“Meeting Code”
In its common form, grounding is typically thought of as no more than a rod driven into the earth in order to provide a safe diversion of lightning strokes.  In a sense, the National Electric Codeâ (NECâ) indirectly provides a basis for this conception.  The Code requires that a single rod or other approved electrode be installed in the soil and tested.  If it tests at 25 W or less, the installation “meets Code”.  If not, a second rod or other electrode is installed at least 6 feet away.  It need not be retested.  The additional electrode can routinely be expected to reduce the measurement by about 40%, but that still says almost nothing about what the final value might be.

However, meeting code is not all there is to ground protection.  The Code is a working directive promoting electrical safety.  It tacitly acknowledges that soil conditions are so variable that to insist on a universal absolute would be impractical and unfair.  A homeowner who has the misfortune to live in an area of high soil resistivity cannot reasonably be expected to drive a ground rod halfway to China.  If one is inadequate, the addition of a second will afford substantial improvement, and in the final analysis, “something is better than nothing.”

Likewise, in “meeting Code”, the implication is that the facility is safe, not necessarily functional.  The Code is concerned with safety, not performance.  The building may be protected from lightning and electrical faults, but still have “noise” on datacom lines.  The familiar 25 W standard is actually very forgiving, selected for practicality and basic protection, not optimal design.  In theory, one would want to ground an electrical system at zero resistance.  But this, of course, is not possible in the real world.  A realistic alternative goal is to get as close as possible, substantially cutting down the enormous gap between “making Code” and theoretical perfection.

On the “downstream” side of 25 W, there can be a nearly inverse relationship between resistance and performance.  Aside from the NEC, no universal ground resistance standard exists.  However, industry practices and insurance recommendations have established some familiar guidelines: 5 W for a typical commercial ground, 3 W for a chemical plant, 2 W (or even less!) for computer rooms and process-control operations, 1W for large utility substations and generating plants.  And if anything, these practices are becoming more demanding.  The increasing reliance on computer operations, process control, and datacom/telecom functions has made the presence of “noise” intolerable.  With data signals as narrow as 3 V, or even 1.5, the differences between “Xs” and “Os” can be scrambled by noise that means nothing to the operation of standard 120/240 equipment.  Voltage regulation is important as never before, and ground is critical in the mitigation of internally generated noise as well as external faults and disturbances.

Therefore, the popular image of a single rod for lightning protection is only the beginning.  This paper will deal with ground testing as it fits in with the implementation of maximum grounding efficiency.  To attain a high level of protection, one must first know how to measure.  The practice of ground testing is much abused, primarily through failure to recognize its unique properties.  Proper measurement breaks down into two elements: equipment and procedure.

The Right Equipment
Good ground measurement begins with proper equipment.  Ground testing presents challenges unlike any other in the arena of electrical testing.  The first line of error is one of faulty logic: to make a resistance measurement requires an ohmmeter.  Wrong!  To perform a ground test, it requires a ground tester; that is to say, an instrument specifically designed to meet the unique factors involved in testing unlimited earth.  A common mistake is to use a generic multimeter, with one lead connected to the test ground and the other to an arbitrary reference ground.  This procedure will provide a measurement, but the critical question is, what is it actually measuring?  This technique suffers from at least three potential sources of error: interference, extraneous resistances, and the arbitrary position of the reference.

The fact is easily overlooked, but the earth carries a lot of “noise” from transients trying to find their way back to transformer secondaries.  Multimeters are DC testers, and their readings will be influenced by whatever voltages may be present in the soil.  The operator may be made aware by destabilization of the display, but there is no specific indicator to provide warning.  Secondly, the reading...influenced or not...is a series resistance that includes the soil and everything else in the loop.  It would be nice if this were zero, but that’s not likely.  The reference ground is assumed to make a negligible contribution, but that is only an assumption, largely untestable.  The most commonly employed reference is the water-pipe system, but if this has been repaired with plastic pipe or couplers, its usefulness is negated.  Finally, even if there is no interference or additional resistance from the reference, the reading still may not be reliable.  Under these latter circumstances, a generic multimeter may provide a good reading of soil resistance between the two points.  This may be an accurate measurement of ground resistance...and it may not.  It can only be accepted on faith.

