AD-AO09 228

ELECTROMAGNETIC-PULSE HANDBOOK FOR

ELECTRIC POWER SYSTEMS
Edward F. Vance
Stanford Research Institute

Prepared for:
Defense Nuclear Agency

4 February 1975

DISTRIBUTED BY:

National Technical Information Service

U. S. DEPARTMENT OF COMMERCE
Destroy this report when it ts no longer
needed. Do not return to sender.

'L
.0
 UNClASSIFIED
SECURITy CCASSIFICATiON OF THIS PAGE (When t)ate Entorvd)

REPORT DOCUMENTATION PAGE READ INSTRUCTIONS

BEFORE COMPLFTING FORM
I. REPORT NUMBER 2. GOVT ACCESSION NO, 2. RECIPIENTrs CATALOG NUMBER

DNA 3466F 11 44
4. TITLE (and SiAItIi.) 5. TYPE OF REPORT 6 PERIOD COVERED
Final Report for Period
ELECTROMAGNETIC-PU
ELECRIC LSE
SYSEMS1[
OWEHANDBOOK FOR June 1973-15,June 1974
ELECTRIC POWER SYSTEMS 6. PERFORMING ORO. REPORT NUMIER
EMP Handbook, SRI Project 2709
7 AU1"SORNA)
O N. C PIACT OR OMANT NUMUEIR(.)

Edward F. Vance DNA 001-73-C-0238

S 'PRPORMINO
z OROANIZATION NAME AND ADD 'I1 to, O LM . TI
,PROJ EC'T TASK"

Stanford Research Institute ARIh oRK UNIT NUMIEMI

333 Ravenswood Avenue Subtask R99QAXEB075-36
Menlo Park, California 94025
IIt. CONTAOLLNO OFFICE NAME AND ADDRESS 12, REPORT DATE
Director 4 February 1975
Defense Nuclear Agency 13. NUMBER OF PAGES
Washington, D.C. 20305 _ ..___
14 MONITORING AGENCY NAME A ADDRESS(if ill #ient(Pam Controlling OffiCt) ". SECURITY CLASS (of this ,epo;Fp
UNCLASSIFIED

"i•SDV(C'i FICATiON/OO*WNORADIN d-

'S. DISTRIUIJTION STATEMENT (of Ithi Report)

Approved for public release; distribution unlimi,Lt2

7T, DISTRiBuTION STATEMENT (oall. shaIatrt enhe.ed In Block 20, It dilhervnO from RepOPI)

I0. SUPPLEMENTARY NOTES

This work sponsored by the Defense Nuclear Agency under

Subtask R99QAXEB075-36.

It. KPv WORnS (CnoinLte "M revorer axir It nrecera r), and Iderntify by block number)

Electromagnetic Pulse Shielding

Transmission Lines Filters
Transformers
Lightning Arresters
20. 'dSTRACT (Co.ntinue nn reverm. ideit n#ece".er), and Identilfy b bihirA number)

This handbook provides formulas and data for evaluating coupling of the
high-altitude EMP to electric power systems and to facilities served with
commercial electric power. The subjects covered Include coupling to power
transmission and distribution lines, transient coupling through transformers,
lightning-arrestor firing characteristics, and coupling through the service
entrance. Grounding, EMP protective measures, and testing are also discussed.

1473 odd byNCLASSIFIED

NATIONAL TECHNICAL
INFORMATION SERVICE SECURITY CLASSIWICATION OP THIS PAGE (37iy nI, Eneod)
US SWdngield,
Departmsent VA.Cof
ctp.,n.
22151
 UNCLASSIFI ED
SECURITY CLASSIFICATION Or THlIS PAGE(Won Valaa KFmrtpd)

20. ABSTRACT (Continued)

The emphasis of the handbook Is on the, EMP effects of concern to the power user,
but much of the information 'eontainedIn the handbook can also be umud Lu .'valuaLe
the effects of the EMP on electric power systems.

UNCLASSIFIEID
SECURITY CLASSIFICATION OF THIS PAC lrlrwlI V1IfR PJI',WN.rI

.{. . . . . .
 SUMMARY

This handbook has been prepared primarily for the power, communications, and
systems engineer who must be concerned with the offects of the nuclear electromagnetic
pulse on his system. The power engineer should be aware of the effects of EMP on his
transmission and distribution system, and the power users must protect their equipment
from the pulse conducted into their facilities on the power lines. The contents of this
handbook draw heavily on the results of research conducted at Stanford Research Institute
for the Air Force Weapons Laboratory under Contract F29601-69-C-0127 and on the
extensive work conducted by D. B. Nelson, J. K. Baird, and J. H, Marable at the Oak Ridge
National Laboratory.

This handbook was prepared under the auspices of the Defense Nuclear Agency
under Contract DNA001-73-C-0238. Project Officers for this task were Mal. Frank Vajda
and Maj, William Adams.

........ ... 1
 PREFACE

The author gratefully acknowledges the 'onstructive criticism of the draft by A. L.

Whitson, L. Schlessinger, J. E. Bridges, J. H. Marable, and R. V. Hugo, and by Maj. William
Adams, Their many suggestions have, without doubt, contributed to a more useful hand-
book. Special credit should also be given to Mai. Frank Vajda who initiated the program
for preparation of this handbook,

2
 TABLE OF Cý)I"`!NTS

Chapter One EFFECTS OF THE EMP ON POWER SYSTEMS

1.1 Introduction ........... ..................... 21

1.2 Characteristics of the Nuclear EMP ................ .... 22

1.2.1 Generation of the EMP .......... ........... 22
1.2.2 Characteristics of the High-Altitude EMP ...... 24
1.2.3 Comparison with Lightning ...... ............ 25
1.2.4 Technology Applicable to EMP Analysis ....... 28

1.3 Description of a Power System .... ............. .... 29

1.4 Contents and Use of the Handbook ...... ............ 34

1.4.1 Subjects Covered ....... ................ 34

1.4.2 Conventions Used In the Handbook ..... ......... 38

1.5 Cited References . .. . . . . . . . . . . . . . . . . . 40

Chapter Two COUPLING TO TRANSMISSION LINES

2.1 Transmission-Line Configurations ... .. .. . .. ... 42

2.1.1 General ................ . 42
2.1.2 Construction of Transmission Lines ........ 43
2.1.3 Transmisslon-LIne Cable . . . . . . .. .. . . . 45
2.1,4 Line Insulators ....... ................. 46
2.1.5 Interaction of the EMP with Transmission Lines . . . . 49

2.2 Coupling to Horizontal Conductors ................ 50

2.2.1 Semi-Infinite Line . . . . . . . . . . . . . . 50
2.2.1.1 General Approach . . . . . . . . . .. 50
2.2.1.2 General Analysis of Transmission Lines 5 1.
2.2.1.3 Open-Circuit Voltage in the Frequency
Domain . . . . . .. . . . . .. . . 56
2.2.1,4 Transient Waveforms for an Exponential
Pulse . . .. .. . . . . .. . . . . 62
2,2,1,5 Parametric Variation of the Open-Circuit
Voltage ........ ............... 62

3
 2.2.2 Line of Finite Length ...... .............. 66
2.2.2.1 Uniform Plane Wave ... .......... ... 66
2.2.2.2 Spherical Wave at Glazing Incidence .... 68
2.2.3 Rate of Rise of the Open-Circuit Voltage ..... 73
2.3 Response of a Vertical Element .... .............. ... 77
2.3.1 Current at Top of Vertical Element ..... ......... 77
2.3.2 Current at Base of Vertical Element ........ 81
2.3.3 Comparison of Voltage In Vertical and Horizontal
Elements ........ ... ..... ..... ... ... 83
2.3.4 General Analysis of Vertical Elements ............ 84
2.3.5 Impedance of a Ground Rod ..... ........... 86
B
2.3.6 Response of Horizontal Conductor with Vertical
Risers ............ .................. 88
2.3.7 Periodically Grounded Line ...... ............ 91
2.4 Transmission Properties of Power Lines ..... ...... ... 98
2.4.1 Attenuation and Phase Characteristics ... . 98
2.4.2 Junctions in Transmission Lines ........ .. . . 106
2.4.3 Bends in Power Lines ...... ......... ...
. 109
2.4.4 Other Effects ....... ............ ...
. 112

2.6 Differential Coupling to Horizontal Conductors ............ 116

2.5.1 Wires in a Horizontal Plane . .. .. .. . .. . . 116
2.5.2 Wires in a Vertical Planet ................ .... 121

2.5.3 Mode Conversion at the Load ..... ........... 125

2.6 High-Voltage Properties of Transmission-Line Components . . . 126

2.6.1 General ..... ................ ...... 126
2.6.2 Insulator Flashover Characteristics ..... ......... 127
2.6.3 LightnIng-Arrester Firing Characteristics ...... 129
2.6.4 Corona Threshold of Conductors .. .. . 131
2.7 Cited References ........ ... ..... ..... ... ..... 133

4
Chapter Three CONSUMER'S SERVICE ENTRANCE

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . 136

3.1.1 Description of the Service Entrance ..... ......... 136

3.1.2 Ground-Based Transformer Installation ....... 138
3.1.3 Pole-Mounted Transformer Installations ........... 140
3.1.4 Effect of Service Entrance on EMP Coupling ........ 143
3.2. Transmission Through Conduits and Cables ....... 145
3.2,1 Transfer Functions for Conduits asld Shielded
Cables . . . . . . . .... . . . . . . . . . 145
3.2.2 Theveiiin-Equivelent Source ...... ........... 153
3,123 Characteristic Impedance of Conductors in a
Conduit ......... ................... 153
3.2.4 Characteristic Impedance of Shielded Cables ..... 157
3,3 Coupling Through Steal Conduit Wells ..... . .. . 168
3.3.1 General Considerations . . ..... .. .......... 15S
3.3.2 Bulk Currents Induced in Buried Conduits ...... 161
3.3.2.1 General ............... 161
3.3.2.2 Current Far from the Ends of a Long
Conduit . . . .. .. .. . . . . . . . 161
3.3.2.3 Current Near the End of a Long Bare
Conduit . . . .... . , 165
3.3.2.4 Current In an Eleutrically Shoit
Conduit ..... .............. .... 170
3.3.3 Curront and Voltage induced on Internal
Conductors . . . .. . . .. . . . . . . . . 173
3.3.3.1 Transfer Impedance of Steel Conduit . . .. 173
3.3.3.2 Internal Voltage Induced in an
Electrically Short Conduit - Uniform
Exponential Curient .. ....... . 176
3.3.3.3 Internal Voltage from Incident Field ... 181
3.3.3.4 Saturation of Steel Conduit . t...... 182
3.4 Cables in Nonmetallic Conduit .,............ 184
3.4.1 General Considerations ........ ............. 184
3,4.2 Current Indut.ud on Cables . .. . .. .. .. . . 185
3,4.3 Transfer Impedance of Cable Shields ........ 194
3.5 Cited References . . . .. . . . . . . .. . . . . . . . 199

5
Chapter Four PROPERTIES OF DISTRIBUTION TRANSFORMERS

4.1 Introduction ..... ... ..................... .... 201

4.1.1 Effects of Transformers and Lightning Arresters . . .. 201
4.1.2 Transformer Construction ............ 203
4,1.3 Lightning-Arrester Construction .... ........ ... 209
4.2 Linear Characteristics of Transformers ........... 211
4.2.1 Analysis of Transformer Coupling Characteristics . . . 211
4.2.2 Distribution of Voltage Along the Windings ... ..... 217
4.2.3 CW Measurements of Transformer
Characteristics . . . . . . . . . . . . . . . . . 219
4.2,4 Linear Transient Test Results .......... 231
4.3 Nonlinear Characteristics of Transformers ............ .... 241
4.3.1 Transformer Responses ............. 241
4.3.2 Lightning-Arrester Firing Characteristics ..... 244
4.3.3 Summery of Transformer and Lightnlng-Arrulw
Properties . . . . . . . . . . . . . . . . . . 248
4.4 Cited References . . . . . . . . . . . .. . .. . . . 249

Chapter Five LOW-VOLTAGE CI RCU ITS

5.1. General Description of the Low-Voltage System ...... 251

5.1.1 Installation Practice ....... ............... 251
5.1.2 Distribution of the EMP-Induced Signal ......... ... 254

5.2 Linear Analysis of Conduit Circuits ................. ... 258

5.2.1 Transmission-Line Analysis (Common-Mode) .... 258
5.2.2 Determination of the Properties of Conduit
Circuits ....... ................... ...... 64
5.2.2.1 Characteristic Impedance ..... ........ 264
5.2.2.2 Propagation Factor and Length ...... 267
5.2.2.3 Stray Inductance of Leads ..... ........ 267
5.2.2.4 Load Impedances ....... ........... 288
5.2.3 Multiconductor-Trensmisslon- Line Analysis ..... .... 273
5.3 Cited References ........... .................... 275
Chapter Six GROUNDING SYSTEMS

6.1 General Description of Grounding Sy,.tems ..... .......... 276

6.1.1 Power-System Grounding ........... . 277
6.1.2 Grounding Low-Voltage Wiring .......... 279
6.1.3 Grounding Electronic Equipment ......... 280
6.2. Evolution of a Facility Grounding System 280
6.3 Principles of Grounding . . . . . . . . . .. . . .. . . 285
6.3.1 Single-Point vs. Distributed Ground ........... ... 285
6.3.2 Ground Impedance ...... ............... 287
6.3.3 Tree Ground . . . . .. . . . . . . . . . . . 288
6.3.4 Distributed Ground . . . . . . . . . . . . . . . 291
6.3.5 Star Ground . . . . . . . . . . . .. . . . . 291
6.3.6 Some Guidelines for System Design ........... ... 292
6.4 Grounding Counterpoise ......... ................ 294
6.5 Cited References ........ .................... 297

Chapter Seven POWER-SYSTEM PRACTICES FOR EMP PROTECTION

7.1 Introduction . . . . . . . . . . . . . . . . . . . . . 298

7.2 Protection from the Power System ....... ............. 300

7.2.1 EMP Protection of the Consumer ......... 300
7.2.2 Voltage Limiters and Filters .... ........... ... 300
7.2.3 Selection and Installation of Surge Arresters ........ 301
7.2.4 Selection and Install.. jn of Line Filters . ... 306
7.2.5 Auxiliary Power Systems ........ .... 310
7.2.5.1 Function of Auxiliary Power Systems. . .. 310
7.2.5.2 The Automatic Transfer Switch ...... 311

7.3 Protection of the Power System .... .............. ... 315

7.4 Cited References ..... ................... .... 317

7
Chapter Eight TESTS OF COMPONENTS AND FACILITIES

8.1 Introduction ........... ..................... 318

8.2 Standard Insulation Tests .......... ............... 320

8.3 EMP Tests of Equipment ..... ................ ... 321

8.3.1 Purpose of Equipment Tests .... .......... ... 321
8.3.2 Direct-Injection Tests .................. ... 324

8.4 EMP Tests of Facilities ....... ................. ... 327

8.4.1 Excitation of the Service Entrance ..... ......... 327
8.4.2 Observation of the System Response ..... ........ 329

8.5 Cited References ...... .................... ... 334

INDEX ................ .............................. 335

8
 ILLUSTRATIONS

1-1 Mechanism for Generation of the EMP ........ .............. 23

1-2 Two-Exponential Representation of the High-Altitute Map Waveform . 26

1-3 Magnitude of the Spectrum of the Two-Exponential Waveform ..... 26

1-4 Last Pole of Distribution Spur Showing Lightning Arresters, Potheads,

and Entrance Conduit ............ ................... 30

1-5 Ground-Based Distribution Transformers: Primary Side and

Distribution Potheads ...... ..................... ... 31

1-6 Ground-Based Distribution Transformers: Secondary Side and

Service-Entrance Weatherheads ..... ................. .... 33

1-7 Typical Rural Power Distribution System .... ............. ... 35

1-8 Block Diagram of a Typical Power Generation, Transmission, and

Distribution System.....................36

1-9 Breakdown of the Power System for Presentation In Handbook. 37

2-1 Wood Pole Construction for Distribution and Subtransmission

Lines ...... ........ ........................... 43

2-2 Wood Pole Construction for Transmission Lines at Voltages up to

161 kV ..... ... ..... ... .......................... 44

2-3 Grounding Methods for Wood-Pole Lines ........ ............. 44

2-4 Steel Tower Construction for Transmission Lines at Voltages of

69 kV and Above ........ ... .. ...................... 45

2-5 Transmission-Line Cable Construction .... .............. ... 47

2-6 Transmission-Line Insulators ......... .................. 48

2-7 Coordinates Defining Azimuth and Elevation Angles of Incidence . . .. 51

2-8 Equivalent Transmission-Line Circuit ....... ............... 52

9
2-9 Comparison of the Transmission-Line Approximation and the Exact
Solution for the Current in a Wire Over a Perfectly Conducting Ground
Plane - Step-Function Incident Field, Horizontally Polarized ... ..... 56

2-10 Power-Line Directivity Patterns for Vertical Polarization ....... 58

2-11 Power-Line Directivity Patterns for Horizontal Polarization .... ...... 59

2-12 Open-Circuit Terminal Voltage Induced in an Infinite Transmission

Line by a Vertically Polarized Incident Field of 1 V/in ... ...... ... 60

2-13 Open-Circuit Terminal Voltage Induced In a Semi-Infinite

Transmission Line by a Horizontally Polarized Incident Field of
1 V/m ....... ....... ........................... 61

2-14 Open-Circuit Voltage at the End of a Semi-Infinite Line for Various

Soil Conductivities . . . . . . . . . . . . . . . . . . . . . . 63

2-15 Open-CIrcuit Voltage at the End of a Semi-Infinite Line for Various

Line Heights . . . . . . . . . . . . . . . . . . . . . . . 64

2-16 Open-Circuit Voltage at the End of a Semi-Infinite Line for Various

Incident Pulse Decay Time Constants .... ............... .... 65

2-17 Transmission Line of Finite Length ................ 67

2-18 Open-Circuit Voltage Waveform at the Terminals of a 150-m-Long

Line Open-Circuited at Both Ends ....... ................ 69

2-19 Physical Arrangement that Produces a Vertically Pola,;.:qd Wave

with End-On Incidence ...... ..... .................... 69

2-20 Open-Circuit Voltage at the Terminals of a Line 1000 m Long and

1000 m from the Source with 1 V/rn Incident on the End Nearest
the Source (Zo - 300 r2, z1 = 1000 m, Z2 - 2000 m, exponential
pulse) ...... ... ........................... .... 73

2-21 Rate of Rise of the Open-Circuit Voltage at the Terminals of a

Semi-Infinite Line ........ ...................... .... 75

2-22 Minimum Length of Line Required to Obtain an Open-Circuit Voltage

of 100 kV with 50-kVm Field Incident at Small Angle .......... ... 76

10
 2-23 Vertical Element at the End of a Transmission Line ..... ......... 77

2-24 Short-Circuit Current at Top of Lossless Vortical Element Terminated

in Soil at the Base ....... ..... ...................... 80

2-25 Current Delivered to a Matched Lead at the Top of a Vertical Riser by

a Vertically Polarized Incident Wave ......... .............. 80

2-26 Short-Circuit Current Induced at the Base of a Vertical Riser by a

, Vertically Polarized Incident Wave . . . . . . . . . . . . . . . . 82

* 2-27 Open-Circuit Voltage Induced at the Base of Vertical Element by

a Unit Stop Incident Wave ............................ 84

2-28 Illustration of a Vertical Ground Rod of Circular Cross Section 87

2-29 Variation of Log (V/'2/-joa) with Frequency and Conductor

Radius ........... ... .......................... 87

2-30 Variation of dc Resistance of Vertical Ground Rod With Length 88

2-31 Transmission Line Terminated with Vertical Risers at Both Ends 89

2-32 Current Components Induced on Horizontc" Line and Vertical

Risers, and Total Current......... ...... 92

2-33 Steps In the Analysis of a Periodically Grounded Transmission

Line . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93

2-34 Open-Circuit Voltage at Terminals of Finite Horizontal

Transmission Line ....... ..... ...................... 98

2-35 Frequency-Domain Response of a Periodically Grounded Line ..... 99

2-36 Time Constant rh as a Function of Soil Conductivity and

Line Height '. ....... ... ..... ... ..... ..... ... ..... 101

2-37 Attenuation Constant a as a Function of Frequency for Various

Line Heights and Soil Conductivities .... ............... .... 102

2-38 Normalized Attenuation Constant a/k and Propagation Factor

j3/k as a Function of Frequency, Line Height, and Soil Conductivity . 103

11
2-39 Wave Front of Unit Step Current After Propagation for a Distance
of 10 km (6.2 miles) Along Copper Conductors of 1 mm and 1 cm
Radius ........ ........................... ... 105

2-40 Wave Front of Unit Step Current Aftur Propagation for a Distance
of 10 km (6.2 miles) Along a Copper Conductor of 1 mm Radius
at a Height of 10 m, for Various Soil Conductivities .... ......... 106

2-41 ,ymmetrlcal Junction In a Transmission Line ...... ............ 107

2-42 Two-Conductor Transmission Line Over Ground Plane (Common-

Mode Excitation) .................. 109

2-43 Reflection and Transmission Coefficient at a Symmetrical Junction,

Where h - 40 ft ........ ....................... ... 110

2-44 rwo-Wire Line with a Bend ......... *.....*. 111

2-45 Comparison of Calculated and Measured Reflection Coefficients of

a Bond as a Fiinctlon of Bmnd Angle ............... 112

2-46 Reflection and Transmission Coefficient of a Bend in a Single-Wire

Line Over Ground as a Function of Bend Angle ........... 113

2-47 Reflection Coefficient for a 900 Bend as a Function of Radius of

Curvature ........ ..... .... .................... 114

2-48 Variation in Characteristic Impedance Caused by Line Sag ........... 115

2-49 Geometry of Two-Wire Transmission Line for Differential-Coupling

Analysis ......... .......................... ... 117

2-50 Open-Circuit, Differential Voltage Induced in a Semi-Infinite

Two-Wire Line by an Exponential Pulse (Wires In a Horizontal
Plane) ........... ......................... ... 121

2-51 Two-Wire Line Lying In a Vertical Plane ...... .............. 122

12
2-52 Open-Circuit Differential-Voltage Waveform Induced in a Semi-
Infinite, Two-Wire Line by an Exponential Pulse (Wires in a
Vertical Plane) ....... ........................ ... 124

2-53 Asymmetrical Load on a Four-Wire Line with a Common-Mode

Voltage . . . . . . . . . . . . . . . . . . . . . . . . . . 125

2-54 Impulse Flashover Characteristics of Particular Types of Apparatus

Insulators on Positive and Negative 1%X 40 Waves at Standard
Air Conditions ........... ........................ 128

2-55 Impulse Flashover Characteristics of Suspension Insulators for

IY/X 40 Waves at 77*F, 30-Inch Barometric and 0.6085-inch
Vapor Pressure ....... ..... ....................... 129

2-56 Breakdown Voltage as a Function of Time-to-Fire for Lightning

Arresters ............. .......................... 130

2-57 9-kV-DistrIbution Lightning-Arrester Firing Characteristics for

Fast Rates of RIe . . . . . . . . . . . . . . . . . . 131
2-58 Corona-Threshold Voltage of a Wire Over Ground with Standard Air
Conditions . . . ...................... 133

3-1 Typical Power Service Entrances . . .............. 138

3-2 Potheads for Single- and Three.-Conductor Cables ............ .... 140

3-3 Shielded Cables for Servlcij Entrance ....... .............. 142

3-4 Pole-Mounted Service Entrance ....... ................. 142

3-5 Approximate Equivalent Circuit for Analysis of Transmission Through

Shielded Cable or Conduit Feeder Circuits ...... ............. 148

3-6 Feeder Transfer Functions for Resistive Loads ...... ........... 147

3-7 Feeder Transfer runctlons for Inductive Loads ............ .... 148

3-8 Feeder Transfer Functions for Capacitive Loads .......... 149

3-9 Voltage Across a Resistive Load Produced by a Unit Step Source . . . 150

3-10 Voltage Across aii Inductive Load Produced by a Unit Step Source . 151

13
3-11 Voltage Across a Capacitive Load Produced by a Unit Step Source.. 154

3-12 Feeder Source Translated to the Load Terminals ..... .......... 155

3-13 Characteristic Impedance of Conductors in a Metal Conduit .... ...... 156

3-14 Variation of Characteristic Impedance of Eccentric Coaxial Cable . . .. 157

3-16 Illustration of Coordinates for Conduit and Angle of Arrival

of Transient Wave ... .... .. .. .......... 162
3-16 Skin Depth 6 in Soil as a Function of Frequency and Soil
Conductivity oa ........ ... ........................ 163

3-17 Waveforms of Incident Field, Cable Current, and Open-Circuit

Voltage Between the Shields ......... .................. 165

3-18 Variation of Peak Conduit Current as Azimuth (ýo) and Elevation (0)
Angle of Incidence Change ...... ................... ... 166

3-19 Peak Conduit Current as a Function of Soil Conductivity and

Incident Exponential Pulse-Decay Time Constant r ...... ....... 167

3-20 Variation of Log (V/'26/,yoa) with Frequency and Conductor Radius . . 168

3-21 Waveform of the Conduit Current Near the End when Conduit End
is Open-Circuit . . . . . . . . . . . . . . . . . . . . . . 169

3-22 Magnitude of the Transfer Impedance of Rigid Steel Conduit ...... 175

3-23 Volta•e Waveforms produced by an Exponential Pulse

Lo e- 17 of Current in the Shield . . .
,... . . . . . . . . . 1/8

3-24 Variation of Peak Open-Circuit Voltage and Rise Time of

Conductor Voltage as a Function of Expinentlai Conduit-
Current-Decay Time Constint T ................ . 179

3-25 Core-to-Shield.Voltage Waveforms Induced in Conductors in a

Buried Conduit by an Incident Plane-Wave Exponential Pulse e-7 . , 182

3-26 Buried Insulated Conductor. ......... .................. 186

14
3-27 Characteristic Impedance Zo and Propagation Factor for Conductors
in a Buried Nonmetallic Conduit ....... ................. 190

3-28 A ..,a. -n Constant a for Conductors in - Buried Non-Metallic

Conduit ...... ..... ......................... ... 191
3-29 Short-Circuit Current and Open-Circuit Voltage Induced In
Conductors in a Buried, Nonmetallic Conduit ..... ............ 193

3-30 Illustration of Parameters for a Single-Leycr Tape-Wound Shield .... 198

3-31 Magnitude of the Transfer Impedance Computed for a Tape-Wound

Shield ....... ... .......................... .. 197

4-1 Typical Power-Distribution Transformers ...... ............. 204

4-2 Photographs of the Windings and Core of the 25-kVA GF.
Transformer .......... ....................... ... 205

4-3 Internal Construction of a Valve-Type Lightning Arrester ....... 210

4-4 LightnIng-Arrester Styles Used In Conjunction with the Transformer

HV Transient Tests . . . . . . . . . . . .. . . . . . . . 211

4-5 Transmission-Line Model for Coupled Transformer Windings .. , 213

4-6 Low-Frequency Equivalent Circuits for a Transformer ..... ........ 215

4-7 High-Frequency Equivalent Circuits for a Transformer .......... ... 217

4-8 Spatial Distribution of Winding Voltage at Early Times .... ........ 218

4-9 Voltage Distribution Along Primary Winding at Late Times .... ...... 219

4-10 Voltage Distribution Along Secondary Winding at Late Times ... ..... 219

4-11 Power-Transformer Configurations and Equivalent T-Networks

Used to Obtain CW Transfer-Function Datae ...... ............ 221

4-12 Magnitude of Transfer Function Relating Common-Mode Primary

Voltage to Common-Mode Secondary Voltage Across a 100-ohm
Load for the Test Transformers ...... ................ .. 223

15
4-13 Magnitude of Transfer Function Relating Common-Mode Primary
Voltage to Differential-Mode Secondary Voltage Across a 100-ohm
Load for the Test Transformers ..... ................. ... 225

4-14 Magnitude of Transfer Function Relating Differential-Mode Primary

Voltage to Common-Mode Secondary Voltage Across a 100-ohm
Load for the Test Transformers ...... .............. ...
. 227

4-15 Magnitude of Transfer Function Relating Differential-Mode Primary

Voltage to Differential-Mode Secondary Voltage Across a 100-ohm
Load for the Test Transformers ......... .............. . 229

4-16 Schematic Diagram of Transformer Test Circuits ......... . 232

4-17 Waveform of Excitation Voltage Applied to Transformer ...... . 233

4-18 Waveform of Pulser Capj.itor Discharge Current (which includes the

current through the 100-ohm pulser load) ............ ...... 234

4-19 Secondary Output-to-Ground Voltage Observed at XI for Common-

Mode Excitation and Loading for Four of the Transformers ..... .... 235

4-20 Secondary Output-to-Ground Voltage Observed at X3 for Common-

Mode Loading . . . . . . . . . . . . . . . . . . . . . . .. 238

4-21 Differential Voltage Appearing Between Secondary Terminals

X1 and X3 for Common-Mode Loading .... . ....... .. 239

4-22 Secondary Output-to-Ground Voltage Observed at X1 for

Differential-Mode Loading .... ................. ..... 240

4-23 Secondary Output-to-Ground Voltage Observed at X1 for

Common-Mode Excitation and Loading ........ ........... 243

4-24 Peak Secondary Voltage as a Function of Primary Excitation

for Various Transformer Configurations .... ......... ...... 245

4-25 Lightning-Arrester Firing Characteristics for Fast Rates of fib ..... 247

16

~'
5-1 Single- Line Diagram of Internal Low-Voltage Circuits ..... ........ 252

5-2 Surge-Protection Techniques for Low-Voltage Circuits .... ........ 254

5-3 Typical Low-Voltage Wiring in Rigid Steel Conduit ..... ......... 255

5-4 Low-Voltage Wiring Inside a Main CIrcult-Breaker Panel .... ....... 256

5-5 Tronsmission-Line Model of Low-Voltage Wiring in

Metal Conduit ....... ....................... .... 259

5-6 Steps in Translating Load Impedances Toward Driving

Source .......... ... ........................... 261

5-7 Steps in Translating Driving Current and Voltage Toward

Loads ......... ........................... ... 265

5-8 Techniques for Determining the Characteristic Impedance

Experimentally . . . . . . . . . .. . /. . . . . . . . . . 266

5-9 Inductance per Unit Length of a Wire Over a Ground Plane ...... 268

5-10 Characteristic Impedance, and Capacitance and Inductance

per Inch . . . . . . . . . . . . . . . * ...... . . . . . .. 270

5-11 Line-to-Case Impedance of 7V/-hp Induction-Motor Winding ..... 271

5-12 Line-to-Case Impedance of Va-hp Induction-Motor Winding .... ...... 272

5-13 Line-to-Case Impedance of 1/4-hpSingle-Phase Capacitor-Start

Motor Winding .......... .................... .... 273

6-1 Typical Grounding Arrangement for a Transmission System 278

6-2 Three-Phase Distribution Transformer Connections ..... ......... 278

6-3 Grounding System for a Telephone Switching Center .......... .... 281

6-4 Examples of Grounds for a Simple Radio Receiver .......... 282

6-5 Simple Receiver Separated into Modules Interconnected by Shielded

Cables ......... ........................... ... 283

17

............
6-6 Single-Point Tree Ground System ..... ................ ... 284

6-7, Loop Formed by Cable Shields and Metal Floor or Walls ....... 285

6-8 Single-Point Ground and One Trunk ..... ............... ... 286

6-9 Impedance of Metal Ground Planes and of a Ground Cable ....... 289

6-10 Tree with Trunk and Two Branches ..... ............... ... 290

6-11 Junction Grounded to Reduce Coupling Through Trunk 290

6-12 Trunk Eliminated to Form a Star Ground ..... ............. 292

6-13 Behavior of a king Counterpoise at Low and High Frequencies ..... 296

7-1 Use of Secondary Lightning Arresters and Filters to Suppress EMP-

Induced Transients on Low-Voltage Circuits ..... ............ 302

7-2 Illustration of Secondary Lightning Arresters at the Main Circuit-Breaker

Cabinet .... ................. . . . ..... . 304

7-3 Impulse Test for Evaluating Power-Line Filtors for EMP Applications , . 309

7-4 Determination of the Insertion Loss of a Line Filter ........ . 310

7-5 Illustration of Power-Line Filters in an Entry Vault on Wall of Shielded

Structure . . . . . . . . . . . . . . . . . . . . . . . .. . 312

7-6 Schematic of a Simplified "Uninterruptible Power System" (UPS) .... 313

7-7 Circuit Diagram of an Automatic Transfer Switch .......... 314

8-1 Impulse-Test Waveform Showing Method of Specifying Rise Time and

Time-to-Half-Peak . . . . . . . . . . . . .. . . . . .. . . 322

8-2 Conductors Entering or Leaving an Equipment Cabinet ........ 323

8-3 Impedance Matrix Used to Inject Common-Mode Voltages 326

8-4 Capacitors Used to Inject Common-Mode Voltages ........ 327

8-5 Current Transformer to Inject Current on a Grounded Cable Shield . 328

18
8-6 Schematic Diagram of Power-Line Pulser and Coupler ..... ....... 330

8-7 Power-Line Coupler Unit ........ ................... ... 331

8-8 Probe Adapters for a Tektronix P-6047 Passive Voltage Probe ..... 332

8-9 Definition of the Excitation for an Internal Equipment Component , . 333

I1

19
 TABLES

2-1 Some Typical Properties of Above-Ground Power Lines . s...... 46

2-2 Tentative AIEE Standard on Insulation Tests for Outdoor

Air Switches, Insulator Units, and Bus Supports ........... 126

3-1 Suggested Withstand Impulse Voltages for Cables with Metallic

Covering ....... ..... .......................... 141

3-2 Common-Mode Characteristic Impedance of Four 300-MCM

Insulated Conductors in a Three-Inch-Diameter Conduit ....... 155
3-3 Characteristic Impedances of Paper-Insulated Shielded Power
Cables .......... ... .......................... 159

3-4 Properties of Standard Galvanized or Enameled Rigid Steel Conduit 174

3-5 Shielding Parameters for Rigid Steel Conduit . . . ...... .......... 180

4-1 Standard Insulation Classes and Dielectric Tests for Distribution

and Power Transformers ............ .................... 208

4-2 Properties of Class OA Transformers Used in CW and Transient

Measurements ............ ........................ 220

4-3 Comparison of Peak Voltage at Secondary Terminal X1 for a

One-Volt, 7.5-ps Exponential Pulse Applied to the Primary . . . ..... 237

5-1 Maximum Number of Conductors In Trade Sizes of Conduit

or Tubing ........ ..... ......................... 257

20
Chapter One

EFFECTS OF THE EMP

ON POWER SYSTEMS

1.1 INTRODUCTION

Until a few years ago, the nuclear EMP community gave little attention to commer-
cial power systems other than to recommend surge arrestors and line filters for the power
lines, It is now recognized, however, that the commercial power system can be a major path
for coupling the EMP into ground-based systems, The power distribution system forms a
very large, completely exposed antenna system that is hard-wired into the consumer's fa-
cility, Thus extremely high voltages may be developed on the power conductors, and even
if the commercial power is not relied on for system survival these voltages may be delivered
to the system either before commercial power is lost, or by the ground or neutral system
after transferring to auxiliary power,

In the last few years, considerable researuh has been performed on EMP coupling to
commercial power systems In an effort to characterize the power distribution lines as EMP
collectors and to determine the effects of major components, such as transformers, lightning

21
arresters, and low-voltage wiring, on the penetration of the received signal into ground-based
facilities. This research has entailed development and experimental verification of the theory
of coupling to transmission lines, and even though considerable research is still continuing
on the EMP interaction with power systems, it is felt that the preparation of a handbook on
the interaction of EMP with commercial power systems is in order so that designers and
systems engineers can benefit from the results of extensive data already accumulated.

1.2 CHARACTERISTICS OF THE NUCLEAR EMP

1.2.1 GENERATION OF THE EMP

The nuclear electromagnetic pulse (EMP) is produced by the products of a nuclear

detonation interacting with air molecules to produce a sudden separation of electrical
charge, This process Is illustrated In Figure 1-1(a) for a surface detonation and In Figure
1-1(b) for a high-altitude detonation, Very-high-energy -y-rays (photons) are produced by
the nuclear detonation and propagated radially away from the burst point. These -Y-rays
eventually collide with air molecules to produce relativistic Compton electrons which also
move in the radial direction, The heavier positive Ions are left behind, so that a separation
of the negative electrons from the positive ions occurs. This Is the charge separation mech-
anism by which the electromagnetic pulse is Initiated. As the Compton electrons travel
through the air, however, they collide with neutral air molecules to produce secondary elec-
trons, and lose their energy in this process, The air soon becomes fairly highly ionized and
the separated charge begins to discharge as a result of current flowing through the ionized
air. At large distances from a surface burst, this charge separation and relaxation has the
appearance of a transient elec•tric dipole.

The charge-separation mechanisms are similar for the high-altitude detonation, but
because the probability of the y-ray (photon) colliding with an air molecule is relatively low
until It has reached the more dense atmosphere,, iost of the Compton electron production
and charge separation occurs In or below the region of the upper atmosphere containing
the ionosphere D-dayer. In this region, the distance that a Compton electron travels

22
 COMPTON
ELECTRONS

SECONDARY
ELECTRONS %2

POSITIVE

",.GAMMA-RAYS

BURST, POINT

(al SURFACE BURST

BURST POINT

GAMMA-R3AYS

SOURCE REGION
1 p,RTWS
SUHFýACE

(b) HIGH-ALTITUDE BURST

Figure 1-1 MECHANISM FOR GENERATION OF THE. EMP

before it loses its energy through collisions is great enough that the curvature of 'cs path pro-
duced by the earth's magnetic field and the resultant magnetic dipole moment, are also sig-
nificant, as is indicated in Figure 1 -(b), As is also illustrated in Figure 1-1 (b) the source
region where the charge separation occurs can cover a very large area, so that the EMP pro-
duced by a single high-altitude detonation may be quite strong over a region with dimen-
sions of several hundred miles.

23
 1.2.2 CHARACTERISTICS OF THE HIGH-ALTITUDE EMP

The detailed characteristics of the high-altitude EMP waveform vary with weapon
chz-,acteristics and the positions of the observer relative to the burst point, but for the pur-
poses of most EMP coupling analysis a simple, approximate wavefolm is adequate. A useful
approximation for the incident electric field strength to be used in coupling calculations is
the two-exponential pulse given by

Eft) =Eo (et/TJ e-et/•2)(-)

where rl is the pulse-decay time constant and 72 Isthe rise time constant, Since "1>> ,2,
E0 is approximately the peak value of the Incident electric field strength, The values given by
Marable et al.l * for E0 , rl1 , and r2 are

Eo* 5.2 X 104 V/m

T1 6.7 X 10-7 s

"2 3.8 Y 10-9 s

A sketch of the waveform plotted from Eq. (1-1) using these values is shown in Figure 1-2.

The Fourier transform of the two-exponential pulse Is

T 1 -r•2
E( -) EoSE(11 +oJw7) ((1 + jJwr2) (1-2)

'References are listed in the last section of each chapter.

24

' . i ,t'. !ia;•i

- I .. I I ' • i i.. .. N ov•ii..
A plot of the magnitude of the field E(w) for the values of E0 , r1 , and r 2 given above is
shown in Figure 1-3.

From this approximate waveform and its Fourier transform, It is apparent that the
EMP produced by a high-altitude detonation is characterized by a very fast rise to a peak
electric field strength of the order of 50 kV/m, and that the duration of the pulse is of the
order of lps. The spectrum of the pulse contains significant energy at frequencies approach-
Ing 100 MHz.

The magnetic field strength (outside the region of ionization) associated with the
EMP Isrelated to the electric field through the intriiisic impedance of air (120,7), so that the
incident magnetic field Is

1207r

1.2.3 COMPARISON WITH LIGHTNING

Because both the nuclear EMP and lightning produce large electromagnetic tran-
sients it Is natural to compare their properties and characteristics. As is well known, light-
ning Is produced by the discharge of static electricity accumulated in clouds, The discharge
occurs as a long arc between the cloud and the earth, in which peak currents of tens of kA
(but occasionally hundreds) and charge transfers of about 1 coulomb occur. 2 , 3 Typical
current rise times are of the order of 1 ps, and the duration of individual current pulses
(time to decay to half the peak value) is of the order of 40 Ms. Each lightning flash normally
consists of several strokes or pulses of current. The lightning stroke-current rise-time and
duration are thus much greater than those of the EMP fields.

The effects of lightning are most severe when a direct strika is incurred. These effects
are often manifested as molten metal, charred Insulation, and exploded timber-the result
of a high energy density delivered by the stroke to Its point of attachment, Lightning

25
 60

40 . f - - - :- --

LL 20-

LU

0
0 on 1.0
TIMr:- •Ut

Figure 1-2 TWO-EXPONENTIAL REPRESENTATION OF THE HIGH-ALTITUDE

MAP WAVEFORM

. .
10-2

t0*

i
• 10,3

U.1 1

10 "4 ..

0o01 0.1 1.0 10 100

FREQUENCY-. MHz

Flgurv 1-3 MAGNITUDE OF THE SPECTRUM OF THE TWO-EXPONENTIAL

WAVEFORM

26
transients propagated along transmission lines often produce peak voltages of several hun-
dred kilovolts with pulse rise-times of about 10 ps and durations of 50 ps or more. This
transient is not necessarily a common-mode (zero.sequence) voltage, since the stroke may
attach to only one conductor; the voltages induced in the remaining conductors are pro-
duced by mutual coupling.4

Because the nuclear EMP arrives at the surface as a plane, propagating wave, rather
than as a stroke channel, the melting, charring, and splintering effects associated with direct
lightning strikes do not usually occur with the EMP. The EMP exerts Its influence through
induced effects; the very large electromagnetic fields of the EMP induce large voltages or
currents in antenna-like elements of equipment. For example, a 5-meter-high vertical con-
ductor (monopole antenna) exposed to a 50-kV/m Incident electric field will have an open-
circuit voltage of 500 kV induced at Its base. The reader should adjust his thinking to
accomodate the fact that the very fast rise time of the EMP implies that conductors over a
few feet long are no longer electrically short. Thup, although a 20-ft-long ground wire on a
transformer pole may be treated as a lumped Inductance for lightning transients, 20 ft Is
twice the distance a wave will propagate during the buildup time of the EMP. Conductors
over a few feet long must therefore be analyzed as transmission lines, rather than lumped
elements, In the investigation of EMP effects, In addition, small inductances and capaci-
tances that are negligible in lightning analysis become important in EMP analysis because of
the large rate; ol change associated with the leading edge of
the EMP.

The primary effect of the EMP is, therefore, the production of large voltages or cur-
rents In conductors such as power lines, buried cables, antennas, etc. These Induced currents
and voltages may then cause secondary effects such as Insulation flashover and electronic
component damage or malfunction. Electronic logic circuits, In which infornation is trans-
ferred as a train of pulses, are particularly susceptible to transients of the type Induced by
the EMP. Even small transients In these circuits can cause a false count or status indication
that will lead to an error In the logic output, and large transients can destroy the junctions
of the solid-state devices used in these circuits. Furthermore, the techniques used for pro-
tecting equipment from the slowly rising lightning transients are not necessarily effective
against the fast-rising EMP-induced transients.

27
 1.2.4 TECHNOLOGY APPLICABLE TO EMP ANALYSIS

Because the EMP is a transient whose spectrum Issignificant at frequencies approach-

ing 100 MHz, the analysis of EMP effects is a transient analysis. Hence, although some dif-
ferences between the EMP and lightning have been noted above, the results of extensive
transient analysis related to lightning are beneficial to the analysis of EMP effects on power
systems and components. One of the problems encountered In the analysis of the coupling
and propagation of EMP-Induced transients is that of determining the electrical properties
of components such as transformers, motors, etc, at frequencies other than their normal
operating frequencies (60 Hz). For many transmission-line components, however, analyti-
cal techniques arid component characteristics have been developed for lightning and switch-
ing transient analysis that are applicable for the frequency spectrum below 1 MHz. 4 "7
Engineers familiar with conventional transient analysis of power systems should therefore
be well equipped to perform EMP analysis,

Some data are also available for system and component characteristics In the 50-to-
150-kHz range used for power-line carrier applications. 8 Although much of the supervisory
control, telemetering, and relaying Is now carried on microwave links, power-line carriers
are still in use and much data on the transmission properties of lines and components are
available to supplement the data available from lightning and switching transient analysis.
Because of the fairly narrow band of frequencies used for power-line carrier operations,
however, these data are of somewhat limited value,

Because the usual treatmont of lightning and switching transients is not concerned
with frequencies above about 1 MHz, high-frequency techniques must be Invoked for ana-
lyzing the power system's response to the EMP. Most useful in the EMP analysis at frequen-
cies above 1 MHz Isa good understanding of the fundamentals of electromagnetic waves and
transmission lines,"' 11 As was Indicated In Section 1.2.3 above, many conductors that are
electrically small at 1 MHz (300-meter wavelength) are large at 100 MHz (3-meter wave-
length), Thus the EMP analyst must be acutely aware of wave propagation times and the
transmitted and reflected waves at discontinuities In conductor configurations. To illustrate
this point, consider a lightning transient with a 1.5-Ms time-to-peak. The toe of this tran-
sient has propagated 1500 ft beyond an observer by the time the peak arrives. For a 1-Is-
wide pulse with a 10-ns time-to-peak, however, the toe of the pulse has propagated only 10
ft by the time the peak arrives, and by the time the toe has propagated 1500 ft beyond the

28
observer the trailing edge of the pulse has passed and propagated 500 ft beyond the observer
(the velocity of propagation is approximately 1 ft/ns). One consequence of this difference
is that the current at the top of a vertical ground wire on a pole may be radically different
from the current only 30 ft away at the base of the pole.

1.3 DESCRIPTION OF A POWER SYSTEM

The power system Is described from the viewpoint of a consumer who Is concerned
about the EMP-induced transients entering his facility on the power conductors, As viewed
by the consumer, the most important parts of the power system are those that are closest
to his facility. Thus the service entrance, the distribution transformer, and the last mile of
distribution line are quite important, but parts of the distribution and transmission system
over a mile away decrease In interest as the distance increases.

Typical service at the end of a distribution line Is shown In Figure 1-4 for service
with ground-based transformers, The 3-phase aerial distribution line ends on the guyed
pole, where the line Is spliced to shielded cable, which enters a conduit running down the
pole and underground to the transformer, The top of the pole is shown in Figure 1-4 to
illustrate the installation of the lightning arresters, disconnect switches, and potheads. Rigid
steel conduit is normally used to protect the conductors on the drop down the pole, but
often fiber duct Is used for a segment of the buried horizontal run between the pole and
the transformers. There is almost always a ground wire for the lightning arresters running
down the pole to a ground rod ur butt wrap at the base of the pole.

The other end of the conduit and the distribution transformers are shown in Figure
1-5. In the case shown, the transformers are mounted outdoors on a concrete pad and
protected by a chain-link fence. The shielded cables exit the conduit through a moisture
barrier and are terminated in potheads, Connecting leads between the pothead terminals
and the transformer terminals can be seen In the figure, Three single-phase transformers
are used to reduce the distribution voltage (13,2 kV) to the consumer's voltage (440 V).
The neutral conductors and the transformer cases are grounded to ground rods just off
the concrete pad.

29

S.
.. . . . . . .. . .. . . . . . . . . .. . .. . . . . . . . . .= • , . . . . . . . .. . . . . . . . . . . . . .. . . . .. . . . . .. . . . . . . .. . .
 (;AIII

Figure 1-4 LAST POLE OF DISTRIBUTION SPUR SHOWING LIGHTNING ARIRESTERS,

POTHEADS, AND ENTrHANCE CONDUIT

30
Figre
ROU -5D-BsED DISRIB TIO TR NSFRMES'PRIMARY SD

.31
 A view of the secondary side of the transformers and the low-voltage conductors
entering the weatherheads is shown in Figure 1-6. In the installation shown in this figure,
several conductors are used for each phase, so that three entrance conduits are required,
These conduits also may be either steel, part steel and part fiber duct, or fiber duct all the
way. The conduits are normally run underground into the building where they terminate in
the main circuit-breaker panel. The length of the conduit between the top of the pole and
the transformers ranges from about 50 ft to about 1000 ft, but a typical distance is about
100 ft. The length of the conduit between the transformers and the main circuit-breaker
panel Istypically 20 or 30 ft, but it can vary from about 10 ft up to about 200 ft.

The outdoor transformer arrangement shown in Figures 1-6 and 1-6 is quite com-
mon, but other common practices include installing the transformers in steel cabinets or

sheds outside the building or in a vault inside the building. Pole-mounted transformers are
common for light loads and at sites where real estate for ground-based Installations Is limited.

Some of the features of the service that are important in the analysis of EMP coup-
ling to the building electrical circuits are:

(1) The height and separation of the conductors of the distribution lines. These
affect the coupling of the EMP to the line.

(2) The lightning arresters, These protect the potheads and shielded cables from
Insulation breakdown,

(3) The characteristic Impedance of the shielded cables. A large mismatch between
the aerial line and the shielded cables limits the voltage delivered to the cables,

(4) The conduit between the pole and the transformers. If the conduit Is all steel,
the cables will be protected from the fields in the ground, but if it Is partly
plastic or fiber duct, additional coupling may occur along this path.

(5) The transformers and connecting leads. These behave as bandpass filters that
limit the very high frequencies and the very low frequencies. If the secondary
leads are exposed for significant lengths, however, some of the high-froquency
spectrum may be restored by direct coupling to the EMP.

32
 RING

CO•NDUCTORS

Figure 14 GROUND-BASED DISTRIBUTION TRANSFORMERS: SECONDARY SIDE

AND SERVICE-ENTRANCE WEATHERHEADS

(6) The conduit between the transformers and the main circuit-breaker parel, If
the conduit Is steel, the conductors are well shielded, but If partly plastic or
fiber, significant coupling to the conductors can occur.

(7) The lead lengths between the aerial lines and the shielded cables and between
the transformer terminals and the conduits. The inductance of these leads
limits the rise time of the transient propagating toward the building.

The distribution system supplying a facility is likely to be unique to each facility. A

typical rural distribution system Is shown in Figure 1-7. The distribution network shown is
that provided by one rural electric cooperutive serving customers within a 10-mlle radius of
Its distribution substation. Note that there is a mixture of single-phase service and 3-phase
service, and the distribution system is interconnected with neighboring systems, Of Interest
is the fact that most customers are at the ends of segments of transmission line that are at
least 1000 ft long, and many are served by spur lines over a mile long. In the network shown

33
in Figure 1-7, only one 3-phase customer is close enough to the substation to be even
remotely concerned with EMP coupled to the subtransmission line. Also noteworthy is the
fact that even in this rather sparsely populated service area, the distance between branches or
spur lines is typically only about one mile. This fact is important in assessing the buildup of
inducvd current propagating into a facility from great distances, since spurs, branches, or
bends tend to limit the buildup and propagation of this current. In more densely populated
urban and suburban areas, of course, the distance between spurs and branches Is much less
than that shown in Figure 1-7.

Although the consumer is seldom concerned with the generation and transmission
system, this part of the power system is exposed to the EMP and may be susceptible to the
EMP-induced transients. The generation and trPnsmisslon system Is 6hown schematically in
Figure 1-8 to Illustrate the hierarchy of the system. In practice, however, there are many
interties at the transmission, subtransmission, and distribution levels, so that the transmission
and subtransmission systems form very complicated networks. These systems also contain
feedback in the form of supervisory control end relaying, so that a disturbance, (e.g, load
shedding, generator shutdown) in one part of the system causes changes In the remainder
of the system,

1.4 CONTENTS AND USE OF THE HANDBOOK

1.4.1 SUBJECTS COVERED

The goal of this handbook is to provide the formulas and numerical examples neces-
sary to evaluate the principal coupling problems a power-consumer might encounter, and to
provide some general information on grounding theory, protection techniques, and test
methods. Becau%i it has beei prepared as a handbook, rather than a treatise, the complete
derivation of coupling formulas has not been provided, although some of the more impor-
tant formulas have been partially derived, and an attempt has been made to provide refer--
ences to sources where the subjects are treated in more detail.

34
 0 1 2 3 4 5 10
SCALE mileslIV

d'1

• 9!
o•3-PHASE CUSTOMERSI

-o • •SINGLE-PHASE TRANSFORMER
/ • 9•SERVING ONE OR MORE
I• e ,• .. =mOCUSTOMERS
wm-7,2/12,47 kV, 3-PHASE, 4-WIRE
DII'RIBUTION LINE
L-.- • • • 7.2-kV DISTr IBUTION LINE
CONSISTING OF TWO PHASE

• ,CONSISTINO OF ONE PHASE

"m
rmTRANSMIS8ION L.INE

G' POINT WHERE THE DISTRIBUTION SYSTEM IS INTERCONNECTED WITH A NEIGHBORING DISTRIBUTION
SYSTEM. THE BREAK CONSISTS OF MANUALLY OPERATED ISOLATING SWITCHES NORMALLY IN THE
OPEN POSITION.

Figure 1-0 TYPICAL RURAL POWER DISTRIBUTION SYSTEM

35
 r'. GENERATOR

TRANSMISSION LINES

LINES TO AND FROM

OTHER GENERATION TRANSMISSION
FACILITIES AND SNBSTATION
TRANSMISSION STATION
SUBSTATIONS

!SUBTRANESMIS LIES

DISTRIBUTION
I SUBSTATIONS

DISIRIBUTION
LINES

CUSTOMERS'
SERVICES

Figure 1-8 BLOCK DIAGRAM OF A TYPICAL POWER

GENERATION, TRANSMISSION, AND DISTRIBUTION
SYSTEM

Coupling of the EMP tu tranmtisslon lines and through the distribution transformer
and entrance con(;uits is covered In Chapters Two, Three, and Four. The partition of the
power system among these chapters is illustrated in Figure 1-9. Chapter Two contains
formulas and results for coupling to aerip.l transmission lines and includes the effects of
soil parameters and of polarization and angle of incidence of the EMP, as well as the
effects of vertical 0lements such as ground wires and service-entrance conduits,

36
 INTERNAL
WIRING
DISTRIBUTION
LINE

TRANSFORMER

SERV/ICE " SERVICE

.EN"TRANC. E ENTRANCE

[-- CHAPTER 2-------.. --- CHAPTER 34CHAPTER 3 CHAPTER '-,

CHAPTER 4

Figure 1-9 BREAKDOWN OF THE POWER SYSTEM FOR PRESENTATION IN HANDBOOK

Chapter Three treats the transmission of the transient through the entrance conduits,
Included in this chapter are transmission-line formulas for passive metal conduit systems and
for fiber conduits along which additional coupling occurs. Also discussed In this chapter are
the shielding properties of rigid steel conduit and tape-wound shields. Examples of the
waveforms delivered to the load-end of transmission lines simulating service-entrance con-
duits are given for resistive, capacitive, and inductive loads.

