CR 6904

NUREG/CR-6904
BNL-NUREG-75208-2005
Evaluation of the Broadband

r Impedance Spectroscopy
Prognostic/Diagnostic
Technique for Electric Cables
Used in Nuclear Power Plants
\
The Boeing Company and
Brookhaven National Laboratory
U.S. Nuclear Regulatory Commission

Office of Nuclear Regulatory Research
Washington, DC 20555-0001
NUREG/CR-6904
BNL-NUREG-75208-2005
Evaluation of the Broadband

Impedance Spectroscopy
Prognostic/Diagnostic
Technique for Electric Cables
Used in Nuclear Power Plants
Manuscript Completed: November 2005
Date Published: June 2006
Prepared by
D. Rogovin1, R. Lofaro2
1
The Boeing Company
Huntington Beach, CA 92647
2
Brookhaven National Laboratory
Upton, NY 11973-5000
J. Vora, NRC Project Manager
Prepared for
Division of Fuel, Engineering and Radiological Research
Office of Nuclear Regulatory Research
U.S. Nuclear Regulatory Commission
Washington, DC 20555-0001
NRC Job Code Y6410
Intentionally Left Blank
ii
ABSTRACT
Aging mechanisms can lead to degradation of electric cables and resultant failures of critical functions, as
well as losses of essential information for the decision-making process. Consequently, it would be highly
desirable to have a single universally effective prognostic, diagnostic technique that can be used in situ to
monitor the condition and predict the remaining useful life of installed electric cables of all types, in
operating nuclear power plant applications and environments.
To address this issue, the U.S. Nuclear Regulatory Commission (NRC), Office of Nuclear Regulatory
Research (RES), in collaboration with the Federal Aviation Administration (FAA) and the Boeing Company,
initiated a research study to evaluate the effectiveness of the broadband impedance spectroscopy (BIS)
technique for use in nuclear power plant cable diagnostics and condition monitoring. The results demonstrate
the effectiveness of this nonintrusive, nondestructive technique, which uses low-voltage signals at varying
frequencies to scan the length of an installed cable system to locate anomalies and degradation. Based
upon this study’s promising results, RES is considering follow-on collaborative research to demonstrate
the effectiveness of the BIS technique in operating nuclear power plant environments.
iii
iv
vi
TABLE OF CONTENTS
Page No.
ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii
FOREWORD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
TABLE OF CONTENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi
LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxiii
EXECUTIVE SUMMARY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxv
ACKNOWLEDGMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxix
ABBREVIATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxxi
LIST OF SYMBOLS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxxiii
1. INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Technical Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.4 Research Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.5 Equipment Used . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2. REVIEW OF CONDITION MONITORING TECHNIQUES . . . . . . . . . . . . . . . . . . . . . . . . . 7

2.1 General Criteria for an Effective Technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2 Mechanical Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2.1 Visual Inspection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2.2 Elongation-At-Break . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2.3 Compressive Modulus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.3 Chemical Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.3.1 Oxidation Induction Time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.3.2 Fourier Transform Infrared (FTIR) Spectroscopy . . . . . . . . . . . . . . . . . . 10
2.4 Electrical Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.4.1 Insulation Resistance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.4.2 Dielectric Dissipation or Loss Factor . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.4.3 Voltage Withstand Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.4.4 Time-Domain Reflectometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.4.5 Partial Discharge Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
vii
2.5 Reasons to Focus on an Electrical Technique . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3. THEORY OF BROADBAND IMPEDANCE SPECTROSCOPY . . . . . . . . . . . . . . . . . . . . 15

3.1 Equivalent Circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.2 Wire Properties and Wire Impedance Spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.2.1 Approach to Extract Wire Properties from Impedance Data . . . . . . . . . . 17
3.2.2 Wire Properties from Impedance Data for Undamaged Wiring . . . . . . . 21
3.2.3 Wire Properties from Impedance Data for Damaged Wiring . . . . . . . . . 23
3.3 BIS Test Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.3.1 Procedure for Making Impedance Measurements . . . . . . . . . . . . . . . . . . 27
3.3.2 Analysis and Interpretation of Impedance Data . . . . . . . . . . . . . . . . . . . 28
4. PREPARATION AND BASELINE TESTING OF CABLE TEST SPECIMENS . . . . . . . . 31

4.1 Selection of Cable Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
4.2 Preparation of Test Samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
4.3 Baseline Testing of Test Samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
4.4 Accelerated Aging of Test Samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
5. RESULTS OF BIS TESTING OF CABLE SAMPLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

5.1 Application of BIS to Thermally Aged Cables . . . . . . . . . . . . . . . . . . . . . . . . . . 37
5.1.1 Detection of Degradation on Cables with Global Thermal Aging . . . . . 37
5.1.2 Development of Models for Cables with Thermal Degradation . . . . . . . 40
5.1.3 Detection of Localized Degradation on Cables . . . . . . . . . . . . . . . . . . . . 54
5.1.4 Locating Localized Degradation on Cables . . . . . . . . . . . . . . . . . . . . . . . 66
5.2 Application of BIS to Cables with Attached Loads . . . . . . . . . . . . . . . . . . . . . . . 71
5.2.1 Test Specimens Used for Testing with Attached Loads . . . . . . . . . . . . . 72
5.2.2 Low Frequency BIS Tests for Cables with Attached Loads . . . . . . . . . . 72
5.2.3 High Frequency BIS Tests for Cables with Attached Loads . . . . . . . . . . 75
5.2.4 Phenomenological Models of Cables with Attached Loads . . . . . . . . . . 77
5.2.5 Locating a Hot-Spot on a Cable with an Attached Load . . . . . . . . . . . . . 83
5.3 Application of BIS to Cables with Abrasion Damage . . . . . . . . . . . . . . . . . . . . . 84
5.3.1 Abrasion Test Specimens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
5.3.2 Evaluation of Low Frequency BIS Results on Cables with Abrasion . . . 85
5.3.3 Evaluation of High Frequency BIS Results on Cables with Abrasion . 101
5.3.4 Detection of Abrasion Damage Using BIS . . . . . . . . . . . . . . . . . . . . . . 110
5.3.5 Location of Abrasion Damage Using BIS . . . . . . . . . . . . . . . . . . . . . . . 113
5.4 Application of BIS to Cables with Cracking Damage . . . . . . . . . . . . . . . . . . . . 122
5.4.1 Test Specimens Used for Testing of Cables with Cracking . . . . . . . . . 123
5.4.2 Low Frequency BIS Measurements of Cables with Cracking . . . . . . . . 124
5.4.3 High Frequency BIS Measurements of Cables with Cracking . . . . . . . 128
5.5 Application of BIS to Cables with Varying Environments . . . . . . . . . . . . . . . . 130
5.5.1 Specimens Tested to Evaluate BIS for Cables with a Varying
Environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
5.5.2 Low Frequency BIS Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
viii
5.5.3 Phenomenological Models of Cables with Varying Environment . . . . 137
6. EVALUATION OF THE BIS METHOD FOR NUCLEAR PLANT CABLES . . . . . . . . . 149

