• Author / Uploaded
• shawnr7376

Citation preview

Tower Grounding and Soil Ionization Report

Technical Report

Tower Grounding and Soil Ionization Report 1001908

Final Report, February 2002

EPRI Project Manager A. Phillips

EPRI • 3412 Hillview Avenue, Palo Alto, California 94304 • PO Box 10412, Palo Alto, California 94303 • USA 800.313.3774 • 650.855.2121 • [email protected] • www.epri.com

DISCLAIMER OF WARRANTIES AND LIMITATION OF LIABILITIES THIS DOCUMENT WAS PREPARED BY THE ORGANIZATION(S) NAMED BELOW AS AN ACCOUNT OF WORK SPONSORED OR COSPONSORED BY THE ELECTRIC POWER RESEARCH INSTITUTE, INC. (EPRI). NEITHER EPRI, ANY MEMBER OF EPRI, ANY COSPONSOR, THE ORGANIZATION(S) BELOW, NOR ANY PERSON ACTING ON BEHALF OF ANY OF THEM: (A) MAKES ANY WARRANTY OR REPRESENTATION WHATSOEVER, EXPRESS OR IMPLIED, (I) WITH RESPECT TO THE USE OF ANY INFORMATION, APPARATUS, METHOD, PROCESS, OR SIMILAR ITEM DISCLOSED IN THIS DOCUMENT, INCLUDING MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE, OR (II) THAT SUCH USE DOES NOT INFRINGE ON OR INTERFERE WITH PRIVATELY OWNED RIGHTS, INCLUDING ANY PARTY'S INTELLECTUAL PROPERTY, OR (III) THAT THIS DOCUMENT IS SUITABLE TO ANY PARTICULAR USER'S CIRCUMSTANCE; OR (B) ASSUMES RESPONSIBILITY FOR ANY DAMAGES OR OTHER LIABILITY WHATSOEVER (INCLUDING ANY CONSEQUENTIAL DAMAGES, EVEN IF EPRI OR ANY EPRI REPRESENTATIVE HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES) RESULTING FROM YOUR SELECTION OR USE OF THIS DOCUMENT OR ANY INFORMATION, APPARATUS, METHOD, PROCESS, OR SIMILAR ITEM DISCLOSED IN THIS DOCUMENT. ORGANIZATION(S) THAT PREPARED THIS DOCUMENT EPRIsolutions

ORDERING INFORMATION Requests for copies of this report should be directed to EPRI Orders and Conferences, 1355 Willow Way, Suite 278, Concord, CA 94520, (800) 313-3774, press 2 or internally, x5379 (925) 609-9169 (925) 609-1310 (fax). Electric Power Research Institute and EPRI are registered service marks of the Electric Power Research Institute, Inc. EPRI. ELECTRIFY THE WORLD is a service mark of the Electric Power Research Institute, Inc. Copyright © 2002 Electric Power Research Institute, Inc. All rights reserved.

CITATIONS This report was prepared by EPRIsolutions 115 East New Lenox Rd. Lenox, MA 01240 Principal Investigators P. White J. Anderson K. King This report describes research sponsored by EPRI. The report is a corporate document that should be cited in the literature in the following manner: Tower Grounding and Soil Ionization Report, EPRI, Palo Alto, CA: 2002. 1001908.

iii

REPORT SUMMARY

Deregulation of the powder industry has increased the need for greater reliability of the transmission system. Unplanned outages can have significant financial implications, and lightning activity is often cited as one of the main reasons. To address this issue, EPRI is conducting research to increase understanding of the lightning performance of transmission lines. This report details the results of one such study. Background The historical challenge of providing reliable electrical service is becoming more important. With electronic equipment in almost all facets of life, even momentary outages and power quality problems can adversely affect customers at home and work. Lightning causes many such momentary customer outages, and EPRI’s TFlash program was developed to help utility engineers evaluate the lightning performance of power systems. Objectives To provide more accurate grounding algorithms for the TFlash program. Approach Published algorithms, and those generally used by the industry for computing surge current dynamic resistance of ground rods and concrete foundations, provide wide divergences in predicted values. Also, in spite of almost a century of experience, the actual dielectric properties of soils are still a matter of debate. These dielectric properties have a profound effect on the lightning performance of transmission lines. Therefore, the grounding research documented in this report encompasses two fundamental activities: determining the dielectric properties of some typical soils and selecting dynamic resistance models of ground rods and concrete foundations as a function of current in these soils. The project team concentrated on ground rods and concrete foundations because these grounding geometries are so prevalent on transmission and distribution lines. Results Study results, together with previous EPRI research, are being used to develop more accurate models to predict transmission line performance. These models, in turn, will be included in EPRI’s state-of-the-art Transmission Line Lightning Performance Prediction Software, T-Flash. The results also will be used to develop guidelines for utilities on how to effectively design, construct, and maintain transmission line grounding systems. EPRI Perspective TFlash is a state-of-the-art design tool that allows engineers to analyze the effect of a specified lightning challenge on a given transmission line as well as specified mitigation techniques such v

as shielding, improved grounding, line arresters, and upgraded insulation. With this software, utility engineers can analyze the degree of protection an existing line has, define changes to the line to improve protection, or design a new line with economical lightning protection. As a result, TFlash has potential to help utilities achieve cost-effective improvements in lightning protection and customer reliability. Keywords Lightning Grounding Soil ionization TFlash LPDW (Lightning Protection Design Workstation)

vi

CONTENTS

1 OVERVIEW ......................................................................................................................... 1-1 2 THEORY OF GROUND ELECTRODE RESPONSE ............................................................ 2-1 Introduction ........................................................................................................................ 2-1 Ground Rod Resistance ..................................................................................................... 2-2 Concrete Foundations ........................................................................................................ 2-5 CIGRÉ Dynamic Resistance Models .................................................................................. 2-6 Liew-Darveniza Dynamic Resistance Model....................................................................... 2-9 The Chisholm-Janischewskyj Model..................................................................................2-11 The Robbins/TFlash Model................................................................................................2-14 Other Models.....................................................................................................................2-14 3 SMALL-SCALE PRELIMINARY EXPERIMENTS ON ELECTRICAL PROPERTIES OF SOIL ................................................................................................................................. 3-1 Introduction ........................................................................................................................ 3-1 Uniform Field Dielectric Strength of Soils............................................................................ 3-1 Soil Resistivity .................................................................................................................... 3-3 Non-Uniform Field Dielectric Strengths of Soils .................................................................. 3-5 4 LARGE-SCALE EXPERIMENTS ON DYNAMIC RESISTANCES OF GROUND RODS ..... 4-1 Introduction ........................................................................................................................ 4-1 Test Configurations and Soils............................................................................................. 4-1 Digital Filtering.................................................................................................................... 4-3 Voltage and Current Waveshapes ...................................................................................... 4-4 Initial Dynamic Resistances................................................................................................ 4-5 Dynamic Ohms vs. Applied Current.................................................................................... 4-8 Effect of Rod Length and Separation.................................................................................. 4-8 Comparison of Tank Test Values with Dynamic Algorithms................................................ 4-9 Liew-Darveniza Algorithm..................................................................................................4-10

vii

CIGRÉ/Weck Algorithm .....................................................................................................4-11 Chisholm - Janischewskyj Algorithm..................................................................................4-12 TFlash Algorithm ...............................................................................................................4-13 Effect of Rod Shape and Artificial Streamers.....................................................................4-13 Response of Concrete Ground Electrodes ........................................................................4-17 Chemical Enhancement ....................................................................................................4-18 Butt-Wrapped Poles ..........................................................................................................4-18 Computer Modeling of Dynamic Resistance ......................................................................4-20 5 SUMMARY .......................................................................................................................... 5-1 6 GOOD GROUNDING PRACTICES ..................................................................................... 6-1 7 RECOMMENDATIONS FOR FUTURE WORK.................................................................... 7-1 8 REFERENCES .................................................................................................................... 8-1 A THE CIGRÉ CRITICAL CURRENT EQUATIONS ............................................................... A-1 B THE LIEW-DARVENIZA ALGORITHM ............................................................................... B-1 C THE GND_ROD1 ALGORITHM .......................................................................................... C-1 D THE CHISHOLM-JANISCHEWSKYJ (C-J) MODEL........................................................... D-1 E THE ROBBINS/TFLASH DYNAMIC RESISTANCE MODEL .............................................. E-1 F DRIVEN ROD RESISTANCES IN TWO-LAYER EARTHS.................................................. F-1 G ADJUSTMENT OF ROD-TANK RESISTANCES TO ROD RESISTANCES IN AN INFINITE PLANE....................................................................................................................G-1

viii

LIST OF FIGURES Figure 2-1 Dimensions for a single vertical ground rod............................................................ 2-2 Figure 2-2 The Wenner four electrode method of measuring earth resistivity.......................... 2-3 Figure 2-3 Dielectric breakdown of soil around rod electrodes. ............................................... 2-4 Figure 2-4 Reduction of rod resistance with time..................................................................... 2-5 Figure 2-5 Concrete foundation in moist soil. .......................................................................... 2-5 Figure 2-6 Replacing a ground rod with a conducting hemisphere. ......................................... 2-7 Figure 2-7 Dynamic resistance of a concrete foundation vs. time............................................ 2-8 Figure 2-8 Liew-Darveniza ground rod surrounded by concentric cylindrical shells of earth...............................................................................................................................2-10 Figure 2-9 Modernized version of the Korsuncev Curve (from Oettle). ...................................2-12 Figure 2-10 Relationship of Parameter S to Ionization Zones.................................................2-13 Figure 3-1 Rogowski Electrode Test Cell. ............................................................................... 3-2 Figure 3-2 Test Circuit for Soil Uniform Field Dielectric Tests.................................................. 3-2 Figure 3-3 Resistivity Test Cell................................................................................................ 3-3 Figure 3-4 Photo of the Concentric Cylindrical Test Cell. ........................................................ 3-5 Figure 3-5 Electrical Diagram of the Concentric Cylindrical Test Cell. ..................................... 3-6 Figure 3-6 Concentric Test Cell Voltage Waves Before and During Breakdown...................... 3-8 Figure 4-1 Large Test Tank Dimensions and Circuitry............................................................. 4-2 Figure 4-2 Photo of Large Test Tank Installation..................................................................... 4-2 Figure 4-3 Unfiltered Data. ...................................................................................................... 4-4 Figure 4-4 Chebyshev Filtered Data........................................................................................ 4-4 Figure 4-5 Typical Voltage-Current Waves During Rod Tests. ................................................ 4-5 Figure 4-6 Initial Dynamic Resistance Computation. ............................................................... 4-6 Figure 4-7 Dynamic Ohms vs. Applied Current. ...................................................................... 4-8 Figure 4-8 Dynamic Resistance at Two Microseconds As a Function of Current Into One 30-Foot Rod or Two 15-Foot Parallel Rods Spaced 10 Feet Apart. ................................. 4-9 Figure 4-9 Liew Darveniza Algorithm Comparison With Dynamic Resistance Test Values. ...........................................................................................................................4-10 Figure 4-10 CIGRÉ/Weck Algorithm Comparison With Dynamic Resistance Test Values. .....4-11 Figure 4-11 C_J Algorithm Comparison With Dynamic Resistance Test Values.....................4-12 Figure 4-12 TFlash Algorithm Comparison With Dynamic Resistance Values. .......................4-13 Figure 4-13 Artificial Streamers Attached to a Ground Rod. ...................................................4-14