The linear and nonlinear characteristics of transformers and lightning arresters are
given In Chapter Four. Although the data available on transformers and lightning arresters
are still somewhat limited, the trends are evident and the bandpass behavior of the trans-
formers and data on firing voltage versus time to fire available are presented In this chapter.

Chapter Five discusse6 some aspects of coupling through internal building wiring.
Because the electrical wiring in a typical building has extremely complex high-frequency
characteristics, detailed analytical procedures for handling this part of the coupling are very
cumbersome and generally not very accurate. Nevertheless, some general guidelines and aids
to estimating major coupling parameters are presented.

Chapters Six, Seven, and Eight treat grounding, protection, and testing, respectively.
Grounding theory arid practice are reviewed in Chapter Six in a more or less qualitative man-
ner, primarily for the user who is not familiar with the subject. Effective and economical
protection methods using arresters and filters are described In Chapter Seven, Techniques

37
for evaluating component susceptibility to EMP-induced transients and for proof-testing
facilities are described in Chapter Eight.

1.4.2 CONVENTIONS USED IN THE HANDBOOK

Throughout this handbook coupling formulas in the frequency domain and the time
domain are used, Because of its wide use by electrical engineers, the Fourier transform has
been used for frequency-domain represuntations. The Fourier transform pair, as used
throughout this handbook, can be written as follows:

F() / f(t)Me"•Itdt (1-4)

0

fit)
M f F(w)eJihtdw (1-5)
2-7

For those more accustomed to the Laplace transform, the transition Is very simple since
the Fourier variable jw and the Laplace variable s are interchangeable for all applications
encountered In this handbook. This is apparent from a comparison of Eqs. (1-4) and (1-5)
with the Laplace transform pair:

F(s) = f f(t)eW'dt (1-6)

+00

f = f F(s)eltds (11-7)

for the limiting case where b -* 0, (However, this simple change of variable cannot be used
for waveforms that are not zero tor negative time.)

38
 The time dependence of alternating fields, voltages, and currents used in the hand-
book is eiw', consistent with common electrical engineering usage. This time dependence
is consistently suppressed (by the Fourier transform), however, so that the expression fM•r a
propagating field that represents

E(t) = Eo ej(lt - kz)

Is written

E(W) - Eo eijkz

The Important effect of the elJ1t time-dependence is that it makes inductive Impedances pos-
itive imaginary quantities (j•JL), and capacitive Impedances negative imaginary quantities (-.J/C),

The rationalized inks system of units is used exclusively in the analysis and formulas
presented, although dimensions and distances are sometimes discussed In the more familiar
English or engineering terms (e.g., feet, miles, mils), In the rationalized inks system, the
following dimensions and constants apply:

Quantlt Unit Abbreviation

potential volt V
current ampere A
impedance ohm 0
electric field strength volt/meter V/m
magnetic field strength ampere/meter A/m
Inductance henry H
capacitance farad F
permeability
(o- 41r X 10-7 H/m) henry/meter H/m
permittivity
= 8.85 X 10-12 F/m)
-eo farad/meter F/rn
conductivity mho/m

39
Conventional abbreviations for factors of 10•" are used:

Prefix Symbol Meaning

p X 10"12
pico
nano n X 10-9
micro x 10-6
X
m X 10-3
mlli
k X 103
kilo
mega M X 106

One exception io this convention occurs In the discussion of cable sizes, where MCM is
used to abbi iviate "thousand circular mils,"

1.5 CITED REFERENCES

1. J, H,Marable, J. K. Baird, and D. B. Nelson, "Effects of Electromagnetic

Pulse (EMP) on a Power System," ORNL-4836, Oak Ridge National Labora-
tory, Oak Ridge, Tennessee (December 1972).
2. N. Clanou and E.T, Pierce, "A Ground-Lightning Environment for Engineering
Usage," Technical Report 1, SRI Project 1834, Contract LS.-2817-A3,
Stanford Research Institute, Menlo Park, California (August 1972).
3. M.A, Uman, Lightning (McGraw-Hill Book Co., Inc., New York 1969).
4. L. V.Bewley, Traveling Waves on Transmission Systems (Dover Publications,
Inc., New York, NY., 1963),
5. R, Rudenberg, ElectricalShock Waves in Power Systems (Harvard University
Press, Cambridge, Mass., 1968),
6. A. Greenwood, Flectrical Transients in Power Systems (John Wiley & Sons,
Inc, New York, N.Y., 1971).
7, E. 0, Sunde, Earth Conduction Effects in TransmissionSystems (Dover Pub-
lications, Inc., New York, N.Y., 1968).
8, Electrical Transmission and DistributionBook, 4th Ed., 5th printing (Westing-
house Electric Corporation, East Pittsburgh, Penn., 1964).

40
 9. S. Ramo and J. R. Whinnery, Fieldsdnd Waves in Modern Radio, 2nd Ed. (John
Wiley & Sons, New York, N.Y., 1953).
10. S. A. Schelkunoff, Antennas: Theory and Practice (John Wiley & Sons, Inc,,
New York, N.Y., 1952).
11. R, W, P. King, Transmission Line Theory (Dover Publications, Inc.
New York, N.Y., 1965).

41
Chapter Two

COUPLING TO TRANSMISSION
LINES

2.1 TRANSMISSION-LINE CONFIGURATIONS

2.1.1 GENERAL

Overhoad lines for power transmission, subtransmission, and distribution are used to
carry 3-plhase, 60-Hz power from the generating station to the consumer. The lines may
vary in length from a traction of a mile to hundreds of miles and may transmit 80-Hz volt-
ages varying frum a few kilovolts to a few hundred kilovolts. Although it is doubtful that
the power lines will be damaged by the EMP, they are very large, exposed collectors of the
EMP that can funnel the EMP energy Into the consumer's facility, the substations that control
line relaying and load control, and the generating stations. A primary consideration in the
analysis of coupling to the power lines is, therefore, the EMP-Induced voltages or currents
available from the ends of the lines that might affect critical components in the terminal
facilities of the consumer or the pcwer system. A second consideration, however, is the
possibility that the EMP-Induced voltages might produce insuldtion breakdown along the

42
 line and thereby induce faults or protective reactions that cause widespread load-shedding o0
similar responses that the power system cannot cope with,

In this section, a brief description of the physical construction of power lines and
power-line components will be given, The emphasis in this description will be on those
aspects of the power lines that are believed to be important in the coupling and propagation
of EMP-induced transients - namely, their geometry and insulation.

2.1.2 CONSTRUCTION OF TRANSMISSION LINES

Wood pole construction Is used almost exclusively for overhead distribution lines
that transmit power from the distribution substation to the consumer. Wood poles are also
very widely used for subtransmission
POLE TOP TWO ARM SINGLE ARM lines that carry power from the bulk-
power source to the distribution sub-
stations. Typical wood-pole line con-
figurations for 3-phase transmission are
shown in Figure 2-1, The conductors
are supported on crossarms, on pin or
post type insulators, A wood-pole line
may carry a single 3-phase circuit as
Iii SINOLE CIRCUIT shown In Figure 2-1 (a) or two 3-phase

circuits as illustrated in Figure 2-1(b).

Wood poles are also usud for

transmisslon lines at voltages up to
161 kV,y1 2
Typical construction of these
lines is illustrated in Figure 2-2, The
conductors are usually supported on
suspension insulators at those voltages,
and at the higher voltages where large
IbI DOUBLE CIRCUIT conductor spacing is required, the H-

Figure 2-1 WOOD POLE CONSTRUCTION FOR frame construction il1itrated in Figure
DISTRIBUTION AND SUBTRANSMISSION LINES

43

,-~--
 f~4i

1w) 44 kV Mb 66 kV (Ie 110 kV

Figure 2-2 WOOD POLE CONSTRUCTION FOR THAI`1M'MISION LINES AT VOLTAGES UP

TO 161 kV

2-2(c) iscommon, Modern transmission lines d~esigned for l,1ghtning protection are provided
with ground wire's above the phase cdndyctqii (It thti.,p bt'66;:h pole In the H-frame con-
struction, for example).' These gi'oiind wires are geunold at'ea~h pole by wrapping several
turns aeopnd1 the butt of the pole
OR06NO below the ground level as illustrated In
WIA IES Figure 2-3(a). Grounding at the pole
riieyailo b~e achieved with a butt plate
attacho) to the bottom of the pole
ond, occasionally, with ground rods
driv~in near the base of the pole as
illustrated In Figuros 2-3(b) arid (c).

Hlgh-voltage transmission lines

supported on steel towers are common
for bulk-iowor trammiss~lsson from
major genetating plants to utility cus-
tom~r.m or within a utility's power net-
work, 1-2 Typical steel tower supports
for single and double circuits are illus-
(bl traied In Figure 2-4, Overhiead ground
BUTT WRAP BUTT PbLATE OflOUND 1HOD wires for lightning protection (at the
Figure 2-3 GROUNDING METHODS FOR WOOD- points atop the towers in Figure 2-4)
POLE LINES~ are widely used on these transmission
44
 lines, Counterpoises at the tower
footings are used to minimize the surge
impedance of the base of the tower.
Counterpoise and insulation are de-
"signedto permit the tower or overhead
ground wires to accorm modate a direct
lightning stroke without the tower-
to-phase-conductor voltage exceeding
the Insulation flashover voltage, Al-
. • though some older transmission lines
AVERAGE SPAN - 600 ft AVERAGE SPAN G600 ft use horn gaps or expulsion-tube light-
ning arresters to protect the phase
conductors, modern lightning protec-
tion design relies on overhead ground
wires and tower footing counterpoise.
to reduce the probability of flashover
of the phase-conductor Insulators,

Some of the properties of

wood-pole and steel-tower tranamis-
sign lines are given In Table 2-1. The
60-Hz transmission voltage is given in
AVERAGE SPAN * 1000 ft AVERAGE SPAN - 900 ft the left-hand column, followed by the
I) SINOGLE CIRCUIT (bI DOUBLE CIRCUIT basic Insulation level, the height (at

Figure 2-4STEEL TOWER CONSTRUCTION the pole or tower) of the lowest con-
FOR TRANSMISSION LINES AT VOLTAGES OF ductors, the span between poles, and
89 kV AND ABOVE typical transmission-line lengths. The
common-mode (zero sequence) surge
is 300 to 600 ohms,
impedance of typical power lines

2.1.3 TRANSMISSION-LINE CABL.E

Conductors for power lines are usually stranded cable of copper, aluminum, or
copperweld (copper-clad steel), although the conductors for the hligher voltages may be of
45
 Table 2-1
SOME TYPICAL PROPERTIES OF ABOVE-GROUND POWER LINES

Basic Huight of Lowest Span

Voltage Insulation Conductor (ft) (ft) Typical
Level WLength Type
(kV)
VkV) Wood Steel Wood Steel (miles)
Poles Tower Poles Tower
2.4 45 25-35 - 160-200' 0-3 Distribution
4.8 60 25-35 - 100-200' 0-5 Distribution
7,2 30-40 .- 100-200' 1-10 Distribution
12,5 95 30-40 - 100-200* 5-20 Distribution
23 150 30-40 - 200' 5-30 &
34.5 200 40 -- 300' 10-40 Subtransinission
09 350 46 40-60 500 600 25-100 Transmission
115 550 50 40-60 600 700 25-100 Transmission

138 650 50 I0-80 600 800-900 25-140 Transmission

161 750 50 50-80 600 900-1000 Transmission
230 1050 - 60-100 .,- 900-1000 45-260 Transmission
287.5 1300 - 70-120 - 900-1000

'Longer spans are often used Inrural areas

the hollow type HH (Hederheim) construction, 1 ,2 Some typical power cables are Illustrated
In Figure 2-5, Cablle sizes range from No, 8 AWG (16,510 cIrcular miIs) for short lines with
light loads, to the equivalent of 1,000,000 circular mils of copper for long, high-power lines,
Cable for high-voltage transmission lines Is of the hollow or rope-core construction Illus-
trated in Figures 2-5(c) through (f) to give the cable a larger radius for a given cross section,
The larger radius is required for high-voltage lines to increase the corona threshold of the
conductor, Some of the larger hollow cables are over 2 Inches in diameter,

2.1.4 LINE INSULATORS

Insulators for transmission lines are madri of glazed porcelain and shaped to provide
long surface leakage paths, wet or dry, Hardware for attachirg the Insulator to the supporting

46
 structure and for attaching the power
conductor to the insulator are pro-
vided. Figure 2-6 illustrates some
typical power-line insulators, The
(l) STRANDED ib) ALUMINUM CABLE pin-type Insulators shown In Fig-
COPPER REINFORCED ACER) p
STEEL ure 2-6(a) and the post-type shown
In Figure 2-6(0) are used for distrl-
bution, subtransmission, and low-
voltage transmission lines. Strings
of suspension-type insulators, such
as those Illustrated in Figure 2-6(c)
are used for all high-voltage trans-
ail STRANDED ROPE CORE (d) HOLLOW CORE mi6sion lineq, Such strings may he
provided with arcInq horns and
gpidlng rings on some of the older
transmission lines. As can be seen
in Table 2-1, the basic insulation
level for transmission lines, which

Is) TYPE HH indicates the peak 1,5 X 40 ps im-

pulse voltage the system can with-
stand without flashover, Is much
higher then the 80-Hz transmission
voltage, even for transmission volt-
If) ANACONDA HOLLOW COPPIR ages greater than 100 kV, This
Figure 2-5 TRANSMISSION-LINE CABLE margin of safety (on the transmis-
CONSTRUCTION sion voltage) permits the use of
lightning arresters whose firing volt-
age is considerably larger than the transmisslon voltage, yet low enough that the lightning
arrester fires before insulator flashover, 1. 2' 3 On older transmission lines, expulsion-type
lightning arresters were used along the line (at each support) to protect the insulators against
direct strikes to the phase conductors. Modern transmission lines use the overhead ground
wire to protect the phase conductors against direct strikes along the line, Lightning arresters
are then used only at the ends of the line to protect the terminal equipment (transformers,
substations, etc.). Many distribution and subtransmisslon lines are essentially unprotected

47
against lightning strikes along the line, but terminal equipment 6 almost always protected
with lightning arresters.

(al PIN TYPI 1b) POST TYPE

HOOK

DOUBLE
POINT
CLEVITYPE ARC HORN •

1
ADAPTOR

BALL-AND-SOCKET TYPE SUSJPENSION CLAMP

It) USPINESION TYPE

Figure 2-6 TRANSMISSION- LINE INSULATORS

48
 2.1.5 Interaction of the EMP with Transmission Lines

The EMP wave induces current in the horizontal conductors strung on poles or
towers, and it induces current in the vertical elements, such as steel towers, ground wires
on wood poles, or vertical runs of cable or conduit at the customer's service, It Is In these
parts of the transmission line that the EMP Induced currents are generated. The amount
delivered to a particular pair of terminals depends on the length and height of these elements
as well as the magnitude, waveform, and angle of Incidence of the EMP and the conductivity
of the soil, In addition, the propagation characteristics of the transmission line and Junctions
and bends in the vicinity of the terminals also influence the manner in which the Induced
currents propagate to the terminals. Formulas for evaluating these effects are contained In
the following sections.

The common-mode currents and voltages Induced In the horizontal conductors are
treated in Section 2.2 for long lines (extending from the terminals to Infinity) and for lines
of finite length, The currents induced In the horizontal conductors usually have the largest
peak values and the longest durations because they can propagate in from great distances,
and for small angles of Incidence of the EMP these currents can build up to large peak value3
(of the order of 10 kA for the high altitude EMP).

The currents and voltages induced In the vertical elements era treated In Section 2.3.
For poorly conducting soil, the current induced in vertical elements such as towers or
ground wires on poles Is smaller than that induced in the horizontal line (unless the line is
very short), but for highly conducting soils, the peak current induced In the vurtical element
may be comparable to that in the horizontal line. However, the latter result Is bocause the
peak current in the horizontal line is smaller for highly conducting soils, rather than because
the current induced in the vertical element Is larger. The Ieak current induced in a 10m
high vertical element by the high altitude EMP is typically a few kA,

The propagation characteristics of transmission lines are treated In Section 24

Included in this section are the attenuation and phase factors for long, uniform lines over

finitely conducting soil and the effects of bends junctions, line sag, and some other devia-
tions from the ideal transmission line. The topics discussed in this section are useful for
refining the estimates of induced current or voltage made from the formulas in Sections
2.2 and 2.3

49
 Differential coupling to transmission lines is treated in Section 2.5. The differential
voltage is the voltage induced between the wires of, say, a single-phase distribution line by
the EMP. This voltage is usually considerably smaller than the common mode voltage
induced between both conductors and ground, The differential voltage is often ot secondary
interest; howevwr, it may be necessary for the evaluation of stress on insulators and meterinq
transformers,

Some general Information on high-voltage propet tes of insulators, lightning ar-

resters, and transmission line conductors is given In Section 2.6, These data are useful In
evaluating the probable effects of the EMP Induced voltages on the performance of these
transmission line components,

2.2 COUPLING TO HORIZONTAL CONDUCTORS

2.2.1 SEMI-INFINITE LINE

2,2,1.1 General Approach

Common-mode (zero sequence) coupling of a plane wave to qbove-ground

transmission lines has been analyzed by several methods, 4 " 17 These include the solut~on of
the boundary-value problem for a cylinder over a plane,5B, the solution for the scattering by
a conducting filament over a ground plane, 9 14
and the solution for a transmission line with
4 7 8
a distributed voltage source. . , The transmission-line solution Is the least complicated,
and it Is quite accurate for EMP transients on power line configurations, 7 ,8 The results
presented here will therefore be based primarily on the transmission-line approximation.

Two additional approximations considerably simplify the calculation of

the current or voltage Induced on the line without much loss of accuracy, The first is the
assumption that the soil behaves as a good conductor (Ie,, a > co ) for all frequencies of
interest. The second is that the rise time of the incident wave may be considered zero if it
Is small compared to the round-trip propagation time between the wire and ground, The
latter approximation permits fast-rising two-exponential pulses to be treated as single-
exponential pulses. The assumption that the soil ,-'ves as a good condoctor causes some

50
distortion of the leading edge of the induced pulse after the ground-reflected wave returns,
but this distortion is often negligible - particularly if the pulse is filtered by the service
transformers and low -voltage circuits before it reaches the customer's equipment.

The coordinate system used in the mathematical analysis of the transmis-

sion lines 's llustrated in Figure 2-7. The angle of Incidence of the wave Is defined by an
azimuth angle p measured from the projection of the wire on the ground, and an elevation
angle ip measured from the horizontal ground plane.

2.2.1.2 General Analysis of Transmission-Lines

In the analysis of the coupling of electromagnetic waves to above-ground

transmission lines and similar structures, the source of the voltage that drives the line Is
distributed along the length of the transmission line. In this section we will develop the
differential equations that describe transmission lines with distributed voltage sources, and
show general forms of the solutions to these differential equations.

A transmission line with a distributed source voltage Is, by definition, one

that has an increment of source voltage in each increment of line length. An element, dz in
length, of such a transmission line Is illustrated in Figure 2-8. Except for the source labeled

CONDrUCTOR

DIRECTION Or _ .- "
PROPAGATION

Figure 2-7 COORDINATES DEFINING AZIMUTH AND ELEVATION ANGLES OF INCIDENCE

51
 EIdz I + d9l dd
FIa di

V Y dz V++ d
di

di

Figure 2-8 EQUIVALENT TRANSMISSION-LINE CIRCUI1

Ez In Figure 2-8, this transmission line is identical to classical transmission lines, and the
techniques for determining the impedance per unit length Z and the admittance per unit
length Y are the same as for classical transmission lines. The source term has been labeled
Ez because it has the units of electric field strength (V/m),

The coupling calculations are based on the assumption that the height of
the horizontal conductor over ground can be considered small, In terms of wavelengths, so
that circuit analysis can be used, This assumption allows the low-frequency characteristic
impedance of the line to be used, In the circuit analysis, radiation from the line and the
uxistence of transmission modes other than the single simple TEM mode are also ignored.

The differential equations for the voltage and current along the trans-
mission line of Figure 2-8 for harmonically varying signals (ejwt) are

dV
- Fz - IZ
(2..la)
dz

dl
-VY, (2-1b)
dz

52

.........
..
........-...
By differentiating one and substituting the other, the second-order differential equations
can be obtained:

d2V =
-- V (2-2a)
2
dz dz

d2 1
-- 21 -YEz (2-2b)
2
dz

where y2 = ZY, Except for the terma containing Ez, Eqs. (2-1) and 2-2) are identical to the
equations for the more classical transmission-line formulation. The solutions to Eqs, (2-2)
4
are ,7

I(Z) [ K1 + P(z)] e-V + [K2 +Q(z)I elf (2-3a)

,Z).. K + P(Z)e-"' [K 2 + o(Z)]°e,' (2-3b)

where

jar
P(z) -
- e•v Ezdv (2-4a)

Ql) 2Z, -V E/dv 12-4b)

1 f,2

K1 and K2 are coefficients determined from the terminating impedances Zl and Z 2 , at the
ends of the line at z - z2 and z - z2 (z2 > z, ), respectively, and Z,, \/1Z7 as in conven-
tional transmission lines, K1 and K2 are constants given by

p 2 P(z2)e-' 12 Q(21)012
K, p- 1e_1•Q (2-5a)

53
 p1Q(Zi)e1Yzl - p(z 2 )e"Z1z
K2 = p2e_7Z2 (2-5b)
e'Y(z2-Z1) _ PlP2e-I(•2-z)

in which the reflection coefficients p1 and P 2 are given by

• Pl
SZ1 Z1 +
-ZoZ0 P Z2
Z2 +
- Z0
Zo 2-6)
2O

The electric field E,4V) In Eqs, (2-4) is the "undisturbed" field that would exist at the wire
height if the wire were not there, This resultant field is written Eu (h, z) below to indicate
that it is the z-component of the undistrubed field at height h, For an incident field of
amplitude El, the resultant field is

Ezu(h,z) - Sriei"kcOICa1 (i - Re-1k2hiiin) sin €cosp (2-7a)

for a vertically polarized incident wave (magnetic vector parallel to the surface), and

EU(h,z)- E sin • (1 Rhe-I2khlini) ekzc°'cot°s (2-7b)

for a horizontally polarized incident wave (electric vector parallel to the surface). The phase
is referred to the phase of the incident wave at tile wire height when z - 0, In these expres-
sions k -w=pc Is the propagation factor for the wave in free space, and Rv arid Rh are
the reflection factors for wave reflection at the air/earth interface, These reflectinn factors

18 19
are qiven by .

sin -V•"r (1 + _2)- COS2•

Rh = -°' "jB e
sin ýi +V- 1 + . cos2 p

and

er(1 + 0.) sin ý -V/c..r_(14 +T'-' CO2o

FRv =r Cos (2-8b)
2
I
Cr(1 + j esin iP +V/•r(1 +j---)- cos

54

..............
 It ,;,'.nvenik-nt to separate the term 1 4 Re'ik 2hhln 0 in Eqs. (2-7) Into
two parts - ,)nl.given by 1 eijk 2mi1 ' that depends on the tine height and a phase-shifted
term given by e-lk2hs;i (1 + R) that depends on the properties of the soil. Here R -Rv
3

for vertical polarization and R Rh for horizontal polarization. For a >> we sin 2 ', the
terms 1 + R reduce to

2 sin
1+Rh A•---- (2-9a)

2
l'Rv~ - .1(2-9b)
a sin

These expressions can be used to obtain two components of the resultant field Eu (h,z)
from Eqs. (2-7) - one, depending on line height h, that would be obtained If the ground
were perfectly reflecting (I R I - 1), and one, depending on the soil properties contained in
1 + R, that is, in effect, a correction term to account for the fact that the ground is an
Imperfect reflector of the Incident wave. These two components can be used in Eqs. (2.4),
and thence in Eqs. (2.3), to obtain two solutions for the current or voltage.

The approximations of line height small compared to wavelength, negli-

gible radiation loss, and no non-TEM propagation mode have been demonstrated to intro-
duce very small errors In the results computed with the transmission-line approximation
given above. Figure 2-9, for example, shcovs the current Induced In an infinite line by a
step function Eo u(t) of incident field calculated by the transmission-line method (dashed
curves) and by an exact method (solid curves). 5 ,. it is apparent from this Illustration that
the transmission-line approximation Is a good approximation to the current induced in a
wire over a perfect ground plane even for the zero-rise-time step function. For realizable
finite-rise-time pulses, the difference between the bxact solution and the transmission-line
approximation is so small that it is difficult to detect eperh~ientally, The exact solution
does pr-vide interesting insight into the behavior of the %catteredwaves before the structure
"settles down" to behave as a transmission line. In Figure 2-9, for example, one can see the
discontinuity that occurs as the ground-reflected wave arrives at (2h sin 0)/c, a second
discontinuity when the scattered direct wlt! reflected from the ground returns at 2h/c, and
a third discontinuity when the ground-tuflected wave scattered from the wire reflected back

55
 40
h 100

4' 4e

20.- o* / -

/I "%'•'•-•"'5APPROXIMATION
TRANSMISSION LINE
10 ""

//
EXACT SOLUTION

0 I I . ..
0 h 2h 3h 4h 5h

TIME

Figure 2-9 COMPARISON OF THE TRANSMISSION-LINE APPROXIMATION AND THE EXACT

SOLUTION FOR THE CURRENT IN A WIRE OVER A PERFECTLY CONDUCTING GROUND
PLANE - STEP-FUNCTION INCIDENT FIELD, HORIZONTALLY POLARIZED

from the ground arrives at (1 + sin i) 2h/c. Subsequent multiply-scattered waves have a
very weak Influence, hut they eventually bring the exact late-time response into coincidence
with the transmission-line solution,

2.2.1.3 Open-Circuit Voltage in the Frequency Domain

The open-circuit voltage developed at the end nf a semi-infinite trans-

mission line of height h above a perfectly reflecting ground (HH 1 1) by a 1 V/m plane
7
Incident wave is

V..= cD(Ww (2-10)

5w

56
where c is the speed of light, = 2irf is the radian frequency, to (2h sin 4)/c, and D(/,O)
is a directivity function defined by
sin L cos 0 (vertical
(, c +S. cos 0 cos P polarization)
i• k
(2-11)
sin p (horizontal

ac +. .cos Cos 0 polarization)

joJ k

where a and 3 are the attenuation and phase factors for the transmission line (for perfect
conductors,/3 k and a - 0). Although a, i3,and k are In general frequency-dependent,
the value of 1(0,0) depends primarily on the angles of Incidence, and It Is almost Independ-
ent of frequency for many power line configurations. Plots of the directivity functions are
shown in Figures 2-10 and 2-11 for/3/k
O 1 and for ce/k negligible or constant. The assump-
tions that [3/k - 1 and a/k Is negligible are valid for finite soil conductivity If h ;0 6 and the
angles of Incidence 0 and ,p are large, For small angles of incidence, however, cos 0 cos 0
-• 1, and the assumption that 0 - k may cause an overestimate of D(i,,), and thus of the
Induced voltage and current, particularly at low frequencies, The value of 3/k for conductors
over soil are given In Section 2,4,1,

The corroction that must be applied to V.. to account for the Imperfect
reflection from the soil when I Rh,v 1 1 and o >> •e is

2 eJj/t- (vertical
AV c 134,0 -sin 0 v/J polarization)

(2-12)

0cV~e D(') 2 W t (horizontal

polarization)

where r@ - co/O,a is the soil conductivity and co Is the permittivity of free space,

57
 24'210' 10' 161f 120,

300

2100
+ g1
202
300'0

Fiue-0 OERLN
DIRECTIVITY
FANTEIOS FOR VERTICAL POLARIZATIONI*migbe

Souce
R10.0S
lo58
 20 0 1I0" 150
r ' . .. 8 ]
sin6
t h~• •) . .... ........... -.... - c s•

'1 - CO•Cc~oS • ,, k
01 C
Cs.~

S•4U 2.1 .1200

270 . . 90 270"' ,,vjol0

go........,

0.0

300'' \ 6/
,04o)i, 3''0,3. A

1.6

330' 0 30, 330u 0 30°

(a) OIRECTIVITY FUNCTION FOR ibh DIRECTIViYV FUNCTION WITh

i.oO,•eUNTAL, POLARIZATION CONSTANT a/k, FOR • 0

Figure k-11 POMR-LINE DIRECTIVITY PATTERNS FOR HORIZONTAL POLARIZATION,

Source; Ref, 8,

The total open-circuit voltage Is than*

VoC V' + .w -c 0(,,P) 0!" - + 2'l(sin 0- I•'° (2-13)

where the exponent +1 is associated with horizontal polarization, and -1 Is associated with
vertical polarization, The voltage given by Eq. (2-13) Is caused by the horizontal component
of the electric field. For vertical polarization, there is an additional component of voltage
produced by the vertical component of the electric field, This component is discussed in
Section 2.3, where the voltage and current induced in the vertical elements of the transmls-
sion "!ieare discussed, Plots of the open-circuit voltage calculated from Eq. (2-13) are
shown ii Figure 2-12 for vertical polarization and in Figure 2-13 for horizontal polarization,

59
 o 10 - _3

1 10- ' 30
10,1

0'

10

E 30

h 10 mih-1m
30"''f 02
0 0

1~ 10 U1.1- 10

Is) VARIATION WITH SOIL CONDUCTIVITY, 0 Ib) VARIATION WITH ELEVATION ANOL,

Figure 2-12 OPEN-CIRCUIT TERMINAL VOLTAGE INDUCED IN AN INFINITE TRANSMISSION

LINE BY A VERTICALLY POLARIZED INCIDENT FIELD OF 1 V/m

The source impedance for the semi-infinite transmission line is simply its
characteristic impedance Z. = V/7 For typical power transmicsIon,.lIne configurations,
20
this impedance deviatus very little from ,21

?to 2h
Z, log - (h>>a) (2-14)

where a is the radius (or effective radius) of the power conductors and p)j =V/'•'/0'- is the
intrinsic impedance of free space, The Thevenin equivalent circuit of the transmissinn line
is thus the open-circuit voltage gitun by Eq. (2-13) in series with Zo. The Norton equivalent
circuit is a current source Voc/Z 0 in parallul wg,' Zo.

60
 100

1.I0"3 i VARIATION WITH SOIL CONDUCTIVITY, a

100

h.- 10•2

x - go*

r,,'",,L ToII I fillll

so II I 1 11TII::t1
I I
3.0*

0., 101 10 t00

Figure 2-13 OPEN-CIRICUIT TERMINAL VOLTAGE INDUCED IN A SEMI-INFINITE

TRANSMISSION LINE BY A HORIZONTALLY POLARIZED INCIDENT FIELD OF I V/m

e1
 2.2.1.4 Transient Waveforms for an Exponential Pulse

For an incident exponential pulse Eoe-t/r whose Fourier transform is

E(w) - E. -2-15)
JW + 1/7

the open-circuit voltage at, the end of a semi-infinite line Is

V otlW ) Eoc D( 4, ,) -W + 1 /)
-w + ' (sir r, )' 1W)

From the Inverse transforms, 2 2 the voltage wavel -i Is

Vac(t) EocTD(is,o) 11 -e.t/T] (o t < to)

(2-17)

X eU'2 du it 10 to)

where t', t + to,, and to (2h sin 0)/c. The first term in the braces In Eq. (2-17) is due to
the geometry (line height and angle of Incidence) and is independent of the properties of the
soil, The second term contains the effect of the soil conductivity in TI) - ro/o, as well as
geometric effects,

2.2.1.5 Parametric Variation of the Open-Circuit Voltage

The open-circuit voltage vot(t) is plotted In Figure 2-14 for both polari-
zations with soil conductivity varying from 10O3 to - mho/m, The wavelurm for ri

62
 VERTICAL. POLAnWhATION 6P~ 01 D,(30". 0") 3.73
- HORIZONTAL POLARIeATION
----- 'ýP
- 96 D1 3 6

60EU

ZTM 30 E D

-4OPNCRCI
Piue VLAE TTE N P EM-NINW IN O
VARIOUS SOIL~ CODCIVTE
- 11f
a', oe
a 0)15produed
bythe fist tem of q. (2-7) alne. Te devAinsothohr

Figrms in1
FiguEN2-CIR4 indicate Ibecus the d OFunto
ENDvt gr
A SEIsIFNT FORvria
INte

polaIstionueforhorizonsta
bha polrizatonE.(21)aoeThdvitnsfteohr
fothtfigre 2-1 and inusother illusfetraeletions indued
waveforIn of Iniefnrtthate bolo
the ampliund.Te ofpigcasdb
oe-ictvlageouis
the noraliedtoIare D~then)the vaulues ofused by

fort
asaparesont end-gonuncd-enfct~ 0)pln wth vertical polarizationand ~ondsideral inciener
than th0deres wopingto
horizontal polarization, aThoug thfeene reuls aen
graer tappie toe wany-
azimut anFglre of1 Incidecaep.Hwvr because the giecii
ound-cetio DteOirmeptend fon etia
polriztin
tanforhoizota poarza 63n

.hrIlutatoso
InFgre21.adI. *,~,li.~Induced waeom. ha olw
 (sin 0)4 1, as well as on D(ip), the results shown are applicable to only the 30 degree
elevation angle of incidence,

The effect of line height on the open-circuit voltage waveform is illustrated

in Figure 2-15, where the waveforms are plotted for vertical polarization incident on soil with
a conductivity of 10-2 mho/m. Also shown as dashed curves in Figure 2-15 are the wave-
forms with o =; only this portion of the waveforms is proportional to the line height, For
the soil conductivity of 10-2 mho/m, the ground effect is larger than the height effect, and
changing the height of the line does not have a large effect on the peak open-circuit voltage,

The effect of incident pulse duration (decay time constant) on the open-
circuit voltage waveform is illustrated in Figure 2-16, where the op•n-clrcuit voltage is

VERTICAL POLARIZATION I1 0)
0D 30', 0') • 3,73

3 0 E QD h - 15lt 0t
E11-i

3 0 1M 0"
00

K15F
,,"0..

6W4
0

w|

0 10,13 10-6
TIME~

Figure 2-15 OPEN-CIRCUIT VOLTAGE AT THE END OF A SEMI-INFINITE LINE FOR

VARIOUS LINE HEIGHTS. Source: Ref. 7.

84
 plotted for incident pulse decay time constants of 0.25, 0.5, and 1.0 Ms. As is apparent from
Eq. (2-17), the waveform for a given soil conductivity, line height, and angle of incidence
can be plotted as a function of a normalized time t/r, in which case the pulse decay time
constant 7 affects only the relative magnitudes of the two terms in braces in Eq. (2-17).
However, it is apparent in Figure 2-16 that the pulse duration affects both the peak voltage
and the time required to reach the peak value. The wider the pulse, the larger the peak
voltage because the wider the pulse, the longer the segment of line near the terminals over
which the ground effect Is integrated. Note that for perfect ground (dashed curves in Figure
2-16) the pulsewidth has little effect on the peak voltage, and the open-circuit voltage

VERTICAL POLARIZATION
li.,L -- ,(t) Eo,"'/)

30 EOD c 0

011
""1 . 10 m

0.
o 110 0.
M

•'• •,•.2B0.21-
' 6
7'101.- ,

0.5 10e 10-6

0

TIME

Figure 2-16 OPEN-CIRCUIT VOLTAGE AT THE END OF A SEMI-INFINITE LINE FOR

VARIOUS INCIDENT PULSE DECAY TIME CONSTANTS. Source: Ref, 7.

65

j
 waveform is essentially the incident-field waveform except for the finite rise time to
(2h sin /)/c.

In Figures 2-14, 2-15, and 2-16, the waveforms for vertical polarization
have been plotted for an elevation angle i = 300 and an azimuth angle 0p= 0*, while the
waveforms for the horizontally polarized incident wave in Figure 2-14 were plotted for
= 300 and o = 900 (broadside), In all three Illustrations, it was assumed that C/k << 1 and
that a/k is not frequency-dependent. It Is also Important to recognize that changing the
angles of incidence In Eq. (2-13) affects the relative magnitude of the ground effect through
the (sin 4 )k1 coefficient as well as affecting the directivity function D(',O) and the delay
time to. In effect, there are two directivity functions - one for the response with perfect
ground, D(•I,), and one for the correction term, (sin 0)11 D(Pp),

2.2.2 LINE OF FINITE LENGTH

2.2.2.1 Uniform Plans Wave

The open-circuit voltage induced at the terminals of a horizontal trans-

mission line of a finite length 2 as illustrated in Figure 2-17 by a uniform piano wave ise

S=V =(( ) 1[11 - i-(y

jk%)] e-t2 _ e-( -- Ik')Q

x D(1,4 + ir) x rn-2J

e- (2-18)

S... .• i • i C...O
il •l l ll I - 1 :i - I tllN '
86
where Voc = (w) is the voltage induced in a semi-infinite line, and

"Y =+ j/c
(X

k' cos cos p

ZQ - Zo
A•=
Z + Zo '

The open-circuit voltage Voc a (w) for the semi-infinite line is given by Eq. (2-13), the
directivity functions D(ý,Op) are given by Eq. 12-11), and the characteristic impedance ZO Is
given by Eq. (2-14). The reflection factor pQ at the end of the line opposite the terminals is
a function of the load impedance ZV at that end of the line (see Eq. (2-8)),

When the line attenuation o Is negligible, the exponential terms In Eq,

(2-18) become of the form exp [-jw(1- cos 0 coo p) 2/c] or exp [-jw 2V/cl, which
transform into delays in the time domain. Thus the open-circuit voltage waveform of the
finite-length line is Identical to that for the semi-Infinite line until the first end-effect
arrives at (1 - cot 0 cosDo) R/c. Between this time and 2Q/c, the waveforrn i that for the

semi-infinite line, modified by the factor -pq Dj0g + After 29/c, the waveform
repeats, as illustrated in Figure 2-18 where the waveform for a line 150 m long, open-
circuited at bQth ends, and having negligible attenuation, is shown for an exponential pulse
of incident field, When the attenuation of the line Is not neglected, the abrupt changes
shown in Figure 2-18 become more rounded, and the amplitude of the oscillations becomes
smaller with increasing time,

ZR / h TRMNALS

FilUim 2-17 TRANSMISSION LINE OF FINITE LENGTH

57
 The source impedance at the terminals of a line of finite length is

1 +
Z(O) = Z - - (2-19)
1 pye-2

and the short-circuit current at the terminals Is Isc(w) - Voc(w) / Z(O). The open-circuit
voltage given by Eq. (2-18) is only that part induced In the horizontal conductor by the
horizontal component of the Incident electric field. If a vertical element Is Included, the
vertical component of the electric field (in the case of a vertically polarized Incident wave)
will induce a voltawe In this vertical element, The voltage induced In the vertical element Is
discussed In Section 2.3.

2.2.2,2 Spherical Wave at Grazing Incidence

Consider a vertically polarized wave is Incident at - p * 0,

0 as Illustrated
In Figure 2. 19, where the EMP Is generated by a surface burst on the transmission line'(or
an extension of the line), Near the source point (in the close-in region), the radiation from
the weaport will Ionize the air sufficiently to cause It to behave as a conductor (0 >> W6o),
and further away (in the Interrrmediate region), the ionization will be weaker but still sufficient
to produce significant attenuation of signals propagating away from the source an the line.
Here we discuss only the coupling that occurs outside these regions, where the EMP wave
behaves as a vertically polarized wave propagating outward from the source at the speed of
light,

The transmission-line axis is assumed to extend from the point z1 to the

terminals at z2 > zi , as illustrated in Figure 2-19. The wave In the air Is assumed to
propagate as though the ground weru a perfect conductor so that E x - TrlHy, where ri0 Is the
Intrinsic impedance of free space, and Ex - (Eoe-ikz)/z. However, for a finitely conducting
ground there will be a z-component of the electric field at the surface whose value Is
E,(O,z) , viHy(O,z), where Tiis the Intrinsic impedance of the soil, and 7b»>>? > 0.

68

.......................................
 0-
30 ED VE1TICAL POLARIZATION
1iltI /I0
E ' Eo --I

I 2•ohL)•.aOu 30 po 0
c/i2 , 150 m

10 Eod h 10 m

2YI

(I
- 'ou Il c c
DTIME

Figure 2-18 OPEN-CIRCUIT VOLTAGE WAVEFORM AT THE TERMINALS OF A 150-m-LONG

LINE OPEN-CIRCUITED AT BOTH ENDS

INTEM MEDIATEfi
REUION
" ,0 AIR
t'0

CL.(.SE-IN H.(-LGIOCN_ • ,/

0 2

SOIL

Figure 2-19 PHYSICAL ARRANGEMENT THAT PRODUCES A VERTICALLY POLARIZED

WAVE WITH END-ON INCIDENCE

69
 7
The horizontal component of the field at the wire height is

Ez (hz) h Eoei-k (2-20)

where tjo TM • and 71-.v/fJU'o/(o + jwe), If we neglect attenuation of the wave due to
losses to the ground, and apply a first-order correction to the field at the surface to obtain
"thefiea:i it the wire height. (Eo is the total vertical electric field, so that it corresponds to
2Eo in Lite ipace-wave case discussed in Section 2.2.11)

The open-circuit voltage induced at the terminals at z = z 2 by this

horizontal electric field is, from Eq. (2-3),

Voc(W)
EohZ(Oie-•)9 -Jk)z
Z ~ 2 *(y-Jk)zj
l
r
+= h1
((-,7Mk)]

(2-21)
e('-P-k)v dv
X [12 .1v

where t, Z(O), and k are tlhe propagation factor, the source impedance given by Eq. (2-19),
and the free-space phase factor, respectively,

The coupling to the wire can be resolved Into three components according
to the nature of coupling: geometric, ground-effect, and phase-disparity, If the wire and
soil were perfect conductors, then qjw 0 and y jk, and the open-circuit voltage on the wire
at z 2 would be

VZ)- -- Z- k* 2 - 1) (' : j). (2-22)

The coupling In this case is the geometric component caused by the nonuniformity in the
magnitude of the incident electric field along the wire, which causes the potential at the
wire height at z 1 to differ from the potential at the wire height at z 2 -

70
 If 3 jk and P?* 0, we obtain an additional ground-effect component

EOi 1Z(O) -jk+ 2 Z2

V,,(z) Zea1 o e Iouy-
"z i (2-23)

which is caused by the finite conductivity of the soil (?I - Vi,5-,), This component is the
integral (between z Ihand z 2 ) of the z-component of the electric field induced in the finitely
conducting soil by the incident wave,

Finally, if -t "*"
1k, both the geometric component and the ground-field
component are modified, and a third component, caused by the disparity In the propagation
characteristics of the Incident wave and the induced responses, Is obtained. This phase-
disparity component is

Eoh Z(O) -/z 2 ,z2 e0'-1k) v

VOj(z 2 ) - Zo e (-y-jk)Jf2 ev dv (2-24)

which exists only if there Is a disparity between the phase factors of the incident wave and
those of the transmission line,

For typical soil conductivities and line lengths, the geometric term given
by Eq. (2-22) Is significant only at very low frequencies unless the line Is very high or very
close to the source, and the phase-discrepancy term given by Eq. (2-24) Is small compared
to the ground-effect term given by Eq. (2-23). The open-circuit voltage induced in the
transmission line by an exponential pulse (Eo/z)e-t/r is thus

"EZ(0)(

(z)log
[ W](2-25)

-yz o
,/rOlor /A
z Ib
1
I
_ -I
_j+11(

where re seo/u, when phase is referred to the terminals at z 2 .

71
 The voltage waveform Is

EoZ(0) Z,2 t/
v z 2,t) • llog-- e-UI t < E'rTu
Zo /C-r? ZI
(2-26)
EoZ(0) 'TO-V221
I 229-
1o VIrr log-2
ZIL
e d.
i
t> 6rse

for a zero-rise-time exponential pulse. For a two-exponential pulse with non-zero rise time,
the voltage waveform can be obtained by superposing two solutions such as in Eq. (2-26),
The waveforms for single and double exponential transient fields are shown In Figure 2-20
for a pulse decay time constant of 1 ps and soil conductivities varying from 10-1 to 10-3
mho/rn, The line is assumed to be terminated in its characteristic impedance at the end zX
opposite the terminals, For the two-exponential pulse (solid curves), the rise-time constant Tr
is4 ns. The early-time and late-time approximations, which do not meet at t merrs for the
lower conductivities, have been joined by a vertical line. The dashed curves are the zero rise
time (single exponential) responses,

The waveforms of Figure 2-18 are such short, high-amplitude pulses that
it was necessary to plot them on a log-log scale to view significant details. Note that the
peak value and the time to reach the peak value both increase as soil conductivity decreases,
because the more poorly conducting soil behaves as a dielectric longer and supports a larger
z-component of electric field when It behaves as a conductor.

Comparison of Figures 2-20 and 2-14 shows that the peak voltages obtained
with end-on illumination of, semi-infinite line can be much greater than those obtained 'Vith
oblique illumination of a line of finite length, Although some of the difference between
these two cases can be attributed to the nonuniform illumination, most of it can be attributed
to the length of the line. On the semi-infinite line, the peak voltage contains voltage that was
Induced at a distance Ctpk/(1 - cos 4) away from the terminals. Therefore, unless the
transmission line is straight and at least Ctpk/(l - cos 4) long, the open-circuit voltage will
never attain the peak value predicted for a semi-infinite line. One should therefore use
caution in applying the currents and voltages predicted for semi-infinite lines with small
angles of incidencu,

72
 2.2.3 RATE OF RISE OF THE OPEN-CIRCUIT VOLTAGE

The rate of rise of the line voltage is important in evaluating lightning arrester per-
formance and insulation flashover characteristics, For typical line heights such that the
transit time h/c Is large compared to the rime time of the EMP and small compared to the
pulsewidth, the rate of rise of the voltage In response to a step-field Isa useful measure of
the early-time behavior of the line voltage, The open-circuit voltage Induced at the end of
a semi.-infinite line by a step of incident field Eou(t) Is

v(t) 2Eoh sin 0 D(Op) -(0 r. t < to) (2-27)

to

1.0 !T~~~~fl-'-'T'- I I L!-1----nT~--11F1

00

AD1004 m0-RO THE SO C WITH V/r INIETONTEEDNERS

1 H

SOURCE 14o 300 12, tj 1000 m, 12 -2000 mn,exponential pulse). Source: Ref. 7.

73
where to = (2h sin t)/c. At t - to, the reflection of the incident wave from the ground
returns to the wire and the form of the voltage changes. At this time, however, the open-
circuit voltage for small , and 0p(where D(P,ýp) is large) can already be 2Eoh, which, for
typical EMP field strengths and line heights, is of the order of 1 MV. The time required to
rearph this voltage Is to for a zero-rise-time step function; for a finite-rise-time EMP, how-
ever, the minimum rise time Is the rise time T r of the EMP. The maximum rate of rise of the
open-circuit voltage is therefore

Av 2Eoh
A - ('r > to) (2-28)
at T'r

For E0 5 X 104 VWm, h = 10 m, and Tr v 10 ns, the maximum rate of rise is 100 kV/ns,
This rate of rise occurs for either horizontal or vertical polarization for grazing, end-on
Incidence on a semi-Infinite low-loss transmission line. (For vertical polarization, the
maximum net voltage between the end of the horizontal conductor and the top of the
vertical element between the ground and thle horizontal conductor at t - to is 2E 0 h; for
horizontal polarization the total voltage of the horizontal conductor is 2Eoh and none is
induced in the vertical element.)

For angles of Incidence such that rr < (2h sin O)/c, the rate of rise of the open
circuit voltage Is

At Eoc D(O,,p) (?r < t < to) (2-29)

whete D(00,ý) is given by Eq. (2-11). For Eo - 5 X 104 V/m, Av/At , 15 D(0,ýp) kV/ns.
Figure 2-21 shows, for various constant values of a/k, plots of Av/At with vertical polariza-
tion incident at an azimuth angle of op- 0, where D(O,ýp) is maximum (dashed curves) and
with horizontal polarization Incident at 0 - 0, where D(',p) is maximum (solid curves). As
Is apparent In Figure 2-21 the rate of rise of the open-circuit voltage Is strongly dependent
on angle of incidence, with the values approaching 100 kV/ns occurring only for small angels
angles of Incidence.

74
 1000

HORIZONTAL POLARIZATION, :

S- o ,Ina' - I-•
k k
VERTICAL POLARIZATION, '0 0:

Osin

- - s0o 10 v/m

0103tm
ANLEOINIENEI,'Ou

10
a 30 so go
ANGLE OF INCIDENCE, ... dug

Figure 2-21 RATE OF RISE OF THE OPEN-CIRCUIT VOLTAGE AT THE TERMINALS

OF A SEMI-INFINITE LINE

AlthouhJll thi0 initial rate of rise of the vol tage is not clet)indent on the hngthl of the
line, the time that this rate of rise lasts and the final voltage obtained are dependent on lino
"length. For a line of finite length ý, thirofore, the rate of rise EocD( •.p) lasts only for a time

tc - k l -- Cos , cosP)/c (2-30)

and the maximum voltage obtained with this rate of rise is

Ek X J sin ý cos - (vertical polarization)

sin ,0 (horizontal polarizatio ) (2-31)

75
 100 if tt < to, If this open-circuit voltage is quite
small, the rate of rise may not be important
-- even though it is quite large, Note that Lhe
50 -open-circuit voltage is smallest for the angles
E of incidence that produee the largest rate of
rise, Figure 2-22 shows the minimum length
of time required to develop 100 kV at the
S20 small angles of Incidence before ond-effects
alter the rate of rise (assuming t! <to), In
Figure 2-22, ip- 0 and P is varied for vertical
I0 polarization, and 0 Is varied for horizontal
- h •,. polarization (I.e., the angle of Incidence Is
0i -h10 MO
rfor vertical polari?,tlon and p for horizontal
00 to
10 - 20polarization),
20
ANGLE OF INCIOINCE -It should he emphasized that the rates

Plgure 2-22 MINIMUM LENGTH OF of rise given by Eqs. (2-28) through (2-31) are
LINE REQUIRED TO OBTAIN AN
OPEN-CIRCUIT VOLTAGE OF 100 kV rates of rise of the open,'circuit voltage at the
WITH 60-kVm FIELD INCIDENT AT and of a long line, The actual rate of rise across
SMALL ANGLE
the terminals of distribution transformers or
across the potheads is much smaller, For
example, the maximum rate of rise of the voltage across the 50 pF bushing capacitances of
three delta-connected transformers supplied from a 300 ohm line Is

Av 2E, h 2 X 5 X 104 X 10
,C
" :300X6X50X10"12

for E,, .. B X 104 V/mni ndhI- 10m, For tht! mime c(onditionr, the rate of rise of the volt-
age across three 20-ohm shielded ctibhis wuold be

,%v _ 2E,h k u,3

/. 2 X 5 X 104 X10 ?0'3 2,2XV/r2s
t "Z 10-1 300 4 20/3

Thu% the maximum rato of rise of thet vr• t lue actually [ppllind to erquippment torminals is of
the order of 10 kV/ns.

76
2.3 RESPONSE OF A VERTICAL ELEMENT

2.3.1 CURRENT AT TOP OF VERTICAL ELEMENT

Vertical elements such as ground wires and service-entrance conduits also interact
with the incident wave and therefore have current induced in them. Because only tile
vertically polarized wave has a component of electric field In the vertical direction, the
vertical elements actively Interact with only the vertically polarized wave; for horizontally
polarized waves, the vertical elements behave as passive Impedances with delay times
associated with their length. The vertical element Is considered to be a biconic transmission
line with Its upper end terminated In Its characteristic Impedance and its lower end short-
circuited to the ground (see niguro 2-23). The current induced In the vertical element
opposes the current Induced In the horizontal conductor when Iý I < It/2 and aids it when
pI > ir/2, Since the coupling to the horizontal conductor Is greatest for nr/2 - when
<I1P
D(ik,p) is large -* the current induced in the vertical element tends to reduce the maximum
current calculated for the horizontal conductor alone.

For a vertically polarized wave Incident at an elevation angle 0 on a vertical element

of height h, the current at the top of a lossless vertical element will e1),8

at

-x
X

Za

I.) (b)

Figure 2-23 VERTICAL FLEMENT AT THE END OF A TRANSMISSION LINE

77
 2h (1-t1
jW
1 1-e C
Eo cco-,
2Zo j 1 + sin

j2W
2h1 -'t,) 2h (2-32)
+1, - sin o•4+ 1/1.
abov theis1 e:

for an exponential pulse Eae-t/r and a semi-infinite horizontal line, In the expression
above, the ground is assumed to be a perfect conductor, and Zo Isthe characteristic Im-
pedance of the biconlc transmission line given by 1 9

n/o 2h !
Zo "(2-.33)
109
I~r a

where a Is the radius of the riser and h Is its height. Note that Zo for the vertical element is
essentially the same as the characteristic impedance of the horizontal conductor of radius a
and height h,. The waveform uf the current Mtthe top of the vertical element is7.B

I(h,t) 224 1 +sin

cos-t r~c C t>O - t>1
(2-34)
1 1 e (t- t )/t- 1 - (ttt3 l ) 1
• •s in t> t 2 a- t

where t1 - 2h/c (1 + sin 0, t2 -(2h/c)sin ,, and t3 = 2h/c,

For a vertically polarized wave at grazing incidence (0 -- 0),

(1 e-(tt3)/r) tt3] (2-3 '

"Z R'-
-c( e-t/r) t> .