6.1 Results for Global Thermal Aging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
6.2 Results for Global Plus Localized Thermal Aging (Hot-Spot) . . . . . . . . . . . . . 150
6.3 Results for Global Thermal Aging with an Attached Load . . . . . . . . . . . . . . . . 151
6.4 Results for Global Plus Localized Abrasion Damage . . . . . . . . . . . . . . . . . . . . 152
6.5 Results for Global Thermal Aging with Cracking Damage . . . . . . . . . . . . . . . . 153
6.6 Results for Global Thermal Aging with a Varying Environment . . . . . . . . . . . 153
7. CONCLUSIONS AND CONSIDERATIONS FOR FOLLOW-ON RESEARCH . . . . . . . 155

7.1 Conclusions on the Effectiveness of the BIS Method . . . . . . . . . . . . . . . . . . . . 155
7.2 Considerations for Follow-On Research to Demonstrate the Effectiveness of the
BIS Method in Nuclear Plant Environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
8. REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
Appendix A: BIS Test Results for Cables with Global Thermal Aging Simulating
20 years and 40 years of Service . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
Appendix B: BIS Test Results for Cables with Localized Thermal Aging (Hot Spots) . . . . . 203
ix
x
LIST OF FIGURES
Page No.
Figure 1 Wire system equivalent circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Figure 2 Low frequency impedance spectra log magnitude for a 10 meter poly-x
twisted pair wire . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Figure 3 Low frequency impedance spectra phase angle for a 10 meter poly-x twisted
pair wire . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Figure 4 High frequency impedance spectra magnitude for a 10 meter poly-x twisted
pair wire . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Figure 5 High frequency impedance spectra phase angle for a 10 meter poly-x twisted
pair wire . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Figure 6 Frequency-dependent wire resistance and inductance . . . . . . . . . . . . . . . . . . . . . . 22
Figure 7 Micro picture of a poly-x wire cross section . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
Figure 8 Real and imaginary components of the wire insulation dielectric

function (open circuited) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
Figure 9 Log magnitude and phase of the impedance spectra for moist and dry
wiring (open circuited) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
Figure 10 Ratio of the impedance spectra magnitude for moist and dry wiring
(open circuited) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Figure 11 Ratio of the resistance and inductance for moist and dry wiring as a
function of frequency (R is S/m and L is H/m - open circuited) . . . . . . . . . . . . . . 25
Figure 12 Ratio of the real and imaginary components of the wiring insulation’s dielectric
function for moist and dry wiring as a function of frequency . . . . . . . . . . . . . . . . 27
Figure 13 Equipment and test setup for making impedance measurements . . . . . . . . . . . . . . 28
Figure 14 Typical aging chamber loading for global and hot-spot aging of cable test
specimens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Figure 15 Measured low frequency impedance phase spectra for cables with global
thermal aging to simulate 60 years of service at various temperatures . . . . . . . . . 38
xi
Figure 16 Measured high frequency impedance phase spectra for cables with global
Figure 17 Expanded view of impedance phase and magnitude per unit length spectra
over the frequency range from 78 MHz to 86 MHz for cables with global
thermal aging to simulate 60 years of service at various service temperatures . . . 40
Figure 18 Characteristic impedance magnitude per unit length and phase spectra
for cables with global thermal aging to simulate 60 years of service at
various temperatures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
Figure 19 Imaginary component and real component of propagation function for

cables with global thermal aging to simulate 60 years of service at various
temperatures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
Figure 20 Resistance per unit length and change in resistance per unit length for cables
with global thermal aging to simulate 60 years of service at various
temperatures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
Figure 21 Comparison of resistance per unit length predicted by analytical model to

that extracted from measured impedance for cable with global thermal aging
to simulate 60 years of service at 70ºC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
Figure 22 Inductance per unit length and change in inductance per unit length for
cables with global thermal aging to simulate 60 years of service at 70ºC . . . . . . . 45
Figure 23 Comparison of inductance per unit length predicted by analytical model to

that extracted from measured impedance for cable with global thermal
aging to simulate 60 years of service at 70ºC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
Figure 24 Capacitance per unit length and change in capacitance per unit length for cables
with global thermal aging to simulate 60 years of service at various
temperatures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
Figure 25 Conductance-to-frequency ratio per unit length and change in conductance-to-

frequency ratio per unit length for cables with global thermal aging to
simulate 60 years of service at various temperatures . . . . . . . . . . . . . . . . . . . . . . . 49
Figure 26 Comparison of capacitance per unit length predicted by analytical model to that
extracted from measured impedance for cable with global thermal aging to
simulate 60 years of service at 70ºC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
xii
Figure 27 Comparison of conductance-to-frequency ratio per unit length predicted by
analytical model to that extracted from measured impedance for cable with
global thermal aging to simulate 60 years of service at 70ºC . . . . . . . . . . . . . . . . . 51
Figure 28 Comparison of characteristic impedance magnitude per unit length and phase
spectra predicted by analytical model to that extracted from measured
impedance for cable with global thermal aging to simulate 60 years of
service at 70ºC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
Figure 29 Comparison of the imaginary component and real component of the