ix

Figure 4-14 Dynamic Resistance With Artificial Streamers in Place. ......................................4-14 Figure 4-15 Dynamic Resistance With Artificial Streamers Removed.....................................4-15 Figure 4-16 Tested Bar Cross-Section. ..................................................................................4-15 Figure 4-17 Rectangular Bar Ground Rod Dynamic Resistance in sand. ...............................4-16 Figure 4-18 One Inch Diameter Round Rod in Sand. .............................................................4-16 Figure 4-19 Voltage-Current Waves on a Concrete-Encased Rebar. .....................................4-17 Figure 4-20 4.5-Foot Ground Rod Surrounded With Eight Inch Diameter GEM Material. .......4-18 Figure 4-21 Photo of Bottom End of Butt-Wrapped Pole. .......................................................4-19 Figure 4-22 Dynamic Resistance of a Butt-Wrapped Pole in Loam. .......................................4-19 Figure 4-23 Comparison of Lowest Dynamic Ohms Calculated by Korsuncev and LiewDarveniza Algorithms. ....................................................................................................4-20 Figure 4-24 Single Rod Korsuncev Dynamic Resistance vs Rod Length and Soil Resistivity. ......................................................................................................................4-21 Figure 4-25 Dynamic Resistance of Two ground Rods in Parallel vs. Rod Length and Separation Distance (Korsuncev). ..................................................................................4-22 Figure 4-26 Variation of Double Rod Low Frequency Resistance With Spacing. (1-in.Diam. Rod, 3 Feet Long, ρ=100 ohm-meters).................................................................4-23 Figure A-1 A thin soil resistance shell surrounding an embedded conducting hemisphere. .................................................................................................................... A-1 Figure A-2 Gradient at the surface of a conducting hemisphere............................................. A-2 Figure B-1 Liew-Darveniza ground rod model surrounded by concentric shells of earth......... B-1 Figure C-1 Collision of Soil Resistance Shells for Two Rod Case .......................................... C-1 Figure D-1 The Korsuncev Curve for Ground Electrodes ....................................................... D-2 Figure D-2 Application of the Characteristic Dimension S. ..................................................... D-2 Figure E-1 Expanding ionization zone around a rod ............................................................... E-1

x

LIST OF TABLES Table 3-1 Uniform Field Dielectric Tests.................................................................................. 3-4 Table 3-2 Breakdown Vs Waveshape: Kiln-Dried Sand........................................................... 3-7 Table 3-3 Effect of Moisture and Tail Time.............................................................................. 3-8 Table 4-1 Large Scale Dynamic Resistance Tests. ................................................................. 4-3

xi

1 OVERVIEW

This report discusses work performed in 2000 and 2001 on lightning grounding research. The work was initiated by the clear necessity of providing better grounding algorithms for the EPRI TFlash transmission line lightning performance program. In 1982 EPRI published a first comprehensive report (Reference 7) on transmission line grounding, but this report was directed primarily to ac grounding and said little about ground electrode dynamic response under lightning surge currents. Published algorithms and those generally used by the industry for computation of surge current dynamic resistance of ground rods and concrete foundations provide wide divergences in predicted values. Also, in spite of almost a century of experience, the actual dielectric properties of soils are still a matter of debate, and these dielectric properties have a profound effect on the lightning performance of transmission lines. Therefore, grounding research in 2001 has encompassed two fundamental activities: determination of the dielectric properties of some typical soils, and the selection of dynamic resistance models of ground rods and concrete foundations as a function of current in these soils. In 2001, attention was concentrated on ground rods and concrete foundations because these grounding geometries are so prevalent on transmission and distribution lines. This report is organized in the following sections: Section 2 of this report reviews present theories and limitations of ground electrode dynamic resistances under high impulse currents. This review is to acquaint the reader with the dynamics of the electronic processes involved. Section 3 reports results of small-scale experiments to clarify the dielectric properties of various soils, particularly sand, clay and loam, and how these results can apply to large scale electrode response. Section 4 shows some results of large-scale high current experiments on rods and concrete electrodes (including conductivity enhancement chemicals) in a special soil-containment tank, and how these results fit various theoretical algorithms. Section 5 then sums the results and makes recommendations for algorithms to apply in the TFlash program. Section 6 makes recommendations - based on the results of the present investigation - for better tower grounding strategies and grounding enhancement procedures. Section 7 outlines needed future work, including extension of the present work to concrete foundations and substations, rods in two-layer soils and development of better data acquisition methods to determine soil constants on rights-of-way that control the dynamic response of tower grounds. 1-1

2 THEORY OF GROUND ELECTRODE RESPONSE

Introduction The simplest practical ground electrode used on transmission and distribution lines is a vertical ground rod. Often this rod penetrates into two or more layers of earth with different resistivities, such as a layer of loam above a layer of clay or glacial till. The total rod resistance to earth is the parallel combination of the rod resistances in each layer. Earth is a poor dielectric, and at high lightning surge currents electrical streamers and glow discharges develop in the soil around a rod as the soil breaks down, effectively enlarging the rod electrical radius. Electrical streamers also penetrate down into the earth off the bottom tip of the rod, thereby effectively increasing the rod length. In addition, as current is injected into a ground rod, magnetic energy is stored in the earth and in the air just above the rod, and at lightning frequencies this energy is evidenced as an inductance in series with the rod. Finally, the relative dielectric constant of earth can be 10 or more, so a rod has a capacitance to earth in parallel with its resistance that is sometimes taken into account at lightning frequencies. The inductance and capacitance are usually assumed constant for a given rod and independent of earth resistivity. However, clearly the rod capacitance increases as the streamers increase the rod diameter. As for inductance, the conductivity of the metal rod is so much higher than the conductivity of the streamers in an axial direction parallel to the rod that inductance is assumed independent of streamer development. This ignores the fact that each streamer is carrying current and also sending magnetic energy into the soil and that the current in the rod is greater at the earth's surface than down at the rod tip. Even for the simplest case of a single vertical rod, the entire electrical event when a stroke current surge enters the rod is very nonlinear. Calculation of rod dynamic resistance vs. time is further complicated by a lack of information on earth resistivities around the rod and uncertainties as to the earth dielectric strength and its ionization time constants. Concrete foundations are in some ways, simpler because of the large surface area of the concrete and its slow variation of resistance with soil moisture. Theories of ground electrode dynamic resistance abound in the literature, and some of the many references are listed at the end of this report. Reference 2 by Mousa provides a particularly comprehensive list of grounding references relevant to this technology. Because of the complexity of the process and the many uncertainties about the soil electrical environment around any transmission tower electrode, this report has, insofar as possible, tried to simplify the dynamics, and to not describe with great mathematical precision that which is in fact known only very approximately, if at all. The simplified models thus developed should be sufficient until more precise information of soil electrical properties are available. 2-1

Theory of Ground Electrode Response

Ground Rod Resistance When a single vertical rod is driven into the earth (Figure 2-1), its resistance, Ro to the flow of low frequency low amplitude current is given by Eq. 2-1 by Dwight:

L d

Figure 2-1 Dimensions for a single vertical ground rod.

Ro =

ρ   8L   log  − 1 2πL   d  

Equation 2-1

where: Ro = low frequency, low current resistance, ohms ρ = earth resistivity, ohm-meters L = rod length, meters d = rod diameter, meters This resistance Ro is to a uniform infinite earth. It assumes that no skin-effect exists near the surface of the earth in spite of the high frequencies contained in any lightning surge current flowing into the rod, no dielectric breakdowns of soil around the rod caused by high voltages on the rod, no effect of retardation time (the current has spread to infinity instantly - an obvious impossibility), no variations in earth resistivity with frequency or with depth, and no effect of the dielectric constant of the earth around the rod. However, in spite of its shortcomings for fast high transient currents, Eq. 2-1 is the standard Dwight equation frequently used in transmission line lightning simulation programs for rod resistance before any soil ionization processes begin, and is used as a starting point for most dynamic resistance calculations. Eq. 2-1 can also be used to determine earth resistivity in some situations. This is done by driving a rod into the earth, measuring its Ro with a ground resistance megger and then rearranging Eq. 2-1 to calculate the equivalent earth resistivity that results in the observed resistance Ro. However, as will be noted in more detail later, driving a ground rod will - in some soils - vibrate the rod sufficiently so that the hole containing the rod is slightly enlarged, making only partial 2-2

Theory of Ground Electrode Response

electrical contact on some of the rod surface areas, and thereby producing a significant resistance error. This effect has been encountered in EPRI tests of ground rods in sand. Also, if ρ varies with depth, the calculated resistivity will not indicate the resistivity at any particular depth. I v

x

Figure 2-2 The Wenner four electrode method of measuring earth resistivity.

A more accepted on-site method of measuring earth resistivity (valid for a uniform earth) is shown in Figure 2-2. An alternating current I is fed between the two outer rods, creating a voltage drop along the surface of the earth, resulting in a voltage V between the two inner rods. The indicated earth resistivity is then:

ρ = 2πx

V I

Equation 2-2

where: x = distance between the two inner rods, meters. ρ = earth resistivity, ohm-meters Wenner resistivity measurements (Eq. 2-2) are often complicated by stray earth currents and by the fact that both V and I are often small in field measurements. The problem with many on-site measurements of earth resistivity is that the earth often consists of an upper layer of sand, loam or clay only a few meters thick on top and another layer of rock, gravel or earth underneath with a much different resistivity. Aerial surveys at frequencies from 10 to 100 kHz can provide useful resistivity data in multiple layer soils. For a two-layer case, Eq. 2-2 will indicate different resistivities as the spacing x is changed, and this variation has been used (Reference 7) to infer the nature of the resistivity layers underneath A special EPRI Fortran program RESIST was developed for this purpose. However, it still required substantial interpretive skills to infer the nature of the subsurface strata. The dynamic resistance of a ground rod will usually be much less than its low-frequency lowcurrent resistance Ro, and this dynamic resistance is used by TFlash to evaluate transmission line lightning performance. If a surge current of 50 kiloamperes flows into a single ground rod 2-3

Theory of Ground Electrode Response

having a resistance Ro of 20 ohms (a condition easily occurring in service) a voltage of 1000 kV will appear on the rod if Ro remains constant. On the rod - particularly at its deep end and along its wall surfaces- the electric gradients will reach magnitudes far greater that the dielectric strength of any soil, and the soil will start ionizing and failing electrically. This ionization consists of a set of radial electrical streamers and glow discharges accompanied by strong void ionizations between soil particles extending out beyond the electrode (Figure 2-3).

I

I

Figure 2-3 Dielectric breakdown of soil around rod electrodes.

This ionization has the effect of enlarging the electrical diameter of the rod, and by Eq. 2-1 is equivalent to reducing the rod resistance. Note in Figure 2-3 that for two rods, the streamers and/or ionization between the two rods will be shorter that the outer streamers because both rods are at the same potential. The reduction in rod resistance can be dramatic, and reduces the likelihood of insulator flashovers on any tower or pole connected to the rod. Ionization of the soil can continue after voltage crest has been reached, so the rod resistance can continue to drop out on the tail of the surge current wave (Figure 2-4) and then gradually return to a value somewhere near its predischarge value as the soil deionizes. Experiments have shown (Reference 5) that the streamer velocities are quite slow, but ionization in the soil voids can proceed rapidly, making the latter play a dominant role in the resistance reduction. A fundamental objective of this EPRI research is to select the best mathematical model that describes how this resistance changes with current and time.

2-4

Theory of Ground Electrode Response

kA kA

Ohms Ohms

Time Miliseconds

Figure 2-4 Reduction of rod resistance with time.

Soil moisture plays an important role in values of electrode resistivity with time. Moisture content near the surface can vary very significantly with weather conditions, while moisture content at depths of several meters will change only slowly with time since it tends to reach an equilibrium condition with long-term environmental conditions. Although two-layer or multilayer resistance calculations in various grounding strategies can improve mathematical precision in line flashover estimates, in reality for ground rods over 10 feet in length, the uncertainties in general knowledge of soil moisture with time can make the additional precision illusionary. Also, there is an implicit assumption in practically all lightning calculations that soil resistivity caused by moisture content is constant for all frequencies, whether 60 Hz or lightning transient current frequencies. One of the concerns in this investigation has been to determine the best adjustment to make in TFlash for soil dielectric response to transients of different waveshapes and moisture content.

Concrete Foundations Soil moisture also has an interaction with the resistance of concrete foundations. In Figure 2-5, concrete surrounds the rebars and steelwork electrically connected to a tower leg.

H2 O

H2O

Figure 2-5 Concrete foundation in moist soil.

2-5

Theory of Ground Electrode Response

When the concrete is poured, its water content will initially be much higher than the surrounding soil. However, present theory (supported by a substantial amount of research ( Refs. J-M) is that after an initial resistance increase, the moisture in the concrete gradually reaches a rough equilibrium with moisture in the soil, so that the reinforcing steel in the concrete eventually acts as if it is more or less embedded electrically in the soil without the concrete present, albeit a "soil" with a different resistivity. To the extent that this is true it still ignores the fact that the reinforcing steel in soil alone would emit electrical streamers, whereas any streamers in the concrete tend to be inhibited, and - if they occur - can conceivably cause damage by creating puncture paths. The National Electrical Code adopted a recommendation that a copper electrode not less than #4 in diameter and not less that 20 feet in length be embedded in a concrete foundation along one or two of the external walls. This would reduce electrical stresses inside the concrete. It is important to reexamine how concrete foundations compare with driven rods as grounding electrodes, and more research work is indicated. Experience in the former Soviet Union (Reference 13) during nine years of impulse and 50 Hz testing of concrete footings found the following: •

A ratio of maximum to minimum annual range of 50 Hz resistance = 1.4 caused by variation of soil moisture. The maximum resistance occurs in the winter due to freezing.