78
 The soUrce irnpedancr, of the vertical element viewed from the top and terminated in soil at
the base is

2h
1 + pe C
Z(h) Za (2-36)
2h

1_poe c

where po is the reflection coefficient at the base and losses in the vertical element are
negligible. For many power-system applications it can be assumed that the soil is a perfect
conductor, for which p0 = -1. (For the impedance of ground rods, see Section 2.3.5.) The
open-circuit voltage developed at the top of the vertical ulement ;s

Vet(h) = I(h) [Z(h) + Zo) (2-37)

and the short-circuit current is

Z(h) + Z
16,(h) - ---- - (h) (2-38)
Z(h)

A olot of the short-circuit current at the top of a 10-m-high element that is ter-
minated *he base by extending the conductor 2 m into the soil is snown in Figure 2-24.
The base resistance was calculated fiorn Eq. (2-50) with V = 2 m, r = 10-2 mho/m, and
a - 2 mm (81 mils). The short-circuit current shown in Fig, re 2-23 is for a uniform
incident field of 1 V/m at all frequencies, rather than for the rxponential pulse spectrum.

Plots of the current waveform are shown in Figure 2-25 for 4 = 0 and 4 = 30' for
an incident exponential pulse with a decay time constant r = 1/is and a riser height h = 10 m.
The waveforms for thuse two angles of incidence are quite similar, but the leading edge of
the waveform for 4 300 isrmore complex because of the difference in the time of arrival
of the upward and downward traveling current waves induced by the direct and ground-
reflected electric-field waves.

It is noted that the current given here is that induced in the vertical element only;
this current must be added to any current induced in the horizontal line with proper regard

79

'II IIIII II
 FE
>!
100

.Z 10 . . . . .. . . . E -

w
4;j
S4mmý= w
LE

C mho/m

0. 1 .. . .. .

10 4 105 100 10 7

FREQUENCY -- Hi

Figure 2-24 SHORT-CIRCUIT CURRENT AT TOP OF LOSSLESS VERTICAL ELEMENT

TERMINATED IN SOIL AT THE BASE

O.04

i@
N

0.04 30
0

0.02

001

002 24 fl I) 0 80
lIt p.1

Figure 2-25 CURRENT DELIVERED TO A MATCHED LEAD AT THE TOP OF A VERTICAL

RISER BY A VERTICALLY POLARIZED INCIDENT WAVE, Source: Ref. 7.

80
for sign. The direction of positive current in the vertical element is toward the top, while
the dhiuction of positive current assumed for the horizontal conductor in Section 2.2 is
toward the terminals. Also note that the total source impedance at the top of the vertical
element is the sum of the Impedance of the vertical element and that of tho horizontal line.
Thus, for a semi-infinite horizontal line, the total source Impedance is Zo + Z(h) between
terminals at the top of the vertical elements,

2.3.2 CURRENT AT BASE OF VERTICAL ELEMENT

The short-circuit current induced at the base of a vertical element by a vertically

polarized wave incident at an elevation angle 0 Is7 '8

Eocr__ e-_ --- 1 - e-w(h? -

,) joZo L s1sin 1+ sin I

(2-39)

F11W

for an exponential pulse Eoe't/T, In this expression, it is assumed that the ground behaves
as a perfect conductor, the vertical element Is terminated in its characteristic impedance at
th,. top (a matched or semi-infinite horizontal line), and the attenuation of the vertical
ulement is negligible. The meanings of the symbols are the same as in Eq. (2-32), and
to - (2h sin p)/c.

The waveform of the short-circuit current at the base of the vertical element is

ls(0,t) = 'I
osin
Zo(Oil l 1 + sin . t' > 0 (2-40)

lW -- tl)/r 1 - t'- t~/l

L 1--sin V1 > t L 1 e sin t' > t2

81
 whre't-
0/ -(hsn li/, ih to h to

whee t - t (h sin 01/c, t

tocwo2=t h t2c' and t2 h + to Time and phase are
2
referred to the top of the element, so that the time t and the phase in the frequency domain
used here are consistent with those used for the horizontal lines In Soction 2.2 and the
vertical element in Section 2.3.1. Plots of the short-circuit current are shown ill Figure 2-26
for an elevation angle of incidence i =300 and 0O(grazing incidence).
0/i

The source impedance observed from the base of the element Is the characteristic
impedance Zo, and the open-circuit voltage at the base is lee(O) Z4. It is noted that lS(O)
is the short-circuit current generated In the vertical element only; the component of current
or voltage Induced In the horizontal conductor as given in Section 2.2 must be ak.Jed to these
these components from the vertical element to obtain the total current or voltage, It Is also
noted that the current in the vertical element is positive when it flows toward the top of the
element (see Figure 2-23).

r1 o0e I " 1A
LUI - 10 m

0.04

S0,02

0.2 0,4 0. CU. 1.0

i',MPt
- -JIs

Figure 2-26 SHORT-CIRCUIT CURRENT INDUCED AT THE BASE OF A VERTICAL RISER

BY A VERTICALLY POLARIZED INCIDENT WAVE

82
 2.3.3 COMPARISON OF VOLTAGE IN VERTICAL AND HORIZONTAL
ELEMENTS

To illustrate the relative magnitudes of the voltages induced in the horizontal line
and the vertical element, the maximum open-circuit voltage induced in the semi-infinite
horizontdl line by a unit step incident field is

v(tC (2-41)

for a perfectly conducting ground and a vertically polarized wave incident at ,0

0c
The maximum open-circuit voltage induced at the base of the vertical element under these
conditions is

Voc (t) ,,-2h ct

h (0 < t ' h

(2-42)
- -2h

and it Is of the opposite pularity,

The total voltage at the base of the vertical element is, therefore,

VetctW = -2h ( c) (O<t<L•)

= 2h ( ('t h) (2-43)

The individual components and the total voltage waveforms are shown in Figure 2-27, The
voltage induced in the vertical element causes a negative response for a time h/c and reduces
the final value of the step-function response of the horizontal line by a factor of 2,0 when
the soil is a perfect conductor. If the soil Is a poor conductor, however, the peak voltage is
much greater than 4h (see Figure 2-14) but the peak voltage induced in thu vertical element
is changed only slightly. Therefore for very long horizontal conductors over average or
puorly conducting soil, the effect of the voltage induced in the vertical element is relatively
luss important.

83
 2.3.4 GENERAL ANALYSIS OF VERTICAL ELEMENTS

The cases given in Sections 2.3.1 and 2.3.2 are generally of widest use in analyzing
power systems, For certain special applications, however, more general formulas may be
required. The general solutions for the current and voltage at any distance x from the base,
from which Eqs. (2-32) and (2-39) were obtained, are

I[x) - [Kl + P(x)le"*h" + (K2 + Q(xl elfx (2-44a)

V(x) - Zo I(Ki + P(x)]e'rx - (K2 + Qlx)]e97 (2-44b)

L " 1,,-,TERMINALS

4 Ii -=ml
-- -==- HORIZONTAL
- -= - -r~m-- ~LINE
-- r -j m- -

3h

STOTAL
Z2h

h 2h 3h 4h 5h~ Sh 7h 8h
Ch C C C C C C

VERTICAL ELEMENT
-2h a nn- ~
Figure 2-27 OPEN-CIIRCUIT VOLTAGE INDUCED AT THE BASE OF VERTICAL ELEMENT
BY A UNIT STEP INCIDENT WAVE

84
where

(2-45a)
P(x) .-2ZJ E.tv)e'ydv
2Z!!

a(x) = 2- f- E, (v)e-VVdv (2-45b)

KI a po Ph P(h)e-7h - Q(O)e0h (2-46a)

e7h - Po Ph e0-7h

po0 (O) - P(h)

K2 = Ph eh (2-46b)
e~ Po Ph 'Th

Z(O) - Zo
o Z(0) (2-47a)

O Z(O) +Z0o

lh
- Z~h) - Zo (2-47b)
-Z(h) +Z

and where Z is the series impedance per unit length of the biconic transmission line, 'Yis Its
shunt admittan("J per unit length, Z(O) is the terminating impedance at the base (x - 0), and
Z(h) Is the terminating Impedance at the top (x - h). Sufficient accuracy for many engineer-
ing applications is obtained by letting
- 7 2h 2h
ZO -7 log -2h= 60 log 2 (2-48a)
27r a a

y - J- (2-48b)
C

85
where c Is the speed of light (3 X 108 m/s) a is the radius of the vertical element, and h is
its height. The current I(x) is positive when it flows in the positive x-direction, and the
voltage V(x) is positive when the vertical element is positive with respect to the ground plane.

The formulas above can be used to calculate the current or voltage at any point
0 < x < h for arbitrary terminal impedances Z(O) (see Section 2.3.5) at the base and Z(h) at
the top, and for arbitrary field Ex(x) incident on the vertical element. For a uniform
vertically polarized plane wave Incident at an elevation angle 0, the electric field is4,/8

2
Ex(x) * E1 cos • (1 + Rvy-i kx %in 0)e-ik(h-x) %in ' (2-49)

where El is the magnitude of the incident electric field, k = -,ýI'oo, and Ry Is the reflec-
tion factor at the air/earth interface given by Eq. (2-8b).

2.3,5 IMPEDANCE OF A GROUND ROD

The surge Impedance of a single, short ground rod of radius a driven into the ground
a depth Qas Illustrated in Figure 2-28 Is4

Z(O) 1 Fog NL26 - 1>u<e) (2-50)

2iroV -Loa (4 >> a)

where a is the soil conductivity, -'f . 1.781 .... and 6 is the skin depth in the soil given by

1 4irXlO 7 ) , (2-51)

The log term in the expression for Impedance is usually of the order of 10, as is illustrated
in Figure 2-29, so that the surge impedance is predominantly resistive and relatively
independent of frequency. This impedance formula is based on the transmission-line model
of the buried conductor. This impedance may be imsed for Z(O) in Section 2.3.4 for the
terminating impedance at the base of a vertical element.

86
 Figure 2-28 ILLUSTRATION OF A
VERTICAL GROUND ROD OF
CIRCULAR CROSS SECTION

20-m
77

10-2
10

pRIEOUENCY H 0
'102

rtADIIS,
FREQUENCY AND CONDUCTOR0 nsi
2-20 VARIATION OF LOG (-,2/-/Tha) WITH fet8 atro
Figure conduclor radiul rdcstylm
(Note that a factor of 10 in
conductivity.)
rj

23
The dc resistance of a gcind rod is
1 4F 1
R a log--
41 >>a) (2-52)
21ru Q a

Plots of the dc resistance of a ground rod as a function of length are given in Figure 2-30.

2.3.6 RESPONSE OF HORIZONTAL CONDUCTOR WITH VERTICAL RISERS

The effect of the vertical elements (risers) on the finite-length horizontal conductor
(see Figure 2-31) is to add additional length to the line at each end for either polarization,
and to also add additional current to the system for vertical polarization. The frequency-
domain solution is obtained by considering the risers as passive loads on the horizontal

iooo1
__ "II I ' "ili M l I- 1 1 1 l ll-7

10 70 0 0

1)011
Il•2b III

-
ZI o (1 3 b ----
U
0

1 .IJ. LLLLLLK.. L. 1IJILI ll.... iL_;_..I...LLI.I

U 10)
1100i

LENGTH OF ROD - m

Figure 2-30 VARIATION OF dc REFISTANCE OF VERTICAL GROUND ROD WITH

LENGTH Sour.e: Ref, 23.

88
 r!
SWAVEFRONT
/, 1,-1ki Cos ocos l

E0

h, el 'h,1

77*r7•7*7 77 * grPrr7r.fr p',7 7F7,077f

r0 f'r,,rr0(

Figure 2-31 TRANSMISSION LINE TERMINATED WITH VERTICAL RISERS AT BOTH ENDS

conductor, witii Impedances Z1 and Z 2 . For perfectly conducting ground, these imped-
ances are:

[•+ p~le- 2,h

Z3 Zo h (2-53a)

and
Z2-~ ~ -o1÷~
e2-yh1

Z- poe.
2
7J (2-53b)

where/p1 and pj are the reflection coefficients at the base of the risers. The reflection
coefficients p, and P 2 in Eqs, (2-5) are then given by

pIle-2"yh Sý (2-54s)

P2 = p 2e 2
" (2-54b)

89
 r|

2
Thus, in Eqs. (2-5), the term iJP 2e "1 becomes p p~e-2 1 (' 12h), When these quantities
are substituted into Eq. (2-3a), the expression for the horizontal conductor with passive
risers becomes: 8

101,011" Rhei1 2khinri

Ei sin 0 (1 +2Zo 2
1 -'Pie2"h

x 1 r,-ik acos 0
cos 0-2)•
+ Pl'jct 1 -•T
,le_.rV+h)l jct
*4j cos= cost°•p),, (2-55)

X 4 ppie-2,(t+2h) + (p),op 2 e"4 7(+ 2 h) ..+ ]

Note that the effect of these substitutions has been to Inject the 2,yh phase shifts at the
proper point to account for the round-trip propagatinn times of the vertical risers,

For a vertically polarized incident wave the current induced in the risers by the
vertical uomponent of the electric field must be superimposed on that induced In the
horizontal conductor by the horizontal component of the field. For an incident signal, the
current at the top of the riser at z - 0 can be shown to be:

EI cOs ý eikhin• I - Pe-2 7(h+V)

12 (h,O)=
2Ze+
0 1-(2h+] p-
'h

x+('+lk~inIh -1 Rv ekln)h (2-56)

7 +jksin
+ " y -jksin i

-1 R -l l++k s in )h i .
+ i e ( ' v,- lk lin )h

The form of the current Ii (h,-V) at the top of the riser at z -• is the same as in
Eq. (2-56),(with p' and pý interchanged), but it contains a phase factor elk vcos'coio to
account for the phase difference between the wave at (h,-Q) and the wave at (h,O). This
current must be transferred to the other end of the horizontal conductor before it is

90
combined with 12 (h,0) and I(h,0), given by Eqs. (2-55) and (2-56), respectively. The current
at (h,0), Induced in the riser at z = -k, will be

El cos 0 -Jkhslni elkeosýCoS• (1 -pie" 2 1h)

11 (h,0 2
2Zo e?(h+Q)= pepD
eP'he
-t
p je1"27
2)V
(2+ 1s
1

X (-f+jksln0i)h - .(.-.ksnh 1 (2-57)

-(7-jksln0lh 1 e-b7Jkhln0)h
+ y3- Jk si +Rv n 0
y + jk sin J "

The total current at (h,0) from the horizontal conductor and the two risers Is then

lt(h,0) - I(h,0) + Il(h,0) -. 12 (h,O) (2-58)

in which 12 (h,0) carries a negative sign because of the convention for the direction of
positive current In the risers, The current at (h,O) produced by each element, and the total
current caused by all three elements, are plotted In Figure 2-32 for an Incident step function
Eou(t) that Is vertically polarized. The earlier conditions that y-s jk, p•1 0 P2 - -1,- 0
300, Rv 1, and t - 2h are assumed, and only the first cycle of the response Is plotted.
-
From Figure 2-32 It Isapparent that the current Induced in the risers Is considerably greater
than that Induced In the horizontal conductor, and even though the currents oppose each
other, the net current is considerably larger than that Induced In the horizontal conductor
by the vertically polarized wave. Notice that tho vertical scale In Figure 2-32(a) Is twice as
large as the scale used in Figures 2-32(b), (c), and (d).

2,3.7 PERIODICALLY GROUNDED LINE

Consider a periodically grounded line as shown In Figure 2-33(a) with a height h and
a spacing between vertical ground leads of k. The line extends to Infinity at the left, and we
will assume that y and Zo are the same for the vertical leads and the horizontal conductors.

91
 2~~~ v h••

'N~l•°•0 •VERTICAL
1 POLARIZATION

8
S2
--
3 4 5 a O
7 a
TIME - hic
IW) CURREINT INDUCED IN HORIZONTAL. LINE

N1 2

a
0 1 2 3 4 5 6 7 8
TIME --. h/c
4 ib) CURRENT INDUCED IN RIOER AL
AT 0L0

6
0

4 0 1 2 3 4 0 S 7 8
TIME -- h/c
Ib) CURRENT INDUCED IN RISER AT a= 0

IC)

N W2

L I 2 3 4 5
TIME --- hic
7
7 a

(d) TOTAL CURRENT'

Figure 2-32 CURRENT COMPONENTS INDUCED ON HORIZONTAL LINE AND VERTICAL

RISERS, AND TOTAL CURRENT AT x i, :: V, Source: Ref, B.

92
 TERMINALS

(a) PERIODICALLY GROUNDED LINE

(b) ONE SEGMENT

fal EQUIVALENT LOADS FOR ONE SEGMENT

1
'Li U r1

(dl FINAL SEGMENT WITH EQUIVALENT SOURCES

Figure 2-33 STEPS IN THE ANALYSIS OF A PERIODICALLY GROUNDED TRIANSMISSION

LINE

93
We wsh to determine the Norton equivalent current source represented by the semi-infinite
periodically grounded line to the left of the terminals.

In Figure 2-33(b) the right-hand loop formed by the last two ground leads and the
horizontal conductor between them is redrawn, and in Figure 2-33(c) the pertinent admit-
tance and transmission-line lengths required to obtain the source admittance are shown. The
source admittance looking into the terminals In Figure 2-33(c) is Yr in series with YL, where
Yr is the input admittance at the ground lead and YL is the input admittance of the trans-
mission line of length 1 terminated in Yr + YL, If Yo - /Zo is the characteristic admittance
of the vertical and horizontal conductors (with ground as the second conductor in each case),
the Impedance Yr of the ground lead is

- p ge "2,Yh
S1
YR 1-- - (2-59)
YO 1 + pge-2lh

where p9 is the reflection coefficient at the base of the ground lead given by

ZO - Zo (2O)
P9 Z9 + Zo

where Zg is the impedance of the ground connection at the base of the ground lead.

The admittance YR at the terminals Is given by

YQ 1 - pe"2 y6
YL (2-61)
•Y0 1 + pe"2•y

where p is the reflection coefficient at the left end of the circuit of Figure 2-33(c), It is
given by

YO - (Y + Yr) 1 - (YL + YR)

p a -" (2-62)
YO + (YQ + Yr) 1 + (YL + YR)

94
Substituting this value into the expression for YL gives

1 -- e-2-!V + 11 +e-27y) (YL+YR) 1263

42-63)
YL=
1 + e"27V + (1-e-2Y') (YL+YR)

which can be solved for YL to give

YR T /1+4 11+ e-2VQl

16I (2-84)
2 NR 2 YR 1 .- Qf

The positive sign Is evidently the correct choice If realizable admittances are to be obtained.
The total source admittance at the terminals is then

YLYR, Y
Yn 0 SYL + YR -YO 1 (2-65)

Consider now the circuit of Figure 2-33(d) which shows the equivalent sources of
the last two ground wires Ir and Irl and of the transmission line to the left of the second
around wire ILI- Suppose we replace the ground wire on the right by a perfect short circuit.
Then the short-circuit current at the terminals will consist of the curfent delivered by the
sources I LI and Irl through the length Rof line plus the short-circuit current IVinduced by
the incident wave in the length 2 of horizontal line. The current delivered to the left end of
the circuit by the sources ILi and Irl is

V1
1(-R) - (ILI + Irl) (2-66)
YR + Yr + Yi

where Yi is the input admittance ot the transmission line of length Qthat is short-circuited
at its terminals. That is,

1 + e-2-
YV Yo (2-67)
1 -e27

95
The short-circuit current IL at the terminals is

2e-•rY

IL (-4) + IV (2-68)
1+ e-27

which is obtained by transforming the current I(-2) to the right end of the line and adding to
it the current I induced in the line by the Incident wave. Now because the line Issemi-
infinite in length, the current IL differs from the current ILI only by the phase of the Inci-
dent wave arriving at the terminals relative to its phase at the next-to-the-lest ground wire.
Similarly, Ir and Irl differ only by this phase. Thus,

Jr " Ire7' - Ire&fR co' o

001
(2-69)
ILI ILe•f' ii

and
YJ
1 (IL + Irle~f' (2-70)
SY• + Yr + :,

Substituting this value Into the above expression for I L gives

IL - OL + In) - V +l (2-71)
Y +Yr +Yi 1 +e-2-f

which can be solved for IL to obtain

Y1 2e- (R-
l2+lr yr+yi 1+e-2-yR
Yi 2e--Y(R- R') (2-72)

Y + Yr + YI 1 + e-27ý

96
 If we now recall that the right-hand ground lead has finite admittance Yr and a
source Ir, we can combine these with IL and its source admittance Yv to obtain the short-
circuit Ise at the terminals:

ILYr - Ir YQ
I= IL~r(2-73)
YV + Yr

The Norton equivalent source that replaces the periodically grounded line to the left of the
terminals is then the cu~rent Is. of Eq. (2-73) with a shunt admittance Y given by Eq.
(2-65). The current sources IVand Ir are obtained from the analysis of horizontal and
vertical conductors of finite length. The current It is the short-circuit current Induced in a
horizontal line of length Qterminated in a short circuit at the right end (z = 0) and
terminated in the impedance 1/(YV + Yr) at its left end (z - .). l•I can be obtained from
Eq. (2-18) by multiplying Voc(w) by YQ. The current Ir is the short-circuit current at the
top of a vertical conductor of height h that is terminated In an impedance ZO at its base
(x = 0). The current Ir is given by Eqs. (2-38) and (2-32) when Z(h) - 1/Yr. The Impedance
Zg may be calculated from Eq. (2-50).

The magnitude of the open-circuit voltage Induced in one horizontal segment Is

shown In Figure 2-34 for 1 V/m incident at 300 elevation and 0' azimuth angles. Note that
at low frequencies, the open-circuit voltage is roughly 2Eoh, if Eo is the magnitude of the
incident field. Peaks and valleys associated with the resonant lengths and with the line
height are also evident.

The short-circuit current and source admittance at the end of the semi-infinite,
periodically grounded line are shown in Figure 2-35. Again the periodicities of the segment
length and line height are evident, but these are commingled in such a way that it Is more
difficult to relate cause and effect, It is apparent, however, that the source admittance is
between a few tenths of a mho and a few mhos. Therefore, the periodically grounded line
does have a low source impedance compared to the ungrounded line. In addition, a com-
parison of the magnitude of the short-circuit current with that of an ungrounded line
(e.g., the open-circuit voltage of Figure 2-12 divided by a characteristic impedance of a
few hundred ohms) indicates that the mean currents re of similar magnitudes to within n
factur ol 2 or 3, The peak short-circuit currents of the periodically grounded line are

97
 . .
104 ",-

Iv

103r
>

10
W V 1W)

10 102 mho/m

106 106 107

104
FRECUENCY -- He

Figure 2-34 OPEN-CIRCUIT VOLTAGE AT TERMINALS OF FINITE HORIZONTAL

TRANSMISSION LINE

considerably larger than the mean, however, so that selected frequencies ;n the pulse spectrum
will be more strongly coupled to the periodically grounded line, :1

2.4 TRANSMISSION PROPERTIES OF POWER LINES

2.4,1 ATTENUATION AND PHASE CHARACTERISTICS

The propagation of a signal along a wire over earth of finite rnnductlvlty has been
analyzed by Sunde. 2 3 The propagation constant y is given by

t= Jk H(jI) - a + JO (2-74)

98
 E to_ 1T00

I-

10"2

(a) SHORT-CIRCUIT CURRENT AT THE TERMINALS

¶ cIJ Y.......*
7

10

(b) SOURCE ADMITTANCE

o,L itJ I I..I - _._..I._.I _
104 t 05 106 107
FRE0UENCY ... A-

Figure 2-35 FREQUENCY-DOMAIN RESPONSE OF A PERIODICALLY GROUNDED LINE

99

!. ..........
where k = w./c is the free-space propagation constant, c is the velocity of light in vacuum,
and H(jow) is a function that contains the earth and the line parameters. "The function H(jW)
is given by

wg-
H(JoJ log 1h-(iW7)11 1/2
I~)
+(2-75) ~ 1/2 (-5
) [1 + (lo3)'(oga(1wr.h) 1/2 (jWTa)11!2

For most cases of interest the second term on the right is small compared to I so that

2h /1 + Ujwrh) 1 /2 \
H(jw) 1 + 1/2 (log-) a 1
log + )
12-76)
a \ (jwi.h) 1 /2 (I•T 1 ) 1/2

2 2
where Th -eh -ola

and h Is the line height over ground, a Is the tine radius, MIs the permeability, a is the soil
conductivity, and oil Is the wire conductivity. The terms VJW/ and VjW• may be
recognized as complex and related to the line height h and conductor radius a normalized to
the appropriate skin depth, since

N(iwrh 0( +J), 6 - soil Isk Indepth -(2-77a)

62

V1 U'• (1 + 1) bi - wire skin depth • -1/2 (2-77b)

The values of rl are plotted as a function of soil conductivity and line height in Figure 2-36.

If the complete expression for -yis manipulated to extract the attenuation and phase
constants separately, rather simple expi'essions result when the wire loss is neglected:

[ an 1 + 2h] (2-78a)
k ltog j~h

100
 S1 + lg (2-78b)

k 2 log 2h
a

The attenuation constant a calculated from Eq. (2-78a) Is shown in Figure 2-37 as
a function of frequency for various values of the time constant "h = /o ah 2 , and the normal-
Ized attenuation constant and phase constant are shown In Figure 2-38. For a typical soil
conductivity (a, 10-2 mho/m) and line height (h - 10i) the value of r' I. about 1O-

(see Figure 2-36), so that 3/k and o/k

6-4 .. are not strongly dependent on frequency.

The asymptotic behavior of .

when h/5 Is very small and very large
000" compared to 1 illustrates how the lIIne
and soil parameters affect the early- and
late-time solutions, When h/6 << 1, the

00 .00inverse tangent term reaches aconstant

op •.value of ir/4, Therefore

kir
a- - (2-79)
8 log

for late times (low frequencies), For this

condition, a is independent of a and only
1o-9 weakly dependent on h through the log
2h/a term; a Is linearly dependent on
frequency.
1010-1
o0 1o03 10"2 10"1 When h/6 >> 1, the Inverse tan-
SOIL CONDUCTIVITY, 0- rnho/m gent term reaches 1/(2h/6) so that

Filgur 2-36 TIME CONSTANT rh AS A k

FUNCTION OF SOIL CONDUCTIVITY a - (2-80)
AND LINE HEIGHT h 4h/6 log-2h
a

101
 1
10
101 f
T-Y-FTTrFtl 1 TmIIl I-l•

,0oo
1 /

+<-. ... 0 . ,-- _

05108 i07 to0

FREQUENCY - Hi

FIlgUr 2-37 ATTENUATION CONSTANT a AS A FUNCTION OF FREQUENCY FOR

VARIOUS LINE HEIGHTS AND SOIL CONOUCTIVITIES

1 1
for early times (hlgh frequencies), Under these conditions a varies with tr- and with1 f .
In addition, is inversely proportional to h when the variation of lg2/ sneglected,

I A good analytic approximation for all frequencies is

8,2 X 10"-
.. .. . nepers/m. (2-81

102
 1.3

1-
_ 1.2
k

10-8

101

ilos

10103
 The step-function response of a line has also been calculated by Sunde, Although
there is no closed-form solution for all times, an analytic solution may he found for both
early times and late times, where the distinction between early and late times depends on
the value of h/S. If h/6 >-> 1, the early-time solution is appropriate; if n1/ << 1, the late-
time solution is appropriate.

For early times, the current at a point z down the line normalized to the current at
z 0 is

Il(z,W) F 2cvj-1\ log/ 2h1 (2-82)

lIexp -- ZIl expl'Z
IN, Vz a
1(0,W) c lg

where

For most cases, r. >> h so that r, - T h, The transform of this expression for
current when a step function of current Is introduced at z - 0 is

I-( f erfc 4vf-"1 c tz1/2 log (2-83)

I(0,t+) oZa

for early times, where erfc (z) = complementary error function of z, and tz t - z/c,

For late times the normalized current in the frequency domain is

I (z,W) - = )X W• [Ip+ 2h log Jjw rh) 2~~ (-4

12-84)
1(0, W) 4c log a2.r"

104
The time-domain solution with stop-function input is

1 +2 tz1/2
I(z,t)
•-- 1 -z -
I(O,t+) 2h (2-85)
4c tz log-a-

for late times.

For many cases 2(tz/rrr,) 1/2 << 1, which further simplifies the solution. Note that
for late times the response Is independent of the value of soil conductivity and only weakly
dependent on the height above ground through the logarithmic term in h/a. This contrasts
with the early-time solution, which is a function of both a and h.

Figure 2-39 shows the wave front of a unitstep function after it has traveled a dis-
tance of 10 km (6.2 miles) along conductors of 1 mm and 1 cm radii, when the earth
conductivity is 10-2 mho/m.

l.
190.9
CALCULATED
FROM
1C06 FORMULAS FOR
SMALL TIMES
0
0.4

S?// CALCULATED FROM

0.2 - FORMULAS FOR
LARGE TIMES

01 10 100
TIME -js

Figure 2-39 WAVE FRONT OF UNIT STEP CURRENT AFTER PROPAGATION FOR A
DISTANCE OF 10 km (6,2 miles) ALONG COPPER CONDUCTORS OF 1 mm AND 1 cm
RADIUS, Soil conductivity - 10-2 mho/m, Height on conductor, h - 10 m. Source:
Ref. 23,

105
 Figure 2-40 shows the wave front for various soil conductivities for a conductor of
1 mm radius. The curves show that, for perfectly conducting soil, the wave-front distortion
due to the conductor resistance is small compared to that resulting from the finite con-
ductivity of the soil,

2.4.2 JUNCTIONS IN TRANSMISSION LINES

When a transmission line branches into two lines or when a spur line is connected to
a main line, an obvious discontinuity in high-frequency characteristics of the line Is formed.
The feed line and the two branches may be treated as transmission lines having individual
characteristic Impedances (which may be equal If the line heights and effective radii are the
same) and propagation factors. Near the junction, however, there will be mutual coupling
among the three lines. The treatment of transmlssion-line currents and voltages in the
vicinity of branches or junctions has been analyzed for junctions forming tees and crosses by
King 2 4 for the case where the line height is small compared to a wavelength. No theoreticul
analysis is available for the power-line EMP case where line height is greater than one wave-
length; however, experimental determination of the reflection and transmission characteristics
7
of symmetrical junctions have been made.

1.0

ud 0-6

0,O

TIME -

Fpiure 2-40 WAVE FRONT OF UNIT STEP CURRENT AFTER PROPAGATION FOR A
DISTANCE OF 10 km (6,2 mIles) ALONG A COPPER CONDUCTOR OF 1 mm RADIUS
AT A HEIGHT OF 10 m, FOR VARIOUS SOIL CCNDUCTIVITIES. Source: Ref. 23.

0. 106 1 ,- 6-3
 For a symmetrical wye junction, such as that illustrated in Figure 2-41, of a long
feed line with two long branch lines, the branch lines can be considcred as two loads of
impedance Zo in parallel at the end of the feed line if mutual coupling is neglected. The
current reflection coefficient and current transmission :oefficient are then

Iref Zo - ZL

linc Zo + ZL
(2-86a)
- (ZL - Zo/2)
3

I., Itrans 2Zo

--

hinc Zo+
(2-86b)
4
- (ZL Zo/2)
3

hh

2at

0h

h
Figure 2-41 SYMMETRICAL JUNCTION IN A TRANSMISSION LINE

107

. ..... ...
where ri indicates the total current transmitted to both branches. The characteristic imped-
ance is, approximately,

Z. - 71 log 2
21r a

where h is the line height and a is the effective radius of the line for common-mode (zero
sequence) propagation. The load impedance ZL represented by the two branch lines in
parallel is Zo/2 if the lines are very long (in comparison to the pulsewidth cr) or are termin-
ated in their characteristic !-,,adance. For the symmetrical junction, therefore, 2/3 of the
incident current is transmitted to each branch line. If h and a are different for each line, the
load impedance will be ZL - Zol Z 2 /1Z4 1 + Zo 2 ), where Zoj and Zo2 are the characteristic
impedances of the individual branch lines, and tl,e total transmitted current will be divided
between the branch lines accordingly.

The effect of mutual coupling between the branch lines of a symmetrical wye
junction can be estimated from the characteristic impedance of two wires over ground given
DV21

ZO log
/ 2h\2\1/2 (h>>a) (2-87
4vr a d

where h is the height of the lines, a is their effective radius, and d is the spacing between
conductors (see Figure 2-42). Note that when the lines are very far apart (d >> 2h), the
common-mode impedance of the two-wire line is half that of Fn isolated single-wire line,
whereas when the lines are very close together (d -a a), the common-mode impedance
approaches that of isolated single-wire line.

The reflected and transmitted signals have been measured at symmetrical junctions
on scale models, 7 The line configuration and results are shown in Figure 2-43. The line
height used was 40 ft, or six times the equivalent length of the rise time (Leoci C-rW).

If two lines forming a symmetrical junction are considered as two lines of character-
istic impedance Zo in parallel and fed by a single input line of Zo, then the reflection coeffi-
cient would be 0,33 and the transmission cuefficlent would be 0.687 at the junction. These

108
 values are indicated in Figure 2-43. The
experimental results show that this model
accurately reflects the interaction at the
junction when the junction angle is large.
However, when the angle is small, the junc-
tion lines are more tightly coupled to each
other, making the net impedance closer to
Zo. This reduces the reflection and trans-
Figure 2-42 TWO-CONDUCTOrf TRANS- mission coefficients. The reflection coeffl-
MISSION LINE OVER GROUNr) PLANE
(common-mods excitation) cdent is shown In Figure 2-43 for early
times (-20 ns) and for late times ('-170 ns).
At early times the reflection coefficient is much lower than would be expected from the
uncoupled branch lines of characteristic impedance Zo. In the limit as 0 - 0, the value of
Zo beyond the junction almost equals the value of Zo for the input line, resulting in only a
very small reflection. This Is shown as the dashed portion of the early-time curve from
0 - 00 to 0 - 22.50, It is apparent, however, that the assumption that the brinch lines
behave as Independent isolated lines gives a good approximation to the reflected and trans-
mitted currents if the branch angle is of the order of 90* or greater.

2.4.3 BENDS IN POWER LINES

The signals coupled to the power lines will not always propagate straight to the load,
In this section, theoretical and experimental results for propagation of signals around bends
are presented, Analysis Is usually restricted to the case where the line height is small com-
pared to the wavelength. However, experimental measurements have been made to see if
significant deviations from theory could be found when the line height was comparable to
the wavelength.

King has treated a simple bend in a two-wire line when 3h << 1 using the geometry
shown in Figure 2-.44,24 His analysis shows that the bend has the effect of lowering the

109
 0.4

0.3

0.2 0

010

REFLECTION COEFFICIENT

SI ... I 1 I

30.7 -- . Zlo. "Ud zo .

0.6

0,15I
TRANSMISuION COEFFICIENT 0

I , J. 'InnI
0 20 40 60 80 100 120 140 160 180
JUNCTION ANG LE, 0 -- dog

Figure 2-43 REFLECTION AND TRANSMISSION COEFFICIENT AT A SYMMETRICAL

JUNCTION, WHERE h - 40 ft. Source: Ref, 7.

inductance L(w) and capacitance C(w) per unit length of line in the region arourld the bend,
The value: of L(w) and C(w) are

Po
L(w) - - (K - F1 (w) + F2 (w)cOs 0) (2-88a)
21r

C(w) - 2re (2-88b)

K - Fl (w) + F 2 (w)

where
2h
K - 2 log-2 (2-8g)
a

110

Omn
 Fj (w) = log w + V/w2 +'a2_ (2-90a)

F 2 (w) . logC + f7T2hý) (2-90b)

w cos 0 + /wl + a7

(2a < 2h << c/f)

and w is the distance from the bend.

Because the characteristic

impedance Zo -VC'W)/C(W) and
the propagation factor y - /L(w) C(w)
-0-___-__- depend on the inductance and capa-
b citance per unit length, both are
-- )affected by the changes in Land C
2 in the vicinity of a bend.
Power
lines do not meet the criterion that
2h << c/f for all frequencies of
Figure 2-44 TWO-WIRE LINE WITH A BEND interest in the EMP spectrum; how-
ever, measurements were made of
the peak reflection coefficient using scale models of power-line configurations and scaled
exponential pulses, The peak current reflection coefficient for a line 20 ft high and 0.2 Inch
in diameter with a 6-na-rise-time pulse is shown in Figure 2-45 as a function of bend
angles. 7 The measured peak reflection coefficient computed using the low-frequency
capacitance and inductance varies with bend angle in Figure 2-45. The experimental data
indicate reflection coefficients slightly larger than those predicted using the static theory, but
for bend angles less than 90 the reflection Is small in either case.

The perturbance due to the bend is time-dependent. The values shown In Figure
2-45 are peak values. Generally the effect of bend has relaxed within a few tens of ns (or
equivalently, a few line heights). This result Is consistent with King's analysis, which shows
that the effects on L(w) and C(w) are strong only near the junction. At later times - or
longer wavelengths - the strong effects near the junction are smoothed, since perturbances
that extend over only a small fraction of a wavelength cannot be resolved.
111
 I-
Z 0.20

L'"
8- oJ.15 -- a- 20fti
Z a - 0.2 In,

i•: 9• MEASiURED

S0.10
U.

'U05,-- CALCULATED

0.

0 20 40 60 so 100 120 140 160

SEND ANGLE, 0 .- do

Figure 2-45 COMPARISON OF CALCULATED AND MEASURED REFLECTION COEFFICIENTS

OF A BEND AS A FUNCTION OF BEND ANGLE. Source, Ref, 7.

Scale-model measurements have been made for a single wire line 20 ft and 40 ft above
ground for a range of bend angles from 00 to 157.5g. A step-function signal with rise time
of about 6 ns was fed onto the line. The reflection and transmission coefficients for these
two line heights are shown In Figure 2-46, where it is apparent that the effect of line height
is quitte small,

The effect of radius of curvature on the reflection coefficient was also briefly inves-
tigated using scale models, 7 The results are shown in Figure 2-47. The reflection coefficient
decreases monotonically as the radius of curvature is increased.

2.4.4 OTHER EFFECTS

There are many deviations from the circular cylinder over a plane, uniformly
conducting ground assumed in the analysis of coupling to, and propagation along,

112
 I 1 I I I I , /
0.25
REFLECTION COEFFICIENT
W1HF HEIGHT 20 .t

0.20.
WIM HF.IGHT 40 fIt

0.10 1

0.05

11-0 I WIREI
I I4 I .....

il, 00 W~EHEIGHT 20 I

0,90
0 20 40 60 80 100 120 140 100
ANGLE. OF BEND, 0- doe

Figure 2-46 REFLECTION AND TRANSMISSION COEFFICIENT OF A BEND IN A

SINGLE-WIRE LINE OVER GROUND AS A FUNCTION OF BEND ANGLE, Source:
Ref. 7,

power lines. These deviations may be classified as scatterers or variations in line

height:

(1) Scatterers

0 Poles or towers

* Trees and other ground foliage

"Buildings, bridges, and other structures

* Cliffs, ravines, and mountains iwith lines in shadows)

113
 008 I 1

WINE HPIGHT 20 ft
90 BEND

* 004

0.02

080 180 240 320 400 480

RADIUS OF CURVATURE, H c.

PFure 2-47 REFLECTION COEFFICIENT FOR A 9O0BEND AS A FUNCTION OF RADIUS

OF CURVATURE, Source: Ref. 7,

(2) Variations In line height

* Line sag between poles

* Undulating terrain
b Aerial ground wires

* Communications cables on shared poles.

The effect of such deviations from the abstract model Is almost always to reduce the signal
delivered to the power consumer's terminals. The scatterers reduce the excitation field at the
power conductors, and the variation In line height produces multiple reflections of the current
and voltage propagating along the line so that line losses are larger than those predicted for a
uniform transmission line. One can conceive of resonances in the power line matching
resonances in the consumer's low-voltage circuits to produce an enhanced overall EMP
response; however, it is expected that the occurrence of such a situation In practice would be
rather rare, particularly In view of the fact that resonances in wire-over-soil transmission
lines tend to be rather weak.

114
 An indication of the effect of variations in line height can be obtained from the vari-
ation in the characteristic impedance caused by line sag shown in Figure 2-48, For this
example a line 10 m high at the crossarms sags to 5 m high midway between poles 100 m
apart, The average height of the line is 6.56 m, ot which height the characteristic impedance
is300 ohms. The characteristic impedance varies from 325 ohms at the poles to 284 ohms
at the midpoint. The reflection coefficient associated with these extremes is only 0.07;
therefore the effect of line height variations is relatively small unless the Induced signal prop-
agates over great distances,

9.0

610

S4.0

2.0

3301 T 1 1
ib) CHARACTERISTIC IMPKIDANCEI
a -0.019m

z
S310

300
w

290

n 10 7'0 30 40 50 60 70 so 90 100

DISTANCE -r

Figure 2-48 VARIATION IN CHARACTERISTIC IMPEDANCE CAUSED B3Y LINE SAG

 2.5 DIFFERENTIAL COUPLING TO HORIZONTAL CONDUCTORS

2.5.1 WIRES IN A HORIZONTAL PLANE

For two conductors separated by a distance d and at a height h above ground as

illustrated in Figure 2-49 the differential-mode current and voltage induced by a plane wave
can be determined by an analysis very similar to that used for the common-mode current
and voltage. 2 6 The differential equations of the current and voltage are

8 + ZI - E" (h,d,z) - Eu (h1,O,z) + - (2-Ola)

az dt •

+ YV YVu (2-91b)
az

where Ell (x,y,z) is the component of the resultant electric field parallel to the wire and at
the wire location but in the absence of the wire. Z and Y are the impedance and admittance
per unit length of the two-wire line, and Vu is given by

d
Vu - / F.y(h,y,z)dy (2-92)
0

The second-order differential equations are

Sd2l
1
- 1 =- Y[Eu (h,d,z) - El (h,O,z)] (2-93a)

d2V
- - f 2 V = -YZ1 Vu (2-93b)
dz16

li
 where-y - .,/ZY and Zi is the internal
,v- , , impedance of the conductors (if
finitely conducting wires are as-
- ,sumed). The form of the field
y E' (h,O,z) is identical to that given
dd in Eq. (2-7) for the common-mode
"analysis. The field EU(h,d,z) differs
Figure 2-49 GEOMETRY OF TWO-WIRE TRANS- from Eu (h,O,z) by a phase term:
MISSION LINE FOR DIFFERENTIAL-COUPLING
ANALYSIS

Eu (h,dz) (h,O,z) ejkd'

EuE (2-94)

where d' = d sin 0 cos 4. The driving field in brackets In Eq. (2-93a) can thus be expressed
as

EU(h,d,z) . EU(h,O,z) - EI sin p e.Jkz' (1 + Re-jk2h') (eikd'- 1) (2-95)

where the upper trigonometric function (sin 0) applies to horizontal polarization and the
lower applies to vertical polarization, The primed quantities are z' - z cos k cos 'P,h' -Mh
sin p, and d' - d sin p cos ý!i, and n is the appropriate reflection factor for horizontal or
vertical polarization. The solutions for the differential current are, therefore, identical to
those for the common-mode current given in Section 2.2.2 multiplied by the term
(ei d -1) if Zo and -f for the two-wire line are used instead of the corresponding quantities

for the single wire over ground. If ground effects and losses in the wire are neglected, Zo
end -f are

Z
t d _
Z - log-- " j-
= (2-96)
7r a c

for the two-wire line.

117
 Separating the current into the part that would exist for a perfectly reflecting
ground and the part caused by a finitely conducting soil, the open-circuit voltage at the
terminals of a semi-infinite line due to Eu is, in the frequency domain,

_I (perfect ground)
V.,(w) - VU(w) - Eic D(i,O) - (eJ-.'/c-1) (2-97a)

ho2h'/c ((grnund effect

)-
E cD(ý,O)2,/';(sin (eJwd'/c-1) only)
A(V-Vu) Eic
__(2-97b)

when all attenuation factors are negligible. The + exponent applies to horizontal polarization,
the - to vertical polarization, and D(i,tp) and r. are defined in Section 2.2.3. The spectral
magnitudes shown In FioUres 2-12 and 2-13 are thus applicable to the differential-mode
voltages also if the spectra are multiplied by e -1,&d'1C~j,
which can be written

eiwd'/c- 1 = 2Jel~d'/2c sin wd'/2c - 2 sin uAd'/2c /d'/2c +- (298)

/W 2

The open-circuit voltage Vu caused by the y-component of the field Is given by

Eq. (2-92) with

Eu(h~yIz) - EiJCOS ýP e-Jkt' (1 + Re-Jk 2h') elky cos ' sin 4 (2-99)
EV(' z) E i-sin ' sinl J

where the upper trigonometric function applies to horizontal polarization and the lower
applies to vertical polarization of the incident field. The open-circuit voltage V1 (w) pro-
duced by the y-component of the field Is then

-e.- jw2h'Ic (perfect ground)

VU(Vl) - -Eic Du(O,o) 71j•d/cj (2-1 00a)

AV -Ec 2
=(w) (sin
Feijw 2
h'/c (elwOd/c- 1 ) (ground effect
only)
____
_-~___ ___ ___ ___ ___ ___ (2-1 0b)

118
where the + exponent applies to horizontal polarization, the - exponent applies to verti'al
polarization, and

Cos 0o
Du(ý,ýp) = (horizontal polarization)
coo 0 sin •p

sin ýun
si
cos 0, sin # (vertical polarization) . (2.101)

The appaient poles In Du (Ovip) at , rr/2 and p 0 do not affect the solution because

i ewd/c(2-102)
2-• cos 0 sin •o=jodc(212
•-0

Note that the form of VU(w) is the same as that of V(ow) - VU(w) except for Du(Oip), so
that the spectra and waveforms for Vu and V - Vu are similar.

Because eiwd'/c and 1 transform into time delays in the time domain, the time-
domain responses are also similar to those obtained for the common-mode voltages. If
vc(t) is the common-mode voltage, the differential-mode voltage due to Eu 1,

v(t) - vU(t) - vc(t + d'/c) (-d'/c < t < 0)

= Vc(t +d'/c)- Vc(t) (t >=

0l (2-103)

and the total voltage caused by the z and y components of the Incident field Is

v(t) - (1 - Du!D) vc(t + d'/c) (-d'/c c t < 0)

(2-1040
- (1 -Du/0) [vc(t + d'/c). Vc(t)] It->4
t )

119
When ce = 0 and y = jk, the -coefficients (1 - Du/D) become

(1 - IDU/D) =- s - C ' (horizontal polarization)

sin 2 cos i

(2-105)
cos i• cos ~, (vertical polarization)

Because the term eIwd'/c-1 places a "window" on the leading edge of the voltage
waveform vc (t), and because the voltage vc (t) continues to increase long after the ground-
reflected wave arrives at the wires, the maximum differential voltage depends not only on
the directivity function (1 - Du/D), but also on the width d'/c of the window. For linearly
increasing voltage, which is a good approximation to the early-time behavior of the voltage
and current, the maximum voltage will be obtained when d'(1 - Du/D) Is maximum. This
product is

d'(1 - Du/D) -d co t -cos p (horizontal polarization)

pi (2-.106)

- d tan p (vertical polarization)

The maximum open-circuit voltage will therefore occur when ýp -* 0 for horizontal polariza-
tion and when p -- ir/2 for vertical polarization. (Attenuation of the line will prevent the
product from going to infinity as •p -* 0, ir/2).

A typical waveform for the open-circuit differential voltage at the end of a semi-
infinite two-wire line 10 m high with 2--m spacing between the wires is shown in Figure
2-50. The waveform is chown for a horizontally polarized incident exponential pulse with
perfectly conducting ground (dashed curve) and for ground of average conductivity (10-2
mro/m) (solid curve). From a comparison of the differential voltage waveform of Figure
2-50 with the cot mon-mode waveform of Figure 2-14 (for o - 10-2, horizontal polariza-
tion) it Is seen that the peak voltage in the differential mode is only about one-seventh the
peak cornmon-mode voltage, and the duration of differential-mode voltage pulse Is only
about 30 ns, compared with a duration of a tew microseconds for the common-mode
voltage.

120
 3 EoD HORIZONTAL POLARIZATION
U
| ~El Eoetr

U 300 0-90 0
- h-t0m d-2m
I AMs D I*1
I
"" GROUND REFLECTION
S2
goo ARRIVES

•
,I- .2

"0 I

0.02 0.04 0105 0,08

d TIME -gs
"L C

Figure 2-0 OPEN-CIRCUIT, DIFFERENTIAL VOLTAGE INDU(ED IN A SEMI-INFINITE

TWO-WIRE LINE BY AN EXPONENTIAL PULSE (wires in a horizontal plano)

For a three-wire line with the wires equally spaced in a horizontal plane, a similar
voltage waveform would be obtained between the middle wire and the third wire at y - -d.
This voltage would be of opposite polarity and shifted by d'/c to the right of t a 0, however.
The voltage between the two outer lines would thus be approximately twice as large as the
voltage between the middle wire and either outer wire, and the time to reach the peak value
would be 2d'/c.

2.5.2 WIRES IN A VERTICAL PLANE

If the two-wire line lies in a vertical plane as illustrated in Figure 2-51, the analysis
Is quite similar to that for the wires in a horizontal plane except that the form of Vu and Eu

121
 ' d

-t

a az
Vh

Figure 2-51 TWO-WIRE LINE LYING IN A

VERTICAL PLANE

are different. The voltage Vu exists only for vertical polarization If the wires are In a vertical
plane, and its value is,

h+d
Vu J Ex(x,O,z) dx

r 12-107)
(eJiwd$/C-1) 1.1 + Rye=•(2h+d')/c
= -Ejc cotP

where d' - d sin , and h' - h sin 4,

The horizontal electric field terms on the right side of Eq. (2-95) are replaced by

Ez(h+d,Oz) - Ez(h,O,z) E cs~din

p eikl'(eikd'-)
(2-108)
X I- Re-ik(2h'+d')]

122
where the upper trigonometric function (sin p) applies to horizontal polarization and the
lower function applies to vertical polarization. When this expression for the field Is used and
the solution to the differential equations is separated into that part obtained with a perfectly
conducting ground and that part caused by the ground alone, the open-circuit differential
voltage for a semi-inflnite line is

1 + e-J(2h+d')/c (perfect
V.w) - V() EcD( -d'/ 1 ground)
(2-109a)
.-
e-jO(2h'+d')/c (ground
effect
A(V - Vu) - -EicD(O,p) (1 + R) - (aJ..d'/C) offy)
TW_ only)
(2- 109b)

where

1+R sin I(vertical polarization)

22uin v/Fw (horizontal polarization)

The expression for Vu for the differential voltage produced by the vertical component of the
incident field can be similarly separated to give (for vertical polarization only)

- -Eic cot - 9- (2h'-d')/C (perfect ground)

S(2-1 lOa)

e-JI(2h'+d')/c d' (ground

,Vu - -Elc cot 0 (0 + R)-- (eic /c-1) effect only)
__ (2-1!Ob)

where 1 + RY ft 2. In all of these expressions, the phase is referred to the phase of the
Incident wave at the end of the lower wire.

The combination of the voltages Induced by Ez and Ex Is somewhat more complicated

for the case when the wires lie In the vertical plane because of the difference In sign of the
term exp [-jwo(2h'+d')/cJ in the expressions for Vu and Vm(w) - VU(w), This is a problem

123

.....
.
for vertical polarization only, however, because Vu - 0 for horizontal polarization. Note
that the term 1I + exp (-iw (2h'+d')/cl I in Eq. (2-109) transforms into the sum of two
responses, one starting at t = 0 and the second starting at t = (2h'+d')/c,. When the wires
were in a horizontal plane, on the other hand, the negative sign of the exponential term made
the second (delayed) response virtually cancel the late-time part of the first response.

The open-circuit differential voltage induced at the terminals of a semi-infinite two-

wire line with one wire 2 m above the other is shown in Figure 2-52. With perfectly conduct-
ing ground and vertical polarization the voltage builds up in two steps (dashed curve) to a
level similar to that shown in Figure 2-60 for horizontal polarization incident on wires in a
horizontal plane. With finitely conducting ground, however, the ground effect reduces the
level of the second step as illustrated by the solid curve of Figure 2-.52, The late-time
response also persists much longer for vertical polarization incident on wires in a vertical
plane because of the lower ruflection coefficient of the soil for vertical polarization, Com-
parison of the differential-mode voltage of Figure 2-52 with the common-mode voltage of
Figure 2-14 indicates that the peak differential-mode voltage is a factor 10 or more smaller
than the peak common-mode voltage,

2 E00

>
f / a - 10"2

0.02 0.04 0,o0o5• ,

VERTICAL POLARIZATION
d' 2h h- lm d,,2 m
C c O - 30U
.EOD
0"- Ijs D -3,73
E - Eo1't

Figure 2-42 OPEN-CIRCUIT DIFFERENTIAL-VOLTAGE WAVEFORM INDUCED IN A SEMI-

INFINITE, TWO-WIRE LINE BY AN EXPONENTIAL. PULSE (wirat in a vertical plans)

124

.. ......... ~~~~~~.
...
II I I
 2.5.3 MODE CONVERSION AT THE LOAD

In addition to directly induced differential-mode currents in the power lines,

differential-mode currents can be developed from the common-mode currents by unbalanced
terminations. Consider, for example, a perfectly balanced four-wire transmission line with a
common-mode voltage induced on it. As illustrated in Figure 2-53, the lines are represented
by their Thevenin equivalent voltage source Vo and equal characteristic impedances Zo, and
mutual coupling between the lines will be neglected for simplicity, The unbalanced loads
Z4 at the terminals cause unbalanced currents IIto flow in the individual conductors. The
relative magnitudes of the currents are

Ii Zo
. * 1,2,3 (2-111)
14 Z. + Z

and for a balanced three-phase, 4-wire system in which I1 ft 12 , 13, the currents in the
phase conductors would be approximately equal, while the current In the neutral might be
significantly different from the phase-conductor currents, It Is apparent that if ZI is com-
parable to Z4, the differential current 311 - 14 - 14(310/14-1) created by the asymmetrical
load operating on the common-mode current can be significantly larger than the differential
current Induced directly by the Incident wave,

"12

V --

SOIFuce '

LOAD AT
TERMINAL$l

Filows 2-53 ASYMMETRICAL LOAD ON A FOUR-WIRE LINE WITH A COMMON-MODE

VOLTAGE251"

•126
2.6 HIGH-VOLTAGE PROPERTIES OF TRANSMISSION-LINE COMPONENTS

2.6.1 GENERAL

The analysis of intera';tion of the EMP with power lines indicates that very large
voltages (several megavolts) may be induced between the line and ground. However, as
Indicated In Table 2-2, subtransmission and distribution lines are not designed to sustain
such voltages. As a result, it can b, expected that the EMP-induced voltages may be limited
by insulation breakdown or flashover or cause activation of lightning arresters. Insulation
breakdown, in the form of line insulator flashover or corona from the conductors may occur,
but little is known about the behavior of these mechanisms for very large rates of rise of the
conductor voltage. Fortunately, the very large voltages occur near the ends of the lines where
protective devices such as lightning arresters may be installed. Limited data are available on
the firing characteristic of distribution-type lightning arresters with large rates of rise, In
this section, the high-voltage properties of line insulators, lightning arresters, and conductors
are reviewed,

Table 2-2
TENTATIVE AIEE STANDARD ON INSULATION TESTS FOR OUTDOOR
AIR SWITCHES, INSULATOR UNITS, AND BUS SUPPORTS

Withstand Voltage (kV)

Voltage Low Freq. Low Freq. Impulse
Rating 1 Min. 10 Sec. 1.5X40 Full Wave
(kV) (Dry) (Wet) (Pos. or Nag.)