propagation function predicted by analytical model to that extracted from
measured impedance for cable with global thermal aging to simulate
60 years of service at 70ºC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
Figure 30 High frequency characteristic impedance magnitude and phase spectra

extrapolated from low frequency spectra for cables with thermal aging to
simulate 60 years of service at various temperatures . . . . . . . . . . . . . . . . . . . . . . . 53
Figure 31 Imaginary and real components of the high frequency propagation function
determined from the extrapolated characteristic impedance for cables with
Figure 32 Impedance magnitude and phase spectra over the frequency range of
14MHz to 22MHz for cables with global thermal aging to simulate 60 years
of service at 50ºC, plus additional thermal aging to simulate hot-spots of
different severity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Figure 33 Impedance magnitude and phase spectra over the frequency range of 39MHz
to 45MHz for cables with global thermal aging to simulate 60 years of
service at 50ºC, plus additional thermal aging to simulate a hot-spot . . . . . . . . . . 57
Figure 34 Impedance magnitude and phase spectra over the frequency range from
different sizes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
Figure 35 Impedance magnitude and phase spectra over the frequency range of
different sizes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
xiii
of service at 50ºC, plus additional thermal aging to simulate hot-spots at
different locations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
of service at 50ºC, plus additional thermal aging to simulate hot-spots at
different locations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
Figure 38 Example of average impedance phase spectrum over the frequency range from
100kHz to 100MHz for cable with global thermal aging . . . . . . . . . . . . . . . . . . . . 63
Figure 39 Schematic model of a cable with a hot-spot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
Figure 40 Impedance phase and magnitude error functions for cable with thermal
aging to simulate a 1 meter hot-spot located 7.0 to 8.0 meters from the
cable end representing 60 years of service at 70ºC . . . . . . . . . . . . . . . . . . . . . . . . 69
Figure 41 Impedance phase zero crossing error functions for cable with thermal aging
to simulate a 1.0 meter hot-spot located 7.0 to 8.0 from the cable end
representing 60 years of service at 70ºC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
Figure 42 Impedance phase and magnitude error functions for cable with thermal aging to
simulate a 1 meter hot-spot 2.0 to 3.0 meters from the end of the cable
Figure 43 Impedance phase zero crossing error functions for cable with thermal aging
to simulate a 1 meter hot-spot located 2.0 to 3.0 meters from the end
Figure 44 Comparison of measured impedance magnitude for cable with and without an
attached load . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
Figure 45 Comparison of the impedance phase spectra for cables with and without a
load attached . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
Figure 46 Comparison of impedance phase in the frequency range from 116kHz to

123kHz for cables with and without a load attached . . . . . . . . . . . . . . . . . . . . . . . 74
Figure 47 Change in impedance phase for cable system with a hot-spot from cable system
without a hot-spot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
xiv
Figure 48 Comparison of A(T) parameter in the vicinity of the first zero crossing
for cable systems with and without a hot-spot . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
Figure 49 Schematic of model for thermally aged cable with a load attached . . . . . . . . . . . . 78
Figure 50 Comparison of low frequency measured and predicted impedance magnitude

for a cable system with no hot-spots and an attached load . . . . . . . . . . . . . . . . . . . 79
Figure 51 Comparison of low frequency measured and predicted impedance phase

spectra for a cable system with no hot-spots and an attached load . . . . . . . . . . . . . 79
Figure 52 Comparison of high frequency measured and predicted impedance magnitude

and phase for a cable system with a hot-spot and an attached load . . . . . . . . . . . . 80
Figure 53 Schematic of a cable system with a hot-spot and an attached load . . . . . . . . . . . . 80
Figure 54 Comparison of measured and modeled impedance magnitude for cable

system with a hot-spot and an attached load . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Figure 55 Comparison of low frequency measured and modeled impedance phase

spectra for a cable system with a hot-spot and an attached load . . . . . . . . . . . . . . 82
Figure 56 Comparison of high frequency impedance magnitude and phase spectra for
cable system with a hot-spot and an attached load . . . . . . . . . . . . . . . . . . . . . . . . . 82
Figure 57 Prediction of hot-spot location for a cable system with a load attached . . . . . . . . 83
Figure 58 Schematic of four cases studied to evaluate BIS for detecting and locating
abrasion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
Figure 59 Low frequency impedance phase spectra for cables with and without abrasion
exposed to an environment of 20% Relative Humidity . . . . . . . . . . . . . . . . . . . . . 86
Figure 60 Low frequency impedance phase spectra for cables with and without abrasion
exposed to an environment of 85% Relative Humidity . . . . . . . . . . . . . . . . . . . . . 87
Figure 61 Impedance magnitude for cables with and without abrasion plus a conductive
contaminant exposed to an environment of 85% RH . . . . . . . . . . . . . . . . . . . . . . . 88
Figure 62 Impedance phase spectra for cables with and without abrasion plus a
conductive contaminant exposed to an environment of 85% RH. . . . . . . . . . . . . . 88
xv
Figure 63 Low frequency characteristic impedance magnitude for cables with and
without abrasion exposed to an environment of 85% RH . . . . . . . . . . . . . . . . . . . 89
Figure 64 Low frequency characteristic impedance phase spectra for cables with
and without abrasion exposed to an environment of 85% RH . . . . . . . . . . . . . . . . 90
Figure 65 Imaginary component of the low frequency propagation function for cables
with and without abrasion exposed to an environment of 85% RH . . . . . . . . . . . 91
Figure 66 Real component of the low frequency propagation function for cables with
and without abrasion exposed to an environment of 85% RH . . . . . . . . . . . . . . . . 91
Figure 67 Capacitance per unit length extracted from measured low frequency
impedance spectra for cables with and without abrasion exposed to an
environment of 85% RH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
Figure 68 Conductance per unit length extracted from measured low frequency
impedance spectra for cables with and without abrasion exposed to an
without abrasion plus a conductive contaminant exposed to an environment
of 85% RH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
Figure 70 Low frequency characteristic impedance phase spectra for cables with and
without abrasion plus a conductive contaminant exposed to an environment
of 85% RH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
with and without abrasion plus a conductive contaminant exposed to an
Figure 72 Real component of the low frequency propagation function for cables with
and without abrasion plus a conductive contaminant exposed to an
environment of 85% RH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
impedance spectra for cables with and without abrasion plus a conductive
xvi
Figure 75 Schematic of the localized cable interaction model for abraded cables plus a
conductive contaminant . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
Figure 76 Schematic representation of abraded cable with a conductive path between