Figure #1:Grounding Electrode & Immediate Surrounding Soil

 

Most electrical testing is performed on discrete circuits of human design.  The elements are known and their properties can be routinely addressed.  Not with ground testing.  Making an electrical connection to a grounding electrode thereby includes the entire planet Earth.  In theory, a “true” resistance measurement would have to be made at “infinite distance”; i.e., including the whole planet.  Of course, this cannot be done, and doesn’t have to be.  The area immediately surrounding the electrode provides 99.999...n% of the resistance, and the rest of the planet is only of theoretical interest. The “test item”, then, is the electrode and its immediate surrounding soil (Fig. 1).  This cannot be manipulated like a piece of apparatus.  Rather, the tester has to be accommodated to the possibilities.

Dedicated Ground Testers
Dedicated ground testers operate with an alternating square wave of a distinct frequency apart from what is likely to be produced by utility harmonics (Fig. 2).

 

Figure #2:Four-Terminal Ground Tester

They recognize their own signal and, unlike a multimeter, disregard “noise”.  If soil transients are extreme, to the extent that the tester’s filtering capabilities are overridden, warning indicators let the operator know that a problem exists so that faulty readings are not blindly recorded.  Furthermore, the AC signal facilitates the use of the virtually limitless lead lengths that are required when testing large grids in poor soil conditions.  Finally, a ground tester is not a two-terminal device but is designed according to the four-wire Kelvin bridge principle.  Having two separate current and two separate voltage terminals enables the operator to have complete control of the test setup.  The reliability of the test is not at the mercy of fixed-position reference grounds.  The operator drives probes exactly where desired, so that it is known precisely what is being measured.  Furthermore, the separate voltage probe enables surveying of the entire test site in order to recognize local anomalies, determine representative conditions for the area, and proof the readings, as will be described under the discussion of methods.  The Kelvin configuration further eliminates all extraneous resistances, as from leads and contacts, so that the tester provides a precise measurement, not an approximation.

Correct Procedure
The right instrumentation must be accompanied by the right procedure.  In no area of electrical testing is procedure more important than in ground testing.  It is not simply a matter of hooking up and pressing a button.  The test item is uncontrolled and uncontrollable...a substantial and unknown volume of earth surrounding the buried electrode in three dimensions.  Electrical circuits are typically made of relatively pure materials with narrow tolerances, but not in this case!

Because soil is almost infinitely variable, both in terms of composition and the temporal effects of weather, there is no way of knowing, prior to testing, what volume comprises the effective resistance at a particular site.  The literature is full of tables that provide guidelines, but these are only suggestions meant to give a fair chance of performing an acceptable test on the first trial.  To simply place probes and take a reading will provide an accurate measurement of soil resistance between the two points, the test electrode and the potential probe.  This may or may not be the effective resistance that a fault current will encounter.  To make that determination, the site must be rigorously proofed.  Operation of the tester alone does not provide this.  It must be augmented by a proper procedure.

Fall of Potential
The basis for all accepted methods is defined by IEEE (Institute of Electrical and Electronics Engineers) Standard 81, and is called “Fall of Potential”.

Figure #3:Fall of Potential Graph

Making use of the separate voltage probe, the procedure consists of plotting the resistance from the test electrode to a regular succession of points in the direction of the current probe.  This procedure develops a profile of the soil, indicates discontinuities and non-uniformity, and provides much more information than would a single measurement.  Ideally, a Fall of Potential test should produce a graph that looks like Fig. 3. 

This shows that if the measurement were taken infinitely close to the test ground, the resistance would be infinitely small, as would be expected.  This is evident from the simple fact that, at for instance one foot, there is very little soil to offer resistance.  Such a measurement would be of no practical value, however (except possibly to fool an unapprised client or inspector!).  As the probe is moved farther out and additional readings taken, the increased travel through soil adds resistance, just as a two-foot wire offers more resistance than a one-foot section of the same wire.  But a funny thing happens on the way to the current probe!  Readings level off and remain essentially flat, until the approach to the current probe constricts the path and superimposes additional resistance.  Hence, the graph rises toward the end.