7.5 36 30 95
15 50 45 110
23 70 60 150
34.5 95 80 200
46 120 100 250
60 175 145 350
92 225 190 450
115 280 230 550
138 335 275 650
161 385 315 750
196 465 385 900
230 545 445 1050
287 680 555 1300
345 810 665 1550
Source: Ref. 1.

126
 2.6.2 INSULATOR FLASHOVER CHARACTERISTICS

As implied in Table 2-2, insulators used in transmission and distribution systems can
withstand an "impulse" voltage of 3 to 12 times tho line voltage (with the higher multiples
applying to the lower line-voltage ratings.) These ratings presumably apply to the "new-
equipment" state, so that deterioration in service may reduce these Insulation limits. Fur-
thermore, tie actual voltage the Insulator can withstand dopends on the polarity and the rate
of rise or duration of the transient applied voltage, Because Insulation flashover Involves the
formation of long (at least several inches) spark discharge channels, the arc formation time Is
significant In determining the magnitude and duration of the transient voltage that can be
sustained without a high current discharge. In addition, because the spark formation in air
is basically an electron avalanching process, in a nonuniform field, the breakdown conditions
are quite different when the wire is positive from when It is negative.

The "impulse" flashover characteristics of some Insulators in the 7.5-to-69 kV class

are shown in Figure 2-54 for positive and negative voltage applied to the top of the insulator.
These insulators display the typicPl characteristics of lower positive-voltage flashover thresh-
olds and sharply increasing flashover voltage with Increasing rate of rise of the voltage, These
data are for the commonly used "impulse" that reaches its peak in 1,5 Ps and decays to half
its peak value In 40 js. Flashover data are not available for applied voltage with rise times
(or times to flashover) less than about 0.25 Ms.

Figure 2-55 shows similar data (plotted In a different form) for strings of suspension
insulator,,.

Insulator flashover voltage also depends somewhat on altitude (or barometric pres-
sure) and temperature, but the dependence on these parameters is relatively unimportant
from an EMP point of view (the altitude dependence may be significant for lines in high
mountainous region.', however). Contamination and adverse weather effects cause deviations
from the flashover characteristics at standard air conditions shown in Figures 2-54 and
2-55. These deviations are likely to be more important and less predictable than those
caused by variations in temperature or by normal fluctuations In surface barometric
pressure,

127

..
....
 300 400 1 1

200 300 10

B-/3200 Z N
:EG
I 0POS -- S
0100

0 2 4 6 a 0 2 4 6 S
TIME TO FLASHOVER .-- s TIME TO FLASHOVER - Mi
fal 7.5 kV CLASS lb) 15 kV CLAUS

500

15
W40040

>300 - 300 -
cc0 - M E
NG -o PO:: NE
12

S200 - PSý 200 -

4 a 8 10 12 0 2 4 e a
IME TO FLASHOVER liMt TIME TO FLASHOVER -Ms

(W1 23 kV CLAUS (d) 34.5 kV CLASS

600 I 1100

400 I 700 25 -

Noo 13-12O- 12-1/2

300 700

200 I E I I3" 13-1/2

300 10oo
0 2 4 6 aB 2 4 a a 10 12
TIME TO FLASHOVER -- HiTIME TO FLASHOVER - /t

(is 46 kV CLASS If) 69 kV CLASS

Figure 2-54 IMPULSE FLASHOVER CHARACTERISTICS OF PARTICULAR TYPES OF

APPARATUS INSULATORS ON POSITIVE AND NEGATIVE 1-1/2 x 40 WAVES AT
STANDARD AIR CONDITIONS. Source: Ref, 1.

128
 19U0 -MICHOSEC TO - MIC0SV O.
FLASHOVEH FLASHOVEH //XR
Lm z46 ' 2 34613'
_1/

--
> 1300
xJ 4

(C

900 - - -~

500

300 L .Li
2 6 10 114 lB 2 6 10 14 18
UNITS UNITS
I&) POSITIVE WAVES lb) NEGATIVE WAVES

Figure 2-55 IMPULSE FLASHOVER CHARACTER-

ISTICS OF SUSPENSION INSULATORS FOR 1-1/2 x
40 WAVES AT 77'"F, 30-INCH BAROMETRIC AND
0.6085-INCH VAPOR PRESSURE, Relative air
density -1,0, Source: Ruf. 1.

2.6.3 LIGHTNING-ARR ESTER FIRING CHARACTERISTICS

The propertius of commercial lightning arresters used in transmission and distribution

systems are usually not specified for very-fast, rising plulses, since the tranbients produced by
liglitning usoully have rise times greater thani 1 jps, Typical firing characteristics of line and
distributiun lightning Hrresters aro shown in Figure 2 -56, These data indicate that thu firing
voltaye Increases 10"l to 15Y4 above the dIc kslow pulse0) firing voltage if the time to fire is
shortened to 250 ns. 1.3 Further reductions in tinie.-to- fire will certainly be accompailiod by
further increasus in firing voltage.

Thu time-to-fire and the firing voltage of the 9-kV distribution lightning arresters
used to protect the transformers are plotted in Figure 2-57 .26 'no maximum rate of rise of
the voltage applied across the lightning arresters was 2.5 kV/ns, or about 25 timets faster than
tnat normally specified for 9-ky arresters. With this rate of rise (also plotted in Figure 2-56),
the firing voltage was 100 kV, or about 2,15 times the static firing threshold, and the time-to*-
fire was only 40 ns.
129
 2

00II ~ TIME TO FIRE!M

fal LINE-TYPO LIGHTNING APIEUTIFIB,

Fgo

so

40

20 3k

0 1 2 3 4 58
TIME To FIRE -ps

(bI OIBTRIBUTION-rYPE LIGHTNING APINUTERS11

Figure 2-5.6 BREAKDOWN VOLTAGE AS A FUNCTION OF TIME-TO-FIRE FOR LIGHTNING

ARRESTERS, Source: Refl. 5.

130
 120

I
100

> so

u* 40 I 0g ••

201%j I, 0 --

0 0 00 11 1
0 50 1O0 150 200 250 300
TIME-TO-FIME ... na

Figiur 2-57 9-kV-DISTRIBITION LIGHTNING-ARRESTER FIRING CHARACTERISTICS FOR

FAST RATES OF RISE, Sources: Refs, 2 and 3,

The trend of increasing firing voltage with Increasing rate of rise of the applled volt-
ago is typical of spark-discharge devices and insulation flashover. Because lightning arrester
firing and line Insulator, transformer, or pothead bushing flashover involve similar air break-
down procusses, it is expected that the lightning arrester that Is designed to protect these
components for slowly rising transients will also protect them for fast-rising transients. If
solid or liquid insulation is protected by the lightning arrester, however, this generalization
may not be valid because the temporal characteristics of solid or liquid breakdown may be
different from those of air.

2.6.4 CORONA THRESHOLD OF CONDUCTORS

Even withouL lightning-arrester breakdown or insulation flashover, there is a high-

voltage limit imposed by the dielectric strength of the air about the power conductors, When

131
the electric-field strength at the surface of the conductor exceeds 3 X 106 V/m, the air will
break down and a corona discharge or arc leader will form. The field strength at the surface
of a round wire at a height h, of radius a, and at a voltage V relative to ground is

V
E -- h (2-112l
a log --
a

A plot of the static voltage V required to produce the corona-threshold field strength at the
surface of the wire 10 m above the ground is plotted In Figure 2-58 as a function of wire
radius. Since most distribution-line conductors are 1 cm or less in diameter, it Is apparent
from Figure 2-58 that the corona-threshold voltage of these conductors will be less than
250 kV.

If the transmission line is made up of three or more conductors, as Is usually the case,
the field strength at the surface of individual wires will be less, for a given potential, then for
a single wire, This decrease In surface field strength or increase in corona-threshold voltage
can, at the most, be In proportion to the number of wires, but it Is usually considerably
less, For three wires on a horizontal cross-arm, the corona-threshold potential of the two
outer wires will be only slightly greater than the threshold potential of a single wire of the
same size, but the threshold potential of the middle wire will be significantly higher because
It is shielded by the outer wires, (Note that the surface field strength does not decrease In
proportion to the increase In effective radius for a multiple-conductor system; the effective
radius for the purposes of computing common-mode capacitance and inductance per unit
length may increase by a factor of 100 if three codIIrtorb spaced a meter apart are used, but
the conductor surface over which the charge Is ,'Istributed has increased by only a factor of
3 over that of a singlq conductor.)

Corona losses on transmission lines have the effect of distorting the leading edge and
limiting the peak of a transient voltage pulse. 2 7"29 Therefore corona losses in combination
with lightning arresters will limit the wire-to-ground potential difference of typical
distribution lines to about 500 kV or less, Because these llmiti•, actions are not equal for
all of the conductors of the trdnsmission line, however, sizable differential voltages may be
generated on the transmission line, These differential-mode voltages, as well as the common-
mode voltages, must pass through the distribution transformer to affect the customer equip-
ments. A discussion of the behavior of the transformer is contained In Chapter Four.

132
 I°? • I ~I M IillIII IIII' II

0
8

8 06 _.. ..

V* 0
106

1oU LLU
V.WU JLL±

104 1 1 1 I11I1
1 1 1 1 I I
1(r3 1 0 .2 10 "1 1
RADIUS OF CONDUCTOR - m

Figure 2-58 CORONA-THRESHOLD VOLTAGE OF A WIRE OVER GROUND WITH

STANDARD AIR CONDITIONS. Source: Ref, 4.

2.7 CITED REFERENCES

1. Electrical Transmission and DistributionReference Book (Wcstinghouse Electric

Corporation, East Pittsburgh, Pennsylvania, 1964).
2. A. E. Knowlton, Ed., Standard Handbook for ElectricalEngineers (McGraw-Hill
Book Co., Inc., New York, N.Y., 1949).
3. W. W. Lewis, The Protectionof Transmission Systems Against Lightning (John
Wiley & Sons, Inc., New York, N.Y., 1950).
4. E. F. Vance and S. Dairiki, "Analysis of Coupling to the Commercial Power
System," AFWI..-TR-72-21, Contract F29601-69-C-0127, Air Force Weapons
Laboratory, Kirtland Air Force Base, New Mexico (August 1972).
5. C. Flammer and H. E. Singhaus, "The interaction of Electromagnetic Pulses with
an Infinitely Long Conducting Cylinder Above a Perfectly Conducting Ground,"
Interaction Note 144, Contract F29601-69-C-0127. SRI Project 7995, Stanford
Research Institute, Menlo Park, Calif. (July 1973, unpublished report).

133
 6. C. Flammer, "On the Scattering of Electromagnetic Waves by a Perfectly Con-
ducting Cylinder Over a Finitely Conducting Ground," Interaction Note 145,
Contract F29601-69-C-0127, SRI Project 7995, Stanford Research Institute,
Menlo Park, Calif. (August 1973, unpublished roport).
7. W. E. S.:harfman and E. F. Vance, "EMP Coupling and Propagation to Power
Lines: Theory and Experiments," A FW L-T R-73-287, Contract F29601-69-
C-0127, Air Force Weapons Laboratory, Kirtland Air Force Base, New Mexico
(May 1973).
8. W, E, Scharfman, K. A. Graf and E. F. Vance, "Analysis of Coupling to Horizontal
and Vertical Wires," Technical Memorandum 22, Contract F29601-69-C-0127,
SRI Project 7995, Stanford Research Institute, Menlo Park, Calif. (January 1973,
unpublished report).
9. S. Frankel, "Field Coupling Parameters for a Single Round Wire Close to a
Ground Plane or Two Large Round Wires in Free Space," HDL-TM-72-14,
MIPR 2,00518, Subtask EB-088, HDL Project E05E6, Harry Diamond
Laboratories, Washington, D.C. (April 1972).

10. S. Frankel, "Externally Excited Transmission Line: Definition of Procedures

for Detormining Coupling Parameters," HDL-TM-72-1 1, AMCMS Code
5910,21.63387, HDL Project EOSE3, Harry Diamond Laboratories, Washington,
D.C. (May 1972).
11. J. A. Auxier, W. S. Snyder, and D. M. Davis, "Health Physics Division Annual
Progress Report," Contract No. W-7405-eng-26, ORNL-4903, Oak Ridge
National Laboratory, Tennessee (September 1973).

12. J. H. Marable, J. K. Baird, and D. B. Nelson, "Effects of Electromagnetic Pulse

(EMP) on a Power System," Final Report, Interagency Agreement No. AEC
40-31-64 and OCD-PS-64-284, Work Unit 2213C, ORNL-4836, Oak Ridge
National Laboratory, Oak Ridge, Tennessee (December 1972).

13. "Civil Defense Research Project Annual Progress Report," Contract No. W-7405-
eng-26, ORNL-4784, Oak Ridge National Laboratory, Oak Ridge, Tennessee
(December 1972),
14. "Civil Defense Research Project Annual Progress Report," Contract No. W-7405-
eng-26, ORNL-4679, Oak Ridge National Laboratory, Oak Ridge, Tennessee
(March 1972).
15. D. B. Nelson, "A Program to Counter the Effects of Nuclear Electromagnetic
Pulse in Commercial Power Systems," ORNL-TM-3562, Part I, Oak Ridge
National Laboratory, Oak Ridge, Tennessee (October 1972).

16. D. B. Nelson, "Effects of Nuclear EMP on AM Radio Broadcast Stations in the

Emergency Broadcast System," ORNL-TM-2830, Oak Ridge National
Laboratory, Oak Ridge, Tennessee (July 1971).

134
17. "EMP and Electric Power Systems," TR-61-D, Defense Civil Preparedness
Ageny, WAshington, D.r. (December 1972).
18. E. C.Jordan, Electromagnetic Waves and RadiatingSystems (Prentice-Hall, Inc.,
New York, N.Y,, 1950).
19. S. A, Schelkunoff, Antennas: Theory end Practice (John Wiley & Sons, Inc.,
New York, NVY., 1962).
20. S. Ramo and J. R. Whinnery, Fields and Waves In Modern Radio (John Wiley &
Sons, Inc., New York, N.Y., 1953).
21. Reference Data for Radio Engineers 5th Ed. (Howard W. Sams & Co., Inc.,
A Subsidiary of ITT, New York, N.Y,, 1969).
22. G. A. Korn, Basic Tables In ElectricalEngineering (McGraw-HIll Book Co., New
York, NY., 1965).
23. E. D. Sunde, Earth Conduction Effects in TransmissionSystems D. Van Nostrand
Co., Inc., New York, N.Y., 1949).
24. R. W. P. King, Transmission-lineTheory (Dover Publications, Inc., New York,
N.Y., 1955).
25. C. D0. Taylor, R. S. Satterwhite, and C. W. Harrison, Jr., "The Response of a
Terminated Two-Wire Transmission Line Excited by a Nonuniform Electro-
magnetic Field," Proc. IEEE, Vol. AP-13, No. 6, pp. 987-989 (November 1965).
26, R. T. Bly, Jr. and E, F. Vance, "High-Voltage Transient Tests of Service Trans-
formers, Lightning Arresters, and an Automatic Switching Unit," AFWL-TR-74-
34, Contract F29601-69-C-0127, Air Force Weapons Laboratory, Kirtland Air
Force Base, New Mexico (October 1973).
27. L. V, Bewley, Traveling Waves on TransmissionSystems (Dover Publications,
Inc., New York, N.Y., 1933).
28. R. Rudenberg, ElectricalShock Waves in PowerSystems (Harvard University
Press, Cambridge, Mass., 1968).
29. A. Greenwood, Electrical Transients in PowerSystems (John Wiley & Sons, Inc.,
New York, N.Y., 1971).

135
Chapter Three

CONSUMER'S SERVICE ENTRANCE

.ISM

3.1 INTRODUCTION

3.1.1 DESCAIPTION OF THE SERVICE ENTRANCE

At the consumer's end of a distribution line, the distribution voltage Is reduced

with transformers to the level required by the consumer (e.g., 120 V, 240 V, 480 V, etc.),
metered, and wired to the consumer's main circuit-breaker and distribution panel. The
properties of this equipment between the end of the aerial distribution line and the low-volt-
age circuit feeding the main circuit-breaker panel affect the coupling of the EMP-induced
signal to the consumer's circuits, In this chapter, typical service-entrance properties and
installation practices are described, and their effect on the EMP coupling is discussed, The
distribution tranmormer, which Is -in important component of this terminal equipment, is
discussed separately in Chapter Four.

136
 For purposes of discusping their effects on EMP coupling, service entrance instal-
lations have been divided into two categories according to whether the service transformers
are pole-mounted, as in Figure 3-I (a), or ground-based, as in Figure 3-1 (b). For pole-
mounted transformer installations, the transformers are usually installed on the last pole of
the distribution line, and the low-voltage service drop is carried along a messenger cable to a
service-entrance weatherhead, where the low-voltage conductors enter rigid steel conduit.

.IN

THREE-PHASE
WEATHERHEAD•
"
TRANSFORMERS
SERVICE DROP

RIGID
CONDUrr

UNDEROROUND TO Lee
(I) SERVICE WITH POLE-MOUNTED TRANIFORMIRS

Figure 3-1 TYPICAL POWER SERVICE ENTRANCES

*iThe service entrance may be on a separate pole, as shown in Figure 3-1 (a), It may be on the
roof or exterior wall of a building, or It may be on the same pole as the transformers. There
Is some variation in the mounting of the transformers, also. For example, they may be
mounted on a platform supported by four poles, rather than mounted directly on the
final pole.

137
DISTRIBUTION LIGHTNING ARRESTERS
FUSED DISCONNECTS

POTHEAD

TRANSFORMERS
(vault, khod, 1111f))

RIGID CONDUIT. ARMOR, ___

OR LEAD SHEATH

RIGID CONDUIT
OR PLASTIC DUCT I I

LOW VOLTAGE
TO CIRCUIT
BREAKERS AND
DISTRIBUTION
PANEL
(b) SERVICE WITH OROUND-BASEO TRANSFORMERS

Figure 3-4 lCuncludud)

3,1.2 GROUND-BASED TRANSFORMER INSTALLATIONS

Ground-based transformer installations are distinguished by the characteristic that

the distribution voltage is carried underground from the last pole to the transformers, The
transformers may be located in a shed or fenced-in area near the consumer's buildings, or
they may be in a vault within the consumer's building, Shielded cables are used to transmit
power from the pothead near the top of the last pole to the transformers, These cables are
usually routed through rigid steel conduit or plastic or fiber duct for mechanical protection
underground and on the pole,

The principal compotients of the ground-based transformer Installations that affect

the EMP coupling and voltage-limiting are the potheads, the shielded cable, and the steel or
fiber duct, (Transformer characteristics are also important; they are discussed In Chapter
Four.) The pothead Is a weatherproof interface between the aerial distribution line and the

138

.. ..........
shielded cable used for the underground service entrance. Typical potheads for use with
single- and three-conductor lines are shown in Figure 3-2. Basically, they consist of insu-
lated high-voltage bushings, to which the aerial conductors are connected, and a
controlled-gradient termination for the shields and/or lead sheath of the underground cables, The
lower cavity of the pothead is usually filled with an asphalt-based compound to weatherproof and
seal the cable termination. The dielectric strength of the pothead should be commensurate
with the basic Insulation level required for the distribution voltage (see Table 3-1), To
prevent damage to the pothead and cable insulation, the pothead is usually protected with
lightning arrestors,

Cables for use at distribution voltages are stranded copper or aluminum insulated
with cross-linked polyethylene and shielded (for gradient control) with spiral-wound copper
tape, An overall lead sheath or vinyl jacket is sometimes provided for mechanical protection,
The construction of typical one- and three-conductor cables is shown in Figure 3-3. Semi-
conducting tape and semiconducting polyethylene are sometimes used adjacent to the inner
and outer conductors for electric gradient control. A concentric wrap of solid copper
strands may also be provided over the shield tape if the shield carries appreciable current
(the shield tape Is usually only a few mils thick arid is used for gradient control rather than
to carry return or neutral currents), In some older installations, paper-Insulated cable may
be foundt

Rigid steel conduit is almost always used at the ends of the conduit run (iLe., at the
pole and at the transformer) but plastic (styrene) or fiber conduit is often used along the
buried portion of the run, The plastic or fiber duct may be of the rigid, direct-burial type
(Type II), or of the concrete-encasement type (Type I) that is partially encased in concrete
after it is Installed in the trench. This encasement is normally accomplished by laying the
conduits on a bed of sand so that they are less than one-third embedded in sand, then pour-
Ing concrete over the conduits until the trench is filled to a few inches above the top of the
conduits. The remainder of the trench is then backfilled to grade level with local soil. From
an EMP coupling standpoint, the rigid steel conduit can provide excellent shielding for the
cables and it provides a uniform return conductor for the common-mode current on the
cables, Plastic and fiber conduits provide no shielding and no return path for the common-
mode current.

139
 0 0

* CONDUCTING
00 GLAZECAL

oil SHIELDING

ill INUL I-CONDUCTOR (b) THREE-CONDUCTOR

Figure 3-2 POTHEADS FOR SINGLE- AND THREE-CONDUCTOR CABLES

3.1.3 POLE-MOUNTED TRANSFORMER INSTALLATIONS

For installations with pole-mounted transformers, the low-voltage conductors may

be exposed to the EMP fields between the transformer and the service entrance. Because
the components in the low-voltage system are not specifically designed to withstand high
voltages, this portion of the installation may be vulnerable to large induced voltages (even if
the transformer does not pass extremely large common-mode voltages), Vinyl- or neoprene-
insulated cables are normally used for the low-voltage service, These cables are typically
loosely spiraled about a steel messengci cable (which may also serve as a ground or neutral
conductor) between the transformer and the service entrance. The service-entrance
weatherhead is a hooded conduit fitting that permits the cables to be brought into the con-
duit without pormItting varmints or excessive rainwater to enter the building through the
conduits. The service-entrance weatherhead contains an insulating spacer for the cables so

140

wool
 Table 3-1
SUGGESTED WITHSTAND IMPULSE VOLTAGES
FOR CABLES WITH METALLIC COVERING*

Basic Solid-Paper Oil-Filled-Paper

Rated Impulse Insulation Insulation
Voltage Insulation Insulation Withstand Insulation Withstand
EkuiLeel f Thicknes Voltage Thickness Voltage
Equipment (mii) (kV) (mils) (kV)
______ (kV) __ _ ___ _ _ _ _ _ _

1.2 30 78 94
2,5 45 78 94
5.0 60 94 113
8.7 75 141 189 - -
15 110 203 244 110 132
23 150 266 319 145 174
34.5 200 375 450 190 228
46 250 469 563 225 270
69 350 688 825 316 378
115 550 - - 480 575
138 Ob0 560 672
161 750 648 780
230 1050 925 1110
*Source: Ref, 1,

that they are separated from each other and from the rough edges of the metal hood,
A typical service entrance Is shown in Figure 3-4.

Although the cable and weetherhead insulation are not designed for high-voltage
applications, insulation breakdown In the low-voltage cable system seldom occurs in the
conduit or weatherhead unless the cable Isold and deteriorated, Breakdown usually occurs
where the Insulation has been removed or compromised, such as in the metering cabinet
where the cable insulation has been removed to make the meter voltage taps, or at the main
cIrcuit-breaker panel where the insulation Is removed to make the connection to the cir-
cuit breaker terminals.

It is also noteworthy that although large voltages may be induced in the low-voltage
service drop, there is some voltage-limiting built Into the secondary terminals of the trans-
former, which normally has a voltage-limiting spark gap built Into the hardware disoclated
with at least one of the secondary bushings, In addition, when long, exposed service drops
are used, it Is not uncommon to install secondary lightning arresters in the low-voltage cir-
cuit at the transformer, service entrance, or main circuit-breaker panel (or all three),

141
 Is) SINGLE-CONDUCTOR SOLID. COMPACT-ROUND CONDUCTOR

IbI) TI4R1-CONDUCTOR SI4ILODI, COMPACT-SICTOR CONDUCTORS

Figure 3-3 SHIELDED CABLES FOR SERVICE ENTRANCE

MESSEN.JER
ANDWFAT
NEUTRAL W•IHRHA
H EMHEAD

3" TFý.L
]', CONDUIT

Figure 3-4 POLE-MOUNTED SERVICE

ENTRANCE

142
 2
3.1.4 EFFECT OF SERVICE ENTRANCE ON EMP COUPLING

The consumer service entra ico is an important part of the power system for EMP
coupling considerations because It is closer to the consumer's equipment then the remainder
of the system, and because of the following:

(1) The service transformer behaves as a bandpass filter (see Chapter Four).

(2) The low-voltage entrance-conduit circuits or shielded service*-entrance cables

have much lower characteristic (surge) impedances than the aerial distribution
lines.

(3) Additional EMP coupling can occur along the unshielded low-voltage service
drop or along service-entrance cables in buried plastic or fiber ducts.

(4) Voltage limiting may occur at lightning arresters and in low-voltage circuits.

The difference In charactoristic Impedance between the aerial transmission lines and
the shielded service-entrance cables used In ground-based transformer installations causes a
large mismatch between these transmission lines, For times les than the round-trip propa-
gation time from the pothead to the transformer, the common-mode load impedance on the
aerial transmission line ls the parallel combination of the characteristic impedances of the
shielded service-entrance cables, Since the source Impedance of the aerial transmission lines
is several hundred ohms and the characteristic impedance of the shielded cables (in parallcl)
Is of the order of ten ohms, only a small fraction of the large open-circuit voltage induced
in the aerial transmission line Is transmitted through the shielded cables; the remainder Is
reflected buck down the aerial line.

At times greater than the round-trip propagation time on the shielded cables, the
transformer and the loads on Is secondary winding affect the Input Impedance at the pot-
head seen by the aerial line, In the frequency domain, however, the shielded cables may
behave as matching transformers that provide maximum coupling between the aerial trans-
mission line and the transformer for a band (or several bands) of frequencies In the pulse
spectrum, Because of the narrow bandwidth of the matching-transformer effect, this effect
Isusually not important in terms of transformer insulation breakdown, but It Is important

143
in determining the spectral content of the signal delivered to the consumer's low-voltage
circuits,

Similar considerat;ons apply to the low-voltage conduit circuit between the service-
entrance and the main circuit-breaker panel in pole-mounted transformer installations, The
principal differences in thin case are: (1) the signal induced in the aerial transmission line Is
filtered by the transformers before it gets to the service-entrance weatherhead, and Q2) the
common-mode characteristic Impedance of the conductors in conduit is somewhat larger
than that of the shlulded service-entrance cables,

In the case of pole-mounted transformers with long, low-voltage service drops, sig-
nificant voltage can be Induced in the service drop by the EMP in much the same way as It is
induced In the aerial distribution line. This voltage will enter the conduit circuit before the
voltage Induced in the transmision line behind the transformer, and it may be larger than
that passing through the transformer (particularly if the transformer is protected with light-
-Ing arresters). Similarly, current may be Induced In the shields of the buried shielded cables
between the potheads and ground-based transformer when these cables are In plastic ducts,
but if the cable shields are effective, the voltage induced between the conductor and the
shield by this means is a major concern only If the cables are quite long (several hundred
feet). If steel conduit is used for the entire run instead of plastic conduit, the voltage induced
In the buried cables will usually be relatively small (compared to that delivered by the aerial
transmission line) unless the run Isseveral thousand feet long.

Intentional voltage-limiting occurs in the terminal installation because of distribu-

tion lightning arresters protecting the transformer or pothead bushings and secondary light-
ning arresters (if present) at the weatherhead and main circuit-breaker panels. Unintentional
voltage limiting may occur because of low-voltage insulation breakdown at metering taps,
circuit-breaker terminals, and similar points where the insulation is weak, Voltage break-
down in the low-voltage system may have deleterious effects such as damage to the watt-
hour meter and damage to power-monitoring relays and control circuits.

144
3.2 TRANSMISSION THROUGH CONDUITS AND CABLES

3.2.1 TRANSFER FUNCTIONS FOR CONDUITS AND SHIELDED CABLES

The conduit and conductors between the service-entrance weatherhead and the main
distribution panel, and the shielded cables between the potheads and the transformers can be
analyzed as transmission-line segments between the source (aerial transmk,sion lines) and the
load (low-voltage circuits or transformer primary). The equivalent circult for this part of the
system Is illustrated in Figure 3-5 where the source is represented by an open-circuit voltage
Vo and a source impedance ZI and the load is represented by ZR, The transmission line has
a characteristic impedance Zo and is of length Q. For a long(semi-infinite) aerial distribu-
tion line connected to shielded cables for transformer service, the source voltage Vo and
Impedance Zi are the open-circuit voltage induced in the semi-infinite line and Its character-
Istic Impedance, The load Impedance ZR Is the input impedance of the primary terminals of
the transformer, Including any lead Inductance and bushing capacitance.

3
The reflection coefficient at the lorid end of the transmission line Is defined by

ZR - Zo
"1P=Z + Zo

and the propagation factor for the line Is a .r j+, The ratio of the voltage V(R) across the
load to the source voltage Vo is

V(.) (1 + p)e-I (3-2)

V0 z (1 - pe-27) + (1 + pe
Zo

Plots of the magnitude of this transfer function are shown in Figure 3-6 for ZR resistive, Fig-
ure 3-7 for Zq inductive, and Figure 3-8 for Z2 capacitive, These results are for a lossless
air-insulated transmission line (, - Jk - jcw/c) 30 m long and a source impedance of 300 ohms
(typical of an aerial distribition line). A characteristic impedance of 10 ohms Is assumed for
the transmission line. However, these data can be applied to any similar circuit with

146
 zi z0

v (Q) ZJ
V0

SOURCE CABLE OR CONDUIT LOAD

Figure 3-5 APPROXIMATE EQUIVALENT CIRCUIT FOR

ANALYSIS OF TRANSMISSION THROUGH SHIELDED CA5LE
OR CONDUIT FEEDER CIRCUITS

ZI /Zo 30 for which the values of R/Z 0 , Lc/AZo, or cZoC/V shown are applicable, if the
upper normalized-frequency (kR) scale is used. (The speed of light, o, may be replaced by
the propagation velocity v - c/V,/ if insulation other than air is used,1

Note that for all load impedances except the matched resistive load (ZQ a R , Zo),
the transfor function is frequency-selective and tends to pass certain frequencies more
readily than others. Because the first few passbands lie in the range 01 < k' < 10, power-
system responses are often characterized by fundamental oscillations in this frequency range,

Time-domain responses of the circuit of Figure 3-5 have been obtained experimen-
tally using RF transmission line to simulate the conduit or shielded cable, This line is not
lossless, and the load elements are not pure resistances or reactances, so that the waveforms
obtained with the analog circuit are somewhat more representative of those that might be
observed in a power system. The waveforms for a step voltage Vo with source impedance
such that Zi/Zo - 30 are shown In Figure 3-9 for resistive loads. It is seen in Figure 3-9(a)
that the voltage V(Q) across the load resistance is a stair-step rising toward R/(R + Zi)
when R > Z4, The length of the steps Is 2R/c, and the "smoothed" stair-step for R > Zo
approaches an exponential function VI1[1 -exp (-t/r)] where

RZ1 1C
7=-... (3-3)
R + Z1

R
V1 -
R +Z Vo (3-4)

and C is the total capacitance of the transmission line (C -/cZo). Wheni R < Zo, as In Fig-
ure 3-9(b), a damped square wave of Initial amplitude 2 RZol/(R + Z0 )(Z1 + Zo) oscillating

146
 0.1 110

Z1 300 v cv()

20

>" ~(R *100 n2)

(1; 30 al

w ¶

0 .

0.3 1 13 10 30
FREQUIENCY M~z

Figure 3-6 FEEDER TRANSFER FUNCTIONS FOR RESISTIVE LOADS

about the value R/(Zj + R) is obtained, For a load R =4Z, the voltage V(M would be a step
of magnitude R/( R + Z)

The response with an inductive load is of more interest in practice because of the
large lead inductance associated with power circuits near transformer or circuit-breaker
terminals. The load voltage V(1?) for three values of inductance is shown in Figure 3-10
for the step-voltage source with ZI /Zo -30. In Figure 3-10(a), the time constant LIZ0 of

147
 -- WQic
(L,1 10

(L - 10. PHI

- Lc
RZ
-

(L -3 PHI

(L "0.3 H) Z, V- (111
vo Ln Z( .io L

0.3 IL 00.1 Q- 30 m

0.1
0.3 1 to 30
FREQUENCY - MHz

Figure 3-7 FEEDER TRANSFER FUNCTIONS FOR INDUCTIVE LOADS

the inductive load is smaller than the round-trip transient time 21/c of the line. During the
first 2k/c of the response, therefore, the response voltage is a decaying exponential with
time-constant L/Z0 , but beyond 2R/c the response becomes very complicated because of the
convolution of the exponential response with multiple reflections from the inductive load
and the resistive source impedance. The waveform of Figure 3-10(a) is quite representative of
the complexity of the power-system response waveforms that are observed in practice, however.

148
 0.1 1 10
I I

--CZoC 1-

0"0 10 PIP

I-

o-oC
i

-Z 30011

V0 Z~o onC
a30 M

0.3 1 10 30
FREQUENCY - MHz

Figum 3-8 FEEDER TRANSFER FUNCTIONS FOR CAPACITIVE LOADS

Note that, although it Ismasked by the fine structure of the reflections, the response con-
tains afundamental oscillation with aperiod of 2rwQ/c2~o, about three cycles of which
can be seen in Figure 3-100(a)

The responses shown in Figure 3-10(b) and (c)are for larger Inductances. In Figure
3-10(b), the time constant L/ZO of the Inductance isabout equal to the round-trip trarnsit
time 22/c, so that the exponential decay of the voltage across the inductance iseasily

149
 220

176

132

88 -
Be ) - -10
>44 -

0--

-44

36

27

> llb

0" 2 4 16 E 10 12 14 16
TIME I•'

Figure 3-9 VOLTAGE ACROSS A RESISTIVE LOAD PRODUCED) BY A UNIT STEP SCJRCE
(Z1/Zo "'30, circuit of Figure 3-5)

A@,
W

150
 0 4

36

27

20

as
Go

-22

-44

-as

132

'a L

-44

-as

-132

TIME -- kc

Figure 3-10 VOLTAGE ACROSS AN INOUCTIVE LOAD PnODUChD BY A

UNIT STEP SOURCE (Z 1 /Z0 - 30. citcuit of Figute 3-5)
identified even after several reflections. The fundamental oscillation with a period of
2ir./qL/cZo, upon which these exponential responses ride Is also easily recognizable. In Fig-
ure 3-10(c), where L/Zo > 2W/c, the multiply reflected exponential responses appear as a
saw-tooth wave superimposed on the fundamental oscillation.

The voltage developed across a capacitive load by a step voltage Vo with source impe-
dance Zl - 30ZO Is shown In Figure 3-11 for a capacitance whose time constant ZoC is
small compared to the transit time 21/c, The early-time voltage across the capacitor
behaves as

v(V (1 - et/Tc) (0 < t • 21/c) (3-5)

Zi + Zo

where 0 ZoC. After the first reflection returns, the waveform Is more complicated, but
generally lncrvases along an "average" curve given by

v(M) , 1 -0e4/Z1C (3-6)

when ZI Is resistive, Therefore, If the capacitance is Increased, both the "average" time
constant ZI C and the short-term time constant ZOC increase so that the size of the ripples is
reduced and the average rate of rise is reduced. Waveforms such as that shown Ir Figure
3-11 are seldom observed in power systems because the capacitances in power circuits are
usually distributed capacitances associated with machine or component windings, so that they
behave as transmission lines rather than lumped capacitors,

152

.t.. .1lU •" -I

 3.2.2 THEVENIN-EOUIVALENT SOURCE

The Thevenin-equivalent source voltage at the load end of the conduit or shielded
cable may be obtained from Eq. (3-2) by letting Zu so that p - 1, resulting in
-,

0 " ZI (3-7)

where

- Zo
p' " (3-9)

A Thevenin-equlvalent source of voltage V' 0 and impedance Z's at the load end of the con-
duit can then be used to replace the source Vo, its impedance Zi, and the transmission line
ai Illustrated in Figure 3-12.

I o
3.2.3 CHARACTERISTIC IMPEDANCE OF CONDUCTORS IN A CON UIT
b

Formulas for the common-mode characteristic impedance of two- and four-conduc-~

tor shielded cables have been derived by Kaden. 4 These formulas are exact for perfect con-
ductors and uniform dielectric between the inner conductors end the shield. The c:able

153
 28-

24

20

.> 18,.
>

a
e-

4
0-

.4-

0 2 4 6 9 i0 12 14 16

TIME - /,

Figure 3-11 VOLTAGE ACROSS A CAPACITIVE LOAD

PRODUCED BY A UNIT STEP SOURCE (Z1 /Z,) - 30,
mcfl, it of Figurst 3-0)

configuration and appropriate formulas for the common-mode characteristic Impedance are
shown In Figure 3-13(a) and (b). The formulas also assume u uniformly spaced, symmetri-
cal array of conductors as illustrated, The effects of eccentricity of the cable bundle in the
conduit can be estimated from the behavior of the characteristic impedance of the eccentric
"coaxial" cable shown in Figure 3-13(c),F If a, in Figure 3-13(c), is the effective radius of

the bundle of conductors, the common-mode characteristic Impedance of the bundle with an
offset from the c¢,-ter of the condfult can he estimated using the formula given in Figure
3-13(c). A plot of this variation with offset is shown in Figure 3-14, where it is apparent
that the bundle can be off-centered up to 50% with loss than 20% reductilon in character..
Istic impedance.

As an illustration of a typical four-conductor circuit in a conduit, the common-mode

characteristic impedance of four 300-MCM conductors in a 3-inch conduit is shown in Table
3-2 for an arrangemeni like that shown in Figure 3-13(b). Air was assumed for the dielectric,
but the maximum and minimum spacings between conductors ("compact" and "loose" in
Table 3-2) were obtained assuming insulation of 0,933 inch diameter over the conductors,
The "mean" spacing is the geometric mean of the maximum and minimum spacings. As Is
apparent in the table, the characteristic impedance changes by a factor of a little more than

154
 2 as the spacing varies from the maxi-
zI zo, i, c mum to the minimum permitted by
-.. =a.. the insulation If allowance is made
voLa
0.0 for the dielectric constant of the in-
sulation, all of the characteristic
(I) ORIGINAL CIRCUIT impedances shown in the table will

Z; be reduced somewhat,

is) NOW THINVININ EQUIVALENT

Figure 3-12 FEEDER SOURCE TRANSLATED

TO THE LOAD TERMINALS

Table 3-2

COMMON-MODE CHARACTERISTIC IMPEDANCE OF FOUR 300-MCM

INSULATED CONDUCTORS IN A THREE-INCH-DIAMETER CONDUIT

Wire Spacing, Characteristic

b (inches) Impedance, Zo (ohms) Comments

0.660 33.3 compact

0,840 25,0 Mean

1,068 14.6 Loose
Notes: O.D. of insulation: 0,933 Inch
OD, of conductor: 0,629 Inch
I.D, of conduit: 3,068 Inch

1.5
 24 /IB- r 4,1 /A2-

hb+ lb4Ar A2

W1) SHIELDID PAIR

""IU -. 4 /I-
-bA5 L4 .3ej j ;, "- , A--"
A2

lb) SHIELDED QUAD

20 1 ,.coth-t F 2 22 WHEN 1) 0.
LV2. ... J ' , 2 611

(a) ECCENTRIC CENTER CONDUCTOR

Figure 3-13 CHARACTERISTIC IMPEDANCE OF O3NDUCTORS IN A METAL CONDUIT,

Source: Ruts, 4 and 5.

156
 I 0.75 0.5

a22

2~~~~~~
~
---.--------- 0.26
I-

UU

0 0.5 1'0

ECCENTRICITY. .

"2 1

Figure 3-14 VARIATION OF CHARACTERISTIC IMPEDANCE OF ECCENTRIC

COAXIAL CABLE

3.2.4 CHARACTERISTIC IMPEDANCE OF SHIELDED CABLES

The hiqh. trequency characteristic (surge) impedance uf shielded cables can be

obtained from manufacturers' capacitive reactance (data. The characteristic impedance is

Zo (3-10)
cc

157
 where c Is the speed of light, C is the capacitance per unit length, and Cr is the dielectric
constant of the insulation. The capacitive reactance uf cables is often tabulated in
ohms/mile at 60 Hz In terms of the reactance per mile, Xc, the characteristic impedance
in ohms is

Zo =2.02 X 10-3V/•-_rXe (Q11

A table of values of characteristic impedance for various sizes and voltage ratings of
single, paper-insulated shielded cables is given in Table 3-3. These values were deduced
from the reactance-per-mile data given In Ref. 1.

Modern shielded cables are commonly insulated with cross-linked polyethylene, al-
though some peper Insulation Is still In use. Polyethyline-lnsulated cables tend to have
similar characteristic Impedances because the ditlectrir constant of polyethylene Is smaller
than that of paper, but the extruded Insulation !s sllghi,:/ thinner than paper Insulation for
the same voltage rating,

3.3 COUPLING THROUGH STEEL CONDUIT WALLS

3.3.1 GENERAL CONSIDERATIONS

Buried steel conduits, such as those between the serv s-entrance woatherhead and
the main circuit -breaker panel, may have current induced in them by the electromagnetic
field In the soil. Tho current Induced i1 the conduit will, In turn, induce a current In the
conductors Inside the conduit. This current is superimposed on the current injected at the
end of the conduit by the exposed aerial conduc=tors. A similar situation exists for the
shielded cables between the potheads and the transformers, although if these cables are
routed through rigid steel conduit, the conductors are shielded by both the conduit and the
cable shields.

158

,AL
 Table 3-3

CHARACTERFISTIC IMPEDANCES OF PAPER,-INSULATED

SHIELDED POWER CABLES

Voltage Rating
Size 1k J_ýJj '2kV3~k
kV.
(A WG or MCM ) . . .
Characteristic Impedance (ohms)

6 16 19 26 30 - -
4 13 16 22 26 33 - -
2 11 13 18 21 2B 35
1 97 12 12 19 26 31 -
0 8.7 10 14 17 23 28 36
00 7.9 9.5 12 15 20 26 33
000 7.1 8.6 11 13 18 23 30
0000 6.4 7.8 9.6 12 16 21 27
250 5.9 7.2 8.5 11 15 19 25
350 5.0 6.2 7.3 9 13 17 22
600 4.2 5.2 8.2 7.8 11 14 19
750 3.4 4.1 5.2 6,5 9,4 12 17
1000 3.1 3.8 4.4 5.7 8.7 11 15
1500 2.5 3,1 3,7 4.7 6.9 9.1 12
2000 . 22 27 3.2 41A 6.1 80 11
Gased an capacitive reactarnce vatues given in Roe. 1, Dielectric constant ik3.7.

The usual approach to calculating the currant or voltage on the conductors Inside
the conduit nonilsts of (1) determining the electric field In the soil parallel to the conduit,
(2) calculating the current In the conduit, (3) calculating the transfer impedance and satu-
ration characteristics of the conduit as a tubular shield, and (4) analyzing the conductors
and conduit as a coaxial transmission line with a dlst~lbuted driving source. 6 . 10 In the case
of shielded cables In plastic conduit, Steps 2, 3, and 4 are applied to the cable shields rather
than to the condultA, In the case of sh't-!,Jed cables inside steel conduit, Step 3 will Include
calculating transfer imroadance of the cable shield, and In Step 4 the coaxial transmission
line analyzed will be that formed by the cable conductor and Its shield,

159
 A rigorous analysis of the signals induced on conductors inside steel conc'uits is
apparently very difficult.6 However, there are several properties of the conduit and of typi-
cal installations that may be used to simplify the problem and obtain good estimates of the
conductor voltage and current, First, it is noted that rigid steel conduit is a very effective
shield, particularly at high frequencies, so that very little of the external electromagnetic
field penetrates through the conduit walls to the conductors inside. Thus, unless the conduit
is very long (of the order of 106 ft or more), the signal induced through the walls of a con-
tinuous steel conduit with tight couplings is usually negligible compared to that Injented on
the conduit conductors by the aerial conductors at the entrance end, Even for long conduit
runs, the rise time of the ,urrent and voltage transient induced on the internal conductors is
very long (of the order of 1 ms if the steel is unsaturated), so that ordinary voltage limiting
and filtering techniques can he applied.

Second, since only the low-frequency spectrum can penetrate the steel conduit
walls, only the low-frequency currant in the conduit and the low-frequency fields in the
soil need be analyzed, 4, 10, " Because only the low-frequency spectrum Is of interest, the
conduit can be considered electrically short (i.e,, short compared to a wavelength of the
highest freqtiency of interest),

Finally, the depth of burial of conduits is ordinarily only a few meters, so that in
"thelow-frequency spectrum of interest In the analysis of the intornal conductors, the
fields in the soil at the depth of burial are approximately equal to those at the surface of
the ground. One can therefore often neglect the exp (-d/8) dependence of the field
strength on depth d and skin depth 8 - (7rfjuoP)"' in the coil.

The simplifications noted above are valid only for conductors in effectively contin-
uous rigid steel conduits (i.e., steel conduit with tight couplings for the entire run). If the
conductors are routed through plastic ',or other nonmetallic) conduit for all or part of the
run, the above simplifications do not apply, because these assumptions are all contingent on
the high-frequency shielding characteristics of the rigid steel conduit, Even shielded cables
in pleatic conduit may require special consideration because the electrostatic shields on
these cables may not be particularly effective at high frequencies (see Section 3A4).

160
 3.3.2 BULK CURRENTS INDUCED IN BURIED CONDUITS

3.3.2.1 General

This section presents formulas for calculating the total current induced in
a buried conduit by an incident uniform plane wave, for bare or thinly Insulated conduits.
The incident electromagnetic pulse is assumed to have an exponential waveform E0 exp
(-t/ The wave arrives on the surface of the earth from a directlun deflned by an elevation
angle pand an azimuth angle ipas illustrated in Figure 3-15. The depth of burial of the con-
duit is small compared to a skin depth In the soil, so that the fields at the conduit depth are
essentially the same as those at the surface. It is assumed that the conduit is a single, Isolated
conduit with no other conductors in its vicinity.

In all the bulk-current formulas, the soil is considered to be a good con-

ductor, as defined by a >> we. The wave-propagation factor in the soil, bnd the propagation
factor for bare conductors, is thus 3

- 1+1
""= -
8
(3-12)

where 5 is the skin depth In the soil, o is the soil conductivity, and go and e are the perme-
abilitv and permittivity of the soil, respectively. A plot of the skin depth as a function of
frequericv and soil conductivity is given in Figure 3-16. The dashied line along the right-
hand side of Figure 3-16 indicates the limit of validity of the approximation 'Y (1 + j)/I.
Although this limit falls between a few hundred kHz and a few MHz for typical soil con-
ductivities, steel conduit walls have very large attenuation at frequencies above 100 kHz, so
that the bulk-current spectrum at these fre•quencies is usually of secondary interest.

3.3.2.2 Current Far from the Ends of a Long Conduit

The conduit-current formulas presented here apply to points far (several

skin depths in soil) from the ends of long buried conduits that have no insulation, or have
insulation that is thin compared to the conduit radius. The results are presented for an
incident exponential pulse that propagates as a plane wave that is uniform (constant

161
 UIHECTION 0F I'IFOFA(iATION
~ oW INCIIAhNI WAVt

I AM
N0

Fiue31 ILUSRAI N
O ORIAE O ODUTADAGEO RIA

Figur 3-16)
i FgueLUTRTION OFca tORiNATE FORtan
CODI AND ANGL Fonnta
pleIlarRIVALm

pared to the soil time constant re~- c/v. The smallI resistance of the cable Is neglected, and
it isassumed that the, depth of burial issmall compared to the soil skin depth at the highest
frequencies of interest.

The total current induced far from the ends of a long conduit by an Inci-
dent field Eoe-t/7 isgiven by10

I (w) ~ 0(3-13)

in the frequency domain, or

i(t) Io~e~/tlr2 7rJ eU2 du (3-14)

162
 ~u
8.85X 10 1

(10(

In th Dtme tomaie cwsherfenidn ples

Peak electric- field strengith of incident puise (V/rn)

0(0,0) cos ýp for vertically polarized wave
sin 'ksin ýp for horizontally polarized wave,

163
 The waveform for l(t) and the incident exponential pulse are shown in Figure 3-17 as a
function of time in incident-pulse-decay time constant.

The peak current induced in the conduit is50

i - oL
1
v J0 eu 2pk
du rna = 0.61 lo (3-15)

and the peak current occurs at

tpk = 0.85 , (3-18)

The variation of the peak current with azimuth angle of incidence ip and elevation angle of
incidence ý is shown in Figure 3-18. The magnitude of the peak current for maximum
coupling 1 D(0, o) - 1 1 Is plotted In Figure 3-19 as a function of soil conductivity for vari-
ous Incident-field time constants, The plots are truncated in the lower left-hand part of
the graph where r a to. The responses given in this section are valid only If r > to , Where the
approximations used here are valid, the peak current Is Inversely proportional to the square
root of the soil conductivity (L~e,, Vld lo - constant).

Useful insight into the physics of the coupling process can be obtained from
8
the frequency form of the current in the conduit given by

I(z,w) ,"uE~jr62
j I(, '2'83-17)

'Yoe

where E, is the component of the electric field (in the soil) parallel to the conduit, a Is the
radius of the conduit, and yo - 1.781. The current density in the soil is oE, (in the absence
of the conduit), and •62 i the area of a circle one skin-depth In radius in the soil. Thus the
total current induced in the conduit is approximately proportional to the current that would
flow in a circular cylinder of soil one skin depth in radius if the cable were not present.
Because E, o Eo/•Vý through (I + R)Eo [see Eqs. (2-9)], the conduit current Is also propor-
tional to Eo
0 •/', The approximation log (%,/Y/yoa) w.10 is valid (within a factor of 2) for e
wide range (if soil conductivities, conductor radii, and frequencies as can be seen In Figure
3-20. This approximation has been used in deriving Io in Eqs. (3-13) and (3-14).

164

L ,.,.. ...... ....... ,

 10 -~ ~ -lIT 1

• -- • (INCIDENT FIELD)

,t 1- 2

0 :f [ 2 i 0U. du (CONDUIT CURRENT)

0 1 2 3 4 5 6
TIME -- t/"

Figure 3-17 WAVEFORMS OF INCIDENT FIELD, CABLE CURRENT, AND

OPEN-CIRCUIT VOLTAGE BETWEEN THE SHIELDS

3.3.2.3 Current Near the End of a Long Bare Conduit

The conduit current formulas presented here are for points near the end of
a long (semi-infinite) conduit, The bulk current In the nondult Is based on the assumption that
the conduit is In contact with the soil and that the end of the conduit is terminated in a
very low impedance (short circuit) or a very high impedance (open circuit) compared to the
characteristic impedance. The short-circuit case might be representative of a conduit that is
terminated in a large counterpoise or similar low-impedance structure, The open-circuit
case is representative of a conduit that is dead-ended or insulated at the end of the run, The
assumptions that a > w' and a < 6 used for the long conduit above apply to these cases as
well. The results are presented for an exponential pulse with decay time constant r that is
large compared to the soil time constant T. - co/o. The electromagnetic pulse is incident
from a direction defined by an Rlevation angle 0 measured from horizontal and al, azimuth
angle p measured from the axis of the conduit (see Figure 3-15).

165
 11) 4) tgo

V~nTWA1 6

UdO Ili

"HMUNA

V. POLARIZTIO

¾AI(Aru
AN01l11

Figure 3-18 VARIATION OF PEAK CONDUIT CURRENT AS AZIMUTH (,,) AND ELEVATION
10' ANGLES OF INCIDENCE CHANGE lII

Tho total current induced nuar tho end of a Iong conduit that is short-
circuitr.d k idlentical to the current folr fromn thu ends and is!livon by Eqs. (3-1i3) mnd (3-1 41,
dndl has the waveform shown in Fiquire 3 17.