•

After stabilization of the concrete over several months, the ratio of impulse impedance to ac impedance ranged from 0.92 to 1.0 up to 1.8 kA impulse current and this ratio drops to about 0.7 at 10 kA. However, for multiple foundations - because of inductance effects - the ratio can reach unity or even higher.

•

Even in very dry surface conditions, the concrete foundations retained most of their moisture through capillary attraction from sub-surface soil.

•

As an example of foundation resistance magnitudes, in a soil of approximately 1000 ohmmeters, four concrete foundations spaced 7.5 meters apart and having a length of 2.8 meters and resting on a buried concrete plate had a combined parallel ac resistance of 11.5 to 14.5 ohms.

As far as the authors are aware, the only EPRI design curves for concrete-encased electrode resistances is that reported in Reference 7, and these are inadequate for foundations with plates or pyramidal foundations and take no account of the dynamics of the impulse resistance. A set of design curves in Reference 9 can be used for ac resistance only. Development work is indicated to incorporate concrete foundation dynamics in TFlash.

CIGRÉ Dynamic Resistance Models Electrical experiments of ground electrode performance have always been limited by the ability of impulse generators to force high surge current magnitudes into the earth. Since the early researches of Bewley, Bellaschi and others, a substantial literature has evolved to attach mathematical models to the observed voltage-current relationships for different soils and electrode configurations. However, the process is electrically complex and simplifications can be perilous. An example is the present CIGRÉ mathematical model for a single rod (Reference 19). As Figure 2-6 shows, a ground rod is represented in ionized soil - not as a rod - but as a hemisphere embedded at the earth's surface. 2-6

Theory of Ground Electrode Response

I

Eo

Figure 2-6 Replacing a ground rod with a conducting hemisphere.

The justification for this concept lies in the argument that streamers from the rod project out into the soil and soil ionization occurs in a kind of radial cloud whose boundary can be assumed to approximate a hemispherical surface for high currents. Neglecting for the moment its validity, a hemispherical ionized cloud with negligible internal resistivity is easy to represent analytically and makes an algebraic representation possible (rather than a digital one). The problem then becomes one of finding suitable parameters for the various soil types and electrode geometries that make an acceptable fit of this model to experimental results. If one accepts a spherical representation as a rough working model, Appendix A shows that the low-frequency low-current resistance Ro of a hemispherical electrode (Figure 2-6) is: Ro =

ρ 2πr

Equation 2-3

where Ro = resistance, ohms ρ = soil resistivity, ohm-meters r = sphere radius, meters Note that this model assumes that the soil inside this hemisphere is completely ionized so that the soil resistivity in that region can be assumed zero. Making this assumption, Appendix A of this report then derives the CIGRÉ Weck equation (Eq. 2-4) for the critical current I0 required to create a soil critical ionization gradient Eo at the surface of the equivalent hemispherical electrode, this electrode having low frequency resistance Ro of the ground rod it replaces: I = o

ρEo 2πR 2 0

Equation 2-4

2-7

Theory of Ground Electrode Response

As derived in Appendix A, using Io the equation for dynamic resistance of a foundation that can be represented by a spherical electrode is: R=

Ro

Equation 2-5

I Io

where R = instantaneous dynamic resistance, ohms with the proviso that it can never be greater than Ro. Ro = low frequency low current resistance, ohms I = instantaneous foundation current, kA Io = a critical current when ionization starts, kA As an example, Figure 2-7 shows a conceptual foundation dynamic resistance vs. time. Having more bulk and a larger diameter that a rod electrode, its dynamic resistance reduction is less. It should be understood that the critical gradient Eo in Eq. 2-4 is not necessarily the actual breakdown strength of the soil. More realistically, it can be considered a soil critical ionization gradient at which the voids in high dielectric strength soils start to ionize, but it might also represent breakdown strength in poor dielectric soils. More generally, it should be considered a working constant that best fits any model that uses it . Ro = 20 Ohms ρ = 300 Ohm-meters

25

εo = 300 kV S = 1.5 meters

Dynamic Ohms *

* Dynamic Ohms Calculated Using the Korsuntcev Curve. 20 15 10

Critical Ionization Current = 22.5 kA

5 0

0

10

20

30

40

50

Instantaneous kA Figure 2-7 Dynamic resistance of a concrete foundation vs. time.

2-8

60

70

80

Theory of Ground Electrode Response

Mousa (Reference 2), in a quite extensive analysis of the electrical dynamics of soils suggests a conservative value of 300 kV/m for Eo. IEEE (Reference 17) has suggested a value of 400 kV/m, but Oettle in her researches in South Africa has proposed values as high as 1000 kV/m. Liew and Darveniza (Reference 1) found a value of Eo = 300 kV/m to fit several of their tests. If Eo is taken as the dielectric (puncture) strength of soil, Oettle (Reference 5) has shown that there is only a very weak correlation between the resistivity ρ and E0. Her proposed equation for this relationship is: E o = 241ρ 0.215

Equation 2-6

Also, the effect of water content in the soil can have an erratic effect on the resistivity, since the chemical composition of the soil (such as salt) will interact with the moisture present. In the EPRI model to be described later, Eo will be the constant that best fits experimental results, recognizing that it should not be higher than the puncture strength of the soil in a uniform or quasi-uniform field. For a rod, Weck (Reference 18) recognized that ionization starts off the tip of the rod at very low surge currents, so that nonlinear resistance is almost continuously present. For this rod case (Appendix A) he proposed a modification of Eq. 2-5 above to: R=

Ro I 1+ Io

Equation 2-7

where Io is a critical ionization current of Eq. 2-4. Note that neither of these CIGRÉ models have time as a dynamic. The greatest resistance reduction always occurs at the instant of maximum current, whereas many tests have shown that - at least for rods - the resistance can continue to fall for several microseconds after crest current, and these few microseconds can be important in establishing the flashover performance of transmission lines.

Liew-Darveniza Dynamic Resistance Model The CIGRÉ assumption that one can replace a 20-foot ground rod with an equivalent embedded hemisphere and have the same dynamic resistance performance is a long stretch of the imagination, and Liew and Darveniza (Reference 1) used concentric shells of earth around the rod as a better representation (Figure 2-8). Assuming relatively uniform flow of current out of the rod in Figure 2-8 (not very likely) the current density flowing through each shell can easily be calculated. If this current density is sufficient to create a critical ionizing gradient Eo across the wall of any shell, Liew and Darveniza assume that the shell wall starts ionizing and its resistivity starts dropping exponentially with time. Note that Eo is assumed everywhere constant. As the resistance of a shell near the rod starts dropping, that impresses more voltage across shells farther away and they start ionizing also. When the surge current decays sufficiently, the current density in some 2-9

Theory of Ground Electrode Response

shells falls below the critical ionizing value, and those shells start deionizing (again exponentially with time) and the resistance starts increasing towards its original low frequency value. Appendix B describes the Liew-Darveniza model in more detail. While a digital simulation of the process is relatively straightforward and can be incorporated in TFlash, it is difficult to represent analytically the reduction in resistance versus time and instantaneous current with a couple of equations as is possible for the CIGRÉ model. I

r

L

dr Eo

Figure 2-8 Liew-Darveniza ground rod surrounded by concentric cylindrical shells of earth

While in many ways the Liew-Darveniza algorithm seems more realistic than the CIGRÉ approach and at low currents provides Ro values that match theoretical formulas quite well, this model also has some dubious characteristics, among them: •

The assumption of uniform current density in each shell wall at high currents and ignoring individual streamer penetrations of any shell walls.

•

The assumption of uniform exponential ionization and deionization.

•

Uncertainty of values of the ionization and deionization constants and Eo to use for each electrode and soil type.

•

The assumption of an idealized hemispherical bottom end for each shell.

•

For any ground electrodes other than rods, describing the shell geometries becomes difficult unless they are assumed to be hemispheres.

However, dynamic resistance in the Liew-Darveniza model usually continues to drop after the surge current has passed crest and is decaying. This is in conformance with many test observations, including tests by EPRIsolutions described in this report. Conversely the CIGRÉ model reaches its lowest dynamic resistance when the current reaches its maximum value, a phenomenon unlikely for high currents into ground rods.

2-10

Theory of Ground Electrode Response

If multiple ground rods are located within a few meters of one another, the shells in the LiewDarveniza model will collide and no current will flow across the collision interface. Again this is complicated analytically, but relatively easy to evaluate digitally, and a special program GND_ROD1 was written to calculate the total low frequency resistance of any set of asymmetrically located ground rods using the Liew-Darveniza algorithm (Appendix C) and allowing for shell collision.

The Chisholm-Janischewskyj Model A dynamic ground resistance model (Reference 3) developed by Chisholm and Janischewskyj (hereafter referred as the C-J model and described in detail in Appendix D) has the advantage of being applicable to a wide range of ground electrode geometries and makes use of a similarity method originated by Korsuncev (Reference 4). In 1958, A.V. Korsuncev published an analysis of research results on dynamic resistances of several different ground electrode configurations, and in 1987, Oettle (Reference 5) extended the Korsuncev analysis to include recent experimental results. Korsuncev plotted experimental ground electrode test results in terms of two dimensionless ratios Π1 and Π2, where: Π1 = Π2 =

sR

ρ ρI Eo s 2

Equation 2-8

Equation 2-9

where: s = a characteristic distance from the center of the electrode configuration to its outermost point(meters). R = the electrode dynamic resistance, ohms ρ = earth resistivity, ohm-meters I = instantaneous current, kA Eo = critical soil gradient, kV/m Figure 2-9 is a modernized version of the Korsuncev curve from Reference 5.

2-11

Theory of Ground Electrode Response 1

π

1

R S P

.5 .2 .1 0.05

0.02 0.01 0.01

0.02

0.05

0.1

0.2

0.5

π

2

1

2

5

10

20

50

100

I P S2 ε o

Figure 2-9 Modernized version of the Korsuncev Curve (from Oettle).

It should be clear that the values of ρ, Eo, R and s used to make this curve were probably subject to large errors, and yet it is remarkable that the scatter of the plotted test points is not larger than it is. Eq. 2-10 fits the straight line in Figure 2-9: (log Π1) = -0.342 (log Π2) -1.515

Equation 2-10

As described in Appendix D, the C-J algorithm assumes a constant electrode resistance as long as Π1 is below an initial value. Above that value, Π2 is calculated using the current I in Eq. 2-9, then using the Π2 and the Korsuncev curve or Eq. 2-10, Π1 can be calculated, and from Π1 the corresponding instantaneous resistance R can be found. The algorithm then works its way up and down the Korsuncev curve as the electrode current rises and falls, calculating new values of instantaneous R at each time step until the algorithm terminates. The parameter s in Eqs. (2-8) and (2-9) deserves further explanation. Since s is the distance from the center of the electrode configuration to the farthest point on the electrode, for a single rod s is simply the length of the rod in meters. Figure 2-10 shows the application of this parameter for some more complicated electrode shapes. If two electrodes are far apart, then s is the greatest distance at each electrode only. However, if they are sufficiently close, their ionization areas will overlap, and s becomes the distance using the combined set. This complication is explained in more detail in Appendix D. The existence of this parameter s permits the algorithm to be applied to a wide variety of foundations, ground rod configurations and even butt-wrapped poles.

2-12

Theory of Ground Electrode Response

S

S

S

Figure 2-10 Relationship of Parameter S to Ionization Zones.

A very important contribution of Chisholm and Janischewskyj in Reference 3 is their clarification of intrinsic ground electrode inductance. Based on their experimental results with nanosecond geometrical model measurements and time domain reflectometry, they show that even if the ground were made of solid copper, lightning surge currents entering the ground plane will see an inductance in series with whatever ground electrode resistance exists. This is because the finite velocity of light makes the currents "pile up" at the tower base and direct electromagnetic coupling exists between tower currents and those out in the ground plane. Therefore this inductance is influenced by tower height. Their experimental results combined with an analytical analysis indicate that this footing inductance is:

L footing = 60τ T log

Tcrest

Equation 2-11

τT

where: Lfooting = footing inductance, microhenrys

τT = tower travel time, microseconds

.