conductors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
Figure 77 Impedance magnitude of conductivity path for abraded cable with conductive
path between conductors at various locations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
Figure 78 Comparison of magnitude for total impedance, interaction impedance, and

conductivity path impedance for abraded cable with a conductivity path at
various locations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
Figure 79 Comparison of conductivity path impedance phase, >(T), cable impedance

phase, 0(T), and total impedance phase, M(T), for abraded cable with a
conductivity path . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
cables with and without abrasion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
the frequency range from 8.4 MHz to 9.6MHz for cables with and
without abrasion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
Figure 82 Comparison of high frequency impedance magnitude and phase spectra in

the frequency range from 64MHz to 65.5MHz for cables with and without
abrasion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
Figure 83 Imaginary component of the high frequency propagation function for

cables with abrasion plus a conductive contaminant . . . . . . . . . . . . . . . . . . . . . . 105
Figure 84 Real component of the high frequency propagation function for cables
with abrasion plus a conductive contaminant . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
Figure 85 Resistance spectra for abraded cables in the high frequency range from
phenomenological model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
Figure 86 Inductance spectra for abraded cables in the high frequency range from
phenomenological models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
Figure 87 Capacitance spectra for abraded cables in the high frequency range from
xvii
Figure 88 Conductance spectra for abraded cables in the high frequency range from
Figure 89 Slope of the equation representing the imaginary component of the high
frequency propagation function for abraded cables . . . . . . . . . . . . . . . . . . . . . . . 111
Figure 90 Comparison of zero crossings in the frequency range 4.4MHz to 4.8MHz

for the second derivative of the impedance phase, A(T), for cables with
and without abrasion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
Figure 91 Comparison of the zero crossings of the second derivative of the impedance
phase, A(T), in the frequency range from 22.4MHz to 23.8MHz for cables
with and without abrasion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
Figure 92 Comparison of predicted and measured impedance phase in the range from
2.55MHz to 10.04MHz for unabraded cable (Specimen 15) . . . . . . . . . . . . . . . . 115
Figure 93 Comparison of predicted and measured impedance magnitude in the range

from 2.55MHz to 10.04MHz for unabraded cable (Specimen 15) . . . . . . . . . . . . 116
Figure 94 Characteristic impedance magnitude in the frequency range from 2.55MHz

to 10.04MHz for cables with and without abrasion . . . . . . . . . . . . . . . . . . . . . . . 117
Figure 95 Characteristic impedance phase spectra in the frequency range from

2.55MHz to 10.04MHz for cables with and without abrasion . . . . . . . . . . . . . . . 117
Figure 96 Imaginary component of the propagation function in the frequency

range from 2.55MHz to 10.04MHz for cables with and without abrasion . . . . . . 118
Figure 97 Real component of the propagation function in the frequency range from
2.55MHz to 10.04MHz for cables with and without abrasion . . . . . . . . . . . . . . . 118
Figure 98 Measured impedance phase spectra in the frequency range from 4.048MHz to
15.04MHz for cables with and without abrasion on one conductor . . . . . . . . . . 120
Figure 99 Predicted location of abrasion damage for cable with one conductor abraded . . 121
Figure 100 Measured impedance phase spectra in the frequency range from 4.048MHz to
15.04MHz for cables with and without abrasion on both conductors . . . . . . . . . 122
Figure 101 Predicted location of abrasion damage on cable with both conductors abraded . 123
Figure 102 Schematic of cable with cracking damage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
xviii
Figure 103 Comparison of impedance phase for cables with and without cracking . . . . . . . . 125
Figure 104 Characteristic impedance log magnitude and phase for cables with and without
cracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
Figure 105 Imaginary and real components of the propagation function for cables with
and without cracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
Figure 106 Resistance per unit length for cables with and without cracking . . . . . . . . . . . . . 127
Figure 107 Inductance per unit length for cables with and without cracking . . . . . . . . . . . . . 127
Figure 108 Comparison of capacitance per unit length for cable with and without cracks . . 128
Figure 109 Conductance per frequency per unit length for cable with and without cracks . . 129
Figure 110 Comparison of second derivative of the impedance spectra, A(T), for
cables with and without cracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
Figure 111 Schematic of test setup for cables with varying environment . . . . . . . . . . . . . . . 131
Figure 112 Impedance magnitude for 100 meter cable with no aging or hot-spot and
external temperatures of 24ºC (75ºF), and 42ºC (107.6ºF) . . . . . . . . . . . . . . . . . . 132
Figure 113 Impedance phase spectra for 100 meter cable with no aging or hot-spot and
Figure 114 Impedance phase spectra for 100 meter cable with no aging or hot-spot and
external temperatures of 24ºC (75ºF) low humidity, and 29ºC (85ºF)
and 85% RH. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
Figure 115 Impedance phase spectra for 100 meter cable with 1 meter hot-spot and
Figure 116 Impedance phase spectra for 100 meter cables with and without a 1 meter
hot-spot, and an external temperature of 24ºC (75ºF) . . . . . . . . . . . . . . . . . . . . . 134
Figure 117 Expanded view of the impedance phase spectra in the frequency range
from 425kHz to 445kHz for 100 meter cables with and without a
1 meter hot-spot, and an external temperature of 24ºC (75ºF) . . . . . . . . . . . . . . . 135
Figure 118 Expanded view of the impedance phase spectra in the frequency range
from 425kHz to 445kHz for 100 meter cables with and without a
1 meter hot-spot, and an external temperature of 42ºC . . . . . . . . . . . . . . . . . . . . 136
xix
Figure 119 Summary comparison of impedance phase spectra for 100 meter cables
with and without a 1-meter hot-spot and external temperatures of
24ºC (75ºF) and 42ºC (107.6ºF) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
Figure 120 Comparison of impedance magnitude in the frequency range from 1kHz to
1MHz for cables aged to simulate 60 yrs @ 50ºC and 60 yrs @ 90ºC . . . . . . . . . 137
Figure 121 Comparison of impedance phase spectra for 10 meter cables aged to
simulate 60 yrs @ 50ºC and 60 yrs @ 90ºC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
Figure 122 Comparison of extracted characteristic impedance magnitude for 10 meter