The total current indJuced near the end of a long conduit that is opii-,
vircuitr.d tit thp end by an inclirinit exponential pulse E0 etI is given I)yl10

T 1 /C(3-18)

166
 Ae

0.0

SOIL GUNUUMrIVIIY, 41 mlth

l

Figure 3-19 PEAK CONDUIT CURRENT AS A FUNCTION OF SOIL CONDUCTIVITY AND

INCIDENT CXPONENTIAL PULSE-DECAY TIME CONSTANT 7(D(0,ý) -11

in the frequency domain, and by

Ie-/-jeP/U)eU O(') du (3-19)

167
 20

10 Sa 10"3 _

02 t'104 o
FR.QUENCY Hi

Figure 3-20 VARIATION OF LOG (V2h/7 0 a) WITH FREQUENCY AND CONDUCTOR

RADIUS, (N(ote that a factor of 10 in conductoi radius produces thu samu Ofect at a Inctor
of 100 in soil conductivity.)

in the time domain, where

11) 10O3/Fr- EoD(hý,p) (ampere-seconds)

-. Time constant of soil (s)

T
rJ

T :Decay time constant of incident pulse (s)

E, Poak electric-field strength of incident pulse (V/m)
D(i,,-} cos ; for vertical polarization
D(ý,;) • sin sin € for horizontal polarization
z Distance from end of catle (m)
c 1/v'vfl Speed of light In freu space

C 4rT e

168
 r1

This current is zero (very small) at the end (z 00) and approaches the value given by Eq.
(3-14) at large distances from the end, Plots of the waveform at distances of 3,16, 10, and
31,6 m from the end of tne cable are shown in Figure 3-21 for a soil conductivity of
10-2 mho/m,

0.7

O.1
0.0
10"2_____

0.4 ____

I V-6

01

0 T 2 3 4
TIMV lit

Figure 3-21 WAVEFORM OF THE CONDUIT CURRENT NEAR THE END

WHEN CONDUIT END IS OPEN-CIRCUIT

12
Since the waveform is determined by -- (z/c)2 _ 0z 2 /eoc 2 ,
the waveforms shown in Fig-
Te
ure 3-21 will represent different distances from the end of :he cable if the soil conductivity
differs from 10-2 mho/m. The magnitude of the current remains inversely proportional to
Ssquare ront of the soil conductivity (vu-Io
0 constant), Again, the approximation log
161 10 iRused in deriving I,.
70a

169
 3.3,2.4 Current in an Electrically Short Conduit

The conduit may be considered electrically short If its length Is short com-
pared to a skin depth in the soil of all frequencies of interest, From the analysis of conduit
shielding properties in Section 3,3.3, it is apparent that even if the steel in the conduit is
saturated, the frequencies penetrating the conduit walls lie principally ;n the spectrum below
6 kHz, and from Figure 3-16 the skin depth in average (a . 10-2) soil Is about 60 m and in
poorly conducting (u - 10-3) soil It is about 200 m, Thus, conduits a few tens of moters long
in soil of poor to average conductivity may be considered electrically short at the frequencies
penetrating to the interior of the conduits,

The conduit may then be analyzod as an electrically short dipole In a

finitely crnducting medium. The current at the center of a conduit of length V is then

EjVY
I 2-
2 (3-20)

where Ej Is the field strength In the soil and Y is admittance of a center-fed dipole of
length Vand radius a In the conductive medium. Neglecting the capacitive suceptance (since
u >> wd), the admittance Y is

: Y
Y • G
G •(3-21)
2 log Va

where u Is the soil conductivity and a is the radius of the conduit, The current at the
center Is thus

%-- (3-22)
4 log Va

170
 The z-component of the electric field In the soil, from Eqs. (2-7) and
(2-9) with h = 0 and the phase kz cos 0 cos 0 neglected, is

E,- Elr~ýsinCosfe
psin {(323

where I - e0 /o, EI iLthe Incident field strength, and the upper and lower trigonometric
functions apply to horizontal and vertical polarization, respectively, For an exponential
pulse of the Incident field Eoe-t/t,

S2oV•aIsin 0 In
Ej 2EO {~ uln (3-24)
jW ÷ 1'

The midpoint current In the short conduit is thus

IruEOfR2 •• i osn•1
Ir-
2 log/ Va
i snW
Cos•1P
~_ (3-25),
jW + m,-

for the exponential pulse of Incident field (upper trigonometric functions for horizontal
polarization; lower for vertical polarization). The ratio of this current to the current In an
Infinitely long conduit is

i~ot 5 log",n
Slog Y/a j(3-26)
S 8 log W...

where rý _=Aot2, and the log term in the numerator is plotted In Figure 3-20. The ratio

Is thus of the order of j(rJJ. By definition of electrical shortness, wrk << 1; therefore the
magnitude of the current at the midpoint of a short conduit Is always smaller than the cur-
rent in an infinitely long conduit (or far from the ends of a long conduit). Observe also
that because multiplication by Jw In the frequency domain corresponds to differentiation
with respect to time In the time domain, the waveform. of the current in the time dompin

171
will correspond to the derivative of the waveform shown in Figure 3-17. This waveform
of the c'urrent at the center of the condLit is thus12

il)W eU'2du (t TOe) (3-27)

where

2log R/a

SD(4,,o) sin o sin 0 (horizontal polarization)

cos p (vertical polarization)
- Time constant of incident exponential field pulse.

This waveform Is infinite at t a 0, but the sclut'on Is not valid for times less than r., because
for such times the soil does not behave as a good conductor.

Equations (3-25) and (3-27) give the current at the center of the short
conduit, Since the maximum current occurs at the center, the current anywhere else on the
conduit is less than the values given by these formulas. The variation of the current with position
along the conduit is frequency-dependent (or time-dependent) also, so that a transformable
current distribution or average current that can be used in the calculation of internal con-
ductor voltages and currents is not readily available, To within a factor of about 2 ior these
calculations, however, it may be assumed that ,he current in the conduit is uniform at the
midpoint value,

172
 3,3.3 CURRENT AND VOL TAGE INDUCED ON INTERNAL CONDUCTORS

3.3.3,1 Transfer Impedance of Steel Conduit

The voltage and current induced on ccnductors inside a steel conduit by

current flowing on the conduit r.ýe obtained from the transfer impedance of the conduit. The
transfer impedance of a cylindrical shield is the ratio of the voltage per unit length developed
between the internal conductors and the shield to the current flow;ng in the shield. In Its
general form, the transfer impedance ZT Is defined by

1 dV
ZT dZ
II dZ 13-28)

where V is the cnnductor-to-shield voltage, I is the shield current, and Z is the distance
along the conduit,

For thin-walled tubular shields of uniform cross section that contain no

holes or cracks, the transfer Impedance Is 4 "

ZT ', Ro s
(I+ j)T/ (3-29)
slnh (1 + I) T/6

where Ro is twle dc iesistance per unit length of the shield given by

1
Ro a,
27raTo (3-30)

end T is the wall thickness of the shield, a is the mean radius of the shield, a?is the conduc-
tivity of thu shield, and 6 is the skin depth in the shield given by

= • (3-31)

173
For ferroenagnetic chields such as steel conduit, the permeability, 1j,is an important factor,

Since

1 + 0-= = ( +1) (3-32)

where

2
rs= poT

ff PaT2 7 •S

the transfer impeeince may also be expressed in terms of the diffusion time-constant r, or
the frequency f6 at which the wall thickness is one skin depth, The dimensions of standard
steel contiult are given in Table 3-4. A plot of the magnitude of the t'ransfer impedance as a
function of frequency is ohown in Figure 3-22 in a normalized form that can be applied to

Table 3-4

PROPERTIES OF STANDARD GALVANIZED 0R ENAMELED

RIGID STEEL CONDUIT
Size Diameter (inches) Thickness Threads pe,' Weight per
inches) External Internal (inches) inch 100 Feet (Ibs)

0,640 0.622 0,109 14 79

1.050 0
0,824 0,113 14 105
1 1,315 1,049 0.133 111/ 153
11/4 1,660 1.380 0,140 11% 201
l1/ 1,900 1.610 0,145 11// 249
2 2.375 2.067 0.154 111/2 332
2'/ 2,875 2.469 0,203 8 527
3 3,500 3.069 0.216 8 682
3% 4.000 3,548 0.226 a 831
4 4,500 4.026 0,237 8 972
5 5.563 5.047 0,258 8 1314
6 6.625 6.065 0.280 8 1745

174

..
.
 CONDUIT
No NOMINAL R
SIZE (Hz 1.l_______
10-, S(inches) P Ur 1 -

LU0.5 1.03 x 10- 11.0 5507

0.76 7 77 . 10-4 10.3 5124

LU 1 6.23 x 10 .40 3609

LL-4 3186 x 10 6.68 333g9
11.2E,
,1.5 3,23 x 104 U.,22 3112
-4
2 2.40 x 10-4 1352 2759
2.
2.N 1.52 x 10 - 3.18 1588

""- 3-- 1,16 x 10"4 2.81

7.1 1403

3,5 9.64 . 10" 2.56 1281

12 8.14x 10"15 2.3 1165]G

ic- 3
011 10 100

f8

Figure 3-22 MAGNITUDE OF THE TRANSFER IMPEDANCE OF RIGID STEEL

CONDUIT (a =Fj x 106 rnho/rn)

any tubular shield, The norialization values of Roandf6 for standard rigid steel conduits
are shown in the table inset in Figure 3-22

Note that for frequencies significantly above f,, the transfer impedance is
quite small, so that only the frequencies below f1 , are effective in inducing current and voltage
on the conductors inside the conduit. Even if the conduit steel is saturated, the cutoff fre-
quency f, is less than 6 kHz, so that the conduit run may be considered electrically short (for
the analysis of internal current and voltage) if its length is

c
V&---' 50 km

175
Since most feeder conduits are only a few hundred feet long, they may usually be considered
electrically short for purposes of analyzing coupling through the conduit walls.

3.3.3.2 Internal Voltage Induced in an Electrically Short Conduit-

Uniform Exponential Current

For an electrically short conduit of length Pwith a uniform current I

induced by external fields, the total induced internal voltage will be

V(W) = IZTQ (3-33)

where ZT is the transfer impedance of the conduit, If the internal conductors are opon-
circuited with respect to the conduit at both ends, half of this voltage will appear between
the conductors and the conduit at each end,

When the current I is that derived from an exponential pulse I 0e-t/T and
the transfer impedance from Eq. (3-29) is used, the open-circuit voltage at one end of con-
10
ductors that are oper-circuited at both ends Is

V(W) - (3-34)
2 (iW +1 sinh VAr

where

Ro (2raoT) 1 = dc resistance per meter of the conduit

2
Ts MoT = Diffusion time constant for the conduit
7 Decay time constant of the exponential pulse of conduit current.

176
The quantities a, u, 1, and T are the radius, conductivity, permeability, and wall thickness
of the conduit, respectively. The voltage waveform is

I 0 R0Q/R /\32(2
v(t) = -- . t/ exp 12n - 1)2 (1 1

(3-35)

v(t) 2- 1 rs) n 1 exp[(2n 1)2] (r )D r") (3-36)

The waveform for r <<,r, is r times the Impulse response of the conductors, and the wave-
form for >> 7s Is the step-function response of the conductors. Plots of these waveforms
and the response when r" -s are shown in Figure 3-23 normalized to IcRo0 /2 in magnitude
and to I,
In time,

The peak open-circuit voltages are

V(t)peak = 2 (r i> -r

5 9 .L o.R ('K rs) (3-37)

.r\ 2 /
0O.77 1 R ( s

177

. . ... ........... 2
 T <
S4
3

> 2

0 0.1 0.2 0.3 0.4 0.5

TIME -t/7

(a) IMPULSE RESPONSE

0.5 - - - . - -

0 0.1 0.2 0.3 0,4 0.5 0.8 0.7 0.8 0.9 1.0
TIME - Or7

1b) STEP FUNCTION RESPONSEE

0. 01 02 03 04 0 . . . . ,

()EXPONINTIAL CRESPNS

Flgur.~~~~~~~~~ RDCDB
3-2 VOTG2AEOM
A XOETA ~~~~
US 0 -
OF~~ ~ ~ ~ ~~~~~~?
CURN1NTE1HED0Tmale
oV/Z 0 0 2

0 Ole 0 171
if the conductors are open-circuited (relative to the conduit) at both ends. The 10-to-90%
rise time for the voltage is

t10.90 =0.236 rs (7 *' rs)

- 0,038 rs (r rj)
7 (3-38)

- 0.15rs (T . rs)

Plots of the asymptotic values of the peak voltage and rise time against r/Tr are shown in
Figure 3-24 as solid lines, and an estimate of their behavior between asymptotes is shown
as a dashed curve. Values of r, and Ro for rigid steel conduit are given in Table 3-5,

LLF -

I- w

CLI

001 V., 0 01

Dl ( 0 1 10

Figure 3-24 VAr"IATION OF PEAK OPEN-CIRCUIT VOLTAGE AND RISE TIME OF

CONDUCTOR VOLTAGE AS A FUNCTION OF EXPONENTIAL CONDUIT-CURRENT-DECAY
TIME CONSTrANT r (norm'Iiped to diffusion constant rd)

179

......
 Table 3-5

SHIELDING PARAMETERS FOR RIGID STEEL CONDUIT

T_
Nominal
Conduit Ro ur - 500 ur = 1.0 Q5
Size (inches) (ohms/rn) (Ims) (ps) (A-9)

1/2 1.03 x 10-3 28.9 57.8 2.1

% 7,77 x 10- 4 31.1 62.1 2.9
1 5.23 43.0 86.0 5,2
1/ 3.86 47.7 96.3 7.4
1V/ 3.23 51.1 102 9.1
2 2,40 57.7 115 13
2V/ 1.52 100 200 27
3 1,16 113 227 38
31/2 9.64 x 10-5 124 248 48
4 8.14 137 273 59
5 6,01 162 324 87
6 4.63 1D1 381 120

The voltage has the same magnitude and shape at both ends of the conduit;
however, the polarity at one end is opposite to the polarity at the other end. If the conductors
are shorted to the r'rndult at one n, '.Aie open-circuit voltage at the opposite end will have
the same waveform Lut it will double in magnitude,

The current through matched terminations Z. at the ends of the conductors

will be

I(o) =V(W)/Zo (3-39)

180
 3.3.3.3 Internal Voltage from Incident Field

The results presented here give the voltage between the conductors and th
conduit induced by an exponential pldne-wave pulse i,;cident on the surface of the ground.
The results apply to buried conduits that are short compared to the shortest wave-length
penetrating thb wall (see Figure 3-22) but long compared to a skin depth in soil. The results
are obtained from the convolution of the conduit current obtained in Section 3.3.2.2 with
the impulse response of the conduit obtained In Section 3,3.3.2.

The open-circuit voltage induced between the conductor and the conduit
by an incident exponential pulse Eoet/1 is given by 10

tIo Rog " 1

v(.=-524,1-- (JQ + 1/0 sInhvi'r (3-40

where

1o = 106 V 4/YEoDb(I,Ip) (see Eq. (3-14))

2
rg = uoT is the diffusion constant for the tubular shield.

The open-circuit voltage waveform is

SIO0 e-st/e /1 a(rl/T)v n

v(t)
2a 7f/e g 3/2 (3-41)

181
 This waveform, normalized to 2-lt- is plotted in Figure 3-25 for several Values of 7,/r.

In deriving Etjs, (3.40) and (3-41), it has been assumed that the conductors
xre oein circuited with respect to the conduit at both ends, and that the conduit run is long
enough that the conduit current may be considered uniform throughout the length of the
conduit (i,e,, end effects such as those described in Section 3,3.2,3 are negligible).

3.3.3.4 Saturation of Steel Conduit

Much of the shielding effectiveness of steel conduit iscaused by the large rela-,
tive permeability of steel, When very large currents flow in the conduit, however, the magnetic

•',~
.2
6 ~

0
) 1et

Figure 3-25 COHE*-TO-SHIELD VOLTAGE WAVEFORMS INDUCED IN CONDUCTORS IN A

BURIED CONDUIT BY AN INCIDENT PLANE-WAVE EXPONENTIAL PULSE r.-/

flux density in the steel may become large enough to saturate the steel mnd drastically reduce
its permeability. It has been demonstrated that most ferromagnetic materials do not become
completely saturated instantaneously.13 As saturation beginis, the surface impedance of the
saturated outer layer is much lower than the surface impedance of the interior unsaturated
region, so that most of the current flows in the saturated layer. I the current is of sufficient
magnitude and duration, it will eventually saturate the entire conduit wall, For transient

182
currents of short duration, however, the entire thickness of the wall may never saturate, so
that the ferromagnetic shielding properties may be maintained even if partial saturation occurs,

Based on an analysis that assumes that the shield material is either comr
pletely saturated (jAr - 1.0) or completely unsaturated, the depth of saturation Is found to
be dependent on the charge transferred by the shield. For complete saturation of a tubular
shield by a single transient, the saturation depth must approach the thickness, T, of the
shield at some time t <o9, giving
Q6 Idt ffaoBsT 2 (3-42)

where B, is the flux density at saturation, a is the radius of the shield, and IIs the conduc-
tivity of the shield. The charge Os is the total charge that must flow along the shield to
saturate the shield material completely. For a current step function, Qs Is therefore a
measure of the time required to ssturate all the way through thu shield, sinco

0s = Iott = fiaoBT:- (3-43)

Any value of 1o greater than that required to begin saturation will oventually saturete the
shield all the way through, but theo smaller the value of 1o, the longer It will take to saturate
the shield completely. For a rectangular pulse of width r and amplitude 1, we ha~le

lo' = jraoBgT 2 (3-44)

and the current 1o required to saturate the shield completely increases as the pulsewidth
decreases. Similar relations can be obtained for other pulse shapes. For an exponential
current loe-t1T, for example,

loT = iraoBjT 2 (3,45)

108
 The values of Qs are tabulated in Table 3-5 for rigid steel conduit with a '. 6 X 106
2
mho/m and Bs = 1.6 w/m (16 kilogauss), Note that for pulse durations of about 1 ps, peak
currents of tens of megamperes are required to completely saturate 2-inch and larger con-
duits. Therefore, complete saturation of rigid steel conduits by the induced current from
the high-altitude EMP is unlikely.

3.4 CABLES IN NON-METALLIC CONDUIT

3.4.1 GENERAL CONSIDERATIONS

As pointed out in Section 3.3, nonmetallic conduit offers no shielding to the conduc-
tors Inside the conduit, Therefore the electric field In the soil induces current and voltage
directly on the conductors In the conduit, and the high-frequency as well as the low-fre-
quency spectra are important, The current induced on the cables may, in fact, be compara-
ble to that induced on the metallic conduits discussed In Section 3,3, The current differs
from that induced in the metal conduit only because the conductors are not In direct con-
tact with the soil, Thus the coupling to the conductors is through the capacitance between
the conductor and the soil and through the terminating Impedances at the ends of the cable,

Because the cables are insulated from the soil by the conduit and by the air and cable
insulation inside the corndult, the attenuation of induced currents is somewhat less (particu-
larly at low frequencies) than it Is for conductors in difect contact with the soil. For conduc-
tors in direct contact with the soil (such as metallic conduits), the current Is attenuated as
exp (-z/6), where 8 is the skin depth in the soil, and z is the distance along the conductor,
At low frequencies such that

we logr1 /a
0 1 (3-46)
log

184
where rl Is the radius to the outside of the conduit and a is the effective radius of the con-
ductors, the admittance per unit length between the conductor and the soil is dcminated by
the capacitance of the insulation, and the attenuation constant is much smaller then 1/6.
At high frequencies where the admittance per unit length is dominated by the admittance of
the soil, however, the buried insulated conductor behaves in approximately the same man-
ner as the buried bare conductor. Because of the lower attenuation at low frequencies, the
conductors in nonmetallic conduits several hundred feet long may support current resonances
at frequencies of several hundred kilohertz (see Section 2.2.1).

When shielded cables are used in nonmetallic conduit, the cable shield provides some
protection for the conductor inside. The shields on power cables are usually designed pri-
marily for electrostatic gradient control, however, rather than electromagnetic shielding.
These shields arc often fabricated of thin copper tape spiraled around the conductor insula-
tion (with some overlap) to form a smooth outer conductor at low frequencies. Because
the contact resistance of the overlapped tape is large compared to the resistance of the cop-
per, the current in the shield tends to flow in the direction• of the tape rather thaih parallul
to the axis of the cable, Thus the tape-wound shield behaves as a solenoid, and at high fre-
quencies the jwLI drop along this solenoid can be quite large (large enough to cause arcing
between the turns of the tape), This jcwLl drop drives the conductor inside the shield, and
because of the w dependence, the shield tends to readily pass the high-frequency spectrum
(in contrast to the steel conduit, which almost completely eliminates the high-frequency
spectrum).

3.4.2 CURRENT INDUCED ON CABLES

The current Induced in buried Insulated cables can be calculated in the same way as
the current in the aerial transmission lines described In Chapter Two, except the excitation
fieid is that at the surface (or just below the surface) of the ground and the characteristic
impedance Zo and propagation factor y of the huried conductor are used to replace those

185
of the aerial transmission line. Consider the buried, insulated conductor shown in Figure
3-26, If the conductor extenus from z - -• to z = 0, the current at the end z 0 is

(01 e-() I k')k ! - jk'/V

I(0'. I• . -2 - 1 - e--- . ...... .• - 2"

(3..47)

x1- p1joe 2"o

TTII
CONDI)U ( f

SOIL (0 - 1 "INSULATION (t )

Figure 3-26 BURIED INSULATED CONDUCTOR

where

y -. Propagation factor for the cable/earth line

k..... cos , Bsin

c

Z. Terminal impedance at z -z

ZO Terminal impedance at z - 0

186
 Z4 - Zo
(3-48a)
ZV + Zo

ZO - Zo
Ho (3-48b)
Zo + Zo

The current Izw is the short-circuit current at thw end z = 0 of a semi-infinite buried con-
ductor extending from z = -= to z - G. This short-circuit current is given by

2E1 I cos@' ý Y ("

SC Zoy
0 sin Vi sin p5 1- jk'/-y

where rp et¢/u (o is the soil conductivity), and the upper trigonometric function applies to
vertical polarization and the lower applies to horizontal polarization of the incident field
Ei (see Figure 3-15 for spherical coordinate system),

The propagation factor y and the characterlstic Impedance Z, are obtained from the
impedance per unit length Z and the admittance per unit length Y with

7o vi , 3-50)

For low frequencies (f -.. o/2,fe), the impedance per unit length is

LL4o /JTh
"Z....W-+ j-I o'"-
lug V3 b1)
8 2rr toa

where p. is the permeability of free space (and the soil), (Sis the skin depth in the soil, a is
the effective radlus of the conductors, and -Y, = 1.781,,.

187
The admittanc; per unit length Is

1
Y = jWCi (3-52)
1+ -
Ys

where CuIs the capacitance rir unit length between the conductor and the soil, and YJ is the
admittance per unit length of the soil, The capacitance Is

Ci = log21rrl/ (3-53)

wnere rl is the radius to the outside of the conduit. The admittance per unit length Y, of
the soil is given by

JWUOdT 21ro
N', W= Q P
-- - - lo g Io 2- (-4
8 2ir C1orl fhorl

dince o•Ci/Yj << I when

"
'/2
lo g
we 'o0 rj

log- rl
a
it follows that

log---2--\;_

(
log
a

188
at low frequencies. The impedance per unit length Z is, therefore, primArily inductive reac-
tance with a small resistive component R = wpo/8, and the admittancr, per unit length Y is
primarily capacitive susceptance jwCi with a small conductive component

6-
27r log
w2e2 forl (3-56) .
G = --- -- )

0 (,or-og-)
a

The attenuation constant a Is,

R GZo
a ft Re(b') .... + (3-57)
2ZO 2

where R and G are given by the real part of Eq. (3-51) and Eq. (3-56k, respectively. The
characterlstic impedance Is

ZO - er log-log (3-58)

where 1o, = 1207r. The phase faentor P'is approximately

er log (3-59)

C logrl/a

The quantities a, O/k, and IZol are platted in Figures 3-27 and 3-28 for a typical 4-
Inch conduit In average soil. It Is apparent that IZo0 Is fairly Independent of frequency, N/k
is moderately Independent of frequency Pnd of the order of 3, and a Isstrongly dependent
on frequency but so small that attenuation in propagating a few hundred feet Is negligible at
frequencies below 1 MHz.
189
 The values of Z and Y can be substituted into Eqs. (3-50), (3-49), and (3-47) to
a
evaluate the current in the conductors for frequencies f < -- , Numerical techniques can be
21rE
used to obtain the inverse transform of Eg. (3-47). Some approximations can be made that
permit estimates of the current to be obtained without extensive mac'hine computations. In
Eq. (3-49), for example, the term Zo-y can be approximated by

0 V21
Zo* Z f j- -log-9 - (3-60)
21r -too 27r

and the quantity jk'/,y Is

Ik' k cos -- cos ,p =(34-1)
k'•

400

UW
z 300--3
uJ
22

"0 0 ,. 10'2 Frho/m

S100 1

U r1 " 2.25 in.

0
1010 106

FREQUENCY Hz

Figure 3-27 CHARACTERISTIC IMPEDANCE Zo AND PROPAGATION FACTOR

P FOR CONDUCTORS IN A BURIED NONMETALLIC CONDUIT

190

l-:
 10-2

E
I
S10-3

1 -4 0 10"2
6
•< r - 1.5

- 225 In,
a 1 In,

lu6- I
10 4 105 10a

FREOUENCY -- Ht

Figure 3-28 ATTENUATION CONSTANT a

FOR CONDUCTORS IN A BURIED NON-
METALLIC CONDUIT

so that Eq. (3-49) becomes

IFIV'r (Cosp 1P
106 1 -kD/3 sin ksin)' 4/F (3-62)

The terms 1 ± jk'/y, in Eq. (3-47) have been replaced by 1 k/fJ.

W

For an exponential pulse Eoe-t/T whose transform is Eo/(Jw + I/r), the short-circuit
current Ilc - above Is the same as that given by Eq, (3-13) for a long, hare conduit. The
waveform of this current is shown in Figure 3-17 and its amplitude can be obtained from
Figures 3-18 and 3-19. For the finite length of conduit, this waveform Is modified by the
exponential terms in Eq. (3-47), which In the time domain represent combinations of the
basic waveform of Figure 3-17 with delay and attenuation.

191
 A plot of the Induced current waveform for an incident exponential pulse of decay
time constantr - 0.5 ,s is shown in Figure 3-29 for a 100-ft-long ceble short-circuited at
both ends. For comparison, the short-circuit current Isc 0. induced in a semi-infinite cable
is also shown as a dashed curve in Figure 3-29. The short-circuit current is normalized to
the quantity

2Eo0 N'T v 1cos• 1

I0 0 k l sin 4psin pj (3-63)

where v - w/j. The upper trigonometric function applies to vertical polarization, and the
lower applies to horizontal polarization. For the waveform of Figure 3-29 the attenuation
a was neglected, and the phase factor 0 was assumed to have a constant value of 3k (see
Figure 3-27). It is observed that the peak short-circuit current induced in the 100-ft-long
cables Is almost as large as that Induced In the semi-infinite cable, but it has more "fine
structure" because of the end effects associated with the finite length, It is also observed
that the value of 1o given above Is the same as that given In Eq, (3-14) for a long buried
conductor if v c, k/P << 1, and the approximation of Eq. (3-60) Is used for Zo/v Zoy/jW.
0

The source impedance of the Norton equivalent source associated with the short-
circuit current is

1 +p~e-2,yQ
Z(0) = Z (3-64)
I - pve-2,yV

where pV is given by Eq. (3-48). The open-circuit voltage at the end z 0, when the other
end has a reflection coefficient pQ is, therefore,

Voc(O) =IC(O) Z(O) (3-95)

192
 I I

0.5

w t

VOLTAGE
p5 I-
U D 0
U

.. i Q - loo ft
X. I

0
S-30'

5 o0

01,O 20 3.0

TI"ý

Figure 3-29 SHORT-CIRCUIT CURRENT AND OPEN-CIRCUIT VOLTAGE INDUCED IN

CONDUCTORS IN A BURIED, NONMETALLIC CONDUIT

where Iscl0i)Is given by Eq, (3-47) when po, - -1. The voltage waveform Is thus similar to
the current waveform except that the reiiected waves alternate in polarity. This can be seen
by examining the last turm (outside the braces) in Eq. (3-47):

1 • Ep~e-27yVn (3-66)
1 PyPoe-2-V F.:0'

i '1,./ l ',....= ....... ... . . ........ . . ...... ....

: '•"1"'•, .... ........ .. .. ..

"000193
When po = -1, as is the case for the short-circuit current at z 0,

1 _. •
p•exi L-Ppe-2ýJ (short-circuit current) (3-67)

1 + pvte-27k! M=-0 .

and when this is multiplied by Z(0) to obtain the .open-circultvoltage,

1 o20
(0) - a Zo E [pe-2'Vn (open circuit voltage).
1 + p~e-2) 1 - pve-2 •• 0
(3-68)

The odd terms in the voltage series are thus opposite in sign to the corresponding terms in
the current series when PQ - -1.

The open-circuit voltage waveform is also plotted In Figure 3-29 (dotted curve) for
the 100-ft-long cable with the end z - -Q short-circuited, The voltage is normalized to
Vo - 1o Zo, For the conditions assumed here and In Figures 3-27 and 3-28, a 50-kV/m
exponential pulse with a 0 .5 -ps decay time constant would induce a short-circuit current of
600 A or an open-circuit voltage of 90 kV at the end of the 100-ft-long cables in non-
metallic conduit.

3.4.3 TRANSFER IMPEDANCE OF CABLE SHIELDS

When the cables In the nonmetallic conduit are shielded cables, as is generally the
case for cables used between the potheads and ground-based transformers, 'the currents dis-
cussed in the preceding section are induced In the cable shields, As indicated in Section
3.4.1 the shields for these cables are often spiral-wound copper tape and the tape or spiral-
wound shield tends to behave as a solenoid wound about the internal conductors. There is,
therefore, strong coupling between the shield current and internal conductors. For large
dlI/dt (jw 0), the voltage developed pe. turn can become large enough to produce arcing
between turns. (When this occurs, the shield actually improves, because the coupling is reduced).

194
 For a tape-wound shield in which the turns are ,oo *•verla~oed and have no contact
between turns,* the geometry of the shield can be -'<.ned by the radius, a, of the shield and
the width, w, of the tape as illustrated in Figure 3-30, This geometry may also be defined
in terms of the spiral ingle 0 or the turns per unit length N, which art related through

w sin O
cos ce , N s - (3-69)
27ra w

and

2iraN tan 0

The shield current I has the same pitch as the tape and is assumed to be uniformly distributed
across the tape width, w.

4
The transfer Impedance of the tape-wound shield Is

ZT . . ZTO Z + WP+ )tan2 0 (3-70)

I dz 47r

yT
where ZTO uR and Zi a RoyT co h yT, T is the tape thickness and y (1 + j)/6
sinh yT
(6 is the skin depth in the tape). The magnitude of the transfer impedance ZT, normalized
to the low-frequency transfer Impedance Re of a cylindrical tube of the same wall thickness,
is shown in Figure 3-31 for various spiral angles 0. The low-frequency transfer Impedance
Re is the dc resistance per unit length of a tubular shield of radius a and thickness T. Under
this normalization we obtain

ZT yfT T1(-1
S slnhT + ,T coth yT + 21 tan 2 0 (3-71)
-T6 Losn TJ

*That it, the contact resistance is large compared to the resiktance of one turn of tape.

195
 (a) TAPE-WOUND SHIELD

(b) ONE TURN DEVELOPED ON A PLANE

Figure 3-30 ILLUSTRATION OF PARAMETERS FOR A SINGLE-LAYER TAPE-WOUND

SHIELD

where 80 Is the skin depth for a nonferrous shield (p -jo), whether or not the material is
ferrous, with a conductivity a, In Figure 3-31, a/T u 10.

From Figure 3-31 It Is apparent that for low frequencies (T/6 0 << 1) the transfer
impedance increases with increasing spiral angle 0, This follows from the fact that a :arger
spiral angle 0 Is accompanied by a narrower tape (smaller w) and more turns per unit length
(larger N). Thus the tape required to wind a unit length of shield Is longer and narrower, and
has a large resistance, for large spiral angles.

At high frequencies (T/6 0 >> 1) the transfer impedance Is dominated by the induc-
2
tance term, which In proportional to (T/6,M) . The magnitude of the transfer impedance
again increases as the spiral angle o increases, because the number of turns per unit length and
the inductance per unit length increase.

196
 0 so80

00

Nw, 5

00

'0 1.0 10
NORMALIZED TAPE THICKNESS,L

Figure 3-31 MAGNITUDE OF 1HE TRANSFER IMPEDANCE COMPUTED FOR

A TAPE-WOUND SHIELD

Similar expressions have not been developed for the overlapped tape-wound
shields, but the high-frequency characteristics of a single-layer-overlapped shield with tape
width w and overlap width w. can be obtained by substituting w - wo for w In Eq. (3-69).
At low frequencies (T/6 << 1)the transfer Impedance obtained with this substitution will
be too large by a factor w/(w - wo), however, and at very high frerniuncies, the capacitance

197
between the overlapped turns will begin to short out the inductance so that the magnitude
of the transfer impedance will reach a maximum and then begin to decrease,

Between the low-frequency (resistive) region and the high-frequency (capacitive)

region, the tape-wound shield behaves as a solenoid of inductance per unit length

L = tan2o (3-72)
41r

which, for a winding angle 0- 45', is 0.1 JH/m. For current rates of rise of tens of kilo-
amperes per microsecond that might be induced in the cable shield by a 50-kV/m Incident
exponential field, the voltage Induced between the conductor and the shield would be

dl
V - Ld-- 10-7 X 1010 103 V/m. (3-73)
dt

Thus, several kilovolts per meter would be Induced on the conductors Inside the tape-wound
shield, compared to less than 1 V/m for conductors Inside steel conduit, Typical service-
entrance cables cannot be considered electrically short at the frequencies penetrating the
tape-wound shield, For this reason, a transmission-line solution such as that described In
Eqs. (2-3 through 2-5), In which Ez - IZT, Is required to obtain the waveform of the
Induced voltage.

Shields formed with a longitudinal overlapping seam (0 - 0) and multilayer spiral-

wound shields with alternating wlinding directions for alternate layers display a much smal-
ler Inductive coupling effect (typically less than 1 nH//m). Even these shields are murh
poorer than rigid steel conduit, however, because they are thin (usually a few mils) and non-
ferrous (copper or aluminum). For example, a 5-mil-thick longitudinal copper-tape shield
on a 3/4-inch-diameter cable has a dc resistance of 2.3 X 10-3 ohm/m and a cutoff frequency
(at which T - 6) of 2.7 X 105 Hz, while a 4- Inch rigid steel conduit has a dc resistance of
only 8A1 X 10- ohm/m and a cutoff frequency of only 2,3 Hz. Thus the gain-bandwidth
product (Rof&) for the thin copper shield is about 3 X 106 times greater then that of the
steel cunduit. As a iusult, one cali usually neglect the effect of the cable shield5 on shielded

198
cables routed through continuous steel conduit. When the cable shield is the only shield
protecting the conductors from the fields in the soil, however, its shielding properties are
important in the analysis of the current and voltage induced on the conductors.

3.5 CITED REFERENCES

1. Electrical Transmission and DistributionBook, 4th Ed. (5th printing)

(Westinghouse Electric Corporation, East Pittsburgh, Penn., 1964).

2. E. F, Vance and S. Dairiki, "Analysis of Coupling to the Commercial Power

System," Technical Report AFWL-TR-72-21, Contract F29601-69-C-0127,
Air Force Weapons Laboratory, Kirtland Air Force Base, New Mexico (August
1972).

3, S. Ramo and J, R. Whinnery, Fields and Waves in Modern Radio, 2nd Ed.,
(John Wiley & Sons, New York, N,Y,, 1953),
4. H, Kaden, Wirbelstrome und SchIrmung In der Nachrichtentechnik (Springer-
Verlag, Berlin, 1959).
5, W, R, Smythe, Static and Dynamic Electricity, 3rd Ed., pp. 77, 78, (McGrvw-
Hill Book Co, New York, N.Y., 1968).
6, E. D. Sunda, Earth Conduction Effects in Transmission Systems, (Dover
Publications, Inc., New York, NY., 1968).
7. DASA EMP (ElectromagneticPulse) Handbook, DASA 2114-1, Chap. 11,
(DASA Information and Analysis Center, Santa Barbara, CalIf., September 1968).
8. A, L. Whitson and E. F. Vance, "Measurement of Fields Near a Vertical
Monopole," Sensor and Simulation Note 152, Stanford Research Institute,
Menlo Park, Calif, (September 1965, unpublished report),

9. E. F. Vance, "Internal Voltages and Currents in Complex Cables," EMP Inter-

action Note VIIlI, Stanford Research Institute, Menlo Park, Calif. (June 1967,
unpublished report).
10. E. F. Vance, "Prediction of Transients In Buried Shielded Cables," Interim
Technical Report, Contract DAEA18-71-1..0204, SRI Project 2192, Stanford
Research Institute, Menlo Park, Calif. (March 1973).

199
11. S. A. Schelkunoff, "The Electromagnetic Theory of Coaxial Transmission
Lines and Cylindrical Shields," Bell Svstem Tech. J., Vol. 13, pp. 532-579
(October 1934).
12, G. A. Korn, Basic Tables in ElectricalEngineering (McGraw-Hill Book Co.,
New York, N.Y., 1965).

13. R. R. Ferber and F. J. Young, "Enhancement of EMP Shielding by Ferromag-

netic Saturation," IEEE Trans. on Nuclear Science, Vol. NS-17, No. 6, pp. 354-
359 (December 1970), Also see Proc. IEEE, Vol. 61, pp. 404-413 (April 1973).

44

200
Chapter Four

PROPERTIES OF DISTRIBUTION
TRANSFORMERS

4.1 INTRODUCTION

4.1.1 EFFECTS OF TRANSFORMERS AND LIGHTNING ARRESTERS

The commercial electric power for operating ground-based facilities is typlcal!y

supplied from three-phase distribution lines with a three-phase ground-based or pole-
mounted transformer bank at the facility to reduce the voltage to 120, 240, or 480 V.
Three single-phase transformers are usually used for the three-phase system. The low-
voltage leads from the transformers are carried into the facility in plastic or rigid steel
conduit. If the power system Is subjected to the EMP of a nuclear detonation, extremely
large voltages will be induced in the long, above-ground transmission lines supplying the
primary side of the transformer. Part of this induced voltage will be passed by the
transformers and will enter the facility on the low-voltage power conductors. In addition,
the induced voltages can be large enough to fire the lightning arresters installed between
the primary terminals of the transformer and ground. However, because of the fast rise
time (tens of nanoseconds) of the induced voltage, the lightning arresters fire at much

201
higher voltages under the EMP excitation than they do with the more slowly rising lightning
transients.

Single-frequency, or continuous-wave (CW), measurements of transformer transfer

characteristics indicate that distribution transtormers of different manufacturers and
winding design are quite similar to each other. All transformers display a bandpass-filter
characteristic for common-mode excitation, with the frequencies below b eow hundred
kilohertz suppressed by the interwinding capacitance and frequencies above about 10 MHz
suppressed by the bushing inductance. Generally, the type of winding, the manufacturer,
and even the kVA rating have no major effect on the transfer characteristics, although each
transformer has its own fine structure - for example, In the voltage transfer functions.

The results of transient tests of the transformers at excitation levels below the
lightning-arrester firing threshold show that the rise times of the primary current and the
secondary voltages are stretched by the transformer bushing inductance, and the late-time
response of the secondary voltage is suppressed by the interwinding capacitance. Thus,
a fast-rising, wide-excitation pulse applied to the primary terminals produces a narrow
pulse with a slower rise time at the secondary terminals. The peak secondary-terminal
voltage for a 1-V common-mode step excitation of the primary terminals is 0.2 to 0.3V,
with a time-to-peak of about 70 ns and a principal response duration of about 200 to
400 ns.

At an excitation level of about four times their rated voltage, distribution lightning
arresters fire and limit the primary voltage. Near the threshold, 150 to 200 ns are required,
after application of primary excitation, for the lightning-arrester spark gaps to ionize and
become conducting. As the excitation level Is increased, this time lag is decreased, but
fast-rising excitation voltages may reach very large values before the lightning arresters
fire. Because of the Inductively limited rate of rise of the primary current, however, it is
theorized that most of the initial applied voltage appears across the bushing inductance,
rather than across the winding insulation. The shortest time-to-fire that has been observed
experimentally was 40 ns, at a firing voltage of about ten times the rated voltage (2.5
times the static firing voltage). It has been observed, however, that in spite of the large peak
volteges that can be attained befoie the lightning arrester fires, the lightning arrester always
fires before any of the transformer insulation (e.g., bushings, winding insulation) fails. Thus,

202

S-~~.--- .. .........- ---

the lightning arresters appear to be effective protective devices for the fast-rising EMP
transients as well as the slower-rising lightning transients.

The effect of the lighttiing-arrester -firing on the secondary voltage waveform has
also been examined. At excitation levels onl' -lightly greater than the firing threshold, the
lightning-arrester firing has very little effect on the peak secondary voltage, because the
firing time-lag Is about as wide as the main secondary response (about 200 ns), At greater
excitatiun levels, where the firing time-lag Is significantly less than 200 ns, the peak
secondary voltage ceases to increase linearly with the excitation level, At still greater
excitation levels, where time-to-fire is equal to or less than the time-to-peak of the secondary
voltage (In the absence of the lightning arrester), the peak secondary voltage becomes virtually
Independent of the excitation level. For the 7.2/12.5-kV transformer exhibiting the
strongest common-mode coupling, this saturation level was about 30 kV at the secondary
terminals. The lightning arresters at the primary terminals are therefore effc.ctlve In limiting
the EMP-Induced throughput of the transformers as well as In protecting the transformer
from damage,

4.1.2 TRANSFORMER CONSTRUCTION

Power transformers are available in a variety of shapes and sizeg; the discussion here
is limited to a presentation of the major characteristics of the oil-insulated distribution
transformers of the type commonly used for single-custonmer service. In these transformers,
the transformer core and windings are submersed In a metal tank of Insulating oil that circu-
lates by natural ur forced convection. Glazed porcelain feedthrough bushings provide access
to the transformer windings; the larger primary bushings are commonly mounted on the tank
lid, while the smaller secondary bushings are usually arranged along one side of the tank (above
the internal oil lovel, Fýigure 4-1 shows a photograph of two pole-mounting transformers.

The intarnal construction of a typical transformer Is Illustrated In Figure 4-2 with

the photographs of a 25-kVA GE. unit taken from the primary and secondary s'ies when
the core and wind~ngs were removed from the housing for a final damage inspi tion, The

203

S~ ~ -,.
Ir
61ra.r

F1I9ure 4-1 TYPICAL POWER-DISTRIBUTION TRANSFORMERS

switch and knob visible In the photographs are part of the primary-side tap changer, The
single primary winding is actually split in the center and several taps aro made on each side;
this permits the number of primary turns to be varied somewhat to compensate for slight
undorvoltages (the adjustment range Is +0, -10% in 2.5% steps). The types of windings and
insulation used in the transformer depend on the time of manufacture of the transformer.

204
 0

LU
U
LJ.J

6L M

N 0,

206
Older transformers are wound with round or rectangular copper wire with paper or cambric
insulation (In addition to oil) between layers. At the tirs of this writing, however, common
practice is to use aluminum strip for the secondary windings. Because economic factors
dictate the design of power transformers, the relative costs of copper and aluminum and
the cost of manufacturing processes at the time of manufacture have strong influences on
transformer construction.

In shell-type transformer construction 4illustrated in the Inset In Figure 4-2) the

transformer core commonly consists of two roughly square cores butted together along
one side, with the windings linking each hollow square. The windings of the transformers
consist of concentrically layered primary and secondary windings with the primary (high-
voltage) winding iandwiched between the secondary windings. This construction contributes
to asymmetry in the coupling between the primary and the secondary windings. A simple
rectangular core with windings on two legs of the core is also used in some power transformer
designs.

Basic types of cooling used for distribution transformers are referred to by the
following designations: 1

0 OA - Oil-Immersed Self-Cooled, In this type of transformer the insulating oil

circulates by natural convection within the tank (see Figures 4-1 and 4-2).

* OA/FA - Oil-Immersed Self-Cooled/Forced-Air Cooled, This type of transformer

Is basically an OA unit with the addition of fans to increase the rate of heat transfer
from the cooling surfaces.

* OA/FOA/FOA - Oil-Immersed Self-Cooled/Forced-Oil Forced-Air Cooled/Forced-

Oil Forced-Air Cooled, OA rating increased by the addition of some combination
of fans and oil pumps.

a FOA - Oil-Immersed Forced-Oil-Cooled with Forced-Air Cooler.

* OW - Oil-Immersed Water-Cooled. In this type of water-cooled transformer,

the cooling water rui isthrough coils of pipe that are in contact with the Insulating
oil of the transformer.

206
 * FOW - Oil-Immersed Forced-Oil Cooled with Forced-Water Cooler. External
oil-to-water heat exchangers are used in this type to transfer heat from oil to
cooling water.

* AA - Dry-Type Self-Cooled. D- ,-type transformers, available at voltage

ratings of 15 kV and below, contain no oil or other liquid to perform insulating
and cooling functions,

* AFA - Dry-Type Forced-Air Cooled. This type of transformer has a single

rating, based on forced circulation of air by fans or blowers.

* AA/FA - Dry-Type Self-Cooled/Forced-Air Cooled. This design hlas one

rating based on natural convection and a second rating based on forced circulation
of air by fans or blowers.

Standard impulse tests are conducted on distribution transformers to ensure that

the bushings and winding insulation meet the basic insulation level (BIL) specified for
their type, power, and voltage class, Transformers in a given BIL clars are tested to
the levels given in Table 4-1, The three basic insulation tests specified by the American
Standards Association are:

(1) The full-wave test with an Impulse that reaches its crest in 1.5 /s and decays
to half its crest voltage in 40 js.

(2) The chopped-wave test in which the impulse Is truncated by an air-gap

firing after the specified minimum time-to-flashover has elapsed. Only the
leading edge of the impulse is applied to the transformer, since the air gap
chops the tall off the impulse after the specified minimum time,

(3) The low-frequency test In which about twice the rated voltage at about twice
the rated frequency is applied for no longer than 1 minute.

The transformer high-voltage bushings and windings must be capable of withstanding these
tests when applied between the two high-voltage terminals and when applied between either

207
 Table 4-1

STANDARD INSULATION CLASSES AND DIELECTRIC

TESTS FOR DISTRIBUTION AND POWER TRANSFORMERS

Impulse Tests
Rated Voltage Between Terminals Low-Frequency Oil-Immersed Trans-
of Power-Transformers(aI Tests formers 500 kVA or
Les
Rated Single-Phase 3-Phase Oil- Chopped Full
Volt- For Y For Delta- Delta or Y Immersed Dry Wave Wave t1)
age| onnecton Connection Connected (o) Type Typelb Minimum -

( )_n 3-Phase on 3-Phase (kV rms) (kV rms) (kV rms) Time to
System(c) System Crest Flash- Crest
(kV ris) (kV rms) (kV) over(ps) (kV)

1.2 0.69 0 , 9l(d) 1.2 10 4 36 1,0 30

2.5 - 2.5 15 10 54 1.25 45
5.0 2.89 2 ,8 9 (d) 5.0 19 12 69 1.5 60
8.66 5.0 5 , 0 0 (d) 8.66 26 19 88 1.6 75
15 8.66 15.0 15.0 34 31 110 1,8 95
25.0 14,4 25.0 25.0 50 - 175 3.0 150
34.5 19,9 34.5 34,5 70 - 230 3.0 200
46.0 26.6 46.0 46.0 95 - 290 3.0 250
69.0 39.8 69.0 69.0 140 - 40U 3.0 350
92 53.1 92 92 185 - 520 3.0 450
115 66.4 115 115 230 - 630 3,0 550
138 79.7 138 138 275 -. 750 3,0 650
161 93.0 161 161 325 - 865 3.0 750
196 113 196 196 395 - 1035 3,0 900
230 133 230 230 460 - 1210 3.0 1050
287 166 287 287 575 - 1500 3,0 1300
345 199 345 690 690 - 1785 3.0 1550

Notes (1) Intermediate voltage ratings are placed in the next higher ;nsulatIntn clats unlis$ otherwise &pacified.

ib) Standard Impulso tests have not been established for dry-type dlitribution and power transformers.
Present-day values for Impulte tests of such apparatus tre sofollows:
1.2-kV class, 10-kV; 2,5-kV class, 20 kV; 5,0-kV class, 25 kV; 8.66-kV gls, 35 kV; 15-kV class,
50 kV. These values apply to both chopped-wavo and full-wdve tests,

(c0 Y-connected transformers for operation with neutral solldly grounded or grnuded through an impedance
may have reduced insulation at the neultral. When this reduced insulation isbelow the level required for
delta operation, transformers cannol be ogerated celte.-'onnected,

Id) These transformers eru Insulated tor the test voltages corresponding to the Y connection, so that a Single
line of apparatus serves for the Y and delta applicationl, The lest voltages for surh delta.connected
single-phase transformers are therefore one step higher than needed for their voltage rating,
4
(el 1,5 x 0-ms wave,
Source: "If. t.

208
terminal and the case. The Impulse tests are to be conducted with normal 60-Hz energization,
and timed to ensure that the impulse is applied within 30" of the 60-Hz crest.

4.1.3 LIGHTNING-ARR ESTER CONSTRUCT ION

valve-type lightning arrester is a series combination of a tpark-gap assembly and

-'ia
a nonlinear resistance element. The non-linear resistance element is voltage-sentitive; Its

resistance decreases exponentially with the voltage across it, The valve elements are non-
linear resistors made of sillcoq-carbide crystals,

The spark-gap assembly Is designed to minimize the time from overvoltage onset to
firing and to minimize the time duration of any 60 Hz current flow, The first objective Is
frequently accomplished by use of a "prelonizer," Thu latter objective Isaccomo•lbhed by
extending the arc's path length and/or by breaking the arc up into a number of parallel,
series, or rotating arcs, This tends to cool (deionize) the arc and help ensure that the arc Is
extinguished at or before the end of the half-cycle In which the surge or overvoltage occurs.
The gap and valve assemblibs of a lightning arrester are generally hermetically sealed In a
glazed porcelain body and provided with an external, open-air spark-gap In series with the
arrester, Figure 4-3 shows the internal construction of a typical valve-type lightning arrester.
Lightning arresters may be designed as a part of another unit, such as a fuse or a cutout
twitch. Figure 4-4 shows three arresters; two are combination units that include open-link
fused cutout switches,

Lightning arresters are Installed with the gap side of the arrester cnnnected to the
line and the opposite end grounded, In the absence of a voltage surge, all of the line
voltage appears across the open gap and none Is impressed on the valve element (which
typically has a maximum resistance of a few thousand ohms). If the arrester is subjected to
an overvoltage wilh a magnitude sufficient to arc across the gap assembly, the resistance of
the gap assembly is reduced to a negligible value, subjecting the valve element to the entire
surge voltage. This causes the valve-element resistence to undergo a rapid decrease, thereby
forming a low-resistance path to ground through which the surge can bfi -",slpated.

209
r1
GAP

ELECTRODE~

PREIONIZING
C:, SURFACE
GRADING
RESISTOR
ARC
SCHAMBER
ELECTRO-
I .... "" .., COIL
MAGNETIC

VALVE BLOCK

Figure 4-3 INTERNAL CONSTRUCTION OF A VALVE-lYPE LIGHTNING ARRESTER

After discharge of the surge voltage, the voltage across the arrester drops back to
the normal line voltage. This voltage is not high enough to initially fire the arrester"s gap
assembly, but It woula be adequate to maintain the exterrial arc initiated by the overvoltage
if it were not toat (1) the line voltage la insufficient to maintain the low resistance of the
valve element (which therefore returns to a high value) and (2) the Internal gap assemblies
are designed to extinguish the lower current permitted by the high resistance of the valve
element,

The rapid irctease in the valve clement's reoswtance causes a substantial reduction
in the 60-.cycle power current; older-style arresters relied on this to reduce the current flow
to a value low enough that the yap assembly ouUld d•t t( extitrguishi the arc at or before the
next system voltage zero. This subjects the valve element to considerable currents for a

210
1Ve

Id) MeGRAW EDION ibi OENIRAL ELECTRIC (I) A. Is. CHANCE

AP 51 0 kV MODEL .L241CXOoo C7oJ0.-01O0
TYPE UC

Figure 4-4 LIGHTNING-ARRESTER STYLES USED IN CONJUNCTION WITH THE

TRANSFORMER HV TRANSIENT TESTS

potentially extended period of time; modern arrewters are designed to transfer much of this
responsibility for 60 Hz current-limiting to the arrester gap assembly Itself. The major
benefit of this design is that the valve elements can have lower discharge resistances, which
decrease the arrester's I R drop, This IR drop is the voltage applied to the devices or circuitry
that the arrester is intended to protect during an overvoltage surge, 1,2

Typical firing characteristics of distrilution-type lightning arresters are shown in

Section 2,6,3.

4.2 LINEAR CHARACTERISTICS OF TRANSFORMERS

4.2,1 ANALYSIS OF TRANSFORMER COUPLING CHARACTERISTICS

The linear, high-freqUency behavior of power transformers has been studied in con-
nection with their response to lightning transients and their effect on the power-line carrier

211
communication systems. 3 '4,5 Unfortunately, the analytical treatments of the transformer,
in ordoir to be tractable, tend to be highly idealized; the models obtained therefrom must be
accepteu with due regard for these idealizations, Two areas in which the models fail to ade-
quately describe the transformer are in the primary bushing impedance (which is probably
not significant for lightning transients, but is for EMP), and in the symmetry of the windings.
In addition, most of the models seem to Imply that a common-mode primary excitation
produces no dlffercntiul-mode secondary response because both the primary and secondary
windings are assumed to be symmetrical with respect to the core and with respect to each
other. In practice those windings are far from symmetrical in either respect, so that
coupling in the above mode is signifIcaimt.