Tcrest = tower surge current time to crest, microsecs While this equation was derived for a straight-rising front, it should be roughly applicable for other waveshapes. It holds provided the time to crest of the tower surge current is several times the tower travel time. A 120-foot tower would have a travel time τT at the velocity of light of approximately 0.12 microseconds, and for a tower surge current cresting at 2 microseconds, Lfooting would be 20 microhenrys. If the front had a rate of rise of 25 kA/microsecond ( not an unexpected value) the footing voltage would be 500 kV, even though the footing resistance were zero. Prior to the researches of Chisholm and Janischewskyj, this inductive component was usually ignored. 2-13

Theory of Ground Electrode Response

The Robbins/TFlash Model The present (2001) dynamic resistance model for rods in TFlash was developed by David Robbins of EPRIsolutions and is described in detail in Appendix E. It is basically an expanding rod electrode model of the Liew-Darveniza type. It utilizes tables of default soil characteristics for sand, loam, clay, gravel and stone, including default resistivities, critical ionization gradients, ionization and deionization constants, and a special IonConstant variable to adjust dynamic resistances to fit test data.

Other Models Several other dynamic resistance models have been proposed. Examples are the Oettle model (Reference 5) and the Geri model (Reference 15, 16). It was not possible in the allotted time in 2001 to adapt these models to existing data, but it is recommended that these models also be carefully examined in any future experimental work, particularly the Geri model.

2-14

3 SMALL-SCALE PRELIMINARY EXPERIMENTS ON ELECTRICAL PROPERTIES OF SOIL

Introduction The dynamic resistance of any tower ground electrode depends not only on surge current and time, but also on several fundamental soil properties such as dielectric strength, resistivity, moisture content and morphology. Particular concerns include the effects of moisture and whether a breakdown path - once initiated - dominates subsequent breakdowns. All mathematical models make assumptions about these effects, and it was important to verify the electrical characteristics of our test soils on a small scale before proceeding to full-scale electrode tests. The basic soils available were sand, clay, loam and crushed rock. Electrical characteristics to be determined included uniform field dielectric strength under lightning impulse voltages, effects of moisture and the permanence of breakdown paths. It should be recognized at the outset that soils occur in infinite varieties and mixtures of particulates, and neither small-scale or full-scale tests can possibly cover all the electrical peculiarities that exist on rights-of-way; only a general idea is possible.

Uniform Field Dielectric Strength of Soils One of the fundamental electrical characteristics to determine for the soils under scrutiny was the intrinsic dielectric strength versus type of soil and moisture content. This measurement requires the soil to be placed in a uniform field gap, and lightning surge voltages applied in increasing magnitudes until the soil fails dielectrically.

3-1

Small-Scale Preliminary Experiments on Electrical Properties of Soil

Figure 3-1 Rogowski Electrode Test Cell.

A Rogowski gap (Figure 3-1), centered in a wooden soil-containment box, was built for this purpose. The electrodes were constructed of solid aluminum, and gap spacings could be set by turning the threaded shaft at the high-voltage end. To fill the cell, the containment box was laid on its side with the fill-door uppermost. Soil was slowly poured into the box and between the electrodes, carefully compacting the soil as it was poured. The cell was then turned upright and connected to the test circuit shown in Figure 3-2. Im pulse G enerator

Voltage Divider

Load Capacitor

Rf Cg

Rt

T es t Cell _

+ Sg

4 c m Ga p

R ch _ H V DC

U +

Figure 3-2 Test Circuit for Soil Uniform Field Dielectric Tests.

3-2

Small-Scale Preliminary Experiments on Electrical Properties of Soil

The grounded electrode was covered with a thin sheet of paper before filling took place. By counting punctures in this paper, the number of separate-path breakdowns occurring during a test could be determined and compared with the total number of breakdowns oscillographically. No two breakdowns ever took the same path. Each breakdown path healed with at least the dielectric strength it had before breakdown. This result is indicative that soils heal after a transient lightning event. Once it was established that dielectric breakdowns in soils follow separate paths, it was not necessary to use the paper.

Soil Resistivity A separate cell was used to measure soil resistivity (Figure 3-3). It consisted of two metal plates separated by a ring of PVC plastic with an inner diameter of 7.875 inches and a height of 2.0 inches. 1/4" X 8" X 8" Steel Plate

2" X 8" PVC Tube

1/4" X 8" X 8" Steel Plate

Figure 3-3 Resistivity Test Cell.

The ring was placed on the bottom plate and overfilled with soil. The top plate was then applied and pressed and turned until it just contacted both the top of the ring and the soil inside the ring. Then by measuring the resistance between the two plates, the soil resistivity is given by:

ρ = 0.618 R

Equation 3-1

where:

ρ = soil resistivity, ohm-meters R = measured resistance, ohms Table 3-1 provides results of soil tests using the Rogowski uniform field test cell and lightning transient waveshapes.

3-3

Small-Scale Preliminary Experiments on Electrical Properties of Soil Table 3-1 Uniform Field Dielectric Tests Soil Type

Resistivity (ohm-meters)

Dielectric Strength (kV/meter)

Dry Sand

17E+6

1346

Damp Sand

2900

787-1102

Clay

124

759-955

Loam

1050

1312

Dry Stone

22E+6

892

Wet Stone

48E+3

761

Note that in these tests the dielectric strength of wet or dry stone is not much different from damp sand. The stone itself is a reasonably good dielectric, but the high dielectric constant of the stone enhances the electric gradients in the air voids between the stone particles. It is the air in these voids that is breaking down. Details of the tested soil materials are as follows:

•

Dry Sand: Kiln dried sand to have a material with very high resistivity and dielectric strength.

•

Damp Sand: 24 parts kiln dried sand thoroughly mixed with one part well water by volume. Sand would clump together when squeezed.

•

Clay: Compacted clay dug from the EPRI Lenox Center test site. Natural state, not sifted.

•

Loam: Purchased and compacted top-soil, sifted to remove vegetable matter.

•

Dry Stone: One-half inch mesh glacial till.

•

Wet Stone: Dry stone sprayed with water and allowed to drip to remove any excess.

When normal dielectric breakdown occurs, the voltage wave at the instant of breakdown chops cleanly to or through zero. Notable in these tests was the absence of such a clean chop at the moment of breakdown except for the stone. As Mousa (Ref. 2) points out, these small sample dielectric strengths can be misleading in that on a long rod electrode, the minimum dielectric strength on the rod is likely to be substantially less than an average small sample measurement, and it is at the minimum dielectric strength location where streamers will initiate. It appears that the soil minimum dielectric strength is roughly half the average small sample measured value for a typical soil.

3-4

Small-Scale Preliminary Experiments on Electrical Properties of Soil

Non-Uniform Field Dielectric Strengths of Soils While the uniform field tests above can provide fundamental dielectric information on the tested soils, additional information is needed to characterize the process of breakdown in the nonuniform radial fields around rods or counterpoise wire. A concentric cylindrical electrode test cell is ideal for this purpose since it can simulate rod/wire radial fields and its electric field gradients are easily computed. Figure 3-4 shows the concentric cylindrical test cell devised for these tests, and Figure 3-5 displays the equivalent circuit. The outer electrode was a metal barrel 22.5 inches inside diameter with metal guard rings on each end to keep divergence of the electric fields at the ends as small as practical. The center electrode was a 0.625 inch copper rod with corona spheres on each end to inhibit streamer discharges off the ends of the rod. The guard rings in Figure 3-5 were separated from the active center of the test cell by solid 1-inch plastic separating panels, and soil was packed tightly on both sides of the panels. Impulse voltages were applied to the center rod, and the center cylinder electrode was grounded through a highfrequency current transformer.

Figure 3-4 Photo of the Concentric Cylindrical Test Cell.

3-5

Small-Scale Preliminary Experiments on Electrical Properties of Soil Impulse Generator

Voltage Divider

Load Capacitor

Rf Cg Rt

_

+ Sg

Rc h

U

+

HVDC

Rx 1 1.06 Test Concentric Cell

Polypropylene

_

Figure 3-5 Electrical Diagram of the Concentric Cylindrical Test Cell.

Dimensions of the cell are as follows: D1=0.625" (0.0159 meters) D2=22.5" (0.57 meters) L=35" (0.89 meters) The electric charge-free gradient Eo on the surface of the inner rod for any applied voltage is:

Eo =

2V D D1 log 2  D1

  

where: Eo = electrode gradient, kV/meter V = applied voltage, kV

3-6

Equation 3-2

Small-Scale Preliminary Experiments on Electrical Properties of Soil

The low frequency resistance between the center rod and the test cell wall is: Rx =

ρ   D2    log 2πL   D1  

Equation 3-3

where: Rx = cell resistance, ohms.

ρ = soil resistivity, ohm-meters L = length of center electrode, meters Since the cell resistance Rx can be easily measured, the soil resistivity ρ can be extracted from Eq. 3-3 to be:

ρ=

2πLR x D  log 2   D1 

Equation 3-4

Tests in kiln-dried sand (Table 3-2) provided the 17E+6 ohm-meters resistivity. There was a pronounced volt-time effect in breakdown voltages (Table 3-2): Table 3-2 Breakdown Vs Waveshape: Kiln-Dried Sand. Waveshape

Breakdown Crest kV

Square Wave

90

20 x 70

259

By Eq. 3-2, the free-field gradient on the inner rod at 259 kV applied voltage would be 9091 kV/meter or 231 volts/mil. This is in the range of transformer oil impulse strength for small sphere-plane gaps and far exceeds the Rogowski dielectric strength of dry sand in Table 3-1, indicating that streamers started across the gap long before crest voltage was reached. Neither the CIGRÉ dynamic resistance model (Appendix A) nor the Korsuncev curve (Appendix D) make any allowance for volt-time effects, whereas the present TFlash model (Appendix E) and the Liew-Darveniza model (Appendix B) do. Soils in the concentric cylindrical test cell showed a significant reduction in dielectric strength as soil moisture was increased. Table 3-3 shows both effect of moisture and tail time on dielectric strength of sand in the concentric test cell.

3-7

Small-Scale Preliminary Experiments on Electrical Properties of Soil Table 3-3 Effect of Moisture and Tail Time. Material

Breakdown Crest kV

Tail Time - Microseconds

Dry Sand

245

70

Dry Sand

221

70

Dry Sand

112

Long tail

Damp Sand

61

Long tail

In Table 3-3, a long-tailed wave (approximately 200 microseconds) decreased the dielectric strength of dry sand to approximately half its 70 microsecond value and adding moisture made a significant further decrease. These tests confirm the need to incorporate volt-time ionization effects in any digital dynamic resistance model such as TFlash, and underscore the influence of ground moisture in lightning performance of transmission structure grounds. As in the uniform field tests, "breakdown" was not accompanied by an abrupt chop of the voltage wave to zero. Rather it consisted of a sudden increase in current and a pinhole puncture of the paper lining on the inside of the outer electrode. Figure 3-6 shows a set of "before breakdown" and "breakdown" voltage waves. 200.0

160.0

120.0 Breakdown kV 80.0

40.0 Withstand 0.0 0.0

1.0

2.0

3.0

4.0

5.0

6.0

Time - Microseconds

Figure 3-6 Concentric Test Cell Voltage Waves Before and During Breakdown.

Apparently-at least in sand-the soil surrounding the breakdown path helps cool the arc and increase the gas pressure sufficiently to increase the resistance substantially above that of a similar path in air and to encourage a gradual drop in resistance as postulated by Liew-Darveniza (Ref. 1).

3-8

4 LARGE-SCALE EXPERIMENTS ON DYNAMIC RESISTANCES OF GROUND RODS

Introduction Much oscillographic information on surge voltage and current response of ground rods has been published, and Reference 1 provides a comprehensive set of oscillograms from a variety of sources. However, several practical aspects of ground rod performance vital to EPRI members have not been considered in published full-scale experiments, including:

•

The efficacy of ground rods or counterpoise wires with sharp edges vs. conventional round rods or wires.

•

Dynamic resistance characteristics of butt-wrapped pole grounds as an effective alternative to driven rods.

•

Efficacy of chemical grounding enhancement materials in handling high frequency high amplitude lightning surge currents.

•

Dynamic resistance characteristics of tower concrete foundations.