The distinctive shape of the graph is generated by soil volume. Soil is a “good conductor” because of its enormity and ubiquity.  Fault current through a grounding electrode isn’t restricted to a straight path from point a to b, as in a designed circuit.  Rather, it radiates in all directions, 360° from the electrode.  The current path spreads out, rather than traveling in a straight line.  Soil in the relatively narrow confines around the electrode offers some resistance, but at greater distance, the area becomes so vast that there is no increase in resistance large enough to be measured.  Soil volume is the reason that the graph eventually reaches a stable plateau, and if that were not so, grounding itself would not be possible.

Figure #4:Non-Ideal Graph, Insufficient Probe Spacing

Constructing a Fall of Potential graph, then, shows the relationship between space and resistance.  The value where the readings stop increasing is the measure of the effective resistance of the test ground.  This could be at virtually any value up into hundreds of ohms.  But if it is above 25, it’s not meeting Code, and not functioning as an effective ground.  The distance at which this occurs marks the volume of soil that is the determining factor.  This could be only a few feet in prime soil, but could be hundreds of feet or more in areas of high resistivity.  Because this relationship...volume versus resistance...is so flexible, both the tester and the procedure must be adapted to meet the demands.

Performing a full Fall of Potential test is rigorous enough to stand up to any scrutiny.  If the test electrode has a large “footprint”, or electrical field in the soil...either from physical size or poor soil conductivity...the current probe may overlap and obscure the point of maximum resistance for the test ground.  In such a situation, as the potential probe is moved, it would run directly into the superimposed resistance associated with the current probe.  This would produce a graph that looks like that in Fig. 4.  One of the strengths of this method is that it affords a built-in proof.  If a graph like that in Fig. 4 is produced, the current probe is moved farther out and the procedure repeated.  No such proof is available with any other instrument than a dedicated ground tester.

Test Methods
A graph as ideal as that in Fig. 3 is not likely to be produced by a real test.  Field experience becomes a valuable ally.  Buried objects can cause dips and bumps.  Soil variations, especially at graded construction sites, can create a wavy plateau.  But an unreadable graph is a clear indication of an unacceptable test.  The operator has to repeat, perhaps in another direction, but will not be led astray by a “bad” reading, unaware.  The limitations of this method are that it is a lot of time and work, and also may require more lead distance than is available, especially at an urban site.  Accordingly, many variations and additional methods have been devised, some for general application and some for specific situations.  Additional methods are frequently based on simplifications of the Fall of Potential concept, and sometimes on other mathematical abstractions.  Test methods serve two purposes: to provide a proof that the reading actually represents the effective resistance and is not some random measurement, and to permit some simplification either in terms of speed or the means of dealing with some specific challenge.

Those aimed at shortening test time are the Simplified Fall of Potential, 62% Rule, “Dead Earth” Method, and one that for want of any real name might be called the “eyeball” method.  Those designed to meet challenges, specifically of limited space as opposed to limited time, are the Slope Method, Star-Delta Method, and “Intersecting Curves”.  Finally, for measuring the electrical conductivity of soil itself, there is the Wenner Method.

Simplified Fall of Potential
Simplified Fall of Potential applies a mathematical proof to the fundamental concept already described.  Rather than taking the time and work of plotting at regular intervals across the full span from test electrode to current probe, the initial measurement is made at the midway point, then two more at 10% of the distance in either direction.  This is a lot less time-consuming, and its success is determined by whether or not the three readings were taken on the plateau of the underlying Fall of Potential graph, or on the rising curve.  To make this determination, a simple mathematical procedure is applied, derived from calculus and based on rate of change of slope.  The calculation yields a percent accuracy of the average of the three readings.  If the readings are not sufficiently grouped within a narrow tolerance, the average falls outside an acceptable accuracy, and they were likely made on the rising curve.  The procedure must be repeated at greater distance.

The worldly-wise will ask, “Why don’t we just look at a few successive readings and see if they agree?”  This represents a non-descriptive procedure that is widely performed, and might be called the “eyeball” method.  If the technician is sufficiently experienced, it may be reasonably reliable.  However, it lacks objectivity, and won’t likely stand up to third-party scrutiny.  The readings may fall on the rising curve or may reflect only random variations, and the operator’s judgment may be tainted by optimism.  Invoking the mathematical proof of the Simplified Fall removes this potential source of error.