The linear response of power transformers can be analyzed by treating the primary
and secondary windings of the transformer as coupled transmission lines,3,4, 5 The windings
are characterized as illustrated In Figure 4-5 by (1) ,iductances Li and L2 and resistances
R 1 and R2 pet, unit length shunted by the turn-to-turn capacitances per unit length, Cx and
Ch; (2) capacitances per unit length C11 and C 22 between the winding and the core or
housing; and (3) a capacitance per unit length C 12 between the primary and secondary
windings and a mutual inductance M between the primary and secondary turns. Coupling
between the primary and secondary at low frequencies (e.g,, power frequencies) Is primarily
through the mutual inductance M, but at high frequencies the capacitances Cx and Ch carvy
most of the current between the terminals, and the coupling between primary and secondary
windings is primarily through the interwinding capacitance C 12. The manner in which one
divides all frequencies Into low and high frequencies Is dependent on the power and voltage
rating of the transformer, since the turn-to-turn capacitances Cx and Cjh and the primary-
to-secondary capaltance C 12 are dependent on number of turns, size of wire, and compact-
ness of windings,. For example, measurements on a 9.5 kV/220-V service transformer have
shown that the primary turn-to-turn capacitance Ch Is dominant above 24 kHz, but the
secondary turn-to-turn capacitance Cx Is not dominant until frequencies well above 23 kHz
are reached.

Thus, d coMplu)to and accurate analysis of the linear, transient behavior of a trans-
former at higher frequencies must take into account the interwinding capacitances, as well
as the self and mutual inductanc ; of the winding, Furthermore, as is pointed out in
Refs. 3 and 4, an important difference exists between the transmission--line model of the
transformer windings and the conventional transmission line, regardinU the self and mutual

212
 rN1

T _ ' T T
MUTUAL
M2
1 mutual inductance bjtween
prinary arnd secondary
C12 capacitance betwerm primary
and itcicndar windingt

PRIMARY (H1-H2)

ch -turn .to-lurn capacitance

L1 sit inductairca of winding

R " winding rnistanci
C1" winding.ln-htomt capacitanct
t, b s Inductanci of bunhing
C bi ' caa•Ictance of bushing

SECONDARY IXI-X2)

ChX atuln-lo-lurm ciabucitnce

L• •. slf inductance of windnlfg

R2 windinlg I lsiltamuI

C22 witidlng-to-fraline capacitance

L b2 inductancel of bushing

ch2I " rioaitcII e of bushing

Figure 4-5 TRANSMISSION-LINE MODEL FOR COUPLED TRANSFORMER WINDINGS

Inductance per unit length, This difference is related to the fact that the inductance (self
or mutual) of a turn Is a function of the position of the turn in the whole winding. For
example, a turn In the center of the winding Is coupled to other turns on either side of It,
while a turn at the end of the winding Is coupled to other turns on only one side, Reference
3 proposes a parabolic variation of magnetic flux partially linking the turns, with a maximum
partial flux linkage at the center of the winding and the flux linkage decreasing toward the
ends of the winding and reaching zero at some point beyond the end of the winding. This
variation in flux linkage leads to self and mutual Inductances that are functions of position,

21
The result is a set of two coupled fourth-order differential equations relating the mutual and
self-inductance voltages of the primary and secondary windings - i.e.,

a 4 V,1 a
---..4 +j --- [LlILI + MIL2j= 0 (4-1)
ax 3

a4 V2 a
- +jw [L21L2 + MIL1]= 0 (4-2)
ax 4

Where V1 Is the voltage on the high-voltage winding and V2 is the voltage on the low..
voltage winding (both referrpJ to the transformer core and frame), and L 1 , L2 , and M are
the self and mutual inductances,

A detailed solution of Eqs. (4-1) and (4-2) appears in Refs. 3 through 5, The
results are

{p 8P692 (LIC1 + L2Cx) +P4 [L1CI 1+ L2 C 2 2 + C 12 (L1 + L2 - 2M)]

+ p4 s4 (L 1 L2 - M2 )ChCx + p2 s4 (M2 - Li L2) (CxC 1I + CxC12 + ChC12 + ChC22)

2
+s4 (L 1 L2 -M ) (C 11 C2 2 +C 11 C12 + C 2 2 C12 )J V O0 (4-3)

where

p=n/axn and sP = (jwu)n or arn/atn.

The solution to this eighth-order differential equation depends on access to a great deal of
information not normally available and not readily determined from measurements made
at the transformer's terminals, In addition, the winding resistances R1 and R2 have not been
Included in the voltage equations. Finally, for distribution-type power transformers, the
lumped bushing capacitance Ch and inductance Lb in Figure 4-5 play an important role In
determining the coupling between the primary and secondary windings, since they behave
as L-section filters between the transformer terminals and the windings at high
6
frequencies. ,7

214
 It Is Informative to examine the behavior of the transformer at the extremes of the
frequency range. As w approaches zero, for example, all capacitive susceptances become
small and we are left with only the self- and mutual-inductance terms in the equivalent
circuit of Figure 4-5 (the winding resistances have been neglected). In addition, because
the windings are short compared to a wavelength at low frequencies, the distributed self
and mutual inductances can be adequately represented by lumped self and mutual Inductances
as Illustrated In Figure 4-6(a), where the equivalent circuit of an Ideal transformer Is shown.
The practical equivalent circuit containing copper losses (winding resistances) R 1 and R2 ,
and core loss Ac, is shown in Figure 4-6(b) with all impedances referred to primary side.

At the high-frequency extreme, ,.; approaches Infinity and the capacitive reactances
carry all of the current, leaving none to flow through the self and mutual inductances of

L1 LI M2
M12 2

H2 0 -- X2

(i) IDEAL TRANSFORMER

Ri L, L21n2 R21n2

H ~ Ii *x I

V1 Lm Rc V2

H2 0 0 X2
n - TURNS RATIO N2 /N1 Lm LI
L1 - L1 - M12 /L 2 L2 L2 - M1 2 /L.1

1b) PRACTICAL EOUIVALENT CIRCUIT

Figure 4-6 LOW-FREQUENCY EQUIVALENT CIRCUITS FOR

A TRANSFORMER

215

.~...
Figure 4-5 (except the bushing inductances), The differential equation for the voltage in
either winding becomes

d4V CxC1 1 + CxC12 + ChC12 + ChC22 d2 V C1 1 C?2 + C 11 C 12 + C22C12

V -0 (4-4)
dx 4 ChCx dx 2 ChCx

and ua high-frequency equivalent circuit becomes that illustrated In Figure 4-7. The stray
Inductance and capacitance associated with the bushings has been retained in Figure 4-7(a)
because these small inductances are significant in the high-frequency response, and they
are not bypassed by capacitances as are the winding inductances, Notice that the distributed
capacitances Ch, Cx, C1 1 , C2 2 , and C1 2 form a capacitive network that couples:

(1) The primary terminals to each other and to the case or core

(2) The secondary terminals to each other and to the case or core

(3) Each primary terminal to each secondary terminal.

A rough lumped-element equivrkent circuit is shown in Figure 4-7(b) in which the distributed
winding capacitances have been replaced by lumped capacitances between windings,
between terminals, and, between terminals and ground (case or core), Even with this
simplification, the relation between primary and secondary currents and voltages is not
simple, but the coupling paths are more obvious and the bandpass nature of term nal-to-
terminal coupling caused by the bushing Inductances and the bushing and winding capacitances
is apparent.

The analytic model of Figure 4-5 is probably a good representation of a bifilar-

wound 1: 1 transformer, but for typical power transformers that have separate primary and
secondary windings the analytic model is only a crude approximation at high frequencies,
It Is nevertheless a useful aid in interpreting the transient behavior of practical power trans-
formers, ft is also significant that, because the primary windings of transformers are
dominated by the turn-to-turn capacitance above a few kilohertz, the high-frequency
equivalent circuit may be applicable throughout the 10-kHz-to-100-MHz frequency range
of interest here.

216

...........................
........... ....
 LH
b 1 l C_h
h H
N 1,. H-,-.,, --.-....----- d fr H2

HIl0' C''' C
(.1 IDEAL. TRANSFORMER WITH BUaHINOS
L-72
HI:2 2

C ~Ic
T f12 C 2f:r T~

PR~IMARY
BUSHINGS WINEJINGS SECONDARY
IBUSHINGS

(bI APPROXIMATE LUMPED-St EMEINT CIRCUIT.;

Figure 4-7 HIGH-FREOUENCY EOUIVALENT CIRCUITS FOR A TRANSFORlMER

422 DISTRIBUTION Or VOLTAGE ALONG THE WINDINGS

The distribution of a translent voltage along a transformer winding can be estimatedI

from the distributed-parameter circuit model of Figure 4-5. The results of calculation with
such a model (without bushings) predict that when a fast-rising voltage reaches a terminal
of one winding of a two-winding transformer, the voltage is initially distributed nonunltormly,
in both windings due to the effect of the turn-to-turn capacitance, the capacitance-to- i
ground of each winding, and the mutual capacitance between the windings. Following thue
initial capacitively coupled transient, both windings go into oscillation due to the interaction
of their inductances and capacitances.

Fi~qures 4-8, 4-9, and 4-10 show plots of calculated voltages In the windings of a
single-phase transformer when stressed at the high-voltage terminal with a fast-rising voltage.
The neutral of the primary winding and both of the terminals of the secondary winding of
the transformer are assumed to rise to a peak of 330 kV in 10 ns.

H ý217

--
 350
-PRIMARY VOLTAGE Figure 4-8 shows the spatial dis-
300EONAYVOTG tribution of the voltage inside the pri-
mary and secondary winding for early
260 - -___ -times during which only the winding
capacitances are playing arole. The
I200 - -spatial coordinate Is the distance along
Uj the axis of the winding, which Isas-
A150 - - -sumned to be cylindrical In shape. The
zero of the spatial coordinate isas-
- - -- -sumed to be at the high-voltage termi-
1,0 n;nal of the winding, with 100% of the
50_,-0n winding having been traversed upon
reaching the neutral. It Isclear from
Figure 4-8 that for times of the order
0 10 20 30 40 so at the rise time of the pulse, the voltage
AXIAL DISTANCE - percent total length
HV TERMINAL NEUTRAL applied to the hIgh-voltago' terminal
does not distribute Itself uniformly over
Figure 4-8 SPATIAL DISTRIBUTION OF the entire winding but tends to pile up
WINDING VOLTAGE AT EARLY TIMES.
Source; Ref. 8. across the first 40% of the turns nearest
the high-voltage terminal. Hence, the
peak applied voltage of 330 kV isImpressed upon the Insulation of these first turns alone
and, depending upon the Insulation strength, could cause breakdown, From Figure 4-8 It
Isalso apparent that -the voltages In the secondary winding due to the various winding
capacitances are much less than those Inthe primary.

After the Initial transients shown In Figure 4-8, which propagate through the wind.-
ings by means of the turn-to-turn capacitance, the capacitance to ground, and the mutual
capacitance, oscillations are set up due to the combined effect of the winding Inductances
and capacitances. This behavior Isshown In Figures 4-9 and 4-10, The oscillations become
apparent only for times of the order of or greater than the first natural period of the wind-
Ings, which lies Inthe range of afew microseconds for the secondary winding, By these
times, the voltage at the high-voltage terminal of the primary Isalso beginning to decay, but
because th~e natural frequency of the primary winding Ismuch lower than that of the seco I~-
dary, only the beginnings of an oscillation are apparent in the primary voltage distribution.

218
 400
The results shown in Figures
200 4-8 through 4-10 are indicative only
NN of the qualitative behavior of a station
-J 0
2 lis4
to-transformer
6 stressed by EMP, because
of the waveform assumed and because
_
~the circuit constants for the
-200
0 10 20 30405060 7080 90 100 transformer cam be estimated only
AXIAL DISTANCE - PS!Qwfi total Iogt crudely. In addition, because the
HV TERMINAL NEUTRAL
Inductance and capacitance of the
Plgus49 DITRIUTIN
VLTAE AONG bushings are neglected in the calcu*-
PRIMARY WINDING AT LATE TIMES. Source: latlon of the voltage distributions
Rot, 8.
shown In Figures 4-.8, 4-9, and 4-10,
the assumed rate of rise (33 kWns) is
I8I0 considerably greater than that actually
Inue0ot0.Fr xm~e 0-
04 0 applied to the winding by an EMP-

0 a ohm line feeding a 100-pF bushing has

a rise time-constant of 30 ns, so that
-40 even a step.-fronted wave would have
-0 - ----- Its rise slowed to about 60 ns upon
0 10 203 40 -0 -0 70 -0 0 entering the transformer.
AXIAL DISTANCE -- percenit total len~gth
HV TERMINAL NEUTRAL

Figure 4-10 VOLTAGE DISTRIBUTION ALONG

SECONDARY WINDIN6 AT LATE TIMES, Soutce:
Ref. 8.

4.2.3 CW MEASUREMENTS OF TRANSFORMER CHARACTERISTICS

Viewed as a multiterminal network, the ypicoi distribution transformer of the types

diucussed here must be considered a 7-terminal device, because there atre two primary termi--
nals, four secondary turminals (two for each winding), and a ground or case terminal, A
large numbv~r of measurements are Iequi1red to specif y the equ ivalent impedances between
the terminals ot a 7-terminal device, and even with these impedances specified,

219
considerable complex algebra is required to apply the equivalent circuit to a practical trans-
former couplinq problem. To avoid this complexity, the complete definition of the equiv-
alent network has been compromised in favor of obtaining readily usable equivalent circuits
that are representative of those encountered in typical transformer Installations. 7 Thus,
instead of defining a 5- or 7-terminal network, we have defined the four 3-terminal
networks illustrated in Figure 4-11.

Low-level CW measurements of the voltages and currents necessary to specify the

elements of the T-networks of Figure 4-11 have been made over the frequency range from
10 kHz to 50 MHz for each of four transformers 7 (see Table 4-2). To Illustrate the simi-
larity of the transfer characteristics of the transformer and to illustrate the dominant
features of the coupling analysis discussed above, the ratio of the voltage delivered to a
100-ohm load across the undriven terminals to the driving voltage has been computed for
each transformer for each of the four configurations of Figure 4-11.

Table 4-2

PROPERTIES OF CLASS OA TRANSFORMERS USED IN CW AND

TRANSIENT MEASUREMENTS

Unit Power Voltage Rating Manufacturer Secondary

No. Rating (kVA) (V) Winding
1 25 7200/12470 Y - 120/240 General Electric (new) Aluminum strip
2 25/28 13200- 120/240 Allis-Chalmers (used) Copper wire
3 25 7200/12470 Y - 120/240 Westinghouse (new) Aluminum strip

4 50 7200/12470 Y - 120/240 A"Is-Chalmors (new) Ahiminum strip

The voltage transfer function for the common-mode configu,'ation of Figure 4-11(d)
is shown in Figure 4-12 for each of the four transformers. The average tansfer characteristic
for all four transformers is also shown in Figure 4-12. It is apparent from Figure 4-12 that
the coupling to the 100-ohm load is dominated by the Interwinding capacitance In the low-
frequency range between 10 kHz and about 1 MHz, The average interwinding capacitance
from Figure 4-12 Is about 1000 pF in 'his configuration, Also apparent is the effect of the
bushing capacitance and the lead Inductance associated with the primary bushings and the
leads to the 6econdary bushings in reducing the voltage transfer function above about
10 MHz. In spite of the fact that two different power ratings, two different winding types,

220
 Il Xl

H2 X7
ZL
H2H

(b)
HI xi

22

HI
HI XI

(YE)

HI xI

40 0 (d)

Figure 4-11 POWER-TRANSFORMER CONFIGURATIONS AND EQUIVALENT T-INETWORKS

USED TO OBTAIN CW TRANSFER-FUNCTION DATA

-2
_
and three difierent manufacturers are represented by the four transformers, their common-
mode voltage transfer characteristics are remarkably similar,

Similar results for the common-mode excitation and diff erential -mode output of
Figure 4-11 (a) are shown In Figure 4-13, Although the four transformers are similar In
this mode also, the deviations ot the Individual transformers from the mean is greater.
As can be seen in the approximate equivalent circuit of Figure 4-7, the coupling for
common-mode excitation (Hi and H2 driven against ground) and differential-inode loading
(X2 grounded arid 100 ohmns between X1 and ground) Isachieved through thle asymmetry
of the windings (the asymmetry occurring naturally due to the construction of the trans-
former as well aq that Introduced by grounding the terminal X2), This voltage transfer
function Isthus more dependent on the winding-to-case capacitance and the turn.-to-turn
capacitance thani are the common-mode transfer functions of Figure 4-12. Thus the type
and form of the secondary winding are probably responsible for thle greater variation affioing
the transformers in Figure 4-13. In spite of this, however, ~he similarity of the four trans-
formers is quiote striking, It is noteworthy that in this coupling mode also, the transformer
clisplays a likeness to a bandpass filter,

Figure 4-14 Illustrates the voltagii transfer characteristics for the common-mlode
secondary excitation and differential-~mode primary loading of Figure 4-11 Wb. The
coupling mechanism for this cunfiguration isslinilar to that in the preceding example, except
that the force(] unbalance (caused by grounding 1-121 is now in thle primary winding, Because
th~ere is only ono primary winding and It is, in oll four cases, placed between the sL'condary
windings, cu-iplinq i- this mode iqu''IdoUbtOdly morp strongly dependent on tile inter..
kvwndingq ctipaci tince thon on the prhI irV-vv~incliny. tu -ground cupacit~nce 'Thle observed
siminlarity among the trwtnsfur characteristics In FigJUre 4-14 Is qJuite strik Ing, and again the
bandpass-filter characteristics tire apparent.

The voltage transfer characteristics for difierentiol-modu excitation and differential-

mode loading I see Figure 4-11 lc)lI are shown in Figure 4-15, For this configuration, as
for the other configuiration involving a differential-*mode secondary, there Is considerable
variation from transformer to translurnier, although there is significant ovarall similarity,
It appears that below about 100 kHz thle tranufer tuiictlon is relatively Independent of
frequency. If H2 andi X! In Frigure 4-7 are grounded, the coupling between Hi anid X1 is
through the capacitive divldor formed by the interwinding (C'12~) and windlng-toqcround

222
 AU

~a~i~~
L u~.
-ii
;ML ~~~~~~C
it..
NOIL~Nl4wi~~vw ~ OI±~fl UdUNW. C aflINOVz

NDOI±Nfl WBAINVd.L jo 3an.LINDVYkA NOIJ.ONMIfl~ NVkil 10 BOnlINOVINIi

223
rrnymf- rr06

224
rr

~ ~

'CA
T 1
T~'L
CC
NOILOJ.4l WSN~VWI AO 3afl.INOViN N01OIJ~Nfl UIISNVWi Ao IOCn.LINDVI

NOI±DNrhd UdIJNVU± JO BOnh.INDvWN NOI±DNfJ W3ANVhUI Ao 3anfj.NDVIN

225
jI I-II -I III I I I T0 -T

'IW

226
 0

o w

-
U,

NOI.LONfl OýSNVHi io icifliINDYN N1I±ONfli U34SNVWJ. 40 3afrmNOW-4 m

f Co

* o U

00

NUI.L;jNfA H3 IqN '4H Jol gorl1IN!)Vh' N OIJY'rij i~ JNJVUI. Iin ,jcnJ.lNr)VIA)

227
 In

w
MC

NOI.ZONniU HIASNVI~i JO 3anhINDVWA

228
 i- 0

cr.-

A .w
"m 0

1-1

cc

0L
NOI1Nfl I~dNV~ ~O~anhN~v~
NO±3Nl~ ~SNV.L O ~Ql±IDLL

Ul2

......................................
,
 N

230
(C' 11, C' 2 ) capacitances. In addition, the transformer, in this configuration, Is behaving
somewhat as a step-down transformer even though the primary current is being shunted
by the turn-to-turn capacitance (the primary resonance occurs below 10 kil-), In either
event, winding resonances that are. unique to the individual transformers occur above 100
kHz.

When the average characteristics of the four configurations are compared, it Is

apparent that the voltage transfer function for a 100-ohm load is greatest for the common-
mode-to-common-mode configuration (Figure 4 12), slightly less for the common-mode-
primary-to-differential-mode-secondary (Figure 4-13), and slightly less yet for the differ-
ential-,mode-secondary-to-common-,mode-primary (Figure 4-14), The smallest high-
frequency coupling occurs for the differential-mode-to-differential-mode configuration,
but in this configuration the low frequencies are not cut off as strongly as they are In the
other configurations,

4.2,4 LINEAR TRANSIENT TEST RESULTS

Transient tests of the four transformers of Table 4-2 have also been conducted at
a low level to obtain the linear response characteristics of the transformers,7 The driving
level used for these tests ranged from about B kV to about 40 kV (the lightning arresters
fired at about 42 kV). At driving levels below the threshold for lightning-arrester firing,
the transformer behaves as a linear device, and the transient responses can be related to
the frequency-domain characteristics. The circulits of Figure 4-16 were used for these tests,

For the low-level tests, the 0,075-/F switched capacitor was used as the driving
source, so that the source-voltage waveform across the 100-ohm resistor at the pulser
output terminals (see Figure 4-16) was an exponential pulse with a 10-to-90% rise 'Lime of
about 20 ns and a decay time constant of about 7.5 ps, The voltage waveform appearing
at the terminals of the transformer is shown In Figure 4-17. A slight oscillation during the
first 400 ns caused by tlhe winding reactance Is apparent in Figure 4-17. Also of interest is
the waveform of the current entering the transformer, shown in Figure 4-18. (Figure 4-18
I

231
 CASE
HI X1 VI 1x0 I4Io
LIGHTNING ARRESTERS- '6 0 a -

SPARK
30 MS2 VC GAP VX

SUPPLY T_ J

(I) COMMON-MODE DRIVE AND COMMON-MOOO

OUTPUT CONPIGURATIONS
(ihown with trenformer weondhry windlinp aonnuctd in suierll

VPARK
3 V2 110 1
VC GAs 1 r

FIgluft 4-16 SCHEMATIC DIAGRAM OF TRANSFORMER TEST CIRCUITS (submaripti on V

and i ndicato test measurement pointsl

shows the total capacitor current, which includes albout 300 A through the 100-ohm load.)
it is observed that the rise time of the current is significantly longer than that of the driving
voltage. This observation is consistent with the contention that significant series inductance
is associated with the primary bushings. It k• also observed that the duration of the current,

232
 8 kV

-.- 50 ms

S~T Is) FAST SWIEP

lii
-'- .---
2°°.
13(b91.0W SWEEP

Figure 4-17 WAVEFORM OF EXCITATION VOLTAGE

APPLIED TO TRANSFORMER

ohen thu loicI- resistor current is neglected, is much shorter than the duration of the im-
I"sed voltale, The primary current is mainly that required to charge the capacitances
pt

assoLciated with the windings and bushings, and once these are charged, no more current,
other than that associated with the oscillations triggered by the transients, can flow, The
waveforms shown in Figures 4-17 and 4-18 were obtained with the 25-.kVA G,E, trans-
former in the common-mode drive configuration, but they are quite representative of the
general characteristics of all the transformers tested.

The waveforms of the voltage across the 100-ohm load on the secondary terminal
Xl are shown in Figure 4-19 for the four transformers tested in the configuration of

233
 Figure 4-18 WAVEFORM OF Pt)LSER CAPACITOR DISCHARG3E
CURRENT (whic~h includes the cutirnt through the 100-ohm InhIsu loatd)

Fiqure 4-16. F~rom tile fast oscillograrns of Figure 4-19 it Is apparent thait the rise times
of the secondary voltage are of thle order of 50 ns although the primary-voltage rise tinme
was only 20 ns, Thus the rise time stretching caused by the primary bushing inductances
ISobservable in the secondary voltages as well as In thle primary current, From the slower
osciliograms, of Figure 4-19 it Isalso apparent that thle late-tIme volzage present in thle
driver hes boun suppressed, so that only about 200 ns of high-level response isobserved Lit
the suooidary terminal, The stretching of the rise time In thle timne-doinain curresponds to
the high-frequen01cy suppression observed In thle frequency domain, and the suppression of
the late. time response corresponds to the low-frecquency roilloff observed in the froquency
domain. Thus the transient waveforn-is are quite consistent with the frequoncy-doilain
data for voltageo transfer functions,

These ha~ic features of the secondary voltages .. the Stretching of the rise timle anld
the suppression of the late-timea response -. were common to all four transformers, and
were consistent with their fruquency-darn in transfer characteristics, There are obvious
differences in the fine structure of the waveforms, but these appear to be of secondary
Importance, as are the size and location of the peaks and valleys of the frequency-domlain
transfer functions, The peak output voltage at Terminal X1 for one-volt peak at the primary
Isalso quite similar for all four transformers, as isappirent in Table 4-3,

The bushing Inductance was responsible for the slow rise time of the current in the tcsts with d low
sourcu impedance; when sourco mrpedances of a few hundred olims are usad, the rise tim~of thme
terminal voltagig ;, Silm11ilrIly slowed by the bushingi capacitance.

234
(A

X0*0

UJ

C).

IJ

235
 CC

wm

~1.6

>WM

>S

23
 Table 4-3
COMPARISON OF PEAK VOLTAGE AT SECONDARY TERMINAL X1
FOR A ONE-VOLT, 7.b-ms EXPONENTIAL PULSE
APPLIED TO THE PRIMARY

[Transformer
Number
Transformer
Type JI at X2
(V)
2 ~~25-.-kVA Allis Chalmers 02
I 3 2L1-kVA West inghouse 0.29
6040-.-,kV.A Allis.C-halme~rs 0,28

A comnparison of the transient responses of the secondary to commnon-mode and

differential-mode excitations at tin' primary can be made from the oscillograms of
Figure 4-20, which show the voltage waveforms at secondary terminal X3 of thle 25-kVA
Westinghouse transformer for these two excitation modes, Although the responses are
generally similar in shape, It Isapparent that the differentiaIlnmode excitation at the primtti y
produces only a little over half the peak secondary voltage produced by the coninon-nmode
excitation, T'his relationship Issignificant, because the common-mode excitation of the
primary by the transmission lines is uuually much greater than the differential-mode exci-
tation; hence, if the weaker differential-mode signal Is further reduced in passing through
the transfoi nier, its Importance may be greatly reduceu,

rhe osciilogrami of Figure 4-19 and Figure 4-2?0 also permit the voltages at Ter-
minials X1 arnd X3 to bu comipared for Trdnsformner No, 1, for thie configuration of
Figure 4-16. It is apparent that th4 peak vii t 'ge at 1erminal X1 is somnewh~at larger than
that. ut Termminl X3 for the comnion-mode primary excitation, This difference is produced
primarily by the asymmetry or unbdlance In the witirlings; in i qymnletrical bif ilat-wound
transformer those two voltages should be identizal, This differential secondary voltage is
shown in Figure 4-21 for the two excitation modes.

All of the waveforms in Figures 4- 1% 4-20, and 4-21 ware obtained with a cnommoii-
mode output configuration. Figure 41-22 slows thc voltage waveforms obtained at the sec-
ondary terminal X1 for a differential-mode output configuration such as that ihown In

237
 LW

•L)

t.

238
IJI

-U-j
ccJ
0

a8

34

wi

239
 03

I- -

0I

kL~

U.J

240
Figurb 4-16(b). The peak secondary voltage produced by the common-mode prinmary exci-
tation with this secondary connection is largpr than the peak secondary voltage produced
by the differential-mode primary excitation, A comparison of Figure 4-22 with Figure 4-19,
however, shows that the peak secondary voltage for the difforential..node output confiqura-
tion is only about half as large as it is for the common-mode output configuration.

These and other data show that the trantformer responses are (1) highly linear,

(2) reasonably similar in comparison with one another, and (3) In good qualitative agree-
ment with the CW transfer function data presented In Section 4.2.3, The frequency-domain
data show that significant differences In amplitude can be expected for different drive con-
figurations, and this result Is confirmed by the transient-response data, Comparison of the
common- and differential-mode drive data plotted here shows the following:

(1) Common-mode Input or output configurations result in peak output signals

that are approximately double those of differontlal-mode Input or output
configurations.

(2) The peak output response varies by a factor of 2 or less from transformer to
transformer,

(3) Different secondary-windlng configurations (i,e., series or parallel) have little

effect on the secondary terminal-to-ground voltage,

7
4.3 NONLINEAR CHARACTERISTICS OF TRANSFORMERS

4.131 TRANSFORMER RESPONSES

At primary excitation levels above about 42 kV, the 10-kV lightning arresters on
the primary side of the transformers fire if the pulsewidth of the excitation waveform Is
large enough. At the threshold voltage of 42 kV, there is a time lag of 150 to 200 nsbetween
the time of application of the voltage and the closure of the lightning-arrester spark.-gap
switch, but as the applied voltage Increases, this time lag decreases. The transformer's
secondary response to a primary-side excitation at levels above about 42 kV Is therefore

241
influenced by the change in the excitation pulse when the lightning arrester fires, and by the
variable time lag between the application of the exciting pulse and the closure of the
lightning arrester.

"The only nonlinear effect observed during the transformer tests with the lightning
arresters installed was the Itghtning-arrester f iring. No other arcing or flashover was observed
ihwide or outsidu the transformer case (except when the secondary was driven), The non-
linear responses discussed here are, therefore, caused entirely by thle Iightning-arrester firing,

The effect of the lightning-arrester firing on the secondary voltage waveform Is

illustrated In Figure 4-23. As was observed in thle discussion of the linear resj)unse of tile
transformers, the duration of the large-amplitude secondary voltage waveform is of the
order of 200 ns, so that deviations of the excitation pulse after 150 to 200 ns should not
have much effect on the peak voltage appearing at the secondary terminals. Indeed, the
lightning-arrester firing did riot have much effect on the peak secondary voltage until the
time-to-fire was reduced appreciably from 200 ns. This is apparent frorn the secondary-
terminal voltage waveforms •hown In Figure 4-23 for excitation voltages aibove the light-.
ning-arrnster threshold, As the lightning arrester is more highly overvolted and the time.-to-
fire decreases significantly below 200 ns, however, the peak secondary voltage beconmes a
smaller fraction of the peak excitation voltage, When the time-to.-fir( is comparable to or
less than the ft 50-ns time-to-peak of the iecondary voltage, thle secondary-voltage limiting
provided by the lightning arrester is quite effective, and the pebk secondary voltage Increases
very little with increasing excitation voltage beyond this point,

In addition to limiting the peak secondary voltage, the lightning-arrester firing tends
to reduce the width of the main socondary-voltage pulse, This L ipparent in Figure 4-23,
where it can be seen that as the excitation level increases, the time to reach the first zero
crossing In the secondary-voltage waveform decreases and the negative undershoot increases,
It Is also noteworthy that the effect of the firing of the lightning arrester on the secondary
voltage waveform Is so subtle that its occurrence Is very difficult to detect In the secondary
waveform,

242
r " III
*]EXCITATION LE EL Itk

1b) XCITATION LEVEL 54 kV

B kV

S Eu.|
T
(S
(iV * r,ýEXCITATION LEVEL 88U

200 ms

Figue 4-23 SECONDARY OUTPUT-TO-GR(OUND VOLTAGE OBSERVED AT

Xl FOR COMMON-MODE EXCITATION AND I.OADING (25-kVA GE unit;
.2!,Ctu J fl , IHill It ',)

243

i 'r :"'i '••..........i""'••i

;I ' "I. .. ". . . ".
................ '•"••"'
.". .. " •"i .. ...........
. . . "'". :•"•:'''
..- . .-
 The secondary-voltage-limiting effect is illustrated in Figure 4-24, where the peak
secondary voltage is plotted as a function of the peak excitation voltage for all four trans-
formers and for both common and differential primary excitation modes. Here it is appar
ent that the secondary responses are quite linear with the excitation level until tile peak
excitation level reaches about 55 kV (30% above the lightning-arrester threshold), at which
level the secondary voltage begins to level out. With -the 250-kV excitation level ithe highest
level used) the peak secondary ý )ltage was only 30 kV. This measurement was made on the
25-kVA G.E. transformer, which displayed the strongest coupling, and it was made In the
configuration that displayed the 6trongest coupling, It should also be noted that, although
the pulser was charged to 250 kV, the lightning arrester fired during the primary voltage
rise and limited the voltage actually applied to the primary to only 100 kV,

These tests were conducted with a low-Impedance source, so that even when the
lightning arresters fired, the excitation voltage at the primary terminals did not immediately
changw, When the source impedance was several hundred ohms, as It would be for an EMP-.
induced voltage wave incident on pole-mounted transformers driven by the distribution
lines, the excitation voltage collapsed very rapidly when the lightning arrester fired, This
value of source Impedance causes the peak secondary voltage to be limited even more
severely than was observed with the low-impedances sources used in the experiments
described above. Since the transformer Itself appears to remain linear, however, the effects
of source impedance and primary excitation voltage can be simulated analytically and the
secondary response determined from the linear transfer characteristics of the transformer,

4.3.2 LIGHTNING-ARRESTER FIRING CHARACTERISTICS 7

The behavior of the 9-kV distribution lightning arresters used to protect the trans-
formers was also obtainud from the transient tests of the transformers with the lightning
arresters installed, The lightning-arrester characteristics of primary importance in this
evaluation are the time-to-fire and the firing voltage, These characteristics permit an
estimate to be made of the peak voltage that might be applied to the transformer by a
fast-rising transient,

244
35

30 - -LIGHTNING ARRESTERS FH
I'AIA L.LE L
sliHI FS SECONDARIES
2b GNAIIS '2U-kVA (ib-NIHAL. FLEucTRI
* ~COMMON M(JUL LOADING

200

* 26-kVA WESTINGHOUSE COMMON-MODC, LOADING

15 1 25-kVA ALLIS-CHALME-1, COMMON-MODE LOADING

2-kAALLIS-CHALMERS DIPFFERENTIAL-MODEh LOADING

>~1 26-kVA WESTINGHOUSF DIrFFEERENTIAL-MODE LOMýDING

NOTE ALL DATA OBTAINFD USING GE MODEL,

91.28=0ARESTIISEXCEPT FON.
6O-k VA ALLIS-CHALMERS o A-U CHANCE MOUEL C70,14071500
0 GOMONMOD~ LADI~i McGHAW-EUISON MODEL AF 5CI TYPE UC
0
(a) COMMON-MODE EXCITATION CONFIGURATION

15

r-LIGHTNING ARRESTCHRS FIRE

2b.kVA GeNtRAL ELtCTRI1C

10I * COMMON-MODE LOADING

ý 2b-kVA WIESTINGHOUSE COMMON-MODE LOADING

I} T-kVA ALLIB.CHALMEr1S COMME.NI.MODC LOADING

WbkAWESTINGHOUSE DIFFERLINTIAL-MOD)E LOADING

*.k15VA
ALLIS -CHALMEHIS DIF[EHENTIAL-MODE LOADING

r1O500IS 200 250

PEAK EXCITATION VOLTAGE .- k
(b) DIFFERENTIAL-MODE EXCITATION CONFIGURATION

Figure 4-24PEAK SECOND~ARY VrlLTAGE AS A FUNCTION OF PRIMARY EXCITATION

FOH VARIOUS TRANSFORMER CONFIGURATIONS

245
 The tests indicated that all three of the lightning arresters shown in Figure 4-4 were
essentially identical in their firing characteristics. A plot of these firing characteristics is
shro-wn in Figure 4-25, The 9-kV arresters would normally be tested with thu transient
rate of rise of 100 kV/ps plotted in Figure 4- 25; during the transformer tests, the maximum
rate of rise achieved was about 2.5 kV/ns, or 25 times faster than that normally used to test
lightning arresters of this voltage rating, With this rate of rise, which Is also plotted in
Figure 4-25, the firing voltage was 100 kV, or about 2.5 times the static firing threshold,
and the time-to-fire was only 40 ns, In general, the trend of the lightning-arrester firing
characteristics displayed In Figure 4-25 is that to be expected of spark-qap devices having
large overvolted gaps,

The ability of the lightning arrester to protect the transformer under fast transients
for which voltages of 100 kV are attained before the lightning arrester fires Is believed to be
related to two characteristics of the transformer. These are the ionization (or spark forrna-
tion) time for bushing flashover and the Impedance of the bushing and winding Insulation,
Because the arc formation tirno for the lightning arrester is less than the arc formation time
for the bushing, the lightning arrester always fires before bushing flashover occurs, It is not
apparent that this relationship should change even if greater rates of rise were encountered,
since the lightning-arrester firing and bushing flashover involve similar gas-breakdown
processest

The ability of the lightning arrester to protect the winding Insulation against break-
down apparently depends on the fact that a finite time is required for the winding voltagfi
to rise because of the winding-to -cure capacitamcu and the b~ushring capacitance and induc"
tancLu For a typical tiansformer primary windint, the witnding- to core cdpacitance is a few
hundred picofdrads, the bushing copacitince is about 100 pF, and the bushing induutance is
a few hundred nanlohenrus. Furthwrinoru, ab illustrated in Fiiqur, 4 7 the bushirng iruluc
tance and bushing cupacitancu form a ow.. pass fi wi that limits thu rate uý rise (of the
voltage applied to the windinq oven if an infinite rate of rise is applied to the termindls,
Therefore, even If the lightnint. arrestor and hushing are subjecIed to very large rates ,,
rise, the rate of rise of the voltage across the winding insulation is much smaller, therehy
permitting the lightning arrester tu reach its firing voltage ibufhre the voltagle across tile
insulation reaches a dangerous level, Note that the CW-transfer- fnrction (dita indicate
that the hi(h- frequency rollioff, which is believed to be caused primarily by the !irimat -y

246

'V. IS........... . .
 bushing inductaoce and capacitance, uugins at about 10 MHz, implying a 10-to-90% rise
time of the order of 50 ns.

The rat, of rise of the open-ci:cuit voltage induced on power transmission lines
oy the EMP is of the order ol a few Eoc, where E, is the peak incident field strength and
c is the speed of light (see Section 2.2,3). For realizable polarizations and angles of inci-
dence, the maximum rate of rise of the open-circuit voltage is of the order of 50 kV/ns and
the source impedance associated with this voltage is of the order of 300 ohms. This rate of
rise is over an order of magnitude greater.than the fastest rate of rise developed in the lab-
oratory; however, the voltage actually applied to the transformer terminals is the fraction
Zi/(Zi - 300), whuro Zi is the paraltel input impedance of the transformers. Since
Zi< 300 ohms, at high frequencies (early timed) only a fraction of the open. circuit voltage
will actually be applied to the transformer windings,

!i 120

IO

L ) 10l)l 100

0 ino Iu 20 2io30
TIME-TO FIRE ",

Figure 4-25 LIGHTNING-ARAFESTFR FIRING CHARACTERISTICS

FOR FAST RATES OF RISE

247
 4.3.3 SUMMARY OF TRANSFORMER AND LIGHTNING-ARRESTER
PROPERTIES

Tests of selected lightning arresters and power transformers have demonstrated

that at rates of rise up to 2.5 kV/ns the lightning arresters are effective in protecting the
transformers from damage. It is believed that this protective capability should be applicable
for even higher rates of rise, since the protection is based on (1) the lightning-arrester spark-
gap ionizing and becoming conducting before other insulation breaks down, and (2) the
bushing inductance and capacitance limiting the voltage between the primary windings and
the core.

Studies of the transient throughput of the transformers with lightning arresters show
that:

(1) The bandpass character of the transformer slows the secondary-voltage rise-
time to about 50 ns and limits the significant duration of the secondary re-
sponse to less than 1 Ms..

(2) When protected by lightning arresters, the transformer remains a linear device;
all nonlinear activity occurs In the lightning arresters,

(3) The key transformer parameters affecting the bandpass properties of the
transformer are the primary bushing inductance and capacitance, and the
Interwinding- and winding-to-case capacitances.

(4) There is little difference in the basic coupling characteristics among the different
secondary-winding types or arriong the transformers from different manu-
facturers.

(5) The peak secondary voltage is limited to a value of the order of 20 to 30 kV

by the voltage-limiting action of the 9-kV lightning arresters for common-
mode excitation.

(6) Transients induced at the secondary terminals by the lightning arrevter firing
are barely detectable.

248
The test results show that the strongest coupling through the transformer occurs for
common- mode excitation arid common -miode loading, such as would occur if !he 3
phase transformer bank were delta-connected at both primary anid secondary sides. The
weakest coupling Occurs for differential miode excitation arid loading, such as would occur
for three phase wye-con nec ti ons at buoth sides,

The properties of the three models of 9-kV lightning arresters were virtually W01n-
tical insofar as tlwir firing characteristics are concerned, As the rate of rise of the voltagje
across the lighitning arresters increases, the timie-to.- fire decreases, but with the maximumn
rate of rise available in these tests (2,5 ky/~nsl the firing voltagle was only 100 kV arid the
time -to-fire was only 40 ns. This rate of rise is 25 times that for which the lightning
arresters are specified, while the firing voltage is only 2,5 times the static firing voltage.
It is significant that as the rate of rise increases, the firing voltage increases, and as the
time-tu.-fire decreases, the impulse fvdt passed by the lightning arrester does not change
vary much, Thus, the transformorfs secondary voltage for fust rates of rise tends to be
the impulse response and is relatively independent of the excitation voltage.

In operational systems, the use of lightning arresters between the primary terminals
of the transformer und groutnd Is recommonded for till service transformers to protect the
transformer, as well as to limit the high voltages Induced~ on the transmission linies anti
coupled through the transformer,

4.4 CITED REF~ERENCES

1, E/owlrical rrai,,s,,fssion,aif( Viftriti/timpi 8i 10k, 4th ed., 5t h priniting

(West inghouse Electric Corporatlion, East Pi ttsliu rlh, Prim., 1964).
2. W.W, Lewis, Th7w Perovti'u ' -)I Traiitpnissirmi SystvinN Aquainst Liqhhdtiiq
(john Wiley & Sunis, lInc., New York, N, Y., 1950).
3. L. V. Bewley, Travv'/itig Wavei. (ipnTrantrismissio Systems (Dover Publr)iCations5,
Inc,, New York, N. -'., 1963).
4. R. Rudenberg, FlecrtricaI Shock Waves in Power Systwims (Harvard University
Press, Camrbridge, Mass,, 1968).

'49

... . .. ...................... ý. W-W

5. A. Greenwood, Electrical Transients in Power Systems (John Wiley & Sons, Inc.
New York, N. Y., 1971).

6. E. F. Vance and S. Dairiki, "Analysis of Coupling to the Commercial Power

System," Technical Report AFW L-TR-72 -21, Contract F29601-69-C-0127,
Air Force Weapons Laboratory, Kirtland Air Force Base, New Mexico (August
1972).
7. R. T. Bly, Jr., and E. F. Vance, "High-Voltage Transient Tests of Service Trans-
formers, Lightning Arresters, and an Automatic Switching Unit," AFWL-TR*.74-
34, Contract F29601-69-C-0127, Air Force Weapons Laboratory, Kirtland Air
Force Base, New Mexico (October 1973).
8. J, H. Marable, J. K. Baird, and D, B. Nelson, "Effects of Electromagnetic
Pulse (EMP) on a Power System," Final Report, Intaragency Agreement No.
AEC 40-31-64 and OCD-PS-64-284, Work Uni* !2l 3C, ORNL 4838,
Oak Ridge National Laboratory, Oak Ridge, Tennessee (December 1972).
9. J. D. Cobine, Gaseous Conductors (Dover Publications, Inc., New York, N, Y.,
1958).

250
Chapter Five

LOW-VOLTAGE CIRCUITS

5.1 GENERAL DESCRIPTION OF THE LOW-VOLTAGE SYSTEM

5.1.1 INSTALLATION PRACTICE

The Iow-volttge circuits of concern In this chapter tire those between the main c(ir.-
cuit breaker and the loads (lights, appliances and equipment) it', the consumer's facility, The
source side of the mainicircuit breaker has heen discusse5d in Chapters Two and Three, This
part of the power system can be represented hy atsingle (though, perhaps, nmulticonductor)
transmission line whose characteristic impedance changes as its form changes from in aer ad
transmission line to a coaxial line at tho service entrance, On the load side of the main cir-
cuit breaker, however, the circuits branch out to serve the various requirements of the con..
sumer. The consumer's system may contain both single-phase and three phase lraadq
as illustrated ii Figure 5-1, and in general the loads may require two or more voltages,
necessitating more than one service entrince or tiar",.ormer bank, The system shown in
Figure 5.1 would be typical of a 120/208-V system in which single phase circuits are operated
on the 120 V between line and neutral and three-phase loadsare operated from the 208 V

251
(line-to-Iine) system. If other three -phase or Ringle-phase voltages are required, the low-vol-
tage system may also contain power transformers or other power-conversion equipment. A
more common practice where multiple voltages are required, however, is to bring the distri-
bution voltuwge into a vault where the transformers required to supply DIl of the consumer's
voltages are installed, Seperate feeders from the transformer vault may then be used to sup-
ply the main circuit breakers for the various three-phase and single-phase systems in the fa-
cility, If the three-phase service is delta-connected, a zig-zag transformer may be Included in

CIRICUIT
BREAKER~S
LIGHTS
RECEPTACLES
O~A
r- ANS

"•MALL APPLIANCES

INTHUSICIN ALAMM4

LIUHTS
DOOR AGtMUATI'S
) INOLE.-PHASE ML)TOH.[,

LAI4GF M0T1tM

AM CfONi]ITIONEIAS

MAIN "-- LAMAE P1WEll ,SLPI'L.IEN

EL.VATOMN
IMLAKL.i
LN rANCECIANS

ELECTIRIC F URNA(!;
-- .=" -. ,LARGE PUMPS

Figure 6-1 SINGLE-LINI DIAGRAM OF INTERNAL LOW-VOLTAGE CIRCUITS

252
the Iow-voltauge system to ground the low-voltage system. For somle Communications facili-
ties the commrciar~l power isusedI tO drive a motor-generator set which, in turn, supplies
the power fur Operating the facility. This system can piovide texcellent isolation 'if the, con
sumier %circuits from transients on the p)ower SYstem if separate machines (connected only
by the shaft) tire used and appropriate separation of the power circuit and the low-voltage
circuits is provided. Some large commercial computer installations use rectifier and storage-
battery systems to provide surge protection and short .-termstandby power,

Other surge-protection equipment that may be found in commercial installations are

secondary lightning arresters, surge limiting capacitors, and line filters, Secondary Ilightning
arresters that operate at a few thousand volts may be installed at the transformer secondary,
at the weatherheadl, tit the main circuit-breaker panel, on a circuit that requires special pro
tection, or sume combination of these, Surge capacitors are used to protect the windings
of large rotatirnq machines fromn the large rates of change in the leading edges of lig',tnlng-
Induced surges. Line, filters arP Used to minimize the transients that can enter sensitive in-
stallations such as computer centers, As Illustrated In Figure 5-2, combinations of these
protective devices are of ten used. Fur examp~lle, a combination of lightning arresters and
surle capacitorl; may be Used to Iimit both the wtak voltage and the rate of rise of the cur-
rent, or a ligh~tning arretltr may be Usedl to linmit the voltage applied to the Inplut term-inallS
of a lino filter.

Tho National Electrical Cude1 l sets standards for materials and practiCes Used inl low*
voltage wiring; inl add~ition, status, countius, and cities have their own 1w ildlni 00odes that are
of ten miore relitrnct ive than the natianal code. In spite of local variatitons, however, there
isconsideraible uniformity ;n the wviring prac tices ase I in conmmurc ia steel and masonry
buildings. The- low voltag irin in such ilitieS Nsusually instlled inl mtalliC condulit
andl raceways ot guLtters as illustrated ill Figure 5 3. All metal conduit, qutters, c~ircuit
blreaker Cabinets, out let boxes, etc. are reqa reid by the national codle to he "effectively
handed~ where necessary to assure electricifl cont inouit y ,. .". The low-voltai-le wiring i's
thas fairly Well endCOSed ill a mat allCIcshield Of SOrtN VXCepIt at p)lug ill app)IltaCC3 (Ccalculators,
typewriters, coffee- m~ikers, etc,), where the conductors are in4sulated buit Unshielded. We
shall limit ourtdiscu~ssion ill this chapter to low-vnltale systems with wiring in mutal conduits,

253
 CHOKE

SURGE SURGE ROTATING

A'ARRESTER ""CAPACITOR MACHINE

(a) LIGHTNING ARRESTER AND CAPACITOR

1 ~I
SURGE " IOAD
L
ARRESTER 9 I

I
I i I

IN6 FILTER

1b) LIGHTNING ARRISTER AND LINE FILTIR

Figure 5-2 SURGE-PHOTECTION TE(CHNIQUES FOR LOW-VOLTAGE CIRCUITS

Vinyl- or neoprene-insulated copper cc•:'ductors are used in applications where exces-

sive heat oi moisture is not a problom, These iductors are fished through the conduit be-
tween outlet boxes or cabinets, and extra wire left at the ends to allow the craftsman to
make the connections to the fixLure terminals, Figure 5-4 Illustrates the wiring inside a typi-
cal main circuit-breaker panel, The maximum number of wires that can be pulled through
standard conduits is given In Table 5-1.1 The dimensions of riqid steel conduit are given In
Table 3-2, Chapter Three.

5,1.2 DISTRIBUTION OF THE EMP-INDUCED SIGNAL

In moderately well-shielded systems such as the electrical wiring in metal conduit,

the signal induced In the exposed conductors (aerial transmission lines or entrance conduc-
tors in plastic conduit) and conducted into the shielded region is usually much larger than
that penetrating directly through the shield to the conductors. Therefore the primary source

254
 I3IAKEi CAD3INEti

Figure 5-3 TYPICAL LOW-VOLTAGE WIRING IN RIGIr, STEEL CONDUIT

265
Al ( lSIN H

VAINI~

HHAK. r

LOA

F CIRC-UITS VLAG IIN NIE AN ICITBEKE AE

256
 Table 5-1
MAXIMUM NUMBER OF CONDUCTORS IN TRADE SIZE OF
CONDUIT OR TUBING'
Conduit Trade Size 1/21/4 2 2// 3 3.. 4 41/2 5 6
(inches)
Insulating Conductor
Type Letters ( SizeM ..
ii~~~~~....
;k....... - 9 ....
TW,
SUW,T, RUH, 14
1247 9 15 25 44 60 99 142, 171
XHHW (14 through 8) 10 5 9 15 2636 60 85 131 176~
8 3 5 . 42.3..4.729712
RHW and RHH 14 6 10 16 29 40 65 93 143 192
(without outer 12 4 8 1324 32 63 76 117 157
covering), 10 4 6 11 19 26 43 61 95127 163
THWTHW .... .•• . B
8 . =....14 6 11 15 25 36 56 "...--'-
75 96 121 152
TW, 6 12 4 710 162336 4862 78 97141
T, 4 1 1 3 5 7 12 17 27 36 47 58 73 106
THW, 3 1 12 4 6,10152331 4 0 63 91
RUH (6 through 2), 2 1 1 2 4 5 13 20 27 34 43 54 78
RUW (6 through 2), 1 1 1 3 4"63- 14 19 25 31 39 57
P EPB (6 through 2), 0 1 1 2 3 5 8 12 16 21 27 33 49
"RHWand 00 1 1 1 3 5 7 10 14 18 23 29 41
RHH(without 000 111 24 6 9121519 24 35
outer covering) 0000 1 1 1 3 5 7 10 13 16 20 29
250
300
11
1
12
1
1 719 13
11
16 23
14 20

3500
500 1 111
701 1 1 1 32I4~ 8 67
0 11
121
164

75 1
.;2 3L 6-- 6 9

Keyt
T Thermoplastic
TW Moisture-resistant tliermnplastic
THW Moisture- and heat-resistant thermoplastic
RHW Moisture- afid beat-resistant rubber
RHH Heat-resistant rubber
RUH Heat-rusistaiit latex rubber
RUW Moisture-resistait latex rubber
XHHW Moisture- and heat-rimsistant cross-linked synthetic polymer.

257
of EMP-induced signal in the low-voltage wiring is usually the signal conducted in on the
service entrance.

At the main circuit--breaker panel, the voltage and current entering from the service
entrance are partially attenuated by the stray inductance of the wiring inside the circuit.-
breaker cabinet, and are partially transmitted to the load circuits. Because of the reflection,
attenuation, and division of the power at the main circuit-breaker panel, the EMP-induced
power transmitted to tile load-circuit conduits Is only a fraction of that Incident from the
service entrance, In addition, the stray inductances in the cabinets and junction boxes tend
to suppress the high-frequency portion of the incident pulse spectrum, so that the rise time
of the transient increases as the pulse propagates further into the low-voltage system, Each
load-circuit is also shock -excited by the incident transient and tends to oscillate at its natu-
ral frequer Jies, This shock excitation leads to voltage waveforms in the load circuits inoder-
Litely similar to the waveform shown in Figlure 3-10(a), Noedless to say, the EMP-Inducod
waveforms at any point In the low-voltage gystem are quite complex nlwivery difficult to
calculate accurately. As a roug h rule of thumb, however, the lurther from the main circuit,-
breaker panel, the smaller the peak voltage and the smller the rates of change in the waveform,

5.2 LINEAR ANALYSIS OF CONDUIT CIRCUITS

5,2.1 TRANSMISSION-LINE ANALYSIS (COMMON-MODE)

Elec.trical wiring enclosed in metal conduit can be visualized as a transmission line,

For a crude approximation, the power conductorý can hi, consideredI the (:enter conductor
of a co•xial transmission line in which the condo it is the outer, or return, conductor. The
low-voltaqe wiring system in the consumer's facility can then be represented as a network
of transmission- line segments such as that illustrated in Figure 5-5, At the ends of the con-
duits where the wiring enters circuit- breaker cabinets, outlet boxes, etc,, the wiring is more
open and can Ihe represented as a lumped inductance for short segments of, say, less than
2 m. Because the low-voltage systems contain many segqTrents of transmission linie with
many branches and junctions, even this crude representation of the low -voltage wiring is
very difficult to analyze without a computer. The vnalytical concepts required are quite

258
 rZ
SERVICE
ENTRANCEk

MAIN

CIRCUITS •, Zn-1

SUBSIVI AtIY
CIRCUI1 S

Figure 5-5 TRANSMISSION-LINE MODEL OF LOW-VOLTAGE WIRING IN METAL CONDUIT

simple; however, they must be applied to each segment and the results for the segments must
be combined In an orderly fashion,

In the transmission-line network of Figure 5-5, -for example, the upper branch of seg-
ments might represent a lighting circuit containing a wall switch lZs) and two ceiling lights (Zk),
This ciruult contains four segments of transmission line downstream from the main circuit-
breaker panel, A common occurrence in large facilities is represented by the bottom branch in
Figure 5-5, In this case one circuit from the main circuit-breaker panel supplies a subsidial y cir-
cuit-breaker panel, which may, In turn, supply additional subsidiary circuits.