•

Determination of the percentage of dynamic ionization currents flowing off the ends of ground rods vs. that flowing off the side-walls. This is of particular importance when rod ends penetrate into the lower layers of two-layer earths.

Finally, it was of particular importance to build into TFlash a dynamic resistance model that best simulates large-scale impulse response of ground rods and tower foundations. For this, comparative dynamic resistance data were needed to supply soil constants and ionization coefficients.

Test Configurations and Soils The ground strata at EPRI Center in Lenox, MA has principally a two-layer morphology, with an upper layer of mixed loam-sand and a lower layer of clay interspersed with a water table. This complex condition does not permit controlled testing of soil dynamic response of ground rods driven into the EPRI center yard, particularly if a ground rod penetrates into the underlying water table. For this reason, an above ground steel test tank was built into which various kinds of soils and ground electrodes could be experimentally investigated. Figures 4-1 and 4-2 describe the tank in detail. The tank was eight-feet high and six-feet in diameter. Any ground electrode to be tested was located at the center of the cylinder to depths ranging from three to five feet, depending on the test desired. Currents collected by the tank walls and by the tank bottom could 4-1

Large-Scale Experiments on Dynamic Resistances of Ground Rods

be measured separately with high-frequency Pearson current transformers Voltages were measures with a compensated resistance divider located so as to minimize voltage induction in the measuring circuit by circulating surge currents. Soils and electrode configurations tested are displayed in Table 4-1: Impulse Generator Rf Rt

Cg

1" Ground Rod _

+

Sg

Rch

_

U

+

HVDC

Voltage Divider

6'

8'

Bottom Plate

Figure 4-1 Large Test Tank Dimensions and Circuitry.

Figure 4-2 Photo of Large Test Tank Installation.

4-2

Large-Scale Experiments on Dynamic Resistances of Ground Rods Table 4-1 Large Scale Dynamic Resistance Tests. Electrode

Round Rod - 3 ft.

Dry Sand

Damp Sand

Loam

Number of Impulse Tests

Test Currents (kA)

5

1 – 12

X

7

1 – 12

X

26

1 – 12

14

1 – 12

X

Round Rod - 4 ft. Round Rod - 5 ft.

X

X

Round Rod - 5.5 ft.

X

X

Round Rod - 6 ft.

X

8

1 – 12

Rectangular Bar - 5.5 ft.

X

54

1 – 11

15

1 – 5.6

Concrete Encased - 5 ft.

X

Rod With Streamers

X

12

8 – 13

Butt-Wrapped Pole

X

17

4 – 14

Rod Next to Pole

X

13

5 – 14

10

1.3 – 4

Chemical Ground

X

While these tests do not simulate long ground rods, they are sufficient to examine the dynamics of ground electrode resistance on a large scale. The six-foot diameter of the test tank limited streamer distances to no more than three-feet from any center rod to the tank wall. This was sufficient for most streamer clearances but did not simulate the infinite distance to zero resistance experienced by a rod driven into an infinite ground plane. The net effect is that resistances between rod and tank wall will be lower than to an infinite environment. To determine the correction factor to adjust rod-to-tank resistance measurements, a program called TANKOHMS was written, and is described in Appendix G. Computed resistances to an infinite radial distance were found to be approximately 44% higher than resistances to the tank wall and this correction was used to adjust all tabulations to an infinite condition.

Digital Filtering During high voltage measurements in outdoor environments, good electromagnetic shielding is not possible, and extraneous electrical noise can seriously distort measurements made with modern transistorized low-signal oscilloscopes. Some digital filtering is often necessary to smooth the waveshapes. The waveshapes in this series of experiments were smoothed with a Chebyshev Type 1 Filter:

4-3

Large-Scale Experiments on Dynamic Resistances of Ground Rods

50000

2000

0

1000

-50000 0

-150000

-1000

-200000

-2000

Current

Voltage

-100000

-250000 -3000 -300000 -4000

-350000 -400000 -0.000001

0

0.000001

0.000002

0.000003

0.000004

0.000005

0.000006

-5000 0.000007

Time-Microseconds Voltage

Current

Figure 4-3 Unfiltered Data.

50000

2000

0

1000

0

-100000 -1000 -150000 -2000 -200000 -3000

-250000

-4000

-300000

-350000 -0.000001

Current

Voltage

-50000

0

0.000001

0.000002

0.000003

0.000004

0.000005

0.000006

-5000 0.000007

Time-Microseconds Voltage

Current

Figure 4-4 Chebyshev Filtered Data.

Voltage and Current Waveshapes In the large-scale tests it was desirable to apply a current wave to any electrode in the tank that would at least roughly approximate in shape a lightning transient in the field. Because of the non-linearities between voltage and current, an approximate double exponential rod current wave required a very distorted voltage wave. Fig 4-5 displays an example.

4-4

Large-Scale Experiments on Dynamic Resistances of Ground Rods

Figure 4-5 Typical Voltage-Current Waves During Rod Tests.

There was usually some ringing on the current wave caused by the long lead inductance, generator inductance and rod inductance, but this ringing was not severe, and did not appear significantly in the measured dynamic resistance.

Initial Dynamic Resistances When the first large scale impulse test was carried out on a driven rod in dry sand in the test tank, it was immediately apparent that the dynamic impedance at low surge currents was much lower than the meggered rod low-frequency resistance. Part of this difference is caused by capacitance charging currents on the front of the applied voltage wave, but it also became apparent that the low frequency meggered resistance was made excessively high by poor soil contact to the rod surface and possibly by electrolytic effects. When a rod is driven - particularly into compacted sand - it vibrates, and this vibration slightly enlarges the hole around the rod, making for poor contact between rod and soil at various segments along its length. However, on application of impulse currents even low magnitudes of surge currents created the few hundred volts necessary to bridge the very small gap between rod wall and the adjacent sand, making a good connection. (It might be noted that when electrolytic tanks were used to calculate electric fields before the advent of digital field plotting, it was found that copper and aluminum did not "wet" as well as iron, making for substantial resistance measurement errors if the former were used as electrodes in water at low voltage levels).

4-5

Large-Scale Experiments on Dynamic Resistances of Ground Rods

To adjust for this poor soil contact, the early beginnings of the surge voltage and current waves were used to derive a meaningful initial dynamic resistance. Figure 4-6 illustrates the initial relation between surge current and dynamic ohms:

Figure 4-6 Initial Dynamic Resistance Computation.

Effectively, the highest dynamic resistance encountered (invariably during the surge current initial rise time) was used as a "working" initial dynamic resistance, and any resistances lower than this value (due to noise) that occurred prior to that value were adjusted to the "working" value. An additional constraint was utilized rejecting the voltages and currents at the first two time steps in making this calculation. This was necessary because of noise levels at these two steps were sometimes strong enough to override the true voltage-current signals, even with filtering, and dividing observed voltages by observed currents to obtain impedances at low signal levels creates major errors. This working initial dynamic resistance evaluation did not include effects of capacitance charging currents on the front of the applied voltage wave. Rod to tank wall capacitance was of the order of 200 picofarads for a soil relative dielectric constant of 10. For a voltage rise time of 500 kV/microsecond, this corresponded to a charging current of 100 amperes. Since most of the surge currents into the rods were in the order of kiloamperes, the charging currents were neglected in the initial impedance evaluations. An effort was also made to evaluate rod inductance. The rod inductance could not be resolved reliably from the voltage and current oscillograms because the dynamic resistance effects obscured the inductive contribution. This inductance is determined by rod to tank geometry and is roughly independent of soil dielectric characteristics. However, in our test configuration any indicated inductance is distorted by the tank iron, by the voltage divider measuring loop and by 4-6

Large-Scale Experiments on Dynamic Resistances of Ground Rods

inductance of the grounding cables leaving the tank. A separate set of outdoor tests in 2002 with driven rods in a uniform earth is recommended. Equation (4-1) from Grover (Reference 14) for inductance of a rod whose length is much greater than its radius is:

L=

µS   2 S   log  − 1 2π   r  

Equation 4-1

where: L = rod inductance, henrys

µ = soil permeability 4π x 10-7 henrys/meter S = rod length, meters r = rod radius, meters However this equation assumes that the current in the rod is constant throughout its length, whereas the actual rod current varies along the rod as it is drained out into the soil. Eq. 4-1 yields a value of 5.7 microhenrys for a five meter long one-inch diameter rod. This is roughly 38 percent of the classical inductance of a 100 foot tubular steel pole, and this inductance plays a major role in determining insulator voltages. Even if the inductance is reduced by half by current leakage into the soil it is still very significant. Further research is indicated to evaluate inductances of ground rods. Historically this inductance has usually been neglected.

4-7

Large-Scale Experiments on Dynamic Resistances of Ground Rods

Dynamic Ohms vs. Applied Current Even at a few kA, the dynamic ohms dropped rapidly with current for all soils tested. Figure 4-7 illustrates a typical relationship between the dynamic ohms of a three-foot rod in sand and the instantaneous current fed into the rod.

Figure 4-7 Dynamic Ohms vs. Applied Current.

A substantial drop in impedance starting at about one kA is displayed. The small loop at the bottom of the trace is from electrical noise. The trace was cut off at six microseconds. If it had been permitted to continue, it would have eventually returned to the initial value as the current died to zero.

Effect of Rod Length and Separation The low-frequency low-current resistance of a single rod in uniform soil can be determined from the Dwight Equation (2-1). In the test tank however, it was not possible to directly compare separations of rods. But as will be shown later, the Liew-Darveniza algorithm matches the tank test values quite well when corrected for the finite distance from rod to tank wall, and this algorithm does permit analysis of separation of pairs of rods.

4-8

Large-Scale Experiments on Dynamic Resistances of Ground Rods

Dynamic Ohms

20

(2) 15 Foot Rods 10 feet Apart

15

10

(1) 30 Foot Rod 5

0

0

10

20

30

40

50

60

70

80

Total Surge Current - Crest kA Figure 4-8 Dynamic Resistance at Two Microseconds As a Function of Current Into One 30-Foot Rod or Two 15-Foot Parallel Rods Spaced 10 Feet Apart.

Figure 4-8 uses the Liew-Darveniza algorithm (Appendix B) to compare the dynamic resistance of a single 30-foot ground rod with two 15-foot rods separated by 10 feet. Soil resistivity was assumed 300 ohm-meters, the ionization constant Ti was 1.0 microsecond, and the soil critical ionization gradient was 300 kV/meter. The crest surge current was that into either one rod or into the pair of rods, and the surge current waveshape was 2-50 microseconds. The low frequency resistance of the single rod was 35.1 ohms versus 35.7 ohms for the pair (essentially the same). As shown in Figure 4-8, the single rod is always better by having the lower resistance, regardless of the separation distance between the two rods. Part of he reason for this involves the ionization dynamics. For the same voltage applied, each of the two rods receives only half the current fed into the single rod, making the ionization envelopes around two rods less than that for the single rod. There is also mutual coupling between the two rods, and streamers traveling from one rod in the direction of the other are inhibited because the two rods are at the same potential (Figure 2-3). While it was not possible to evaluate rod length effects greater than 5.5 feet in the test tank, the test tank observation that dynamic resistance of the rods and tested soils followed the LiewDarveniza algorithm (Appendix B) reasonably well made it possible to use this algorithm to examine rod length effects up to 30 feet.

Comparison of Tank Test Values with Dynamic Algorithms Dynamic resistance values versus time were measured in the test tank for sand, damp sand and loam for ground rods ranging from 3.0 feet to 5.5 feet, the range of lengths permitted by the dimensions of the test tank. The results were then compared with that predicted by various dynamic resistance algorithms. (Nevertheless this limited range of rod lengths was sufficient to bring into play all the ionization effects that occur in practice, and to permit comparisons to be made with predicted values). In the following examples, all observed dynamic resistances were corrected for the proximity effect of the tank wall, and the initial (low current) resistance was that extracted from the front of the voltage and current waves, since - as pointed out above 4-9

Large-Scale Experiments on Dynamic Resistances of Ground Rods

meggered low frequency values were in error because of slight gaps created between rod walls and the surrounding sand when the rods were driven.

Liew-Darveniza Algorithm Figure 4-9 shows an example of a rod surge current, observed dynamic resistance and predicted dynamic resistance using the Liew-Darveniza algorithm (Appendix B).

Figure 4-9 Liew Darveniza Algorithm Comparison With Dynamic Resistance Test Values.