 “62% Rule”
Simpler still is the widely known “62% Rule”.  This is a single measurement, taken at 62% (actually 61.8) of the distance to the current probe.  Remember that on a classic Fall of Potential graph, the current probe superimposes its own sphere of resistance.  Therefore, at some point, the graph must coincide with the value of the true theoretical resistance, if it were possible to measure the entire planet from the point of the test ground.  That coincidence has been mathematically determined to occur at the 61.8% distance.  Why isn’t all ground testing done to that point, then?  The reason is because the determination is based on an ideal model.  In actual practice, the current probe may not be far enough away, there may be a pipe or power cable directly under the potential probe, or the spot may have been filled with some non-representative material.  For these reasons, the 62% Rule should not be relied upon at unknown sites, but is a good backup test at areas that have already been rigorously proofed.

 “Dead Earth”
The final simplified method is again one of limited reliability, and should not be employed generally; that is the “Dead Earth” Method.  This technique is quite popular because of its simplicity.  Only two leads are used, one hooked to the test ground and one to a reference ground (Fig. 5).  This is essentially the same as using a multimeter, and was described earlier in this article.  The use of a ground tester can eliminate the interference problem for reasons already presented, but the unreliability of the reference ground and the problem of insufficient separation still exist.  Accordingly, IEEE 81 advises, “...This method is subject to large errors for low-valued driven grounds but is very useful and adequate where a ‘go, no-go’ type of test is all that is required.”

 

Figure #5:“Dead Earth” Test “Clamp-on” Method

Ground testers of the clamp-on style have only been available for about ten years.  These units combine form and function so that the tester and method are one and the same.  “Clamp-on” jaws contain both voltage and current windings.  They can be clamped over a ground rod or grounding conductor and will inject a test current into the system.  The current travels through the soil, seeks its own return through a parallel ground, and loops back through the system neutral.  The voltage transformer senses the voltage drop around the loop and the tester calculates resistance.  It’s disarmingly simple to use, and thereby lies both its strength and danger.

The astute reader will already have noticed the parallel with the multimeter technique.  By injecting a current signal, the clamp-on takes advantage of the electrical system to provide a return, thereby removing dependency on an arbitrary “dead earth” reference.  Some peril still remains in the return path and, lacking control of test probes, the operator is at its mercy.  An oscillator produces a distinctive frequency and filters help suppress interference.  The resultant measurement is a series resistance, which should be comprised almost entirely of the contribution from the soil.  The method also simultaneously checks the bonding of grounding conductors, so that an open or corroded bond would show up in a high measurement.

Early models had shortcomings regarding accuracy, especially at the lower readings associated with commercial grounds, and were even banned by some authorities.  However, technological improvements have corrected these failings to the extent that modern units are highly accurate and reliable.  Some caveats remain, however.  A traditional model can be utilized anywhere, but not a clamp-on.  They obviously cannot be used for commissioning of new grounds that have not been connected.  The operator must have a thorough understanding of the electrical plan, so that there are no feedback loops that short-circuit the soil altogether.  Multiply connected grounds can circulate test current within their own structure, providing no more than a continuity test.  Yet operators may be all too willing to accept these as “proof” of a good ground.  Finally, the method does not afford the protection of a reference to an independent standards organization.

Slope Method
Other methods have been devised to cope with the other major problem of ground testing limited space.  To develop a full Fall of Potential graph, leads may have to be run out hundreds of feet, and tests have even been performed for miles!  Such practice may be unacceptable or impossible for a variety of reasons: property lines, obstructions like highways, waterways, and railroads, congested urban sites, or large ground grids, especially in areas of poor soil.  The most prevalent method of dealing with this problem is the Slope Method.  Again, a mathematical simplification derived from calculus is employed, this time to find the point on a graph such as shown in Fig. 4 where the resistance of the test ground stops and that of the current probe takes over.  This obviously cannot be read from the graph, so a mathematical exercise is employed to find it.  Readings are taken at 20, 40 and 60% of the distance, and a calculation made to determine a slope coefficient.  Tables found commonly in the literature are referenced for a corresponding ratio of potential probe distance to current probe distance (dPT/dC), and when multiplied by the known distance to the current probe gives the position at which the potential probe should be placed to derive the correct reading.  If the current probe is so close that it is within the field of the test ground, the mathematics will prove unintelligible and indicate to the operator that a better test position must be found.