cables aged to simulate 60 yrs. @ 50ºC and 60 yrs @ 90ºC . . . . . . . . . . . . . . . . 139
Figure 123 Comparison of extracted characteristic impedance phase spectra for 10 m

cables aged to simulate 60 yrs.@50ºC and 60 yrs@90ºC . . . . . . . . . . . . . . . . . . 139
Figure 124 Comparison of imaginary component of the propagation function for

10 m cables aged to simulate 60 yrs.@50ºC and 60 yrs@90ºC . . . . . . . . . . . . . . 140
Figure 125 Comparison of the real component of the propagation function for
10 meter cables aged to simulate 60 yrs. @ 50ºC and 60 yrs @ 90ºC . . . . . . . . . 140
Figure 126 Comparison of resistance per unit length and relative difference in
resistance per unit length for 10-meter cables aged to simulate 60 yrs.
@ 50ºC and 60 yrs @ 90ºC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
Figure 127 Comparison of inductance per unit length and relative difference in
inductance per unit length for 10m cables aged to simulate 60yrs.@50ºC
and 60yrs@90ºC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
Figure 128 Comparison of capacitance per unit length and relative difference in
capacitance per unit length for 10m cables aged to simulate 60 yrs
@50ºC and 60 yrs@90ºC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
Figure 129 Comparison of conductance per unit length and relative difference in
conductance per unit length for 10m cables aged to simulate 60 yrs
@50ºC and 60 yrs@90ºC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
Figure 130 Schematic of model for 100-meter cable containing a hot-spot . . . . . . . . . . . . . . 144
Figure 131 Error in predicted impedance phase as a function of assumed hot-spot

location . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
xx
Figure 132 Error in predicted impedance phase as a function of assumed hot-spot
location over the cable length from 90 meters to 99 meters . . . . . . . . . . . . . . . . . 147
xxi
xxii
LIST OF TABLES
Page No.
Table 1 Cable Test Specimens for Broadband Impedance Spectroscopy Research . . . . . . 33
Table 2 Specimens tested to evaluate BIS for detecting global thermal aging . . . . . . . . . . 38
Table 3 Parameters for analytical model of resistance for cables thermally aged
to simulate 60 years of service at various service temperatures . . . . . . . . . . . . . . . 44
Table 4 Parameter values for analytical model of inductance for cables aged to
simulate 60 years of service at various service temperatures . . . . . . . . . . . . . . . . . 46
Table 5 Parameter values for dielectric function analytical model for cables
thermally aged to simulate 60 years of service at various service temperatures . . 50
Table 6 Specimens tested to evaluate BIS for locating localized areas of thermal aging . . 55
Table 7 Average impedance phase magnitude for cables with global thermal aging to
simulate 60 years of service at 50ºC, plus additional thermal aging to simulate
hot-spots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
Table 8 Differences in zero crossing frequencies between cable with no hot-spot

(cable 26) and cables with hot-spots (cables 27 to 30) at odd number zero
crossings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
Table 9 Impedance phase zero crossing line spacing for cables with global thermal
aging to simulate 60 years of service, with and without hot-spots . . . . . . . . . . . . . 66
Table 10 Specimens tested to evaluate BIS for testing cables with attached loads . . . . . . . . 72
Table 11 Specimens tested to evaluate BIS for detecting abrasion . . . . . . . . . . . . . . . . . . . . 84
Table 12 Fitting constants for equation representing the real component of the high
frequency propagation function for cables with abrasion plus a conductive
contaminant . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
Table 13 Fitting constants for equations representing the high frequency electrical
properties of abraded cables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
Table 14 Specimens tested to evaluate BIS for testing cables with a varying environment 131
xxiii
xxiv
EXECUTIVE SUMMARY
Aging of electric cables has been studied for decades and it is well known that aging degradation
can lead to failures of electric cable systems. Periodic replacement of installed cables is not
practical due to the costs involved, and the potential for introducing unforseen anomalies due to
installation deficiencies and modifications. As such, a prognostic/diagnostic technique that can
be used in situ to monitor the condition of installed electric cables and predict their remaining
useful life is highly desirable. While a number of prognostic/diagnostic techniques are currently
available to test installed cables, each has its limitations on the types of cables for which it can
be used and the extent of the age-related degradation it can detect. Currently, there is no single
viable in situ technique available that can be used to monitor the condition of all types of electric
cables in a nuclear power plant, especially the low-voltage instrumentation and control (I&C)
cables.
Existing regulatory requirements relevant to electric cables:
• 10CFR, part 50, Appendix A, GDC 18 - Inspection and testing of electric power systems.
Electric power systems important to safety shall be designed to permit appropriate

periodic inspection and testing of important areas and features, such as, wiring,
insulation, connections, and switchboards, to assure the continuity of the systems and the
condition of their components. The systems shall be designed with a capability to test
periodically the operability and functional performance of the components of the
systems, such as onsite power sources, relays, switches, and buses, and the operability of
the system as a whole.
• 10CFR, part 50, Appendix B, Criterion XI

