to be several typo errors in the Appendix: Eq. (4): First term has incorrect numerator. Eq. (7): Should have a negative sign in front of 16. a2, in the denominator n2, between summation symbol and kn Eq. (8): Should have P1 between 4 and summation symbol