Star-Delta
If this latter condition prevails, and room is so limited that an acceptable spacing cannot be derived even with the Slope Method, it may be necessary to resort to Star-Delta.  Named for the configuration of the test probes and lines of measurement (a graphic of it resembles the familiar symbols for “delta” and “star” windings), this method saves space by employing a tight configuration of three probes around the test ground (Fig. 6).  Separation of potential and current circuits is abandoned, and a series of two-point measurements made between all pairs of probes and probes to test ground.  This results in six measurements that are then put through a mathematical “crunch” of four series equations to calculate the resistance of the test ground.  As with all mathematical methods, a faulty or unreliable setup produces unintelligible calculations (e.g., “negative” resistances), and so the operator knows that the procedure needs to be spruced up.

Figure #6: Star-Delta Method

“Intersecting Curves”
A very difficult and tedious method, but ideal for those who welcome a challenge, is that of “Intersecting Curves”.  Very large grids are not only likely to have commensurately large electrical fields that require impractically long leads in order to test, but also compound the problem by having indeterminate electrical centers (which do not necessarily coincide with the geometric center).  Obviously, for measurement purposes, a single rod can be treated as a point, and a small grid or array doesn’t offer enough of an error to be significant.  But with large grids, the uncertainty as to the actual distance of separation from the current probe can become a complication.  In the Intersecting Curves method, a number of resistance-versus-distance graphs are first constructed at different lead lengths.  The unknown distance between the convenient point of attachment and the electrical center is assigned a number of arbitrary values, and using the 62% rule, the corresponding positions are calculated for the potential probe.  The corresponding resistances for each of these points can then be read from one of the graphs.  These resistances are then plotted against their arbitrary unknown distances on another graph.  This process is repeated for each of the measurement graphs, and additional lines constructed on the second graph accordingly.  If the test setup were ideal, these derivative graphs would all intersect at a single point.  In the real world, they are more likely to form a tight triangle, the center of which corresponds to the actual resistance of the test ground (Fig. 7).  The correct unknown distance can also be read, and if this is plugged back into the equation for distance to the potential probe, a measurement can be taken at that actual point and should agree with the one read from the graph.

Figure #7:Intersecting Curves Graph

Soil Resistivity
If measured resistances are unacceptable, the test ground must be improved.  This is accomplished by driving deeper rods, adding more rods, or applying various chemical treatments or backfills.  This can be done by trial and error, but a more efficient method is to first take electrical measurements of the properties of the soil itself, and then apply this data to the construction, location, or improvement of grounding electrodes.  A number of standard procedures exist, but by far the method of choice is the “Wenner” Method.  Four probes are arranged at equal distances, and driven 1/20th of their horizontal separation (Fig. 8).  Such an arrangement will measure average soil resistivity to a depth equivalent to the spacing.  The tester is then energized and a reading taken.  This is plugged into the “Wenner Formula”:

r = 2paR

where:

r = average soil resistivity (typically in units of ohm-cm)

a = probe spacing (typically in cm)

R = resistance reading

The data is called the resistivity of the soil, as opposed to the resistance of a buried electrode.  It indicates how the soil will respond to the flow of electric current, and is invaluable to any effort to establish maximum grounding protection.

Figure#8:Wenner Method

Conclusion
While the NEC is indispensable to electrical safety, its recommendations are not all inclusive.  For commercial applications, it is wise to establish ground protection to its maximum level, not simply conform to a minimum.  This cannot be accomplished without testing, and it cannot be accomplished properly without proper instrumentation and procedures.

Jeff Jowett is a Senior Application Engineer with Megger.  He has been with Megger for over 30 years specializing in electrical testing applications in the areas of insulation resistance testing, ground resistance testing and low resistance ohmmeter testing.  He is a regular speaker at industry events and dozens of articles authored by him have been published by various magazines worldwide.