The general approach for analyzinq such wiring as transmission-line netwnrks Is to

transfer impedances along the segments from right to left In Figure 5-6 to obtain the load im-
pedance on the service.-entrance conduit, When this load Impedance is known, the voltage or
current in the load can be determined from the techniques described in Chhpter Three. The
voltage and currents can then be translated from left to right toward the loads at the right
ends of the conduit circuits. At each junction of two or more conduits the Impedances
translated from the right must be combined with the lumped impedances at the Junction to

259
 Fi

ol)tiain the load impedance for the condIuit to the left, and the currlnt translated from the
left must hie divided amon. these impedances to obtain the co'I unt to translakte to the ;iqlht
toward the lUadi%,

Prior to performing this analysis it is necessary to determine

(I) The characteristic impredunces Z0, of the n conduit segments

(2) The propagation factors y,, for the n conduit segments

(3) The lengths ,, of the condul° agments

(4) The load impedances Z Ln at the right ends of the right-hand segments

(5) The stray inductances Lt. at the junctions of the conduit segments

(6) The Thevenin or Norton Equivalent source representing the service

entranice.

If these properties of the transmission line are determined, the load impedances tit the right ends
of the right-hand segments In Figure 5-6(a) can be translated to the left ends of the segments
by

1 )11ee2 )1
Z,. ,. kIi Z .1 ..... (5-li)

where

ZLn -Zo
. . . . (5-2)
ZL +Zon

This translatiori converts tile original circuit of Figure 5-.6(a) to the equivalent circuit of
Figure 5 6(hi, The inl)ut imt)edtanc Z,11 at the left end of the CIonduit it then added to the
impedance joL, of the stray inductance at the junction, and the impedancesZi, f j1Ln

260
 L zL 1 Z,

L,, L RIA CICI zb LO.D I.PEDNCE

L L EN LF

4 211.1 4 i4

ANDIRAUI
Is) LORIIAD (b) LOAD IMPEDAN4CE FO

IMPEDANCES COM- SEGMENT LEFT OF

13INED IN PARALLEL JUNCTION

Figure 5-6 STEPS IN TPANSLATING LOAD IMPEDANCES TOWARD DRIVING SOURCE

are combinud in pairallel as illustroted in Figjure 5-13(c), The loadI impedance on the right end
of the conduit S091ment to the left of the junction (see F igure 5-6(d) I is then

ZLo - jwl..,1 4 (ilj5-3)

where

--
- --------------.
. ------------ )

Zip PoL, + Zi, jwL 2 1Z2juwL, + Zl

This cycle can then livý repeated to translate the impedances toward the next junction to the
left and so forth, until theý load impedance at the end of the service -entrance conduit
is establlshe.d.

261
 The object of the Impedance translation is to combine the impedances of the branches
to obtain the lumped equivalent load for the service entrance. The voltage and current trans-
lations involve a reversal of this process, in that the starting point is the Thevenin equivalent
source representing the signal entering from the service entrance, and the object of the trans-
lations of the voltage and current from this source to the right Is to determine how this sig-
nal fans out Into the conduit network. Therefore we start with the Thavenin equivalent
source and its load ZoL In Figure 6-7(a) and calculate the junction voltage VI and total cur-
rent li as shown In Figure 5-7(b), The junction voltage Is

zip
V Zl + Zip + jLo Vo (5-5)

and the total current Is

VO

Zi + Zip + jWLo

where Vo and Z 1 are open-circuit voltage and source Impedance represented by the service
entrance at, for example, the main circuit-bieaker panel (see Chapter Three, Eqs, (3-7) and
(3-8) ). The voltage applied to each conduit to the right of the junction Is a fraction of the
junction voltage, however, as can be seen In Figure 5-7(c), The voltage V, applied to the
left end of the conduits is

Zin

Vn =LZln + JwLn V1 (5-7)

and the curient In In the conductors Is

Vi
Zin + jwLn

262
 V,

v r,
(iu LOAD DRIVEN BY SERVICE (b) JUNCTION VOLTAGE AND
INTRANCE CURRENT

L i v .... .. V 1L.

•3

tu) CURRENT AND VOLTAGE (cil CURRENT AND VOLTAGE TRANSLATED TO RIGHT ENUS
LAUNCHED INTO CONDUITS OF CONDUITS

Figure 5-7 STEPS IN TRANSLATING DRIVING CURRENT AND VOLTAGE TOWARD LOADS

Thu wurrunt onld vultoige ot tho riiht e;ndis oI the conduits ii F iqjure 5-7(d) are then

• I,,
-. .... (6-9)
1 / o1I In{

and

V V,510)

263
This cycle may be repeated for further translations to the right if it is observed that the source
impedance ut the right er.d of the conduit is

v11(511
Vn

and the open-circuit voltage V'o at the right end is

V• = In' (ZI-n 4. Z'1).( -2

This Thevenin-equiwlent source drives the junction (if any) at the right ends of the conduits.

5,2.2 DETERMINATION OF THE PROPERTIES OF CONDUIT CIRCUITS

In the Introduction to the transmission-line analysis above, it was stated that six
properties of the low-voltage system must be determined prior to performing the analysis
of the low-voltage circuits. One of these properties - the equivalent source representing the
signal entering on the service-entrance conductors -- is the subject of Chapter Throe. An-
other, the lengthv of the conduit segments, can often be measured directly, The remaining
four, the characteristic impedance Z0 , the propagation factor y, the load impedance ZL, and
the stray inductance I may be measured, or they can be estimated by techniques discussed
in this section.

5.2.2,1 Characteristic Impedance

Accurate formulas for the characteristic Impedance of multiconductor

transmission lines are available for only a few special geometries such as the symmetrical
shielded pair fnd shlelded quad shown in Figure 3-13, The electrical wiring fished through
conduit rarely assumes the uniform symmetrical geometry postulated in these formulas; how-
ever, the formulas can provide estim-ites of the characteristic impedance of the conduLtors in

264
a conduit, The formulas for concentric or eccentric cylinders can also be used to estimate
the characteristic impedance of a bundle of conductors in a conduit. The coaxWa -cylinder
formulas require an ustimated effective radius for the wire bundle, but since the characteristic
impedance varies as the logarithm of this radius, sizable error in tle effective radius can he
tolerated,

Simple experimental techniques can be used to determine the characteristic

impedance of Irregular geometries of uniform cross section, Resistance p3per or a shallow

electrolytic tank can be used as illustrated in Figure 5-8(a) to measure the resistance between
the conductors (all in parallel) and the corduit, This resistance is related to the Oharacteristic
impedance of a transmission line having geometrically similar cross section by

R 1Ir f207

where R is the measured resistance between the simulated conductors and the conduit, and
Rs is the resistance per squaru of the resistance paper or electrolyte, The resistance per square
is the resistance (for uniform current) of the medium (paper or electrolyte) one unit wide
and one unit long,

The time-domain reflectometer (TDR) is very useful for determining

characteristic impedance and velocity of propagation in situ (with the 60-Hz power off). The
TDR is essentially a fast.-rising pulse source and an oscilloscolp so arranged that the wave-
form of the voltaqe at the terininalb of the pulse source can be observed as the reflections
from discontinuities arrive. The oscilloscope is normally calibrated in units of reflection
coefficient relative to a base impedance such as 50 ohms. From the TDR oscillogram (see
Figure 5-8(b) ), therefore, the reflection coefficient p and the round trip propagation time
along the conduit circuit can be determined. For a base impedance of 50 ohms, the character--
istic impedance of the unknown transmission line is

1+
Z) 50 (5-14)

285
 CONDUCTIVE
SILVER PAINT RESISTIVE PAPER
N ,R. OHMS/SQUARE

1201ff

(a) 11RnSTANCE-PAPER METHOD

so
wV

U.

e li e l- - --

TIME

Ib) TOR METHOD

Figure 5-8 TECHNIQUES FOR C{LIERMINING THE CHARACTERISTIC

2*1
IMPEDANCE EXPERIMENTALLY

where p is the reflection coefficient obtained hy attaching the LinktlioWn' line to the mid of
the 50.-ohn line. Tthe veloc'ity of p~ropi~gationi is

V (5-15)

where V is the lenyth of the unknown lihe alid SI is the time elapsed between the arrival of

266
the reflections from the beginning and the end of the unknown line. The TDR method has
the important advantage that it can measure the characteristic impedance and velocity of
propagation regardless of how complex the geometry and the dielectric, between the conduc-
tors arid the conduit are.

5.2.2.2 Propagation Faotur and Length

As Indicated above, the velocity of propagation can be obtained from TDR

measurements when the length of the conduit circuit Is known. Since the attenuation of the
short conduit circuits ordinarily found In low-voltage wiring Is usually negligible, the
propagation factor for these circuitu is

'Y
; jo .M -- (I1
v

When the conduit Is burled or embedded In concrete, it may be difficult to

determine Its length. Then the TDR trace can give

2r =(5-17)
v

but a separate measurement Is required to obtain V and v explicitly, This measurement can
be made on a similar circuit of kc-,own length, such as on one section of conduit with the
same number and size of wires. Then the velocity of propagation is determined from TDR
measurements on the sample of known length, and the length of the unknown circuit can
be determined from Eq. (5-17) above.

5.2.2.3 Stray Inductance of Leads

The stray inductance of wiring in circuit-breaker cabinets, junction boxes,

etc., can be estimated from the formulas for the inductance per unit length of a wire over a
ground plane, The inductance per unit length of a wire over a ground plane is

267
 P 0 h Mo 2h I'l
L cosh. 1 - log -1(5-18)
21r a 27r a

where h Is the height of the wire above the ground plane and a is the radius of the wire, A
plot of the inductance per unit length as a function of the height-to-radius ratio h/a is
shown in Figure 5-9. For h/a between 10 and 100, the inductance is between 0.5 and
1,0 MH/m, A convenient rule of thumb, allowing for some mutual coupling In coiled or
folded wire in lunction boxes, is 1 MH/m for lead Inductance,

5.2.2.4 Load Impedances

The load Impedances represented by appliances and equipment are usually

quite difficult to specify accurately over the spectrum of frequencies contained In the EMP. This
difficulty arises from the fact that these loads are ordinarily spocified and designed for their
60-Hz, three-phase or single-phase
"---T77-"- 1
T"T-T'- - -- properties, whereas their common-
1,0 __•mode, higli-frequency properties are
of primary interest In the EMP analy-
sis, Therefore, an accurate determina-
tion of the high-frequency impedances
of such equipment can be obtained
- only t)y direct measurement, The
purpose of this section is to discuss
the general nature of the common-
.- mode impedances of selected loads
and to present some examples of the
impedances of three-phase and single-
0._._ .L±...L phase motors,
. 10 100
h
In general, the
common-mode impedances of most elec-
Figure 5-9 INDUCTANCE PER UNIT LENGTH trical equipment can be represented as
OF A WIRE OVER A GROUND PLANE

268
transmission-line segments. If tile total length of conductor protruding from the conduit
is short and it is ungrounded, as is the case for an incandescent lamp mounted on an outlet
box, the load will be capacitive. For such loads, the impedance is

ZL cjw• 1
jwC• (short, ungrounded) (519)

where k is the length of the protruding condututor, Z. is Its characteristic impedance, C is Its
capecitance per unit length, and c is the ,ipeud of light, If the length of the cunductor is
short and it Is grounded, the Impedance Is

Zo
ZL • jw -- -• jc L (short, grounded) (5-20)

where L Is the Inductance per unit length of the protruding conductor,

If the protruding conductor Is long and ungrounded, Its Input impedance is

ZL *-Zocot kv (long, ungrounded) (5-21)

wherr, k 'w/c, and if the conductor is long and grounded .t its end, its input impedance is

ZL - jZ, tan kV, (long, grounded), (5-22)

The Inductance and capacitance per unit length, and the characteristic Impedance for paral--
lei-wire, conic, coaxial, and parallel plate transmision-line configurations are plotted in

269
Figure 5-10. These data, the inductance data of Figure 5*-9, and the formulas above are use-
ful for estimating the common load impedlanceu of power equipment such as lamps or heaters
that have relatively small exposed conductors.

Becaust circuit elements with time constants shorter than 10 ns, will
probably have no effect on the response of the low-voltage system, we may neglect the
Inductance or capacitance of protruding conductors shorter than

1113m (ungrounded)
2~(15-23)

(grounded)
L

where Z0 Is the characteristic Impedance of the conduit circuit, and L is the Inductance per

1000 -l-- tV1 ' V ~ rl

0: cc

10 bo
RATIOb

Figure 5-10 CHARACTERISTIC IMPEDANCE, AND CAPACITANCE AND INDUCTANCE

PER INCH
270
unit length of the exposed conductor. Since Z0 is typically of the order of 30 ohms for the
cf)nduit, and L is ,"f the order of 20 nH/inch for openly protruding conductors, grounded
conductors shorter than about 15 inches can be treated as short circuits. Ungrounded con-
ductors shorter than 5 ft (10 i-s round-trip transit time) can be treated as open circuits.
(The round-trip transit time, rather than the lumped capacitance, determines the "time
conistant" ot ungrounded conductors unless the capacitance per unit length exceeds about
5 pF/inch.)

Three-phase and single-phase motors used to power pumps, blowers, etc., often
represent the loads on conduit circuits. Measurements have been made of the line-to-case imped-
ance and line-to-neutral impedance of two three-phase induction motors. The impedances be-
tween the three-phase conductors and the motor frame have been measured for both the
7-.1/2-hp and the 1/2--hp three-phase motors. This impedance is shown in Figure. 5-11 and
5-12. It is apparent that this Impedance behaves as a capacitance at frequencies up to almost
10 MHz, although the value of the capacitance changes near 100 kHz. At 10 kHz, the capaci-
tance of all three windings to the case is 0.0051 MF for the 7-1/2-hp motor and 0.0013 piF for

10' = 1 1 1 1i i7 7 - I Il' I 1 11lll 11 1 1 l 1l1 1 1 1 I l i l t'S

MorOR: 7-112 hp 3 PHASE

- 220 V 60 Ht A
1150 !]PM
r hr'anks-
F, Morse) e

S102•

-0.JLLLLWL..J ...llJJJL I IU ll IIll~

10o4 0o 106 107 10a

FREUUENrCY Hi

Figure 5-11 LINL-TO CASE IMPEDANCE OF 7-1/2-hp ILNDUCTION-MOTOR

WINDING

271
 IT,-T-lT11 Ty T -I I 7T

.C' 0.0013 pF

- iW)TOH.
No 1' hp 3 PHASE _
270 V 60 fieL'

NCU P

dif~~~ct
w~~~ndinr;(
ho (lh LoL'S
CtI1C0) SlIi( I( ~ fhltt ltta h us

alo dl ~l h ioiq IelriuI~ritertr ttulr. Ti OlitII

0 A 1 C, 11210
 10 4m1 TT 7TTFTrrY__-

"• 13 •C 0,0014 F T4

10

MOTOR: 1/2 hp SINGLE PHASE I

S~1110 V 60 Hr
-- 1 72B RIPM
S~(Dayton Eluctric)
0

104 10 5 10 0 10 1 10a
FREQUENCY ... Hi

Figure 5-13 LINE.NTO-CASE IMPEDANCE OF 1/2-hp SINGLE-PHASE

CAPACITOR-START MOTOR WINDING

of measurement is no concern for the 3-phiase induction motors, but for the single--phase :
capacitor-start motor it means that the measurements were made with tihe starting winding

connected, and that the results may be different on a running motor with the starting wind-
ing disengaged,

5.2i3 MU LTICONDUCTOR-TRANSMISSION-LIN E ANALYSIS

A potentially more exact analysis of power conductors in conduits than tne common-
mode analysis described above treats the conductors as a multiconductor transmission line.2"4
Each conductor is then coupled to every other conductor in the cnnutrit through mutual
capacitanc•s and mutual inductances, and loads at the ends of the conductors become
networks of elements, rather than single elementso a ssumed in the comrion-modt model.
Transmission-line voltages, currents, impedances, and admittances are represented by matrices

in the multiconductort--ransmission-line analysis. Evaluation of the voltages and currents

273
at points in the low-voltage wiring isvery difficulIt if there are more than two conductors
in each conduit. Because of the complexity of the rnUlticonnuctor -transrmission -I inti
analysis, all applications of this method to power wiring have involved developing computer
programs for performing the matrix manipulation and numerical eva~luations,

Although the multiconductor-transmlssion i~ne analysis has the potential for piro-
vidim. more arcurate results and for providing voltages between conductors as well atscom-
mon-mode voltages, this potential can only be realized if the properties of the power circuits
can lie accurately specified. In practice, wiring fished through conduit meanders somewhat
randomly along the circuit, so that the cross section of multiconductor circuit isnot uniform
along its length, Furthermore, to specifiy the load impedance of simple devices such as three-.
phase induction motors, a large number of measurements over a wide range of frequencies
are required, It isquostionable, therefore, that sufficient increase in accuracy can be
achieved to warrant the large increase in complexity (or effort) required tc* poi orm the
miulticoniductor tranismissioni-line analysis, Sinice power,-system anialyse ar~e usuailly p)er-.
formed primarily to determine p~eak voltages, that might damage insulation or components
and] to determine tho Lype andl quality of protection required for- sensitive eq~i ilment, a
very rigorous analysis is seldom required.

A less complex version of the inulticonductor transmirifin Iinno analysis is somnetimnes

used. Because the neutral of a three-phase system or the conter tap) of a 120/240 V single-
phase system isoften g~rou~nded at the load as well as at thp :uwer gr-ound near the set-vice
eft rance, these condluc tors muly he)cl;lve quite differently from thle u11olou nded "line"' con
doctors, TIhus if thle neutral or (Aroundl conriuctors are I reate(I as one conductor and oll tho
" Ii nv" co nductors inrcom mon are t reated ou a sop1vnr;its, coniductor, thei mItWcm Id uctOr
circuits Can1 he analyzed as three-conductor (two inner conductors and one C ndlu it) tranMS-
mission lines, This meathod of analysis offers som-fe improvement over tile COMMOrn-1mOth*!
anadlysis, Yet because PCact conduit is presurmed to contain only two conduIctors, it iNmu1Lch
simp~ler than the complete miulticonductor -transniissiun - line analysis. Snme di Ifico Ity is,
encountered, however, where conductors from one conduit aur separated ind ted irto two
or miore conduits, because siicl i branncii does n10t constitute a solid junctionl Of three-Cun -
(ILI(tor transmuissioin lintes,

274
5.3 CITED REFERENCES

1. National Electrical Code 1971, (National Fire Protection Association,

Boston, Massachusetts, 1971).

2. E. F. Vance and S. Dairiki, "Analysis of Coupling to the Commercial Power

System," Technical Report AFW L-T R -72-21, Contract F 29601-69-C-0127,
Air Force Weapons Laboratory, IKirtland Air Force Base, New Mexico (August
1972).

3. S. Frankel, "The Differential Equations of a Transmission Line Excited by an

Impressed Field with Axial Magnotlc Component," Interaction Note 125, Air
Force Weapons Laboratory, Kirtland Air Force Base, New Mexico (September
1972, unpublished report).

4. S. Frankel, "TEM Response of a Multiwire Transmission Line (Cable) to an

Externally Impressed Electromagnetic Field: Recipe for Analysis," Interaction
Note 130, Air Force Weapons Laboratory, Kirtland Air Force Base, New Mexico
(February 1972, unpublished report).

275
 ri
Chapter Six

GROUNDING SYSTEMS
I

6.1 GENERAL DESCRIPTION OF GROUNDING SYSTEMS

6.1.1 POW•,ER-SYSTEM GROUNDING

Grounding systems, although seemingly simple, are In fact quite complex and often
controversial, The grounding system designed by the utility must be compatible with the
lightning protection system, the protective relaying system, the system Insulation, and the
grounding systems of other utilities with which it is intertied, 1' 2 Th!)power grounding sys-
tem rniquired of the consumer by national or local electrical codes is specified primarlly for
personnel aIri property protection. The consumer's equipment grounding system may be
designed primarily for interference reduct,,'n so that objectional crosstalk hetw'en subsys-
tems is minimized, Therefore, there are at least three ground systems serving essentially
different functions within the utility's and consumer's combined system.

276
 Grounding in the transmission and distribution systems is necessary to prevent the
conductors from "floating" to high potentials and inducing insulation failure, High poten-
tials on floating systems can be generated by lightning, transients associated with switching
or faults, leakage f¶om a high-voltage system to a lower-voltage system, or from static
electrification during ,now, sleet, or dust storms. A system ground also facilitates fault
detection and relaying because a fault on any phase can he detected and localized more
readily if the grounding system is properly designed, A typical neutral grounding system for
a three-phase transmission system is illustrated in Figure 6-1, Tile generators feeding the
trantmission bus are either ungrounded, or grounded through sufficient impedance that
fault current through the generator is limited to a safe level. The transmission line is grounded
at the neutral of the wye-connected transformer secondary, as is the subtransmlsslon line
and the distribution line, Tronsmission and subtransmission are normally over three-wire
lines, while distribution lines serving both three-phase and single-phase loads are usually
four-wire line%,

Not shown in Figure 6-1 is the overhead ground-wire system used for lightning pro-
tection on transmission and subtransmission lines. The overhead ground wires are usually
grounded at each pole or tower, but they are not connected Lo the neutral except through
the soil, The overhead ground wires thus form a separate ground system, essentilvly inde-
pendent of the three-phase neutral,

At the consumer end of the distribution system the distribution transformers miy he
connected in any of the four possible delta and wye combinations shown in Figure 6-2,
When the primary side is wye--connected and a four-wire distribution line is uged, however,
the neutral line is usually not connected to the neutral of the transformer bank, This is Ibe-
cause a phase-to-groind fault would then place line-to-line voltage ,L-ross tile transformers
on the unfaulted )hases, with the co',sequent overvoltage und probable damage to the trans-
former and consumer equipment. To avoid this situation, thc neutral of the distribution line
is sometimes grounded one pole bar k from the transrormers and the transformer neutral is
grounded at the transformer pole so that the soil resistance limits the neutral current flow.
However, this method hias the disadvwntage that large voltage gradients in the soil near the
consurner's facility may accompany a phase-to- ground fault.

277
r

sun-
GENERATOR TRANSMISSION TRANSMISSION
BUS 1 skis Bus

- Ii

Ii

GENERATOR TRANSMISSION SUBTRANSMISSION DISTRIBUTION

Figure 6-1 TYPICAL GROUNDING ARRANGEMENT FOR A TRANSMISSION SYSTEM

HV LV HV LV

N N

Ia) WYE-WVYE Ib) WYE-DELTA

HV LV HV LV

4L!~-ZAG

(I) DELTA-WYE (d) DELTA-DELTA

FIgur. 6-2 THREE-PHASE DISTRIBUTION TRANSFORMER CONNECTIONS

278
 The delta-connected secondary shown in Figures 6-2(b) and 0 -2(d) may be pro-
vided with a grounded neutral through a zig-,zag transformer or some other means to meet
the electrical codes, The delta-connected secondary is not uncommon, however, where
460 V (or higher) 3-phase service is required in addition to 120/240-V single-phase service,

6.1.2 GROUNDING LOW-VOLTAGE WIRING

The National Electrical Code requires all 4.-wire, three phase circuits of 480 V or
less to be grounded if the neutral is used as a circuit conductor. 3 In general, If ground on
any ac circuit can be achieved so that the voltage of the ungrounded conductors does not
exceed 150 V the system must be grounded, Thus, one side of 120-V single-phase systems
and the center tap of 120/240-.V single-phase systems are grounded, A single-point ground
Is recommended (but not required), with the ground electrode being one of the following:

(1) A water pip(e or well casing at least 10 ft long

(2) The metal frame of a building If effectively grounded

(3) Gas piping (where permitted)

(4) Other underground piping, tanks, etc,

(M1 Concrcte.-encased reinforcing bar of underground footings

(6) Specially installed rod, pipe, or plate glround electrodes wnerc resistance to
ground does not exceed 25 ohms,

All metal conduit (rikqi( or flexiblh), raceways, gutters, metal equipment enclosures, and
boxes must be horded fnr ;'lectriu:al continuity. Although the naticnal code establishes
these bonding and grounding recluirements, there are few (lual iy-control checks to ensure
that the reqoirements arev met, ant11,
the requirements are often so vaue alb tO be subject to a
wide range of interpretutions. The phrimary purpose of these requirements is to prevent per-
sonnel hazords aod *il•Imunwt
,l( ilmnatqrl' thut might result if un(JIoun0ded equipment cabinets

279
or outlet boxes inadvertently became energized; as they are practiced, the code require-.
ments appear to be satisfactory for this purpose,

6.1.3 GROUNDING ELECTRONIC EQUIPMENT

If the consumer's facility contains extensive electronic equipment (e.g,, a communl-

cations center or computer facility) additional grounding systems may be necessary. In
these facilities It Is not uncommon to find two to four ground buses that are interconnected
only at one facility ground point (such as a well casing or a pipe driven expressly for ground-
ing purposes), 4 This grounding system is required to permit substantially balanced circuits
to be used and segregated according to function served, so that mutual interference between
ac power, dc power, and low-level-signal circuits can be reduced to a tolerable level. A sep-
arate tree system for each circuit class is usually claimed, although in practice there are
usually some rlosed loops in the branches of the ground trees. The principal trunks and
branches of d ground system for a telephone electronic switching center that contains four
ground systems In addition to the ac power ground are illustrated in Figure 6-3.

6.2 EVOLUTION OF A FACILITY GROUNDING SYSTEM

As is pointed out in Section 6.1 above, the grounds for the transmission system, the
building wiring, and the electronic equipment in the building are for quite different purpobes,
In general, the reasons for grounding are varied, and it would be presumptuous to attempt to
specify grounding procedures without first establishing the reasons for grounding and the
goals that the grounding system should achieve. These reasons and goals are usually based on
system functional, safety, and RF interference considerations and aro inherently prescribed
by the system specifications, When th,. EMP is included as a consideration in the ground-
system design, at least one more goal hiv been added (EMP hardness), but the reasons for
grounding may remain unchanged. It i- therefore prudent for the designor to avoid becoming
so engrossed in the method of grounding the system that he forgets why it is grounded.

280

i Eli
 ql(I
IIii
.... t''Ni
....
I...
I ....
. .....

Lf., 1(..
NI ; MJOUND

iH I n
I(1
(MO ' UNtD

SI V 1 Moi.1**IuI(I i'i.N ovP!

jill IUj I;,1IcJI I. IC. 'lI U T 161

1"HANi5MIM8ION ,iHOUND)
WI I

Figure 6-3 GROUNDING SYSTEM FOR A TELEPHONE SWITCHING CENTER

The basic reason for providing a "ground" in electronic equipment is to establish a

firm reference potential against which signal and supply voltages are measured (or established),
For small systems such as the home radio receiver illustrated in Fi(gure 6-4(a), a metal chassis
can be used as the reference even though it may not be "grounded" (there is no need to
ground the chassis if it is so encased that there is no possibility of a shock hazard), The
chiassis potential is the same (almost) througqhout the set, SO that the power supply, the RF
amplifiers, the mixer, the IF and audio amplifiers, etc. share a common reference elec-
1l1
trode and only a signal-carrying conductor and a power-supply conductor are needed to
Interconnect these stages, If the chassis were dielectric instead of metal, it mlght be neces-
sary to provide two wires between stages for the signal and two for the dc power as Illustrated
In Figure 6--4(b), If one of the wires is used as a common reference for Ioth power and signal,

281
 R . IFDET 51 POWER
LO.
I AP AUDiO PL
AMP.MIXER APAMP, SPL
.01 0010 1 0- 1- ll CHASSIS '1 0,o 0.-1-0l0 0-1

(l METAL CH4ASSIS AS GROUND FOR POWER SUPPLY AND RECEIVER CIRCUITS

R bFEPRT ROUND IFORSA ADE d& POWER

6-4i
AMP.r O SMPLE
A
AXMLSMFPOUD SADO
PPCLVY

flucuatins I pocr*-~ppl
curentmigh indce di/t vot~qS1o ojetiNAL ppr

power-tiosupnplsetonwere-l builetand packaged

h as searte modltaes tha wberton physical-

ltvels in the r"ceiver" quThusml, shieldsemsvled interconnecting

wire. dbles wouldferrequired.t

prevente ambient noise, piut uo

th e instemayrionnoecateingh
carles frommodngthro performt

anevl of the receiver, aridquthe smodleswullaeob

shielded teconctn
prevent thud
e rentererence

on the cable shields from coupfling to the signal conductors. Now a third reference system,

282
the shield, has been created, It must be connected to the signal and power-supply reference
conductors to prevent excessive potential differences between the shield and the internal
circuits, but if the shield and the other grounds are connected together at more than one

SIGNAL GROUND

... ..._-j. L-ET

( ~ - ~ . ~ 9 ----
IE - - - - r - - I AMP

SHIELD
GROUND POWem-SUPPLY GROUNP

Figure 0-5 SIMPLE RECEIVER SEPARATED INTO MODULES INTERCONNECTED BY

SHIELDED CABLES

point, voltage drops along the shield may induce malfunction of the receiver, In addition,
because the shield Is presumably exposed to operating personnel, it must be grounded to
earth or building structure to prevent shock hazards, The or!glnally simple receiver is now a
complex, Interconnected system with three "ground" systems,

The simple radio receiver, made complex by regressive evolution, provides an excel-
lent example of the advantages of the popular single-point, tree ground system illustrated In
Figure 6-6, The entire ground system is grounded at one point, and each subsidiary ground
branches out from this trunk attached to the single ground point, If any of the branchor
had been connected to each other at more than one point there would be a reasonable doubt
that the system could achieve the desired performance (even in the absence of extraneous
in:luences such as tho EMP). However, If we examine the shield system of Figure 6-5 care-
fully we observe a situation that poses a paradox for the tree concept of grounding. Although

283
the signal and power-supply grounds follow the single-point, tree concept, the qhirold system
does riot, because there are loops between the shielded cables interconnectinq each pair of
modules, Thus, the shield system is a mesh, rather than a tree.

While it would be easy to redesign the simple radio receiver of Figure 6-5 to elimin-
ate the loops in the shield system, in more complicated systems It is sometimev virtually
impossible to eliminate dlI loops in all of the ground systems, Consider, for example, a
large system with a feedback loop
such that the output of one module
...iis operated on by a second module
and fed back to an upstream module
interconnected with the first, In this
case it is likely that the shield, signal
common, and power-system common
will ail contain loops. The problem
Flgute 6-6 SINGLE POINT TREE GROUND can be even further complicated If
SYSTEM the module enrlosures are installed in
a steel building so that each enclosure
is grounded through its mounting

hardware as well as through the shield system. The latter problem is illustrated in Figure
6-7(a) where two modules are interconnected through the cable shield and the mounting
structure, Even it the cabinets are insulated fromn the metal floor, the capacitance between
the cabinets and the floor would close the loop, as is suggested by the dashed capacitances in
Figure 6-77(a), Opening the loop by breaking the shield, as indicated in Figure 6-7jb), does
not eliminate the problem; it merely changes its form. Whereas the short-circuit current
induced in the loop of Figure 6-7(a) flows in the cable shield, the open-circuit voltagle
induced In the loop can drive the internal (circuits when the shield is hrokwn as in Figure
6-7()).

284
6.3 PRINCIPLES OF GROUNDING

6.3.1 SINGLE-POINT VS. DISTRIBUTED GROUND

The examples in the preceding section illustrate thdt implementing the single-point
tree ground concept in systems with many racks mounted on a common metal structure or
in systems with complex feedback loops may be very difficult or even undesirable, In this

1a) GROUND LOOP FORMED BY CABLE SHIELD

vo' of

(b) GROUND LOOP OPENED BY COMPROMISING SHIELD

Figure 6-7 LOOP FORMED BY CABLE SHIELDS AND METAL FLOOR OR

WALLS

section, an attempt will be made to evaluate the relative merits of the tree ground system
and some violations of the tree ground, For this discussion, two aspects of the grounding
systems will be considered: how it affects normal system performance, and how it affects
coupling to the system from external effects such as the EMP.

Consider a single-point ground consisting of only the ground point and trunk as
illustrated In Figure 6-8. The ground point is assumed to be a point on a metal plane, such
as the steel liner of a shielded building, and the trunk is of length V and of height h above
the ground plane. If this ground trunk is illuminated with a plane wave, incident at an ele-
vation angle 0 to the ground plane and an azimuth angle p to the axis of the wire, the open-
circuit voltage developed between the open end of the wire and the ground plane can be

285
 h 10C v

Figure 64 SINGLE-POINT GROUND AND ONE TRUNK

calculated from Eq. (2-18) with pq - -1 and r. ft 0. The impedance looking into the open
and of the wire is

Z(0) jZo tan 02 (6-1)

If perfect conductors are assumed and R>> h.

At frequencies such that Q<<c/f, the Impedance and open-circuit voltage expres-
sions reduce to

sin ýpsin

Z(0) * jW cosh_1
2ir a

where Ho is the incident-magnetic-field intensity, po Isthe permeability of free space, a is

the radius of the wire, and the upper and lower trigonometric functions apply to vertical and
horizontal polarization of the incident electric field, respectively. The voltage is Just that
induced in a loop of area hQ by the magnetic flux 2 Io Ho X(angular function) linking it. For
a 1-V/m plane, vertically polarized wave incident at 0 - 0, the open-circuit voltage induced
at 1 MHz at the end of a conductor 2 cm in diameter, 2 cm high, and 3 m long is 2.5 mV,
and its impedance is 15 ohms, If the right end of the wire were connected to the ground
plane to form a closed loop, a circulating current of 0.5 mA would flow in the loop at 1 MHz.
Thus, from the standpoint of induced effects, the question is, which is less desirable, the 2.5-
mV potential difference between the ground tree and structural ground or the 0,5-mA cur-
rent circulating in the ground loop?

286
 The answer to this question will depend on the system requirements. If the ground
tree is that formed by the interconnecting cdble shields, the circulating current will usually
be less objectionable than the potential difference because the cable shields eliminate much
of the effect of this current on the conductors inside the shield. If the ground tree is the
signal common, however, the same potential difference will be induced in the signal load as
is induced in the common lead, so that the potential difference between the signal and com-
mon leads is ntqligible. Unless the terminating impedance is balanced and the terminal cir-
cults have adequate common-mode rejetinn, however, part of the common-mode induced
voltage may be converted to differential voltage.

6.3.2 GROUND IMPEDANCE

From the standpoint of minimizing the Impedance to ground of the equipment at

the right end of the ground trunk shown in Figure 6-8, a direct connection to the ground
plane Is much superior. The impedance between two studs 2 cm in diameter welded to the
ground plane 3 m apart is shown in Figure 6-9 for steel, aluminum, and copper ground
planes that are thick compared to a skin depth. For comparison, the impedance of a ground
cable 2 cm in diameter, 2 cm high, and 3 m long is also shown. The impedance of the cable
is 10 to 100 times greater that that of the ground plane in the megaher t. quency range
shown (except at the half-wave resonance frequencies for the cable), Thus if the objective
of the ground is to firmly connect chassis or equipment frames to a low-impedance common
bus, connecting these parts firmly to the ground plane by the shortest path possible provides
the best solution.

In general the impedance of a wire ground connection varies widely with frequency.
At low frequencies, such that the ground lead inductance is small compared to Its Internal
impedance, the impedance of the ground is determined by the internal impedance of the
lead and/or ground plane plus any contact resistance (in fact, the contact or junction resis-
tance is often larger than the resistance of the metal). The low-frequency resistance applies for

f R (6-4)
27r L

287
 r.z

where R is the resistance of the metal in the ground conductor and L is its inductance.

At high frequencies, the ground impedance is dominated by the inductance of the

ground lead, which can be estimated from Figure 5-10 or from

L ( R<
f < (6-5)

where Qis the length of the ground lead, Zo Is its characteristic impedance as a transmission
line, and c is the speed of light (3 X 108 m/s), Since wires over ground planes have charac-
teristic impedances of a few hundred ohms, ground lead inductances are often about 1 AH
per meter of length, The upper limit on the lumped-inductance behavior of the ground
lead Impedance Indicated above is the frequency at which the length of the ground lead
approaches a quarter wavelength. For high frequencies such that f > c/4Q, the impedance of
the ground lead changes radically with frequency, alternating between a very large value and
a very small value as is Illustrated by the cable impedance In Figure 6-9 for frequencies
above 10 MHz. The impedance of the ground lead is then

'Zg JZO tan (21rf - (6-6)

for perfect conductors, which varies between - and + oc as the frequency increases.

6.3.3 TREE GROUND

Suppose that two branches are added to the ground wire of Figure 6-8 so that we
obtain a ground wire with a trunk and two branches as shown in Figure 6-10. Now if this
ground tree is the signal ground for the system, and signal return currents I1 and 12 are
flowing to the single ground point, their sum I1 + 12 must flow through the single trunk. The
voltage Vo developed by these return currents Is

Vo = (11 + 12)Zc (6-7)

288
 -1 1000

22
(3

2 c

2 Prn

1 10 100
FREQUENCYV- MHz

Figure 6-9 IMPEDANCE OF METAL GROUND PLANES (left scale) AND OF A

GROUND CABLE (right scale)

where Zc isthe Impedance of the line between the Junction and the ground point, In
addition, the voltages V1 and V2 between the right ends of the tree branches anld the
ground plane are

V1 11I(ZI + Zc) + l 2 Zc " 8)

289
 V2 = 12 (Z 2 + Z:) + l1 Zc (6-9)

so that the signal-ground voltage at the end of Line 2 depends on the return current I1. If
the numerical values obtained earlier are used, the impedance of the trunk and each branch
is 5 ohms at 1 MHz, so that 1 A of return current in either branch will produce 5 V in the
other branch, and vice versa. Thus, the common trunk has introduced mutual coupling
between the subsystem at the end of Branch 1 and the one at the end f Branch 2. Such
mutual coupling may be undesirable (or intolerable) from the point of view of system
performance.

. -V2
+ ~1
.,=----11
•
"--

.V 1

Figure 0-10 TREE WITH TRUNK AND TWO BRANCHES

.2
vi

Figure B-11 JUNCTION GROUNDED TO REDUCE COUPLING THROUGH TRUNK

290
 6.3.4 DISTRIBUTED GROUND

The mutual coupling in the trunk can be greatly reduced by grounding the junction
to the structure as illustrated in Figure 6-11. If the junction grounding lean is of length h,
and radius a, its Impedance will be

ljoh 2h.
Zn W=-2- 10 7o- (6-10)

which is about h/V times the impedance of the trunk of length Q( >> h > a). This reduc-
tion of coupling through the trunk has been achieved at the expense of forming a "ground
loop" containing the trunk. This loop may be quite acceptable for many systems; however,
it may be objectionable under either of the following conditions:

(1) Grounding the signal common merely forces a common-mode signal conver-
sion to take place at the junction (for either the desired signal or the externally
Induced signal),

(2) The loop currents in the trunk can induce significant signals in the branches
(e.g., if the trunk and branch are in the same cable or bundle),

6.3.5 STAR GROUND

An alternative method of reducing the coupling between the branches caused by the
common trunk is to eliminate the common trunk by forming a 'star" ground with three
spokes as Illugirated in Figure 6-12, For the simple example being considered here, the
single-point star ground system would be preferred for a signal ground system In which there
Is negligible coupling between the branches (spokes), It should be pointed out, however, that
If the three branches are In the same cable, there Is a possibility of coupling (crosstalk) among
them from both mutual capacitance and mutual inductance between the conductors.

291
 2I;In practice, twisted pairs (or, if neces-
v1 sary, twisted shielded pairs) can be
used for the signal and signal common
leads to minimize cross-talk between
// / /• /4 / //////// /4 the circuits in a common cable. For
more complex systems involving feed-
Figure 6-12 TRUNK ELIMINATED TO FORM
A STAR GROUND back loops, the star ground with
twisted pairs does not offer a solution

to the problem of potential differences between separate parts of the system 1e.g,, V 2 , V1 ).

6.3.6 SOME GUIDELINES FOR SYSTEM DESIGN

It is evident from these discussions that one cannot a priori specify an optimum
grounding system -that is applicable to all systems or even to all parts of a given system. Each
system and each part of a complex system has its own peculiar requirements based on func-
tional and safety considerations, The design of the grounding system Is therefore an integral
part of the system design; It cannot, In general, be designed separately and added to the
remainder of the system, Neither can rigid specification of a particular ground type ensure
that the system will function properly or be immune to external interference such as the
EMP. In the final analysis, it is the circuit (system) designer's responsibility to ensure that
the system performs Its Intended function reliably and safely In the environments specified.
The following guidelines are offered to assist the designer In achieving his objectivos:

(1) Use a tree or star ground system with twisted pairs and balanced signal circuits
wherever practical,

(2) When loops are formed by the Interconnecting cables between separate modules,
use balanced input and output stages with large common-mode rejection for
the signal pairs.

(3) Use separate ground returns for circuits that have large signal return currents or
large rates of change of signal return currents to avoid mutual interference In
common trunks or branches,

292
14) If the system is installed inside a welded steel (or other metal) shielded enclo-
sure, hono all cabinets anti equipment racks to the shield. The 9hields of shielded
interconnecting cables should he connected to the cabinets (or racks) at hUth
un(N evuen if this forms a loop.

(5) If the system is Inside a closed shield, the walls of the shield should be used as
the common ground for the system. The signal common should be connected
to the inside of the shielded enclosure, and the ac power ground should be con-
nected to the outside of the shielded enclosure (e.g,, in an entry vault contain-
Ing all other penetratlor.s such as communications cables, antenna leads, etc.).

(6) Avoid connecting the power neutral and external cable shields to the same
point on the room shield that the signal ground is connected to. This will min-
imize the possibility of coupling large EMP-induced signals on these external
conductors to the signal common through the ground-point impedance (see
Items 3 and. 5 above).

(7) The building or site ground for the ac power (as required by the electrical code)
should be connected to earth external to the facility shield. This earth connec-
tion may also serve as the external ground for the facility shield if the voltage

drop between the shield and the earth Is not a safety hazard or is not other-
wise objectionable.

(8) The use of balanced signal circuits with twisted pairs for interconnecting the
subsystems and common-mode rejecting schemes (e.g,, Isolation transformers)
on those circuits subject to externally Induced signals, together with adequate
overall shielding, can usually be implemented to minimize the effect of the
EMP on the system.

(9) The use of nonzero ground impedances (capacitances for high frequencies,
inductances for low fiequencles, and resistance for statlc-electricity bleeders)
can often be used to advantage, particularly In view of the fact that zero-impe-
dance grounds can be approached only at low frequencies. Often devices such
as surge filters, chokes, and high-resistance bleeders can contribute to better
overall system performance,

293
6.4 GROUNDING COUNTERPOISE

The use of a grounding counterpoise to reduce the surge impedance of the ground
(earth) connection is a common practice for transmission lines and stations. For transmis--
slon lines, the counterpoise mny be an array or grid of buried conductors fanning out from
the tower footings, or it may be one or more continuous buried conductors under and par-
allel to the power conductors to which the towers are connected. 1' 2, 3 For the transmission
lines, a primary consideration Is to obtain an Impedance to ground sufficiently low that when
lightning strikes the tower, the IZ drop at the tower footing does not exceed the flashover
voltage of the line Insulators. Another consideration Is that gradients In the soil be mini.
mized for safety reasons (particularly in populated areas) In the event of a line fault. Station
grounds may be large buried grids, rings, or metal pipes. The station counterpoise serves the
same general purpose as the transmission-line counterpoise as well as providing the reference
for relaying and other protective and control systems.

Large communication stations such as telephone switching centers often use a

counterpoise in the form of a well casing or a buried cable "ring" around the building, or
both. Such a counterpoise serves a function similar to that served by the power-station
ground; It provides a tow-impedance ground reference for the communication equipment
and It minimizes the gradients In the soil due to surges Induced on the power and communi-
cation cable systems,

In an EMP environment, the counterpoise (or at least the common ground point) may
have some effect In reducing potential differences between cable systems and power systems,
but because of the lengths of the grounding conductors Involved, this affect is usually second-
ary. In addition, the EMP induces gradients directly in the soil, so that the counterpoise
does not reduce soil gradients In the same sense that it does for lightning surges; it merely
dlstorts the gradients induced by the EMP. Finally, because the attenuation of currents con-
ducted by buried conductora is very large at high frequencies, only a few meters of the
counterpoise may be effective in grounding the system in the high-frequency part of the
EMP spectrum. The attenuation constant o is

294
 -- -- (0 Wt ) (66.12)
2

where a is the soil conductivity, = 41 X 10-1, and (; is the permittivity of the soil, Fol
avE~rage soil ((; = 10-2 mho/m, (: = 8.85 X 10-11 F/m), the attenuation constant is 0.14 m- 1
1
at 1 MHz (v>>dc)and 0.60 m- above 18 MHz (o, ), Since 1/(Y is the distance in
which the current is attenuated by e-1, this distance is only 7 m at 1 MHz and about 1.7 m
at frequencies where the soil b',huves as a lossy dielectric.

Because of this large attenuation, the input impedance of a single cable or well casing
used as a counterpoise s essentially the characteristic impedance of the buried conductor,
For a horizontal conductor a few feet deep, this impedance is

11 +j J26
log - ((I w(.)
uP (6-13)
i ' 21rob ^toa

Zo log ---- ( w•c)

.:'. (6-14)
2m; Y0 a
'j /vi,7 2

where 6 is the skin depth in the soil, ii is the radius of the conductor, o is the soil conduc-
tivity, and -ro - 1,7811.... For a horizontal conductor with the attachment point along the
run, the impedances will be smaller by 1,1since there are then two impedances in plarlth.l.
Fur average soil ((u- 10-2 niho/m, t = 8.85 X 10-11 F/m), the impedance of a 1 inch -dii
meter horizontal cable is 26 ohms at 1 MHz ((j "--> wc ) and 320 ohms at 100 MHz (f) ' W,
if the cable is long compared to 1/o (i.(., long compared to 7 m at 1MHz). The surge impe-
dance of the counterpoiso at high frequencies is therefore large, independent of length, and
relatively independont of conductor radius, Although the counterpoise impedance can be
reduced by using multiple conductors in a grid, the high-frequency impedance of the
counterpoise will still be tens of ohms in the vicinity of 100 MHz, so that 100-A surges will
produce kOovolt potentials at the ground point,

295
 The difference in the behavior of the counterpoise at hi qh frequencies is illustrated
in Figure 6-13. intriure 6 13 (u), the dimensions of the counterpoine urn smallr than a
skin-depth in the soil, so tlhit the entire ring is eilective in carrying uwidI dissipatinog grolind
currents, Its surge impedance ;- low, and its "region of influence," taken as the area within
one skin depth of the counterpoise conductors, is large. At high frequencies, however, the
skin depth (or 1/A if u -X< •(e) may be small compared to the dimensions of the counter-
poise as illustrated in Figure 6-13(b). Then the surge impedance is large and the "region of
influence" is small, Furthermore, only that part of the counterpoise within about a few
skin depths in the soil is effective in carrying ground current; the remainder of the ring has

REGION OF
INFLUENCE

HING COUNTERPOISE

,
GROUND
POINT
I

to) LOW FREQUENCY

RHING
/ COUNTERPOISE

REC3ION OF
INFLUENCE

SGROUND POINT

Wb) HIGH FREQUENCY

Figure 6-13 BEHAVIOR OF A RING COUNTERPOISE AT LOW AND HIGH

FREQUENCIES

296
virtually no effect on either the surge Impedance of the counterpoise or the ground-current
distribution in the soil. Thus if two or more attachment points separated by several skin
depths are used, each mcy behave as essentially independent ground points at high frequencies.

6.5 CITED REFERENCES

1. Electrical Transmnission and DistributionReference Book (Westinghouse

Electric Corporation, East Pittsburgh, Pennsylvania, 1964)
2. A, E. Knowlton, Editor, Standard Handbook for Electrical Engineers, (McGraw-
Hill Book Co., Inc., New York, N.Y., 1949),
3. Notional ElectricalCode 1971 (National Fire Protection Association, Boston,
Massachusetts, 1971).
4. R, Morrison, Groundingand Shielding Techniques In Instrumentation (Joh n
Wiley & Sons, Inc., New York, N.Y,, 1967).
5. R, Rudenberg, "Grounding Principles and Practice," Electrical Engineering,
Vol. 64, pp. 1-13 (January 1945).

297
Chapter Seven

POWER-SYSTEM PRACTICES
FOR EMP PROTECTION

7.1 INTRODUCTION

The protection of electric power systems from the effects of the EMP requires U
determination of the EMP .induced transients in the system and an undurstanding of the tol-
erances of components of the system for these transients. At present, the transients induced
by the EMP are better understoad than are the tolerances of the power system to these tran-
sients - particularly the tolerances of the generation, transmisk, .n, and distribution systems.
The tolerances of the distribution transformer, insulators, and lightning arresters, as presently
understood, are discussed in Chapters Two and Four. The effect of the EMP-induced tran-
sients on supervisory-control systems, relaying equipment, and turbine control systems conl-
taining solid-state devices, or on other equipment in the generation and transmission portions
of the power system is unknown, These components are apparently sufficiently tolerant to
withstand normal switching and lightning transients (although early versions of some equip-
ment containing silicon- controlled rectifiers were apparently susceptahle to switching tran-
sientsl. However, one it inherently suspicious of eqtoipment L.sing solid- state electronics

298
unless the electronic circuits are heavily protected from the direct EMP fields and from
induced currents conducted into the electronic circuits. 1 ' 2

The behavior of the consumer's end of the power system is considerably better
understood because this portion of the system has been analyzed and tested in connection
with several facility hardening and evaluation programs. The object of these programs has
always been to protect the consumer facility from the effects of EMP on the power system,
however, rather than to protect the power system from the EMP. In most such assessments,
the commercial power system is considered expendable In the event of nuclear warfare; the
primary concern Isthat EMP transients conducted into the facility on the power conductors
do not cause vital equipment in the facility to malfunction. Such a hardening philosophy is
the only prudent one to profess for critical communication and weapon systems that must
function during or immediately after a nuclear engagement. For surface or air bursts of nu-
clear weapons, the blast and thermal damage will undoubtedly incapacitate power transmis-
sion and distribution equipment within a few kilometers of the burst,3

Weapons detonated at high altitudes do not produce significant blast and thermal
effects at the surface, however. Their principal effect at the surface will be Ohe EMP, which
will be experienced over a large area (hundreds of kilometers) under the burst. Thus, while
it will remain prudent for the designer of critical facilities to consider the commercial power
system expendable, it may also be prudent to consider making the power system invulnerable
to the EMP from a high-altitude weapon detonation - particularly If such hardening can be
achieved in the course of providing protection from lightning and switching transients.

As has already been stated, there are insufficient date on the behavior of generation
and transmission equipment to prescribe specific protection measures. For this reason, the
protective measures described in this chapter are separated Into two categories: measures
that protect the consumer from transients conducted Into his facility on the power lines,
and some rather general measures for protecting the electric utility system from the effects
of the EMP.

'Studies are curt antly in progress under the auspices of the Defense Civil Preparedness Agency. Date from
these studies may help to fill the presont void,

299
7.2 PROTECTION FROM THE POWER SYSTEM

7.2.1 EMP PROTECTION OF THE CONSUMER

From the consumer's EMP-protection viewpo,nt, the power system is a large collec-
tor of the EMP that penetrates his facility. The objective of consumer EMP-protection is,
therefore, to eliminate the EMP-induced transients on the power conductors before they
reach sensitive components In the facility, Inasmuch as the open-circuit voltage induced on
distribution lines may be of the order of megavolts, while some electronic circuits may be
upset or damaged by volts, 1,, clear that some kind of protective measures will almost al-
ways be required for an electronic system that is to survive and operate immediately after
the EMP,

As is evident from Chapters Two, Three, and Fuur, some of this protection will come
from the power system itself, If the distribution transformers or potheads are protected with
lightning arresters, the voltage delivered by the secondary of the transformer will be reduced
to tens of kilovolts, Mismatches at the service entry also reduce the voltage that enters the
low-voltage wiring, and the stray Inductance of wiring in metering boxes and circuit -breaker
panels slows the rise time of the transient, Thus, for facilities served by their own lightning-
protected distribution transformers, the transient delivered to the main circuit breaker may
have a rise time of 30 to 100 ns and a peak voltage of 10 to 50 kV. The task for the facility
designer is then to suppress this remaining EMP-Induced transient and provide a standby
power source to carry the load If the commercial power sourco fails,

7.2.2 VOLTAGE LIMITERS AND FILTERS

The residual 10 to 50 kV of EMP-induced signal can be adequately suppressed

with a combination of voltage limiters and filters. Secondary lightning arresters that fire at
voltages of 1 to 5 kV are available commercially for use on low-voltage circuits, The second-
ary arresters contain spark gaps and nonlinear resistancet similar to the distribution type
arresters, but they fire and extinguish at lower voltages. The arresters serve two purposes:

300
they prevent insulation breakdown in the low-voltage circuits (including the power-line fil-
ters) and they limit the low-frequency (late-time) content of the transient propagating be-
yond the arresters. Line filters suitable for use on circuits carrying up to 100 kVA are also
available commercially, These filters are usually low-pass ir-sections containing a series in-
ductance with shunt capacitors across the input and output terminals. These filters usually
have some specified minimum attenuation (e.g., 100 dB) at frequencies above a corner fre-
quency such as 100 kHz so that they suppress the high frequencies passed by the lIght-
ning arresters.

The placement of the secondary arresters and line filters will be determined to a
large extent by the nature of the facility, Two possible applications are suggested In Figure
7-1. In Figure 7-1 (a), a portion of the raw power Is used for station-keeping equipment that
Is either expendable or Insensitive to transients. Only the portion that is essential for sensi-
tive and essential equipment is filtered, but all of the low-voltage circuits are protected by
the voltage-limiting action of the secondary lightning arresters Installed on the incoming con-
ductors at the main circuit-breaker panel, since Insulation damage must be provided even
for the parts of the system that are not susceptible to transients. In Figure 7-1 (b), all the
Incoming power is filtered upon entering the facility, A standby generator to provide power
In the event of commercial power failure Is also shown in Figure 1-7(b). Some or all of the
external equipment shown In Figure 7-1 (b) may be enclosed in the entry vault along with
the line filters In some installations.