Using a critical ionization gradient of 300 kV/meter as recommended by Mousa, the actual versus predicted dynamic resistances are quite close.

4-10

Large-Scale Experiments on Dynamic Resistances of Ground Rods

CIGRÉ/Weck Algorithm Figure 4-10 shows an example of the same comparison as in Figure 4-9, but using the CIGRÉ/Weck algorithm (Eq. 2-7 ). The critical ionization current was 0.8 kA as is apparent from the observed current and resistance waveshapes.

Figure 4-10 CIGRÉ/Weck Algorithm Comparison With Dynamic Resistance Test Values.

The theoretical and observed values do not compare very well. However, if instead of the critical ionization current of 0.8 kA indicated by the oscillograms, the value proposed by Eq. 2-4 had been used (0.21 kA), the observed and calculated values would have been quite good. Note however that the calculated dynamic resistance is dropping prematurely.

4-11

Large-Scale Experiments on Dynamic Resistances of Ground Rods

Chisholm - Janischewskyj Algorithm Figure 4-11 compares the same rod dynamic resistance with that predicted by the ChisholmJanischewskyj (C-J) algorithm (Appendix D).

Figure 4-11 C_J Algorithm Comparison With Dynamic Resistance Test Values.

The match is quite good for the final dynamic resistance at 5 microseconds, but - as is the case for the CIGRÉ algorithm - the dynamic resistance decays prematurely

4-12

Large-Scale Experiments on Dynamic Resistances of Ground Rods

TFlash Algorithm Figure 4-12 compares the same rod dynamic resistance with that predicted by the algorithm that is presently in TFlash (Appendix E). This algorithm was devised by David Robbins of EPRIsolutions, and is based on the Liew-Darveniza algorithm, with the addition of a table of scaling constants.

Figure 4-12 TFlash Algorithm Comparison With Dynamic Resistance Values.

The match is again quite good with some premature drop in calculated dynamic resistance compared with measurements. In general the above trends were common throughout the range of rod lengths and soils tested. Any one of these algorithms can be made to match measured values reasonably well if the constants are chosen properly, but the constants will be different for different algorithms.

Effect of Rod Shape and Artificial Streamers If dynamic resistance is reduced as the number of streamers from a rod are increased, then it can be reasonably assumed that the addition of artificial metallic streamers to a ground rod should 4-13

Large-Scale Experiments on Dynamic Resistances of Ground Rods

further reduce the dynamic resistance. Figure 4-13 shows a set of artificial streamers made from copper wire and attached to every six inches along the bottom three feet of a four foot rod. The rod was inserted at the center of the test tank, and loam carefully packed around it until the tank was completely filled.

Figure 4-13 Artificial Streamers Attached to a Ground Rod.

Figure 4-14 displays an oscillogram of dynamic resistance for a 10.5 kA peak current with the streamers in place, and Figure 4-15 shows the same test result with the streamers removed. There is essentially no difference

Figure 4-14 Dynamic Resistance With Artificial Streamers in Place.

4-14

Large-Scale Experiments on Dynamic Resistances of Ground Rods

Figure 4-15 Dynamic Resistance With Artificial Streamers Removed.

The apparent reason for a lack of improvement using artificial streamers on a ground rod is that there are so many natural breakdown streamers and ionization pockets along the rod in any case, and they extend so far beyond the length of the artificial streamers, so as to mask out any artificial additions. This lack of improvement was also apparent in changes in rod cross-sectional shape. Tests were made comparing the dynamic resistance provided by a rectangular bar with sharp edges as compared to a round rod with no edges. Figure 4-16 shows the cross-sectional area of a tested rectangular copper bar with a total surface area of five inches. 2"

1/2 "

Figure 4-16 Tested Bar Cross-Section.

4-15

Large-Scale Experiments on Dynamic Resistances of Ground Rods

Figure 4-17 shows a dynamic resistance oscillogram of this electrode as compared with Figure 4-18 for a round rod. The dynamic resistances of the rectangular rod with sharp edges and the round rod with substantially less surface area and no edges are essentially the same.

Figure 4-17 Rectangular Bar Ground Rod Dynamic Resistance in sand.

Figure 4-18 One Inch Diameter Round Rod in Sand.

4-16

Large-Scale Experiments on Dynamic Resistances of Ground Rods

As far as dynamic resistance is concerned, the cloud of streamers and ionization off either the round or rectangular rod masks the rod cross-sectional geometry, and the shape of the rod crosssection makes little difference in dynamic resistance. This suggests the following rule: Select ground rod cross-section geometry for mechanical reasons. Variations in shape or size will have little influence on the rod dynamic resistance.

Response of Concrete Ground Electrodes In the six-foot diameter test tank, it was not possible to bury concrete tower foundations and still maintain sufficient clearance to the tank walls. However, an attempt was made to examine the transient response of a concrete incased ground rod. Figure 4-19 shows one result of a set of tests on a 5/8 inch rebar five feet long incased at the center of an eight-inch cylinder of concrete and buried in damp sand.

Figure 4-19 Voltage-Current Waves on a Concrete-Encased Rebar.

Until 2.4 microseconds the dynamic resistance process was progressing normally, but at that time a breakdown started from a section of the rod outside the concrete which dived under the sand near the surface to the tank wall. The current increase after 2.4 microseconds is the result of this failure. Note that again the voltage does not chop suddenly to zero as it would do in air. The sand appears to increase resistance of the arc by cooling and compressing the failure channel. There was no evidence of concrete fracture from the severe dielectric stresses in the concrete around the rod. A useful calculated dynamic resistance could not be obtained from this test because of the complete breakdown from the concrete to the tank wall.

4-17

Large-Scale Experiments on Dynamic Resistances of Ground Rods

Chemical Enhancement The limited size of the test tank prohibited complete tests of effects of chemical enhancement materials on the dynamic resistances of ground rods. The chemicals appeared to migrate out into the damp sand and initiate failure paths, obscuring dynamic resistance measurements. Figure 4-20 shows one example.

Figure 4-20 4.5-Foot Ground Rod Surrounded With Eight Inch Diameter GEM Material.

Dynamic resistance reduction does occur with time, but – because of the limited size of the test tank – a comparison with damp sand alone was postponed to a future date when an outdoor test could be made.

Butt-Wrapped Poles One set of tests was completed embedding a butt-wrapped grounding electrode in sand. Figure 4-21 shows the butt-wrapped pole electrode. The butt-wrap was 14 turns of #6 stranded wire on the butt of an eight-inch-diameter pole embedded in loam.

4-18

Large-Scale Experiments on Dynamic Resistances of Ground Rods

Figure 4-21 Photo of Bottom End of Butt-Wrapped Pole.

A dynamic resistance oscillogram of this grounding arrangement is shown in Figure 4-22. In general, the dynamic resistance was roughly the same as for a five-foot ground rod.

Figure 4-22 Dynamic Resistance of a Butt-Wrapped Pole in Loam.

4-19

Large-Scale Experiments on Dynamic Resistances of Ground Rods

Computer Modeling of Dynamic Resistance It is clear from the above tests that relying on meggered low-frequency resistances of towers to access transmission line lightning performance can be very misleading. It is the dynamic resistance of grounding electrodes that governs the flashover process, not the meggered resistance. It is clear from both the Korsuncev and Liew-Darveniza algorithms that major efforts to improve meggered resistances can - in some cases - have little effect on dynamic resistance, and conversely that in some cases, the dynamic resistance will be sufficient although the meggered resistance appears inadequate. Dynamic resistance changes with time, but by six microseconds the dynamic resistance has usually stabilized for most lightning currents and flashover has taken place if it is going to. Dynamic resistances depend on soil resistivity, but so do meggered resistances. Dynamic resistance depends on a critical ionization gradient of the earth around the electrode, but the analysis in this report on various soils was fairly successful if a gradient of 300 kV/meter was used (as recommended by Mousa, Reference 2), rather than the IEEE 400 kV/meter value. If soil resistivity is known, the Liew-Darveniza algorithm can provide the dynamic resistance of rods (Appendix B) and the Korsuncev C-J algorithm (Appendix D) can provide values for either rods or tower concrete foundations. The C-J algorithm is very easy to use. A comparison of the Liew-Darveniza and Korsuncev C-J algorithms is shown in Fig.4-23 as a function of rod length. 40

Lowest Dynamic Ohms

30

ρ = 500 Ω / meter

20

Lieu Darveniza

Korcuncev

10 Korcuncev

0

Lieu Darveniza

ρ = 100 Ω / meter 0

2

4

6

8

10

12

Rod Length - Meters Figure 4-23 Comparison of Lowest Dynamic Ohms Calculated by Korsuncev and Liew-Darveniza Algorithms.

4-20

Large-Scale Experiments on Dynamic Resistances of Ground Rods

At a soil resistivity of 100 ohm-meters the six-microsecond dynamic resistance provided by the Liew-Darveniza algorithm matches the Korsuncev C-J algorithm quite well. At a higher resistivity of 500 ohm-meters, the C-J algorithm is more optimistic at long rod lengths. In the tests made in this report, both algorithms yielded similar results at six microseconds. Based on the above, it was decided to examine how target dynamic resistances could be utilized during line construction. Because of our lack of concrete foundation data at this time the procedure was restricted to ground rods. Figure 4-24 displays dynamic resistance (at 6 microseconds) of single rods in uniform soil as a function of soil resistivity and rod length up to 30 feet, using the C-J algorithm. One can interpolate between the curves to arrive at a target value. If a single rod is not sufficient, or if two rods are required anyway (H-frame construction), Figure 4-25 provides improvement in percent if two rods are used instead of one. Note that for short rods, increased separation distance can make a substantial improvement, but for 30-foot rods there is little improvement by driving two rods instead of one even if the separation distance is 30 feet. In no case does two rods yield half the resistance of one rod.

Dynamic Resistance Ohms

50

32 kA 2 X 50 Wave,

40

ε

c

= 300 kV/m,

30

ρ = 1000 Ω / meter ρ = 700 Ω / meter ρ = 500 Ω / meter ρ = 300 Ω / meter

20

10

0

ρ = 100 Ω / meter ρ = 50 Ω / meter 0

5

10

15

20

25

30

Rod Length - Feet Figure 4-24 Single Rod Korsuncev Dynamic Resistance vs Rod Length and Soil Resistivity.

4-21

Large-Scale Experiments on Dynamic Resistances of Ground Rods

Percent of Resistance of One Rod

100% 3 90%

6

80%

10

70%

15 20

60% 30 Foot Spacing 50% 0

5

10

15

20

25

30

Rod Length - Feet Figure 4-25 Dynamic Resistance of Two ground Rods in Parallel vs. Rod Length and Separation Distance (Korsuncev).

4-22

Large-Scale Experiments on Dynamic Resistances of Ground Rods

If the C-J algorithm is correct (and it seems to have the blessing of CIGRÉ) , then the dynamic resistance reduction for two rods is a lot less than the low-frequency reduction for two rods compared to one. Figure 4-26 shows a meggered resistance calculation for one and two threefoot rods as a function of spacing. Even with a separation of five-feet, a Liew-Darveniza calculation shows a great improvement in meggered resistance for two rods in place of one, and the reduction is asymptotic to half the single rod resistance as spacing increases. In the EPRI "Red Book" (Reference 20), a two-rod formula by Rudenberg was included in Chapter 12. As can be seen in Figure 4-26, the Rudenberg formula is in error for large rod spacings. 90

Single rod resistance - Rudenburg Single rod resistance - Lieu Darveniza

80

Single rod resistance - Dwight

Low Frequency Resistance Ohms

70 60

50

Lieu Darveniza

40

Single rod resistance / 2 Rudenberg

30 20 10 0

0

5

10

15

20

25

30

Spacing Between Rods - Feet Figure 4-26 Variation of Double Rod Low Frequency Resistance With Spacing. (1-in.-Diam. Rod, 3 Feet Long, ρ=100 ohm-meters).

4-23

5 SUMMARY

I.

The ground rod and tower foundation dynamic resistance models examined in this research project will only be as good as the soil constants that govern the model response, particularly soil resistivity. These soil constants are given default values in TFlash, but are subject to wide variations in the field, and these variations represent a fundamental limitation to the precision of any evaluation of line performance. Better field measurement technologies are needed to determine in the vicinity of any tower, as follows: A.

Earth resistivity as a function of depth and its anticipated seasonal range.

B.