A test program shall be established to assure that all testing required to demonstrate that
structures, systems and components will perform satisfactorily in service is identified and
performed in accordance with written test procedures which incorporate the requirements
and acceptance limits contained in applicable design documents.
• 10CFR, Part 54, License Renewal, GALL Report - XI E3 Inaccessible medium voltage
cables not subject to 10CFR 50.49 environmental qualification requirements
Testing and condition monitoring based aging management program is acceptable to the
staff.
The utilization of the research results and the applicability of an effective diagnostics and
condition monitoring method for installed cable systems in operating nuclear power plants:
• The research results show that the BIS method has potential for being a non-intrusive,
non-destructive and an effective in situ diagnostics and condition monitoring method for
installed cable systems in operating nuclear power plants.
xxv
• The BIS method has potential for satisfying the aging management requirements of
inaccessible/underground cables for the current license term and for license renewal
considerations.
• The research results provide technical bases for:
(a) evaluating the licensees programs for aging management of cables within the scope of
license renewal (10CFR Part 54),
(b) supporting the development of appropriate IEEE standards, and
(c) for the development of a dedicated regulatory guide on cable diagnostics and
condition monitoring
The BIS technique - applied to electric cable diagnostics and condition monitoring:
The Broadband Impedance Spectroscopy (BIS) technique was developed by the Boeing
Company under the sponsorship of the United States Federal Aviation Administration (FAA) to
monitor the condition of installed aircraft wiring. In the study reported herein, which was
sponsored by the Office of Nuclear Regulatory Research (RES) of the U.S. Nuclear Regulatory
Commission (NRC), the BIS technique was evaluated for application to electric cables used in
nuclear power plants. Cable samples, which are representative of a commonly used type of low-
voltage instrumentation and control cable in nuclear power plants, were prepared and received
accelerated aging to simulate various types of degradation expected in actual plant service
conditions, including the following:
• Global thermal degradation,

• Global thermal degradation plus localized hot-spots,
• Thermal degradation with cracking, and
• Abrasion damage
The cable samples were then tested in the laboratory using the BIS method, and the data were
analyzed to draw conclusions on the effectiveness of the method. Several test configurations
were evaluated, including:
• Constant temperature and humidity along the cable with no load attached,
• Constant temperature and humidity along the cable with a load attached, and
• Varying environments along the length of the cable with no load attached
For the cable type tested, pre-aging measurements were used to develop models that represent
the impedance response of the cable. These models were then used to determine the post-aging
condition of the cables. In applying the BIS method, a model is useful for each type of cable to
be tested.
xxvi
The results demonstrated that the BIS method can be used on nuclear power plant cables, and
may represent a breakthrough in the prognostics and diagnostics of installed cable systems. The
technique provides a non-destructive means of monitoring cable systems in their installed
configuration. Age-related degradation can be detected in an incipient stage prior to failure.
Specific conclusions from this research are:
• The BIS method was clearly able to detect the presence of thermal degradation
associated with the cables used in this study. Specifically, the impedance phase
spectra of the cables tested were observed to increase as the amount of thermal
degradation on the cable increased. This increase can be used as an indicator of
global thermal degradation.
• The BIS method was able to detect the presence of localized thermal degradation,
or hot-spots on the cables. Specifically, a shift in the zero crossings of the
impedance phase spectra was observed when a hot-spot was present on the cable.
• An approach was developed for locating hot-spots within a cable using models of
the cable’s electrical properties. The models were able to predict the hot-spot
location within ±10%.
• The BIS method was not sensitive enough to distinguish between the different
severities and sizes of hot-spots simulated at low frequencies. However, high
frequency data were able to distinguish between the severity levels. Additional
research is warranted to establish the sensitivity limits for this technique.
• The BIS method was able to detect and locate the presence of abrasion-related
damage on a cable. The models and approach used are similar to that for
detecting and locating thermal hot-spots.
• The BIS method was demonstrated to be effective for detecting and locating
degradation on cables with an attached load. This is important since it is
desirable to have a technique that can test cables in their installed configuration,
without having to disconnect them from attached equipment.
• The BIS method was able to detect simulated cracking damage on cables.
However, the simulation method used in this study was determined not to
accurately represent the cracking phenomena. Additional research is warranted to
more accurately evaluate the BIS method on actual cable cracking.
or hot-spots on the cables even with a varying environment along the external
surface of the cables. For the example presented herein, the model’s prediction of
hot-spot location is within a few percent of the correct position.
While the BIS method shows great promise as a prognostic and diagnostic technique for installed
cable systems, additional research is desirable before it can be applied in the field. Specifically,
the following additional items should be considered for future research:
xxvii
• Research on the detection and location of cracking damage in cables using more
realistic simulation of the cracking damage,
• The technique should be demonstrated on additional types of cable to determine
its usefulness for other materials and cable configurations,
• The technique should be demonstrated in an actual plant environment to
determine the impact of the various environmental factors that may impact the
BIS measurements, and
• The technique should be demonstrated on blind test samples in which the type,
severity, size, and location of the degradation are unknown
Based upon the promising results obtained from this research effort, the Office of Nuclear
Regulatory Research is considering follow-on research, preferably in a collaborative manner, to
demonstrate the applicability of the BIS method in nuclear plant environment.
xxviii
ACKNOWLEDGMENTS
The authors would like to thank the NRC program manager, Jitendra Vora for his technical
guidance and support throughout this project.
We would also like to thank Martin Kendig of Rockwell Scientific for his assistance in the
performance of the impedance measurements, and the processing of the data.
We also thank Michael Villaran of Brookhaven National Laboratory for his assistance in
reviewing the report and providing technical comments.
The assistance of Louis Gerlach in the preparation of the cable test specimens is also greatly
appreciated.
We also express our appreciation to Susan Signorelli for her assistance in the preparation of this
manuscript.
xxix
xxx
ABBREVIATIONS
AWG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . American Wire Gauge