7,2.3 SELECTION AND INSTALLATION OF SURGE ARRESTERS

The secondary lightning arresters used for limiting transient voltage on the low-
voltage wiring should be reliable, maintenance-free units designed for use on 60-Hz power
circuits. Such components are commercially available and relatively Inexpensive, They are
designed to fire at voltages 5 to 10 times the peak 60-Hz voltage and will extinguish with
tho rated 60-Hz voltage applied. The latter property Is Important; surge arresters that con-
duct appreciable 60-Hz follow-on current tend to be short-lived because of the excessive
electrode erosion and deformation caused by the follow-on current (failure then often occurs
as a destructive short-circuit). In addition, the transient caused by the follow-on current

301

L NSA
 ST ATION-KEPING
CIRCUITS
SECONDARY
LIGHNIN
ARRESTERS

ENTRY
ENTRANCE VAL

POWER TO
T ENS ITI VE
_.j MI :.L EQIPME NT

I FILTER
0 AND NI

UNSIELOID01 AMIA st4liftL0* ARIA

(al SECONDARY ARRESTERS ON ALL POWER. SELECTED CIRCUITS FILTERED

ENTRACE TODISTRBUTIO

TENRATOFR

ENTRANUCED TRNOT ONISW-VOTABUCTRUIT

PANE
flowing through surge arresters may be almost as objectionable as the original surge that
triggered the arrester. The secondary lightning arresters should be essentially trouble-free
for the life of the installation (unless a direct lightning strike to the low-voltage system
i. incurred).

The minimum firing voltage of secondary lightning arresters Is determined by the 60-
Hz voltage of the circuit being protected. Thus, for a 120-volt circuit the firing voltage of
the secondary lightning arrester is usually at least 1000 V. The time-to-fire and the rate of
rise of the current through the arrester are somewhat more variable In that they depend on
the construction of the surge arrester and the wiring through which It Is connected to the
power conductor and ground, Adequate protection is usually obtained If the tIme-to-fire Is
less than 20 ns at twice the rated firing voltage of the arrester and If the effective time con-
stant for the current buildup is 50 ns or less,

The time-to-fire is a property of the lightning arrester that can be determined exper-
Imentally by applying a fast-rising voltage step across the arrester and observing the time lag
before conduction begins. However, the time constant for current buildup depends mainly
on external-circuit properties, although the Internal construction of the lightning arrester
may also be a factor. For example, if the lightning arrester has an effective Internal Induc-
tance of 60 nH and Is connected to a long entrance conductor, whose characteristic Impe-
dance is 20 ohms, through 30 Inches of lead (with 10 nH/Inch), the current rise time constant
is determined by the 360 nH of inductance and the 20-ohm source:

L. .. 360 nH
. .• 1l"".ns ,(7-1)
R 20 il

Only the 60-nH internal inductance is attributahle to the lightning arrester; the controlling
resistance and external inductnce are properties of the external circuits. hi is evident, there-
fore, that the method of installing a secondary lightning arrester has a strong influence on
Its ability to dissipate short transients of current, A typical installation of secondary light-
ning arresters at the main circuit breaker panel is Illustrated in Figure 7-2.

303
81ICONVARY
LIGHTNING MINIMIZE
ARRESTERS LEAD
LENGTH

MAIN CIRCUIT
BREAKER OR
SERVICE
SVVTCH

LOAD

Figure 7-2 ILLUSTRATION OF SECONDARY LIGHTNING ARRESTERS AT THE MAIN

CIRCUIT-BREAKER CABINET

304
 The following guidelines are provided for the installation of secondary lightning
arresters for EMP protection:

* Use short leads. The leads that connect the lightning arrester to the power con-
ductor and to the ground should be as short as possible to minimize the external
inductance in series with the arrester. This is particularly important for low
source-Impedance installations such as the main bus shown in Figure 7-1 (a) where
the lightning arrester is fed by many conduit conductors in parallel,

0 Use external grotind, The ground side of the lightning arrester should he connected
to a ground point ouki;Je the shielded area (preferably to the circuit-breaker Vab-
Inet or service-entrance conduit), The surge-arrester discharge current should
never be allowed to flow directly into the internal ground point used for sensitive
electronics equipment inside the shielded ý.rea.

a Install arresters on all phase conductors. Arresters should be installed between

each phase conductur and local ground, If the neutral c" single-phase common
are grounded (to the conduit or cabinet) In the main circuit-breaker cabinet, they
will not need secondary lightning arresters. If the neutral or common ground-
point is quite remote from the main circuit-breaker panel, or if the ground con-
ductor is directly exposed to the EMP, secondary lightning arresters should be
installed between these conductors and the cabinet or coalduit in the main circuit-
breaker cabinet,

0 Follow lightning arresters with filters. Power supplied to sensitive electronic sys-
tems should always be filtered between the lightning arresters and the sensitive
equipment, The lightning arrester only limits the transient voltage; it does not
eliminate transients on the power conductors (sometimes the rates of change of
the voltage or current are actually increased by the lightning arrester), Low-pass
filters (either in the power system or in the equipment) are requiied to protect the
equipment from the transient passed by the lightning arresters.

305
 The discharge-current requirements for the secondary lightning arresters are also
determined by the external circuit. For times less than the round-trip propagation time on
the service-entrance conduit or cable, the current through the surge arrester is limited by
the characteristic impedance of the conductor, Thus, for example, a 20-kV (open-circuit)
transient entering on a 20-ohm conductor can deliver only 1000 A to a short-circuit until the
discharge-current wave has propagated to the opposite end of the conductor and back to the
discharge point, At later times, the discharge current will depend on the source impedance
at the opposite end of the entrance conductor, This impedance Is often tens or hundreds of
ohms on phase conductors, however, so that peak surge-arrester currents seldom exceed the
value obtained by dividing the open-circuit voltage by the characteristic impedance of the
entrance conductor (see Chapter Three for more exact techniques for determining the short-
circuit current at the end of the service entrance),

Since the duration of the high-level transient passed by the distribution transformers
is usually less than 1 jus, the total charge transferred through the surge arrester by an EMP-
induced transient with a peak open-circuit voltage of tens of kilovolts is only tens of milli.-
coulombs, The current and charge transfer ratings required of secondary lightning arresters
for EMP protection are thus lower than those required for lightning protection, for which
longer pulses of similar magnitudes are usually specifled.

7.2.4 SELECTION AND INSTALLATION OF LINE FILTERS

As has been discussed in the preceeding sections, secondary lightning arresters can be
used to limit the peak transient voltage to a few kilovolts so that insulration breakdown and(
flashover in the low-voltage wiring is limited. Transient voltages of a lew kiluvolts are per
mitted to propagate along the conductors beyond the lightning arresterts, however, and fast
acting lightning urresters can cause fast-changinc, voltage and current transients in these cir.
cults. The purpose of power-line filters is to remove these fast-changinq components of
the transient that remains after the lightning arresters have acted. Commprcially availhihle
line filters are typically low-pass v-section filters with a Inrqo insertion loss at frequencies

306
well above the 60-Hz or 400-Hz power frequency, but fairly low-loss at the power frequen-
cies. These filters can thus greatly suppress the large rates of change in current and voltage
conducted past the lightning arresters.

Power-line filters should, like the secondary lightning arresters, be reliable and main-
tenance-free for the life of the Installation except in extenuating circumstances such as a
direct lightning strike to the low-voltage conductors or a short-circuit In the filtered circuit
that Is not cleared by circuit breakers or fuses. Most filters will tolerate sufficient temporary
overcurrents that faults on properly protected circuits will not damage the filters. Thus, for
example, a 100-A line filter on a circuit protected by a 100-A fuse or circuit breaker should
not be damaged by the fault current required to blow the fuse or open the circuit breaker.

The primary factors, other than line voltage and current ratings, that Influence the
choice of line filters for EMP protection are the high-frequency attenuation and the Insula-
tion breakdown or flashover characteristics of the input terminals. The filter must be able
to tolerate the peak voltage, rate-of-change of voltage, and rate-of-change of current passed
by the secondary lightning arresters, i-section filters with shunt capacitors can usually tol-
erate large transients of short duration since the input capacitance of the filter can absorb
the transient charges without large voltage changes, Their ability to do so, however, depends
on the quality and size of the Input capacitor. Very large capacitors that could, in principle,
absorb large currents without big voltage changes, may, In fact, behave as Inductors at high
frequencies. Therefore, for EMP applications in which large amplitude, short-duration volt-
age transients may be delivered to the filter, the filter Input characteristics must be care-
fully examined.

It is recommended that power-line filters for facility power applications such as

those illustrated in Figure 7-1 be tested with a short, high-voltage Impulse to determine
that the input capacitance and Insulation will withstand the fast, high-voltage transients
passud by the secondary lightning arresters. A suggested test circuit and Impulse waveform
are shown in Figures 7-3(a) and (b) for filters expected to operate on 60-Hz voltages of
480 V or less. Note that the voltage waveform In Figure 7-3(b) is the open-circuit voltage
behind the 25-ohm source impedance. Thus if the Input capacitance of the filter is 1 JF,
the test pulse will increase the filter terminal voltage to only about 100 V. If the Input
capacitor behaves as in inductance at early times, however, the terminal voltage may rise to
several kV and ptrmlt the terminals to flash over or cause the capacitor dielectric to break-

307
down. Because a dielectric breakdown in the input capacitor would be a catastrophic fail-
ure of the filter, it is recommended that the filter be tested with a fused 60-Hz source of
rated voltage applied to the output terminals so that insulation failure resulting in 60-Hz
follow-on current can be detected, A suggested pulser for generating the test pulse Is shown
in Figure 7-3(c).

The high-frequency attenuation of power-line filters is specified In terms of the

Insertion loss when the filter is placed in a 50-,ohm circuit. The tests for measuring the Inser-
tion loss prescribed by MI L-STD-220-A are widely used In the manufacturing industry for
specifying and evaluating line filters. The rudiments of this test are Illustrated In Figure 7-4,
where it is seen that the current through a 50-ohm load on a 50-ohm source is measured
before and after the filter is inserted between the source and the load. The ratio of the unfil-
tered load current to the filtered load current (usually uxpressed in decibels) Is the insertion
loss of the filter, Insertion losses of 80 to 100 dB at frequencies above 100 kHz are common
for commercially available line filters, It is widely recognized that the insertion loss measured
in a 50-ohm circuit is not readily applicable to any other circuit Impedance levels unless
additional information on the filter is available, For high-current, 60-Hz, 7r-section power-
line filters, however, the series inductance must be small (millihenrles or less), to minimize
the insertion loss at the power frequency, which Implies that the shunt capacitors must be
large (- 1 yF). In addition, many of the conduit-and-conductor transmission lines used In
electrical wiring have characteristic impedances within a tactor of 2 of 50 ohms, so that for
many of the fast transients associated with the EMP, the 50-ohm Impedance level Is an
appropriate mean volue. Thus, for power-line filters, one can infer some of the properties
of the fimter, and it can be deduced that the attenuation with the 50-ohm impedance level is
a reasonable approximation uf the attenuation to be expected for the very rapidly varying
components of the EMP.-induced transients.

ror EMP-protection of a shielded facility, filters should be Installed on the neutral

(or common) as well as on all phase conductors entering the shielded area, The ground sides
of the filters should be connected to the outer surface of the shield - preferably by bolting
the filter case directly to the shield. Only the output terminal of the filtr should be per-
mitted to enter the shielded area, A common practice is to install the filters in a closed metal
vault attached to the shield as illustrated in Figure 7-5 so that exposed wiring and filter In-
put terminals are shielded from the external environment.

308
V(li .1ITR
I TO 60 H
RATED
VOLTAGE

INPUT OUTPUT
TERMINALS TERMINALS
Is) TEST CIRCUIT

10 kV

5 kV

(b) TEST WAVEFORM

as ft-

SPARK-GAP
5M41 SWITCHES

RGOB/U

20 kV s0of
dcT
- 0

Il SUOGGESTED PULSE SOURCE

Figure 7-3 IMPULSE TEST FOR EVALUATING POWER-LINE FILTERS FOR EMP
APPLICATIONS

309
 INSERTION LOSS - 20 logo1 -- dB

50 .50,

11 12
V0o V0 FILTER W0 1

Figure 7-4 DETERMINATIOI4 OF THE INSERTION LOSS OF A LINE FILTER

7.2.5 AUXILIARY POWER SYSTEMS

7.2.5.1 Function of Auxiliary Power Systems

Installations requiring continuous and reliable power are usually operated

from commercial power and provided with standby sources to provide power in the event of
commercial power outage. Such systems are used in critical facilities such as hospitals, com-
munication centers (civil and military), strategic military systems, power stations, and computer
centers, and In similar facilities where power failure would cause loss of life, critical operat-
ing capability, or essential data. In many installations, even momentary loss of power or
transients in the power system are intolerable. For example, computer centers and electron-
ically controlled communication centeirs may lose stored data during a power outage or be
upset by transients. Therefore the auxiliary power systems are usually designed to protect
the facility from interference (lightning and switching transients) conducted in on the power
lines as well as to provide stable and continuous electric power,

Auxiliary power systems for these applications usually consist of a combina-

tion of storage batteries and one or more engine-generators. The storage batteries are kept fully
charged from the commercial power, so that when an ojtage occurs they can supply the
facility operating power for a brief period (from a few minutes to a few hours). If the com-
mercial power outage lasts longer then a few minutes, the engine-generator is started (usually

310
automatically) and the facility load is transferred from the commercial power service to the
engine-generator, The enrgine-generators can supply power for days, or weeks, if necessary,

A schematic diagram of an "uninterruptible power system" is shown in Figure

7-6. The 60-Hz commercial power may be used directly for certain station-keeping functions
surh as outdoor lighting, and operating pumps, air conditioners, etc. The "uninterruptible"
rectified power may also be used directly as the dc source for operating electronic equip-
ment. "Clean" 60 Hz power for sensitive equipment is obtained from a motor-generator
driven by the de supply, In the event of commercial power failure, the storage batteries con-
"tinueto supply dc power and drive the motor-generator to provide 60-Hz power while the
engine generator is being started and brought up to speed, A properly designed power system
of this type can provide a higl, degree of immunity to power outages and line transients for
the system operated from the dc or "clean" 60-Hz power,

7,2.5.2 The Automatic Transfer Switch

A key component 6f an auxiliary power system is the automatic transfer

switch, The function of an automatic transfer switch is to transfer a load from the normal
(or preferred) power source to an emergency supply if normal voltage fails or is substantially
reduced. Once thle normal source is again in proper operating condition, the transfer switch
should automatically restore the load to its original feeder lines, regardless of the condition of
the :mergency supply. A scheematic of the automatic transfer switch is shown in Figure 7-7.
This model is designed to provide full protection in 60-Hz, three-phaso, 3-wire or three-.
phase, 4-wire solid neutral power services.

A properly desijned transfer switch should be rugged, dependable, and

rated for coltinuJous duty, since failure of the unit might well create the hazard it is intended
to eliminate - complete service outage. For similar reasons, it is des;rahle that the switch
operate from the powpr source that is to be connected. Also, failure of a relay coil or control
circuit in the transfer switch should riot leave both normal and emergency contacts in the
open position,

It is also necessary to provide suitable time lags in the switch's operation to

allow for generator starting and warm up, and for the return of stable normal power or the

311
 SHIELD WALL

IJNSHIELDED AREA SHIELDED AREA

FILTERED

POWER

ENTRY
VAULT
FILTERS

POWER

lnmno
grounded
to wall; phase
and neutral
mop conductors
filtered)

Figure 7-5 ILLUSTRATION OF VOWER-LINE FILTERS IN AN ENTRY VAULT

ON WALL OF SHIELDED STRUCTURE

shut-down of an emergency power source after a return to normal puwer. Time delays
should be used to prevent unnecessary transfers due to momentary dips or transient outages,
Switching, once Initiated, should be fast (less than 10 cycles) to reduce contact burning and

312
coMM,,CI,...___-______I®60 Ht

POWER POWER

ENGINE AUTOMATICj 6HR

O5EAT~ L .
SWITCH .CLEAN
~_ - "..
CLEAN

Figure 7-8 SCHEMATIC OF A SIMPLIFIED "UNINTERRk,'OTIBLE POWER SYSTEM" (UPS)

pitting during make and break operations. Other useful attributes of an automatic transfer switch
include minimum maintenance, quiet normal operation, and minimum spice requirements,

Because the automatic transfer switch must be located in the main or second-
ary branch circuits, the switch and its component parts may be subjected to the full magnitude
of any EMP-Induced voltages and currents coupled through the service transformer bank and
service-entrance condult to the main circuit-breaker panel. Therefore, the transfer switch
must be resistant to the EMP-induced transients coming In on the power conductors. Trans-
fer switches that contain only electromechanical relays for sensing and logic functions are
usually relatively immune to EMP-Induced transients that do not exceed the Insulation
strength of the switch compnnents. However, there is a tendency to use solid-state elec-
tronic circuits for sensing and for providing time delays in modern transfer switches, Such
components are much more susceptible to failure induced by the EMP transients, Therefore
the transfer switch itself must be protected from transients on the commercial power system
by voltage limiting and filtering if the switch is to function during or after an EMP environment.

313
 N8 INA NORMAL N
0V2 TRi TEST SWITCH

UVI uv2

L- IF - - - - -- - - - - - - - J

TBE TE T

Too

LBO D S LOAD

SE TRC

LA +A C2PONC c

SHOWN CLOSAD

3.eT TSNIATSOE;!PIGCOTCS
4.TO IDCAESTMEDEAYO COSN; OOINIATS II ELY NOPNIO

INTS. TIM CHRCeRSI ~ TIHSLAG.

LO A . S NITV
RE
an O V)IT G N F E U N Y
7.CLSEDIFRETIL
EAYIRl CID O ALPHSE PIVILDBYTE E ETY

3143
7.3 PROTECTION OF THE POWER SYSTEM

As was indicated in the beginning of this chapter, the tolerance of the power system
to the effects of the EMP are largely unknown. It it suspected that transmission and distri-
bution lines that are adequately protected against lightning will be relatively immune to the
effects of the EMP. 4 , s Similar Immunity is suspected in the case of transformers and elec-
tromechanical switch gear installed in accordance with modern lightning-protection practice,
The principal uncertainty In these components is the ability of the solid dielectric to with-
stand the large EMP-Induced voltages that might be developed before the lightning arresters
fire. Insufficient testing has been conducted to provide assurance that the solid insulation is
adequately protected by conventional lightning arresters, although the results to date suggest
that conventional protection may be adequate. For the present, therefore, the primary EMP-
protection policy for the transmission and distribution lines, transformers, and switchgear is
an effective lightning-protection policy, 6 , I For utilities in areas of high lightning incidence,
lightning protection is probably already practiced; utilities In areas of low lightning incidence
may need to upgrade their lightning-protection systems to reduce vulnerability to the EMP.

Because very little is known about the vulnerability of many crucial partsof the power
system, however, only the most general recommendations can be made. Modern supervisory con-
trol systems, turbine control systems, and relaying equipment, for example, use solid-state
electronic components in varying degrees. It is strongly suspected that some or all of this
equipment may be susceptible to damage from the EMP-induced transients, but no data on
the nature or levels of susceptibility are available.

The utilities can, without expensive EMP testing, make a crude assessment of the vul-
nerability of their equipment by observing the types of equipment that fail and the nature
of the failures in an environment of transient activity such as that accompanyin~g thunder-
storm activity or fault-clearing operations, Components that malfunction or are damaged
in these environments may also be susceptible to the EMP transients. An occasional mal-
function under severe lightning or fault conditions may be tolerable in one part of the system,
but in the EMP environment, the entire utility system and most of its neighboring utilities
may be exposed to the high level fast transient excitation. Thus, if a component is vulnerable,

315
all such components in the power network may malfunction simultaneously to cause a dis-
astrous disruption of the system, Therefore in assessing the vulnerability of the system to
the EMP, even rare malfunctions associated with severe transient activity may be significant.

Based on prosent understanding of the characteristics of high-voltage components

and fast-pulse technology, it is assumed that the high-voltage components of the transmis-
slon and distribution systems are probably the least vulnerable to the transients induced by
the EMP - particularly if good lightning-protection practice is employed, Low-voltage con-
trol circuits - particularly those with sensors (current or potential transformers) on trans-
mission leads and those containing solid-state electronics - appear to be the most vulnerable
to the EMP. Low-voltage circuits connected to instrument transformers are particularly
suspect because the fast rise.time of the EMP-Induced transient will be coupled through
these transformers much more efficiently than the slower lightning and switching transients,

Improvements in the EMP protection of these low-voltage circuits can be achieved by

following the principles used to protect sensitive equipment from transients conducted In on
power leads,1, 2, 8 For systems already in existence, the basic hardaning techniques are:

(1) Making maximum use of existing shielding afforded by cabinets and housings
through operating and maintenance procedures,

(2) Installing voltage-limiting surge arresters on conductors that penetrate these

shields.

(3) Installing low-pass line filters on low-voltaqe conductors penetrating these

shields,

(4) Modifying operating characteristics of logic systems so that lockouts or lockups

cannot be triggered by a single, wide-spread event,

For new systems and components, considerable EMP immunity can be incorporated
in the design by taking precautions to ensure that housings, cabinets, etc, are designed for
shielding integrity as well as mechanical protection, etc, In addition, surge filters and surge
limiters can often be incorporated into new designs at little additional cost. Power-system
components have the advantage, from the EMP-hardening point of view, that they must be

316
Inherently immuno to lightning and switching transients, Hence, the odditional protection
required to reduce their vulnerability to the EMP-induced transients may be quite minimal.

7.4 CITED REFERENCES

1. "DNA EMP Awareness Course," DNA2772T Notes prepared by lIT Research

Institute, Contract OASA 01-69-C-0095, Defense Nuclear Agency, Washington,
D. C, (August 1971).
2. "EMP Protection For Emergency Operating Centers," Protection Engineering
Management Note 8, Lawrence Livermore Laboratory, Livermort Cal:?.
(Originally issued by the Defense Civil Preparedness Agency as T R-CIA,
July 1972),

3, S. Glasstone (Ed,) The Effects of Nuclear Weapoons, U. S. Government Printing

Office, Washington, D, C. (April 1962),
4. "EMP and Electric Power Systems," TR61-D, prepared by Oak Ridge National
Laboratory for the Defense Civil Preparedness Agency, Washington, D, C,
(July 1973),

6, G, E, Barker and W, 8, Hull, "Characteristics of Commercial Power Systems,"

AFWL-TR-74-27, Contract F29601 -89-C-0127, Air Force Weapons Laboratory,
Kirtland Air Force Base, New Mexico (Septembur 1973).
6. Electrical Trainsmission ond Distribution• Book, 4th Ed, (5th printing)
(Westinghouse Electric Corporation, East Pittsburgh, Penn,, 1964).
7. W, W. Lewis, The Protectiottof rransmission Systems Aqainst L~qhtnitkq
(John Wiley & Sons, Inc,, New York, N, Y., 1950),
8. DASA EMP (Electromaqnetic Pulse) Handhook, DASA 2114-1 (DASA
Information and Analysis Center, Santa Barbara, Cal;f,, September 1968),

317
Chapter Eight

TESTS OF COMPONENTS
AND FACILITIES

8.1 INTRODUCTION

The tests of power-system components and consumer facility and consumer com-
ponent sensitivity to EMP-induced transients on the power system can be divided into three
categories:

(1) Standard tests - established by the IEEE and the American Standards Asso-
ciation and designed to determine transmission-system component insulaton
levels and tolerance to lightning and switching transients.

(2) Component tests - with simulated EMP to determine the tolerance and trans-
fer characteristics of power system and consumer components or Items of
equipment,

318
 (3) Facility tests - with simulated EMP transients injected at the service entrance
to evaluate the tolerance of the facility wiring and components for the EMP-
induced transients on the distribution system and to determine the excitation
of internal equipment.

A fourth category of test could be postulated to determine the response of the power
transmission and distribution network to the EMP. At present, however, no practical method
of performing such a test has been proposed. Until such a method is developed, therefore,
power network response must be assessed by a combination of direct-injection tests at the
"nodes" (consumer facilities, switching centers, substations, etc.) and network stability

analysis. It is quite likely that network assessment will never progress beyond this state
because of the very serious technical and political problems associated with wide-area
illumination of a power transmission network, In any event, much work remains to be done
at the node end network analysis level before the implications of EMP Interaction with
power networks can be evaluated,

Tests of equipment and facilities are usually required to (1) determine the coupling
between the incident EMP and a component or subsystem, (2) evaluate the sensitivity of
the equipment to the coupled signal (iLe., threshold for damage or upset), and (3) assess the
effectiveness of designs for reducing vulnerability to the EMP. The complexity of EMP tests
varies from fairly simple laboratory tests of components to very complex tests of entire
facilities, The nature of the test requirements is also Influenced by the state of evolution of
the system, Existing facilities that are to be upgraded usually require all forms of Lesting
listed above, while new systems designed to be immune to the EMP may require primarily
the design-elfectivenuss tusts,

Experimental determination of the EMP tolerance and throughput is the only reliable
means of obtaining these characteristics of most electrical systems because the broadband
and nonlinear pruperties of most components are not well known, To illustrate this,
consider a typical electric appliance designed to operate on 120 V, single-phase, 60 Hz,
Its functional characteristics at 120 V, 60 Hz are usually well understood, but what can be
said of its electrical properties at frequencies between 100 kHz and 50 MHz, and what will
happen to the applicance if it is subjected to a 5-kV common-mode voltage transient on
the power conductors? The nameplate data offer little help in answering these questions,

319
and often even the design data are of little value in evaluating the high-frequency and high-
voltage characteristics of the appliance.

Because many items of equipment that are designed to operate from low-voltage,
60-Hz power are not designed to have specific high-voltage or high-frequency characteristics,
these characteristics may alio vary considerably among supposedly interchangeable Items.
This is attributable to the fact that manufacturing techniques and tolerances that affect the
high-voltage and high-frequency characteristics are not controlled unless they also affect
the 60-Hz performance. Because of this variability from unit to unit, it is important that
the basic coupling mechanisms (or malfunction mechanisms) for each piece of equipment
be understood, and that more than one item be tested, If possible, to ascertain that the
same mechanism prevails in each case,

8.2. STANDARD INSULATION TESTS

Th( urpose of standard Insulation testing Is to demonstrate that the design, work-
manship, a, materials of electrical equipment are adequate, Such testing originated from
the need to design and certify transmission and distribution system components capable
of withstanding the transients associated with lightning and line switching. These tests are
not nuclear EMP tests; they are quality-control tests esttablished by the industry. They are
described briefly here to illustrate the insulation standards used in the design of transformers,
bushings, insulators, etc.

The Insulation tests conducted by the power-component testing laboratories are of

three types: (1) full-wave impulse tests with a pulse having a rise time of typically 1.5 Ps
and a decay time (to half the peak value) of typically 40 ps, (2) chopped-wave impulse
tests in which the test waveform Is truncated (chopped) after a few microseconds by
shorting the source through a spark-gap switch, and (3) a low-frequency test at twice the
rated voltage and at about twice the rated frequency for up to 1 minute, 1 These tests are
not routinely conducted as a part of the manufacturing quality control; rather they are
conducted as a part of the design qualification procedure, on samples of a new or altAred
component design. Thus, although a component ha4 been designed and manufactured for

320
a basic insulation level, each individual component is not necessarily subjected to the tests
specified for that basic insulation level.

The full-wave impulse test is conducted with a high-voltage pulse applied across the
initilation. The pulse shape is basically an exponential pulse with a finite rise time such as
that illustrated in Figure 8-1. The pulse is specified in terms of its crest value V, the time to
crest T, and the time to half-crest T,,,,, The method of determining the time to crest and
the time to half-crest Is illustrated in Figure 8-1, The commonly-used impulse for insu-
lation testing is the 1.5 X 40-ps pulse, which means Tr - 1.5 .s and T/, - 40 As. The crest
value V is a function of insulation class (voltage rating) of the equipment. Crest voltages
for full-wave impulse tests are given in Tables 4-1 and 2-2 for most transmission adrd dis-
tribution voltages.

For chopped-wave tests, the waveform of Figure 8-1 Is foreshortened hy shorting

the pulse source out after a predetermined time with a spark-gap switch. Tests with the
chopped wave are conducted with slightly larger crest voltages, as can be seen in Table 4-1.
However, the duration of the pulse is usually 3 ps or less.

The low-frequency tests of transfortrers and similar equipment are conducted at

twice the rated frequency and at least twice the rated voltage of the equipment. At twice
the rated frequency, the copper and iron losses are lower, so that the component can
withstand the higher voltages without overheating, This test provides an evaluation of
sustained ac overvoltage (test time is 1 minute or less) effects on the insulation between
turns and between the winding and case.

8.3 EMP TESTS OF EQUIPMENT

8.3,1 PURPOSE OF EQUIPMENT TESTS

EMP tests of power-system or consumer facility equipment are usually conducted

for one or both of the following purposes: (1) to establish the threshold for equipment
malfunctions or failure, and (2) to determine the transfer function, or EMP throughput,
2 3
of the equipment. ,

321
 v

0.9 V .. . .

0
TIME

Figure a-1 IMPULSE-TEST WAVEFORM SHOWING METHOD OF SPECIFYING

RISE TIME AND TIME-TO-HALF-PEAK

The threshnld for equipment malfunction is a measure ,nf the tolerance of the equip-.
ment for the EMP..induced transient, This tolerance for the transients induced through the
power system may be determined by injecting transients on the power leads supplying the
equipment, In general, however, there may be several other coupling modes to which the
equipment is sensitive, so the susceptibility to transients conducted on the power leads
should not be considered the only susceptibility of the equipment. As is illustrated In
Figure 8-2, the equipment may also have ground conductors and signal Input and output
conductors, and the equipment and Its interconnecting conductors may be exposed to the
incident EMP field or some fraction thereof, The equipment may be susceptible to the
fields and to EMP-induced transients conducted on any (or all) of the other conductors.
Thus although the emphasis of this handbook is on susceptibility to transients related to
the power system, it Is Important to recognize that these other paths also exist,

322
 POWER

SIGNAL~-HU~O SIGNAL
INPUT OUTPUT

CHASSIS

]ROUND

Figure 8-2 CONDUCiRJRS ENTERING OR LEAVING AN EQUIPMENT CABINET

The determination of the transfer function, or throughput, of certain types of equip-

ment is often necessary to determine the nature of the EMP-Induced transient passed
through the equipment to more sensitive components downstream, Chapters Two through
Five of this handbook describe techniques for determining the transfer functions for trans-
mission lines, service entrances, and transformers so that, given the Input fields or source
characteristics, the voltages or currents at the output (downstream) terminals can be
evaluated. Similar characteristics for other components of the power system or consumer
facility may be required to evaluate the transients deliverod to sensitive components, In
Figure 8-2, for example, it may be necessary to determine the transient delivered to the
input, output, and ground conductors by the EMP-Induced signal conducted into the equip-
ment on the power conductors.

323
 8.3.2 DIRECT-INJECTION TESTS

Tests of equipment responses to transients conducted on the power conductors

usually make use of some form of direct injection of the transient on the conductors, Some
of the important considerations in performing such a test are:

(1) What are the Thevenin or Norton source characteristics of the signal to be
injected?

(2) How are the other conductors (i.e., input, output, ground) terminated?

(3) Is it important that equipment be energized when it is tested?

(4) Is it necessary to establish and hold a particular state or operating mode during
the test?

(5) Is the equipment part of a system with feedback such that its output affects
its input?

(6) Are other environmental factors such as ambient pressure, temperature,

humidity, illumination, or combustible vapors important to the test results?

(7) Do transients conducted on other conductors or Induced by the EMP fields

affect the responses to the transients o:, ..hc power conductors?

Most of these questions relate to specific properties or functional characteristics of

the equipment to be tested end can be answered only when a particular item of equipment
and Its operating characteristics and environment are specified. This section will be confined
primarily to techniques for Injecting signals onto power conductors for the pUrpose of
evaluating the tolerance or transfer characteristics of the equipment, However, the source
characteristics of the excitation source chn be defined from tests suc'h as those described
in Section 8,4,2,

For the case where one end of a power circuit is accessible for injection of test
signals, a common-mode test voltage can be directly injected at the end through an imped-
ance matrix as illustrated in Figure 8-3. The impedance matrix may simulate the

324
impedances usually connected between individual conductors and between -theconductors
and the conduit. In the case of a very long conduit, the voltage delivered to the conductors
is:

Zin
Vn -- V (8-1)
Z + ZIn

where V Is the source voltage, Zin is the input impedance of the conductcrs with their
normal load on the right-hand side (see Figure 8-3), and Z is the common-mowe Inpedance
of the terminating resistors between the source and-the conductors. It should be noted that,
If the righthand end of the conduit circuit is also terminated In Its characteristic impedance,
only helf of the source voltage is applied to the conduit.

This method of driving power conductors Is perhaps the most straightforward and
commonly used of all the direct-Injection methods. It can also be used with unshielded
cables that are routed along a metal structure or are placed In metal cable trays. With
unshielded cables of this type, the conductors are driven against the metal structure or
trays rather than against the conduit, One disadvantage of this method'is that the conductors
being driven must be dlsconnnected at one end; hence, the equipment may not be operating
in its normal state during the test,

When one end of the circuit is not accessible, a different injection method must be
used. Such cases arise where disconnecting the power precludes operating the system in its
normal mode. In these cases, it may be necessary to accept some compromise In the quality
of the simulation to perform tests economically. One approach that can be used under
certain conditions is illustrated in Figure 8-4. At some suitable junction in the power system
the excitation source is capacitively coupled to the conductors and permitted to drive them
with respect to the local ground or conduit. As illustrated in the figure, however, the current
injected at this point is divided into two parts, one flowing in each direction from the Injec-
tion point. Because this method of distributing the current differs radically from the current
dist,'ibution that would have resulted from the EMP excitation of the system, some care is
required in designing a valid test using this approach.

A test usinq the injection method shown in Figure 8-4 will be valid only if the
equipment response is not affected by the attachment of the energy source. Because the

325
 TEIMINATI 40G
HFS1!TOHS CONDUIT
z

A00~
SOURCE

Figure 8-3 IMPEDANCE MATRIX USED TO INJECT COMMON-MODE VOLTAGES

equipment to the right of the injection point in Figure 8-4 is of primary interest In the test,
for the test to be valid the portion of the current that flows to the left from the injection
point must not be reflected back into the right-hand circuitry during the period when the
system is being observed to determine its response to the EMP. This condition implies that
no significant reflections should return from the left end of the circuit, Thus the circuit to
the left must be very long ( a round-trip transit time that Is longer than viewing time), or
very short ( a round-trip transit time that is shorter than any response of Interest), or It
must be terminated in a matched load (no reflections).

If the driven conductors cannot be made very long, the stipulation that attachment
of the excitation source should not significantly affect the system response usually Implies
that the coupling between the direct-injection source and the system conductors must be
so loose that the system impedances are not significantly affected, It is important to
observe that the loose-coupling requirement applies to the differential-mode Impedances
as well as to the common-mode, or line-to-ground, impedances. That is, the attachment
of the energy source should, in general, disturb neither the line-to-ground impedances nor
the line-to-line impedances of the power circuit at the injection point. The loose-coupling
requirement usually requires that much of the source voltage be dissipated in the coupling
network.

An alternative to the capacitive coupling method illustrated In Figure 8-4 is the

inductive coupling method illustrated In Figure 8-5. In this method, the power conductors
are made the one-turn secondary winding of a transformer by threading them through the
window of a toroidal ferrite core with the conductor carrying the source current.

326
 7AiIN IiC

F~gure
8-4 CAPACI TORSUEMOIC OMO-OEVLAE

8,4 EMPTESS O FACLITIES

8igure EXCPCITATONSOF THE

S NERVCE ENTRANCMOE VLAE

Tests at the facility luvel are also performed primarily to extablish thresholds for
malfunctions and to determine transfer functions from the power service (e~g., the dis-
tribution lines ur service entrance) to points of interest in the facility, For existiriq faicilities,
diagnostic testing may be performed to locate and rank malfunctions as well ais to l)roof-test

327
 -SHIELDED CABLE

TOROIDAL CORE

STRUCTURAL GROUND
Iboth enids

Figure 8-5 CURRENT TRANSFORMER TO INJECT CURRENT

ON A GROUNDED CABLE SHIELD

the facility after modifications have been made to increase its tolerance for the EMP-Induced

transients. New facilities designed to be immune to EMP must also be tested to evaluate the
success of the design.

In almost all cases in which EMP coupling into facilities is to be examined experi-
mentally, it is necessary to perform the test without disconnecting the power. There is
usually a significant change in the state of the circuits and equipment inside the facility
when the commercial power is disconnected, because of relays becoming deenergized, auto-
matic transfer switches becoming activated, and active electronic systems becoming dormant.
Thus, unless the facility is very simple or totally passive in its function, it will probably be
necessary to test with the power on to obtain valid results, For this reason, the capacitive
or inductive coupling methods have been used for injection of the simulated EMP on the
service-entrance conductors.

As with direct-!njection testing of equipment, the effect of the excitation source

impedance on the system response must be considered, At the facility level, however, there
is often sufficient attenuation between the excitation source and the internal eqlupment
that the source Impedance Is not a dominant factor. For economic reasons, the source
impedance is often compromised to avoid the expense of loosely coupled megavolt pulse
sources. Thus, for example, a 100 kV,-tightly coupled source might be used where strict

328

L Iii!-!iw
adherence to preservation-of-impedence concepts might requIre a 10-MV loosely coupled
source,

Because the insulation and coupler design requirements arc more severe and the
personnel hazards are greater if the pulse is coupled to the primary side of the distribution
transformer, it is usually preferable to perform the injection on the secondary side. This
often has the additional advantage that a substantially lower voltage is required, Because
of the filtering action of the transformer and the mismatch between the aerial line and the
conductors in the conduit, only a fraction of the open-circuit voltage induced in the dis-
tribution lines is transmitted into the service-entrance conduit. This is another reason that
a low-voltage source (e.g., 100 kV) can be used to simulate the effect of several megavolt.
induced on the distribution lines,

The schematic of a capacitor discharge pulser with capacitive coupling to the power
conductors Is shown in Figure 8-6, This direct injection pulser was designed so that the
high-voltage energy storage capacitor could be placed at ground level, The capacitive
coupler unit Is mounted near the service-entrance weatherhead and connected to the energy
storage unit through four 50-ohm coaxial transmission lines operated in parallel, The reac-
tance of the coupling capacitors is large enough that little 60-Hz current flows through
them, but throughout most of the EMP spectrum, their reactance is so small thet virtually
all of the pulser voltage is applied to the power conductors, The coupler installed at the
weatherhead (but not yet connected to the power conductor) is shown In Figure 8-7,

8.4.2 OBSERVATION OF THE SYSTEM RESPONSE

For the simplest form of proof-test, the system response may by judged on the
basis of whether or not the system functions properly during and after the simulated EMP
is injected into the service entrance, Visual and aural observations to detect any spurious
arcing are usually included in such a test. It is more common, however, to provide some
instrumentation to make quantitative measurements of selected internal currents and
voltages so that the margin by which the system passed or failed the test can be determined.

329
 T H U•
Xi, R

-I -(•~~-
bU!U WAX ..
PP ... - -( NUN orF•AA
POWEA I2 ___o_____ kv I
;5 rnA 04 M .
•'r ~~~POW
FE o-. .. 'I

SU PPL Y I to
I 4 7-.
. .7"0
. . .P H ASt.

I~I
MSZI

iH-+ PHASC C

UNIT
PULSEH -0- TO CONDUIT

CQUPLV"
UNIT

Figure B-6 SCHEMATIC DIAGRAM OF POWER-LINE PULSER AND COUPLER

Such Instrumentation Is a necessity for those tests whose primary purpose is to

determine the excitation levels to be used in the equipment tests described In Section 8.3.

The basic instrumentation required for measuring the system response Is a selection
of current probes, high-voltage probes, and a low-power, wideband oscilloscope with a
camera capable of recording at fast writing rates 01 cm/ns), Commercial wideband
(;;P100 MHz), low-power (<150 W) oscilloscopes and cameras are readily available, If
shielded operation Is necessary, Inverters for operating the oscilloscope In a shielded
enclosure from a 12-V storage battery are also available. (One of the advantages of direct-
Injection testing Isthat large external fields and the instrumentation problems associated
with them are not produced).

Curre*nt probes with a variety of window sizes, amplitude and frequency character-
istics, and attachment provisions are also available commercially. Most of these current
probes have transfer impedances (ratio of output voltage to current through the window)
that decrease with decreasing frequency below a few kilohertz so that they tend to suppress
the response to 60-Hz current in the power conductors, This property Is desirable for
making measurements of transients whose peak current value is equal to or loss than the

330
 Figure 8-7 POWER-LINE COUPLER UNIT

60-Hz current in the conductors. The hinged, clamp-on type of probe is also very conve-
nient because it eliminates the necessity of disconnecting conductors to thread them
through the window of the probe.

Passive wideband voltage probes are also available for measuring voltages ranging
from a few hundred volts to about 30 kV. The high-voltage probes are excellent for
measuring voltages with peak values greater than a few hundred volts on 120/240-V power

331
 IA 11I1111,

10 o~ 10 ppF

id ki 10o 0OFt
pm
PR M1.1b
110M
Prw4 '~

6bTo
V u i

LOW-VOLTAOE ACIAPTEn
R
ADA~rRS
Figure FOR A TKTRONIX P-00?PSIEVOTG
84 ROBE

0-332
 REMAINDER conductors. For smaller transients,
OF FACILITY however, the 60-Hz voltage on which
-COMPONENT the transient is superimposed makes
INTERIEST the measurement more difficult. This

vot probleri can be ailelated by making

I Ca high-pass-filter adapter for the probe

--- so that the 60-Hz response of the

L-__ .- . probe is suppressed. Such adapters

Figure 8-0 DEFINITION OF THE EXCITATION are shown in Figure 8-8.

FOR AN INTERNAL EQUIPMENT COMPONENT
When the purpose of the test
is to establish the excitation levels for later use in dIrect-injection tests of internal equip-
mint, some consideration should be given to methods of specifying these excitation levels,
These levels can be specified in terms of the in vint current and voltage on the conductors
entering the equipment, or they can be specified in terms of a Thevenin (or Norton)
equivalent source (see Figure 8-9). in the first case, the excitation sourco contains the
unknown characteristics of the remainder of the facility (which must be simulated in a
bench test of the component equipment), In the latter case, the characteristics of the
remainder of the facility are contained in the Thevenin equivalent source voltage and
impedance. The in s/tu measurements are usually easier to make than those necessary to
specify the Thevenin equivalent source, but they may be inadequate to define the exci-
tation source for laboratory or bench tests of the component, On the other hand, to define
the Thevenin equivalent source, both the short-circuit current and the open-circult voltage
(or one of these and the source impedance of the circuit) at the excitation point must be
measured, Measurement of open-circuit voltage and short-circuit current with the power
on is often impossible, however,

This paradox can usually be resolved by acceptable estimates or approximations to

the characteristics of the Thevenin equivalent source based on power-off measurements or
circuit analysis. If it is necessary to specify the Thevenin source impedance more accurately,
series inductance and shunt capacitance can be inverted at the terminals of the component
to measure the open-circuit voltage and short-circuit current in the EMP spectrum with
the power on,

333
8.5 CITED REFERENCES

1. Elec•rical Transmissiooi and DistributionBook, 4*h ed., 5th printing

(Westinghouse Electric Corporation, East Pittsburgh, Penn., 1964).
2. A. L. Whitson, "Engineering Techniques for Electromagnetic Pulse Testing,"
DNA 3332F, Stanford Research Institute, Contract DNA 3332F, Stanford
Research Institute, Contract DNAO01 -71 -C-0087, Defense Nuclear Agency,
Washington, D.C. (November 1973).
3. DASA EMP (ElectromagneticPulse) Handbook, DASA 2114-1, DASA
Information and Analysis Center, Santa Barbara, Calif. (September 1968).

334
 INDEX

Angle of incidence . . . . .... .. .... .561 Conductors, low voltrrye . . ... ..... 254

Attenuation consln tant...... 57.98 Conduit ,.... .......... I.. 29,161

burled cable....................1869 dimension of .. ,............ 174
transmission line
.... ,...,...... 57,98 number of wires In . .. .. . .....
. 257
wave in loty media . .. . . ... . , 294 plastic . ,,... . .,,. . ... ..... 139,254
Autmatc
swtc . .. ..
.ranfe . 11steel . . . . .....
. ,. .. . .. ..... 139
Auoatctrnfr
wtc .* 31transfer impediance . .. . . . ...
. , 173
Auxiliary power systems ........... 310 Core, trunsformer . . .. .. .. ,.., 202,207
Bend in tranismission gins 109 Corona .,...,, ...
,, . . ...
. I. I146,125
Bre~ich In transissilon line . ., . . 106 threshold of conductors . . . ... 133
Bushing, Itansfoirmer .. ... ......... 203 Coupler, tast ,. . ,. . . ... , 326,326
capacitance ,... .. . .. ... ..... 76,213 Coupling ...
. . . . ... .. . ... .. .. 56
indutane
. .. .. . .. .. .216234to vertical element .. . .. .. . .. 177
Butt wrap, plate . . . . . . ........
. 29,44 to conduit . . . . . . ..
. .. .. .. 161
Cable.
.. . . . . .. . . .. 43to shielded cable
. . . ... .. .. . .. 185
Cbet5h'.ruoh servIce entrance . .. . .. .... 145
aerial . . ... .. .. . .. ... . .. 47 through traforitumers. .. .. .. .. ..... 219
shieldeid . ..... .. .. .. .. ...... 139
Counterpoise ... . . .. .. .. .. ...... 46
Caacta~e
.. .. * ,. . 9high-frequoncy performance . . .. ... 294
couplinrg . ... . .. . ... ........ 217
motor windings ,. . . . . ...
. .... 271Ciit olae , . , ,, , , ,. 20
%tray .,. . . .. . . . .... 270 Cutrrent transformer ,,. . .. ... , 33,310
surge , . . .. .. .. .. ......... 254
traniformer windmars . .. .. ....... 213 Dlflot oirrlal-mode ....
.. . .. . .. .. 116
transmission line. . .. ....... .. 157 coupling to line .*. .. .. .. ....... 121

Carrier, power line ................ 28 In transforrners , .. ý, . .. .. .. ... 222

in unbalanced ...... 2
Charge %aprpartiun . 22lri
22
Dill usran constant . .. ,,. .. ,.., 176,18~0
Crroppod-wmiva test . .. . . .. ..... 207,320 ielvt fucon..... .5 ,2
Cii nut liriraskm . .. . ... .. .. .. . .. ..32
MUM'. .. ........... 136,251 Dischdrigo . .,.
.. .. . . .. . .. .... 231
protection at ,. .. .. . . . .. . ... 253 )rol .. , ,. ., . ,. . 132
lightning. .. .. .. .. .. .. ...... ..... 25
Common-mode .. . ... ... . ... . .. .. 49
Dist I ibu~ted sooi ci...... .. .... .. ...... 1
rixiltatiori of ti irrsfor irels ., , 222
on %irjnal line%.. .. .. .. .. .. ....... 292 Distirbut ion system . .. ... . ........ 33

Cumpton eletirtonti. .. .. .... .. .. ..... 22 lightning arresters ... .. .. . ... .. , 209

lInir. ... ,. .. .. .... .. ....... 43
Conrductivity . .. . .. . .. .. .... 39 liansfoimor.... .... .. .. .... ..... 204
tdil . . .... . ... . .. .. .. ...6722 Duct, fiber .. .. .. .... .... ....... 138

conduit . .. .. .. .. .. ...... ...... 173

335
 Electr ic fie~ld . 24 Impedance, intrinsic . . . .. .. . ... ... .860

at wire height . 4Impedance, transfer. .. .. .. .. ..... 173, 194

in ground. .. .. .. .. .. .. .. .. ..... 171 o odi 7
oil j ronducti . .. .. .. ....... . ..... ..... 132 fC dit . .. . .. .. . 17
of timpp-wound s511101( .,. . .. ... ... 196
Electromagnetic pulse (EMP) 22.24Imus., ...... ,1720

Electronic equipment . .. .. .. ....... 27

Impulse, test .*... . . . ..
. , ... 209,320
grounding . .. .. .. .. .. 282,293
.. ..
I.susceptibility .. . . .. .*. .......... 27 Inductance * . . . . . . . . 39

Entrance, service (s,,i. Service bushing ,. . . ..

. .. ... ,..,... 246
erittmncell .. . .. ... . .. ... . .. .. 29 shielded Cdble ... . . .. . . .*.. 199 .
%troy ... ,...., 268, 260,267
Induction motor ....
. . . . . ......
. 271
action of transformer .,.... .....
. 248
tust of *.. . . . ... .........
. . ., 307 Insulation . .. .. . ... . .. .. .. .. 126

[ ~~~Firinog voltage .. .. .. . .. .. .. .. 244,302lovlae ... ,. .. 23

iFlashover ..... .. ... . .. .. ..... 46 transformer ,, , , . . , . . , 20.

of insulators . . . . . ..
. .. .. ,.. 126In ltrs . , . .. , .. , ,

Fourier trasnsform .,. . . .. . . . ...

. 38 flashovest ,,....... .. ....
, 120
poit .. .. .. .. .. ..... . . 48
Gomm. vs . . .,,. .. . . . . .. .... 22 . . . ...

Gap, %park . . .. .. .. . ... ... .. .. 209 uiesn4

Gradient control . .. .. . . .
...... 139 lonini tioin .,.,,.,..,.. ,. ... .... , 22

.. .. .. .276by
.. .. .. ... ...
Groud Compton electrons . .. .. . . .. 22,68
Grud.........26of l,uhtning arrester *,. .. . ,. 209,245
butt wiap, plate . . .. ... . .. .. .. 44 Lg t ig......................2
counterpoise . .. . . . . .. .,.........292
, Lih.n
impedance ....... . ....
. .. . 7Lightning
. . . .. atrestisrs . .. .. . ...... 29, 210
rod ... .... .. .I I. .114283 distribution . ... . .. .. .. . ... ... 141
ogngl .p
...... .. .. .. .. .. . . .24 286 firing character istics ... .. ....
. 129, 144
ongre..nt....... .... . .......... 4428 secondairy ,. ... ,... . .. .. .. ..... 301

High-altitudle EMP . . . .. .. .. . 22,24 Low -f requency test .. .. . . ... .. ... 209

Impiedarnce . . ........
. ......... 3 Messenger cable . . . . .. .. . . .. ....
. 137
Of gI ound cdhHable ..... 289 Multlcoriductor transmission lines . ... 153
. .
of gioumrd connuction .. .. .. .. .. ... 290 aerial ... .,. .. . . .. . .. .. .... 106
of motor windings ............. 271 condlull. ...
., . .. .. . . .. .. . . 273
of transformers .. .. . .. . ... 219 model of transformer . . . .. . .. .. , 213

Impedance, characteristic ,. . .
,.,.. ,.. ,. . 0properties of. .. .. .. .. .. .. .. ..... 156
calcul"Itioll of . .. ... . . .. 108, 117,270 Neutral . .. .. .. . . . . .. . . . 29,308
measurement of .. ... . .. .. ..... 265 grounding .,.,,.,,.,.,,,,I . 29,.278, 305
of buried cohle . .. .. .. ...... . . '87 conductor In conduit . .,,.,..,..,.,,., 274
of power litle . .. .. ........... 60
of shielded cable .... 153, 159 Nuclear detonation. .. .. .. .. .. .. .. ... 22

336
Periodically grounded line ............. 91 of incident pulse ..... .......... 64
soil , . .... . . ..... . 57,62,162
stray inductance ................
303
Pothead ...... ..... .. ....... 29,138 Tower, steel .. .... ..... .
.. ...... 44
Propagation factor ................ 98 description ........ ........... 45
in free space ......... ......... 54 grnunding ........... . .. 45,294
in toll .......... .. .......... 151 Transfor function .............. 145,223
for aerial line .... ......... 67,98 n u 1457,3
for buried cable . .............. 181 Transforms . ................... 38
Protection ......... .. ......... 298 Transformer ,. ............... 29,202
lightning .. ... ......... 45,248 classes ......... ...... ... . 206
of low-voltage circuits ... ....... . 254 construction .......... ....... 203
distribution .. ... ............ . 204
Reflection coefficient transfer characteristics ........ 219,248
Reflectometer, time-domain ......... 265 transient tests of .. .......... . 207

Sag in transmission line ........... . 114 Transmission line .*.............. . 42

biconic ........... ,...... .. 79
. . . . . . . .. . 18 2 di st rib . o.r. . l . . . . . . . . . . . . . .. 7
S a t u r a t io n o f co n d u it
12distributed source *.. .. . ... ,... 51
Service entrance ... ......... . 136 formulas for ..................
multiconductor 52
...............
116
analysis of .. .... ... ..... .... 145
weatherhead . .......... .. 137,140 properties of .. ..... .. .. . 43
affect on low-voltage circuits ...... 143 Transmission system ............... 43

Shielded cable 139

1............ grounding ............... 278
characteristic impedance .......... 157 Uninterruptible power systems ..... . 311
transfer impedance ............. 194
Velocity of propagation ....... 29
Skin depth ................ 100,173 of light ....... . ... .... . . . . 86
defltidtion of .................... 1 1 measurement of ............... 267
in soil . . . . . . . . . . . . . . ..... . . 163
Vertical elem ents ......... ........ 77
Source region .. ... ... ... ........ 24
Wave . . . .... .. .... . . .. .. . .. . . 27
Step function ........ . ......... 104 incdent ..... ... ...... .. 54
response of entrance cabins ........ 146 reflected ............... 54
Surge arrester (see Lightning Winding, transformer ......... .. 204
arresters). ....... ... ... .... . 210 motor .............. I ..I. 271
Switched capacitor ..... ......... 231 Zero-sequence coupling ..... . .... . 50

Three-phase . .... ..... ... .... 33

transformer connections .... ...... 279
grounding cunsiderations ....... . 252,277

Tim e constant .................. 24

bushing capacitance ............ 246
diffusion . . . . . .. .. . .. . .. . . . . 174

337