The critical earth ionization gradient Eo around the grounding electrodes and concrete foundations.

C.

The low frequency resistance of concrete tower foundations after the moisture inside the concrete has reached equilibrium with changing moisture conditions of the surrounding earth.

Item A requires either efficient transfer of aerial survey data or onsite test data into TFlash data files. Item B can be found by on-site tests, as an alternative to present published generic values of ionization gradients for different soils. Item C also will require on-site tests.

II.

Measurement of low frequency ground rod resistance can be in considerable error because vibration as the rod is driven can enlarge the hole around the rod slightly, reducing wall contact with the soil. Dynamic resistance will not be greatly effected by lack of wall contact since electrical breakdowns will take place across any small gap.

III.

Both the Korsuncev and Liew-Darveniza algorithms yielded a fairly good agreement with test values for the range of rod sizes and soils tested.

IV.

The cross-sectional shape or presence of artificial streamers on a ground rod made little difference in dynamic resistance of the rod. The rod cross-section should be selected for mechanical requirements only.

V.

A butt-wrapped pole in loam had approximately the dynamic resistance characteristic of a five-foot ground rod.

5-1

Summary

VI.

Tests of chemical enhancement of dynamic resistance of ground rods were not successful because of test tank size limitations, and should be replicated in future in an outdoor test. Some dynamic resistance improvement was evident.

VII.

Dynamic resistances of ground rods have been shown in this report to have very different relationships with soil resistivities, rod lengths, diameters and separation distances than do low-frequency meggered values. Based on preliminary results, EPRI should complete development of dynamic resistance specifications of transmission towers to be applied during construction. These would be more relevant to transmission line reliability that meggered resistance targets.

5-2

6 GOOD GROUNDING PRACTICES

From results of the grounding research documented in this report, the following grounding practices are summarized:

I.

Driving a ground rod (particularly one with a coupling on it) may vibrate the rod in some soils. This vibration may slightly enlarge the hole around the ground rod. making for poor electrical contact along some of the rod surfaces, thereby increasing the initial measured resistance. Repeating the measurement a few months later when the earth has had a chance to settle will supply a more realistic resistance value.

II.

Concrete tower foundations with reinforcing steel inside can serve as good ground electrodes. It will require a few months for the moisture environment inside the concrete to reach equilibrium, and this depends on the moisture content of the soil surrounding the foundation. At least three months after pouring may be required for a realistic measurement of the foundation resistance to be taken. It is recommended that at least one ground rod be driven parallel to the foundation to help equalize electrical stresses inside the concrete when severe lightning surge voltages appear.

III.

In uniform soil, one 30-foot ground rod will provide a lower dynamic resistance to lightning currents that two 15-foot rods. In general, keep the number of rods as small as practical to attain a required low-frequency (meggered) resistance. However, if ground rods are used on H-frame structures, it is important to drive at least one ground rod at the base of each pole.

IV.

If multiple ground rods are used, separate them as far as practical so that mutual effects that increase the total resistance will be minimized.

6-1

7 RECOMMENDATIONS FOR FUTURE WORK

Based on test results in 2001, the following recommendations are made for possible future work: 1. Devise efficient methods to survey soil morphology at transmission towers during construction, including presence of multiple layers of soil of different resistivities, so that lightning flashover simulations will have improved accuracy. 2. Complete chemical enhancement tests of rod dynamic resistance. 3. Complete concrete tower foundation tests of dynamic resistance in an outdoor environment. 4. Develop an effective procedure for EPRI members to set targets for tower dynamic resistance values during line construction. 5. Continue the present tank tests of rod dynamic resistance with soils not yet tested, such as clay, gravel and crushed rock. 6. Apply the Liew-Darveniza dynamic resistance algorithm to two-layer soils and compare with test results. 7. Supply to EPRI members, a comprehensive computer program to guide them through all phases of grounding strategies.

7-1

8 REFERENCES

1.

A.C. Liew, M. Darveniza "Dynamic Model of Impulse Characteristics of Concentrated Earths", Proceedings of the IEE, vol 121, no. 2, pp. 123-135, Feb. 1974.

2.

A.M. Mousa "The Soil Ionization Gradient Associated With Discharge of High Currents Into Concentrated Electrodes", IEEE Trans. on Power Delivery, vol. 9, no. 3, July 1994, pp. 1669-1677.

3.

W.A. Chisholm, W. Janischewskyj "Lightning Surge Response of Ground Electrodes". IEEE Trans on Power Delivery, vol PWRD-4, no, 2, 1989, pp. 1329-1337.

4.

A.V. Korsuncev "Application of the Theory of Similarity to Calculation of Impulse Characteristics of Concentrated Electrodes", Elektrichestvo, no. 5, 1958, pp. 31-35.

5.

E.E. Oettle "A New General Estimation Curve for Predicting the Impulse Impedance of Concentrated Earth Electrodes", IEEE Transactions on Power Delivery, vol. 3, no. 4, 1988, pp. 2020-2029.

6.

J.G.Anderson "Transmission Line Reference Book: 345 kV and Above," Second Edition, Chapter 12, Electric Power Research Institute, Palo Alto, California.

7.

F. Dawalibi "Transmission Line Grounding", EPRI Research Report 1494-1, Final Report EL2699, October 1982.

8.

J. Nahman, D. Salamon "A Practical Method for the Interpretation of Earth Resistivity Data Obtained From Driven Rod Tests", IEEE Trans. on Power Delivery, vol 3, no. 4, Oct 1988, pp 1375-1379;

9.

B. Thapar, O. Ferrer, D.A. Blank "Ground Resistance of Concrete Foundations in Substation Yards", IEEE Trans on Power Delivery, vol 5, no. 1, January 1990, pp 130136.

10.

B. Thapar, V. Gerez, A. Balakrishnan "Ground Resistance of Concrete Encased Electrodes - Field Tests", Proceedings of the American Power Conference, vol. 52, 1990, pp 421-425.

11.

Julius Preminger "Evaluation of Concrete-Encased Electrodes", IEEE Trans on Industry Applications, vol IA-11, no. 6, November-December 1973, pp 664-668.

8-1

References

12.

Paul Weiner "A Comparison of Concrete Encased Grounding Electrodes to Driven Ground Rods", IEEE Trans. on Industry and General Applications, vol IGA-6, no. 3, May-June 1969, pp. 282-287.

13.

A.L. Vainer "Current Off-Flow From Reinforced Concrete Footings in Poor Conducting Ground", Elektrichestvo, no. 11, 1970, pp 74-77.

14.

F.W. Grover "Inductance Calculations" (book) D. Van Nostrand Company, Inc., New York, 1947.

15.

A. Geri "Behavior of Grounding Systems Excited by High Impulse Currents: the Model and Its Validation", IEEE Trans. on Power Delivery, vol.14, no. 3, July 1999, pp.10081017.

16.

E.Garbagnati, A. Geri, G.Satorio, G.M.Veca "Non-linear Behavior of Ground Electrodes Under Lightning Surge Currents: Computer Modelling and Comparison With Experimental Results", IEEE Trans. on Magnetics, vol.28, no.2,March 1992, pp.14421445.

17.

IEEE Standard 1243-1997 "IEEE Guide for Improving the Lightning Performance of Transmission Lines", 1997.

18.

K.H. Weck "The Current Dependence of Tower Footing Resistance", CIGRÉ 33-88 (WG 01), 14IWD, 1988 and 33-89 (WG 01), 7IWD, 1989.

19.

CIGRÉ Working Group 33.01 (Lightning), "Guide to Procedures for Estimating the Lightning Performance of Transmission Lines", CIGRÉ Technical Brochure 63, Oct. 1991.

20.

EPRI, "Transmission Line Reference Book: 345 kV and Above," Second Edition, Revised, Electric Power Research Institute, Palo Alto, California, 1987.

8-2

A THE CIGRÉ CRITICAL CURRENT EQUATIONS

ro

r

dr

Figure A-1 A thin soil resistance shell surrounding an embedded conducting hemisphere.

Assume that a perfectly conducting hemispherical electrode of radius ro in Figure A-1 can be used as a mathematical equivalent of a rod or tower foundation. The total resistance dR of a thin shell of soil of radius r and thickness dr surrounding this electrode is simply:

dR =

ρ 2πr 2

dr

Equation A-1

Integrating (A-1) from the electrode radius ro out to infinity to find the total soil resistance surrounding the electrode:

Ro =

ρ 2π

 dr  ∫ 2  r r 

Equation A-2

o

or

Ro =

ρ 2πro

Equation A-2

Eq. A-2 is idealized for the resistance of a hemisphere of radius ro in meters for a soil with uniform resistivity ρ ohm-meters.

A-1

The CIGRÉ Critical Current Equations

I

r

J E Figure A-2 Gradient at the surface of a conducting hemisphere

In Figure A-2, the current density J in amperes per sq. meter entering the earth from the surface of any conducting hemisphere is: J=

I . 2πr 2

Equation A-3

where: J = current density, amperes per sq. meter I = current injected into the hemisphere, amperes r = hemisphere radius, meters The electrical gradient at the surface of the hemisphere is governed by the voltage drop in the soil at the surface, or: E = ρJ =

ρJ 2πr 2

Equation A-4

But rearranging Eq. A-2:

r02 =

ρ2 4π 2 Ro2

Plugging A-5 into A-4, with r = ro and I = Io and solving for Io

A-2

Equation A-5

The CIGRÉ Critical Current Equations

Io =

ρE o 2πRo2

Equation A-6

Eq. A-6 is the CIGRÉ equation for the critical ionization current Io in kA at the surface of a hemispherical electrode having a low frequency resistance R0 for a soil having a resistivity ρ in ohm-meters and a dielectric strength Eo in kV per meter. As the current injected into the hemispherical electrode exceeds the critical current I0, the boundary of the hemisphere expands and the equivalent radius r becomes some new value. Note however that from Eq. A-6 that the product of the current and the square of the resistance will always be constant, so IR 2 = I 0 R02 .

Equation A-7

where: I = any injected current exceeding the critical current I0 R = the new resistance for the current I R0 = the low frequency resistance for any current equal or less that I0 I0 = the critical ionization current defined by Eq. A-6. Then from Eq. A-7:

R=

Ro I I0

Equation A-8

Eq. A-8 is the CIGRÉ equation for dynamic resistance as a function of surge current for an equivalent hemispherical breakdown around a tower foundation or large ground electrode. However, Weck (Ref S) presumably recognized that, for a rod, ionization starts almost immediately, making the initial ionization current much lower than specified by Eq. A-6 for a hemispherical electrode. In any case, he concluded that - for a rod - when the current reaches the value specified by Eq. A-6 the rod resistance has already been reduced by approximately the square root of two, or:

R=

Ro I 1+ Io

Equation A-9

A-3

The CIGRÉ Critical Current Equations

where: R = rod dynamic resistance for any surge current I exceeding I0 R0 = rod low frequency resistance at currents less than or equal Io Complexity increases rapidly for two or more ground rods or foundations having a conducting hemisphere around each one. Foundations are usually far enough from each other for the CIGRÉ hemispheres not to collide, making the dynamic resistances independent for each foundation. But ground rods can sometimes be close together, and some of their hemispheres can collide at high currents and Equations (A-8) and (A-9) depart even more from reality.

A-4

B THE LIEW-DARVENIZA ALGORITHM

In 1974, Liew and Darveniza (Ref. 1) reported results of an early series of tests on ground rod dynamic response, and they developed a soil ionization model to simulate the impulse test results, both from their own work and from tests by Bellaschi and others. In this model, a ground rod is surrounded by a set of concentric shells of earth extending from the ground rod to infinity (Figure B-1).

5' 8'

6' Figure B-1 Liew-Darveniza ground rod model surrounded by concentric shells of earth.