BIS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Broadband Impedance Spectroscopy
BNL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Brookhaven National Laboratory
CM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Condition Monitoring
DZC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Detection Zero Crossing
EAB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Elongation-at-Break
EPR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ethylene Propylene Rubber
FAA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Federal Aviation Administration
FTIR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fourier Transform Infrared Spectroscopy
I&C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Instrumentation and Control
IR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Insulation Resistance
NRC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nuclear Regulatory Commission
OIT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Oxidation Induction Time
OSTP . . . . . . . . . . . . . . . . . . . . . . . . . . . White House Office of Science and Technology Policy
RES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . U.S. NRC, Office of Nuclear Regulatory Research
TDR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Time Domain Reflectometry
XLPE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cross-Linked Polyethylene
xxxi
xxxii
LIST OF SYMBOLS
d . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Diameter of cable conductors (cm/in.)

l . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . total length of cable (meters/ft.)
la . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . length to start of defect (meters/ft.)
lb . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . length to end of defect (meters/ft.)
lc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . length to abraded portion of cable (meters/ft.)
ri . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Inside radius of coaxial cable insulation (cm/in.)
ro . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Outside radius of coaxial cable insulation (cm/in.)
r . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Radius of cable conductor (cm/in.)
s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Center-to-center distance of cable conductors (cm/in.)
x . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Distance along cable specimen (meters/feet)
C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Capacitance (Farads)
G . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conductance (mho)
I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Current (Amperes)
L . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Inductance (Henries)
Lk . . . . . . . . . . . . . . . . . . . Inductance for cable aged to simulate service temperature k (Henries)
R . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Resistance (Ohms)
Rs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Resistance of a single conductor (Ohms)
Rk . . . . . . . . . . . . . . . . . . . . Resistance of a cable aged to simulate service temperature k (Ohms)
V . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Voltage (Volts)
Z . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Impedance (Ohms)
Z0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Characteristic Impedance (Ohms)
* . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Difference between parameters

*s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Depth of conductor skin effect (cm/in)
g0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Dielectric constant of vacuum
g1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Real component of the dielectric function
g2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Imaginary component of the dielectric function
(i . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Imaginary component of the propagation function
(r . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Real component of the propagation function
H . . . . . . . . . Slope of linear function representing imaginary component of propagation function
J . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Time constant (seconds)
7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Structure factor
A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Second derivative of the impedance phase
M . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Impedance phase (Degrees)
Q . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Characteristic impedance phase (Degrees)
T . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Frequency (Hertz)
Tk . . . . . . . . . . . . . . . . . . . . . Frequency for cable aged to simulate service temperature k (Hertz)
xxxiii
xxxiv
1. INTRODUCTION
1.1 Background
Condition monitoring of electric cables is of particular interest for application to nuclear power
plants, in which miles of electric cables are used to perform numerous operational and safety
functions. Aging studies have shown that, as these cables age, there is the possibility that their
performance may degrade making them susceptible to failure (Lofaro, et al., 2002; Villaran and
Lofaro, 2003). Since replacement of all the cables would be expensive and impractical, an
effective technique to characterize their condition is highly desirable. To be useful, the
technique would have to be unobtrusive, non-destructive and able to be performed in situ.
Currently, there is no single in situ condition monitoring technique that has been proven to be
effective at determining the installed condition and the remaining useful service life of all the
various types of electric cables, particularly low-voltage instrumentation and control (I&C)
cables, in a typical nuclear power plant.
The issue of cable system safety was brought to the forefront with the issuance of the U.S. White
House Office of Science and Technology Policy (OSTP) Report containing a review of Federal
Programs related to the safety of installed electric cable systems, which was published in
November 2000. The report concluded that cable system safety is an important public health and
safety issue that transcends government agencies. One of the recommendations from that report
is to perform collaborative research to develop effective condition monitoring techniques.
In response to the White House OSTP report recommendation, the U.S. Nuclear Regulatory
Commission, Office of Nuclear Regulatory Research (NRC/RES) organized the International
Conference on Wire System Aging in April 2002 [Lofaro, November 2002] to provide a forum
for researchers from around the world to meet and discuss issues related to wire system aging,
including promising new condition monitoring techniques for cable systems. One of the
techniques discussed was broadband impedance spectroscopy.
Broadband impedance spectroscopy (BIS) is a technique being developed by the Boeing

Company, under the sponsorship of the U.S. Federal Aviation Administration (FAA) to monitor
the condition of wiring in aircraft. It utilizes impedance measurements of the wiring to obtain
information on the dielectric properties of the wire insulation. Impedance measurements are
made in situ at relatively low voltages and over a broad range of frequencies. Using algorithms
and computer models developed by Boeing, the dielectric properties are extracted from these
impedance measurements to characterize the condition of the wiring. The information extracted
can be used to identify and locate areas of degradation.
Based on the initial success of the FAA research, NRC/RES sponsored a research program to
evaluate the feasibility of using BIS to monitor nuclear power plant cables. While the
construction of nuclear plant cables is different than aircraft wiring, conceptually there are
similarities; therefore, condition monitoring techniques developed for aircraft may be suitable for
use in nuclear power plants. The feasibility study demonstrated that BIS was able to distinguish
1
between aged and un-aged nuclear plant electric cables. Also, the technique was able to locate
localized areas of degradation. Based on these results it was determined that BIS is a promising
new technique for application to nuclear plant cables and further research was initiated to
evaluate this technique in more detail. This report documents the results of this research.
In this report, an overview of the types of electric cables used in nuclear power plants is
presented. The theory behind BIS is then discussed, along with the procedures for performing
the test and evaluating the data. Finally, results from tests on actual nuclear plant cables are
presented, from which conclusions regarding the effectiveness of the technique are drawn.
Recommendations are made for future research, as well as demonstrations of the BIS method in
nuclear plant environments.
1.2 Technical Issues
There are several important technical issues related to condition monitoring of cables in nuclear
power plants that should to be addressed through research. These include the following:
• Detection of age-related degradation that could impact cable performance,