In Figure B-1, the resistance dR of any shell of thickness dr is: dR =

(

ρdr

2π r 2 + rL

)

Equation B-1

B-1

The Liew-Darveniza Algorithm

where: dR = total shell resistance

ρ = soil resistivity, ohm-meters r = shell radius, meters dr = shell thickness, meters Assuming no ionization and a rod radius ro, one can integrate Eq. B-1 from ro to infinity to find the low frequency low current rod resistance:

R=

r + L ρ  . log o 2π  ro 

Equation B-2

Liew-Darveniza found that Eq. B-2 matches the Dwight conventional rod resistance equation reasonably well over the range of rods tested for no ionization, giving confidence that their shell model was suitable. The current density J flowing through any shell wall at any time t is simply the rod current at that time divided by the shell area, or J=

(

I

2π r + rL 2

)

Equation B-3

and the voltage gradient E across the shell wall is: E=

Iρ 2π r 2 + rL

(

)

Equation B-4

where: E = shell wall voltage gradient, kV/meter I = rod input current, kA r = shell radius, meters L= rod length, meters

ρ = soil resistivity in the shell, ohm-meters When this voltage gradient E reaches the critical value Eo at a time to, the soil in the shell wall starts ionizing, and the resistivity in the shell wall is assumed to drop exponentially: B-2

The Liew-Darveniza Algorithm

 − (t − t o )     τ1

ρ = ρ o exp

Equation B-5

where:

ρ = shell resistivity at time t, ohm-meters ρo = resistivity before ionization occurs to = time at which ionization begins, microsecs t = time, microsecs Eventually the current density will drop below the critical value and the soil will start deionizing, causing the resistivity to increase. When this occurs, Liew-Darveniza concluded that the resistivity would start increasing as follows:  − (t − t1 )  J  1 −  ρ = ρ i + ( ρ o − ρ i )1 − exp τ 2  J c  

2

Equation B-6

where:

ρ = resistivity at time t ρo = resistivity before the onset of ionization ρi = resistivity at time t1 when the resistivity starts increasing τ2 = a deionization constant J = the current density at time t Jc = the critical current density for this shell Since each shell can have a different resistivity at any time t depending on its state of ionization, the total resistance seen by the rod is the sum of all the shell resistances at time t. Note that the algorithm assumes that when the current density reaches some critical value Jc, ionization begins and when current density drops below this Jc value, deionization begins. Gases do not act this way, nor do liquids - once ionization has occurred, the current density can drop substantially while ionization continues - and it is also likely that aggregate combinations of minerals and air would also not have the same critical ionization and deionization current densities.

B-3

C THE GND_ROD1 ALGORITHM

The GND_ROD1 program was developed to calculate the low frequency resistance of any symmetric or asymmetric combination of ground rods in a constant resistivity soil. It is based on the Liew-Darveniza algorithm (Ref. 1). For the special case no soil ionization. Liew-Darveniza found that their algorithm gave calculated values close to those computed by Dwight for single rods, and EPRIsolutions calculations also yielded values close to those provided by classical formulas (Ref. 6) for multiple ground rods arranged symmetrically. Referring to Figure A-1 of Appendix A, this algorithm represents the earth around a ground rod as a set of concentric shells extending out to infinity. Each shell has a hemispherical bottom, and a constant wall thickness dr. For a single rod, an analytical integration is easy, and Eq.A-2 shows the result. Over the range of dimensions, this equation yields results close to the Dwight equation (Eq.2-1). However, for multiple rods, some of the shells will collide (Figure C-1) or intermix. GND_ROD1 makes the assumption that no current can flow through any collision boundary as long as each rod is at the same potential, a condition almost invariably the case. The total area of all the collision boundaries can then be calculated, and the current that would normally flow through this region is diverted to flow through that part of the shells where no collision exists.

Figure C-1 Collision of Soil Resistance Shells for Two Rod Case

C-1

D THE CHISHOLM-JANISCHEWSKYJ (C-J) MODEL

As summarized in Section 2 of this report, the Chisholm-Janischewskyj (C-J) algorithm for dynamic footing resistance utilizes the Korsuncev curve (Figure D-1) which displays over several orders of magnitude the relationship between two dimensionless variables:

Π1 = Π2 =

sR

ρ ρI Eo s 2

Equation D-1

Equation D-2

where: s = a characteristic dimension of the electrode, meters R = electrode dynamic resistance, ohms

ρ = soil resistivity, ohm-meters I = surge current, kA Eo = critical soil ionization strength, kV/meter

D-1

The Chisholm-Janischewskyj (C-J) Model 1

π

1

R S P

.5 .2 .1 0.05

0.02 0.01 0.01

0.02

0.05

0.1

0.2

0.5

1

π

2

2

5

10

20

50

100

I P S2 ε o

Figure D-1 The Korsuncev Curve for Ground Electrodes

Figure D-1 was compiled from reports for a variety of tower ground electrode tests, and updated by Oettle (Ref. 5) in 1988. The electrode characteristic dimension s is the distance from the center of the electrode at the surface of the earth to its farthest active point. For a rod, s is simply the rod length. For sets of rods, s can vary as shown in Figure D-2, depending on the degree of ionization. The algorithm starts by first determining a value of Π 1 at the low frequency low current resistance of the ground electrode. This value is defined as Π 10 . It can be determined in one of two ways: if the low frequency resistance and earth resistivity are known, then Eq.(D-1) can calculate it directly, However, if only the earth resistivity and the geometry of the electrode are

S

S

Figure D-2 Application of the Characteristic Dimension S.

D-2

S

The Chisholm-Janischewskyj (C-J) Model

known Π 10 and the low frequency resistance of the ground electrode has not been measured, it can be estimated from Eq. (D-3):

 s2  1 Π = 0.4517 + log  2π  A 0 1

Equation D-3

where: A = the electrode area in sq. meters (as if it were wrapped in paper and A is the area of the paper). Using this computed value of Π 10 one can then estimate the low frequency resistance of the electrode by:

Ro =

ρΠ 10 s

Equation D-4

At this stage, working values of Ro and Π 10 have been established. Next, the critical ionization gradient Eo must be selected. If possible, this value should be supplied from test data. In the absence of test data, two options are available. The first option is to use the experimental equation found by Oettle (Ref 5): E o = 241ρ 0.215

Equation D-5

recognizing that this is very likely a pessimistic value, yielding values that tend to limit the spread of ionization around the electrode and yield high values of dynamic resistance. The second option is to adopt 300 kV/meter as the working value of Eo, as recommended by Mousa (Ref. 2). Better still would be a field method developed by EPRIsolutions to measure Eo, ρ and Ro.

D-3

E THE ROBBINS/TFLASH DYNAMIC RESISTANCE MODEL

When TFlash was first developed, some kind of conceptual ground dynamic resistance model was necessary in the program until experiments could verify which algorithm most closely matched field data. After a review of published research, David Robbins - the programmer responsible for all TFlash encoding - combined several dynamic resistance models into what was called the "Robbins/TFlash Model" and this model was encoded into TFlash. The following is a brief description.

I

ρ

ρ

Eo L

Da

Figure E-1 Expanding ionization zone around a rod

E-1

The Robbins/TFlash Dynamic Resistance Model

The current density out at the wall of the FigureE-1 ionization shell is: J=

Eo

ρ

.

Equation E-1

where: J = current density, kA/meter2 Eo = ionization gradient at the wall, kV/meter

ρ = earth resistivity, ohm-meters The total current flowing through the wall (which must be equal to the applied current) is then: I=

E o Π Da L

or Da =

ρ IKρ Eo Π a L

Equation E-2

where: Da = an average intrinsic diameter of the ionized shell, meters L = rod length, meters K = a scaling constant to fit experimental results Note that in Figure E-1, no hemispherical bottom end of the ionizing shell was included. Instead, this will be included in the scaling constant K which can be different for different soils. Initially, the new diameter of the ionized shell will be: NewDiameter=RodDiameter

Equation E-3

At each time step t, the new current I is found and Da then calculated. However, the new diameter does not encompass completely ionized soil because time is required for the ionization to take place. The actual equivalent new diameter assuming completely ionized soil will be less that the new value of Da and the new diameter (provided it is increasing) will be: NewDiameter=LastDiameter +C1(Da-LastDiameter)

Equation E-4

or if the diameter is receding because the current is lessening: NewDiameter=LastDiameter-C2(LastDiameter-Da) E-2

Equation E-5

The Robbins/TFlash Dynamic Resistance Model

where: C1 = Ionization constant for a particular soil C2 = Deionization constant for a particular soil All the soil within NewDiameter is assumed to be perfectly conducting. The new dynamic resistance Rt is then found by the Dwight equation:

Rt =

ρ 

8L    log  −1 .  2ΠL   NewDiameter  

Equation E-6

after which: LastDiameter=NewDiameter

Equation E-7

The main difficulty with this dynamic resistance model - as with all the models - is determining default constants for each type of soil.

E-3

F DRIVEN ROD RESISTANCES IN TWO-LAYER EARTHS

Driving ground rods into two-layer earths is a common occurrence, particularly in the EPRIsolutions environment where a sandy loam soil overlays a layer of clay with a water table under the latter. There are several published algorithms for calculation of rod resistance in a twolayer earth , most of considerable complexity. One simple and reasonably accurate set of equations for calculation of rod resistance in a two-layer earth is that published by Nahman and Salamon (Ref. 8). The variables used in these equations are defined as follows: R = rod low frequency resistance, ohms L = rod length, meters Le = equivalent length of the rod, meters D = rod diameter, meters h = depth of the upper layer of soil, meters

ρ1 = upper layer soil resistivity, ohm-meters ρ2 = lower layer soil resistivity, ohm-meters There are three possible rod positions: A.

Rod lies entirely in the upper soil layer and ρ1 > ρ2. R=

B.

Equation F-1

Rod lies entirely in the upper layer soil and ρ1 < ρ2:

R=

C.

ρ1 4L log 2πL D

ρ1 ρ 4 L ρ1 log + log 2 2πL D 2πh ρ1

Equation F-2

Rod lies in both soil layers

R=

4L ρ2 log e 2πLe D

Equation F-3

F-1

Driven Rod Resistances in Two-Layer Earths

where:

Le = L2 + L1

ρ2 ρ1

Equation F-4

Accuracies of these formulas where reported to be usually less than 10%, which - considering the uncertainties of soil resistivity measurement and rod contact - is about all to expect.

F-2

G ADJUSTMENT OF ROD-TANK RESISTANCES TO ROD RESISTANCES IN AN INFINITE PLANE

For a ground rod in the center of the test tank, the tank wall is only three-feet away, whereas the resistance of the earth from a driven rod in an open ground plane stretches to infinity. Although most of the earth resistance around a rod is created in the first meter of radial distance (caused by current concentration in this area), additional earth resistance extends beyond that region, and if the tank resistance data is to be equated to any of the applicable rod resistance equations a correction must be made. To determine this correction, a program called TANKOHMS.EXE was written. The program sums up the resistances of 50 cylindrical shells of soil between a center rod and the tank wall, An example of these shells is shown in Figure B-1. The bottom of each shell was adjusted in thickness so that as the final shell reached the tank wall the bottom shell just reached the bottom of the tank. Thus the bottom of each cylindrical shell had a different thickness depending on the height of the rod tip above the tank bottom. The radial resistance of each shell was the product of the specified soil resistivity and the shell thickness divided by the shell area. Similarly the resistance of the bottom of each shell was the product of the soil resistivity and the bottom thickness divided by the bottom area. These two resistances are essentially in parallel, and were combined to find the total shell resistance. Each shell resistance between the center rod and the tank wall was then summed to obtain the total rod-to-tank resistance for a specified soil resistivity. The algorithm makes the tacit assumption that the shell walls are equipotential surfaces. This is not strictly true, but by making this assumption Liew-Darveniza (Ref. 1) achieved results that matched the Dwight equation (Eq. 2-1) quite well, so the same assumption was used in TANKOHMS as a reasonably accurate approach.

G-1

Target: Overhead Transmission

About EPRI EPRI creates science and technology solutions for the global energy and energy services industry. U.S. electric utilities established the Electric Power Research Institute in 1973 as a nonprofit research consortium for the benefit of utility members, their customers, and society. Now known simply as EPRI, the company provides a wide range of innovative products and services to more than 1000 energyrelated organizations in 40 countries. EPRI’s multidisciplinary team of scientists and engineers draws on a worldwide network of technical and business expertise to help solve today’s toughest energy and environmental problems. EPRI. Electrify the World

© 2002 Electric Power Research Institute (EPRI), Inc. All rights reserved. Electric Power Research Institute and EPRI are registered service marks of the Electric Power Research Institute, Inc. EPRI. ELECTRIFY THE WORLD is a service mark of the Electric Power Research Institute, Inc. Printed on recycled paper in the United States of America 1001908

EPRI • 3412 Hillview Avenue, Palo Alto, California 94304 • PO Box 10412, Palo Alto, California 94303 • USA 800.313.3774 • 650.855.2121 • [email protected] • www.epri.com