• Location of localized damage due to “hot-spots” along the cable route,
• Handling of connected loads to allow testing cables in situ without de-termination,
• Identification and location of localized damage, such as abrasion or cracking, and
• Handling of varying environments along a cable due to routing through different areas
and environmental zones of the plant
The first issue relates to the ability to determine the installed condition of a cable. Typically,
electric cables used in nuclear power plants perform numerous operational and safety functions
in the plant. The service environments to which the cables are exposed vary from benign indoor
conditions, to adverse localized environments, including high temperatures, humidity, radiation,
and pressures. Some safety-related cables are expected to operate under harsh environments and
accident conditions. These cables must be qualified through testing to provide assurance that
they can perform their safety function for the licensed life of the plant. Once qualified, the
cables are typically installed and receive minimal attention thereafter. For license renewal,
acceptable options are provided for demonstrating continued qualification for qualified cables.
Aging management programs are required for non-qualified cables that are in the scope of
license renewal.
As plants approach the end of their original licensed life, many utilities are applying for license
extensions to continue operation of the plant. The basis for the license renewal is that aging
degradation will be managed through the implementation of effective aging management
programs. For electric cables, current aging management programs rely on visual inspection of
cables to determine their condition and whether they are fit for continued service. While visual
2
inspections provide some assurance that the cables are not damaged, these inspections provide
very little information on sections of the cables that are not visible. For these hidden sections of
cable, techniques that can interrogate an entire cable system would be useful to detect
degradation that could potentially cause a loss of function.
The second technical issue that must be addressed is related to areas along the cable route that
expose the cable to adverse localized environments. Such areas can be caused by routing of a
cable near a hot steam pipe, or through a high radiation area. Installations that place the cables
in areas with localized adverse environments, or “hot-spots,” can result in localized damage to a
cable. Even though the majority of a cable may appear to be in good condition, a hot-spot can
cause localized degradation of the cable that can impair its performance. Therefore, a viable and
preferred condition monitoring technique must be able to locate localized damage caused by
these hot-spots.
Another important issue related to condition monitoring of cables is the ability to test the cables
in situ without disturbing them. Ideally, the cable would not have to be de-terminated from its
load for the testing process since this could introduce undesirable stresses on the cable and its
connectors due to repeated disconnecting and re-connecting of the terminations. Since each
cable will have a load attached, it is desirable that the condition monitoring technique be able to
correct for the impact of the load on any measurement data obtained from the cable.
Studies have shown that exposure to aging stressors can degrade the insulating materials
commonly used in the construction of electric cables. This degradation can be global in nature,
such as a reduction in ductility of the insulation, or it can be localized, such as cracking.
Mechanical damage has also been observed on installed cables, such as due to abrasion. This
localized damage can adversely impact the ability of a cable to perform its function, particularly
under abnormal operating or accident conditions when moisture is present. Therefore, it is
important that a condition monitoring technique be able to detect localized damage, such as
cracking and abrasion.
Finally, the cable runs in nuclear power plants can be very long, and one cable can traverse many
different areas of the plant. As such, the environment along the cable may vary. Therefore, it is
important that the condition monitoring technique be capable of accounting for this varying
environment along the cable.
Each of these technical issues was addressed in the research program discussed herein.
1.3 Objectives
3
The objectives of the research program reported herein were to evaluate the effectiveness and
sensitivity of the BIS technique with respect to the technical issues discussed above. To
maintain a manageable scope for this effort, the issues were reviewed and selected ones were
chosen to be addressed in this program based on their importance to demonstrating the
usefulness of the BIS method for nuclear plant cables. For this research effort the objectives
were to evaluate the BIS technique capabilities for the following:
• Detecting and locating areas of localized degradation (hot-spots) on nuclear plant

cables caused by stressors commonly found in nuclear plants.
• Detecting and locating localized damage on nuclear plant cables with a load
attached
• Detecting and locating localized damage on nuclear plant cables in the form of
abrasion and cracking
• Detecting and locating localized damage on nuclear plant cables with a varying
environment along the length of the cable
Additional issues not covered in this program are intended to be addressed in future research.
1.4 Research Approach
The research approach used to meet the project objectives involved the testing of cable samples
that were exposed to accelerated thermal aging to simulate degradation caused by exposure to
service conditions. A test matrix was developed that included tests of cable specimens to
address each of the issues identified above. The testing was divided into five sections; 1)
detection and location of thermal degradation (both global and localized), 2) detection of
degradation with attached loads on the cable, 3) detection of abrasion damage, 4) detection of
cracking damage, and 5) detection of degradation with varying environments along the cable. In
each case, appropriate test specimens were prepared and broadband impedance measurements
were made.
Analytical models were developed, the data were analyzed, and correlations were established.
The cable test specimens were prepared and baseline tested at BNL. Once preparation of the
cable test specimens was completed, the specimens were shipped to the Boeing test facilities in
Thousand Oaks, California. There, BIS testing of the specimens was performed and the data
obtained were analyzed. Measurements were made on un-aged cable specimens, and the data
were used to develop models that represent the impedance response of the un-damaged cables.
The models were then validated by comparison of the model predictions to the actual data
obtained. These models can be used to identify the level and location of degradation on other
cables constructed of similar materials and with similar conductor configurations. The theory
and process for developing and using the models is discussed in Section 3 of this report.
4
Once the models were completed, testing of the aged specimens was performed. The models
were then used to analyze the data obtained and predict the aged condition of the cables.
1.5 Equipment Used
The impedance measurements were made using a commercially available instrument, and an
instrument built for the FAA for BIS measurements. The former instrument was used for low-
frequency measurements ranging from 100Hz to 1MHz. The later instrument was used for the
high-frequency measurements ranging from 1MHz to 100MHz. Both instruments were
controlled by laptop computer using a LabVIEW® software graphical interface. Data were
downloaded and stored on the laptop for subsequent distribution and analysis.
5
(This page intentionally left blank)
6
2. REVIEW OF CONDITION MONITORING TECHNIQUES
In this section a discussion on what constitutes effective cable condition monitoring is presented,
along with a brief overview of several currently available condition monitoring techniques. The
techniques are categorized as mechanical, chemical, and electrical based on the properties they
monitor. The advantages and limitations of each technique are discussed. The purpose of this
discussion is to present the current state-of-the-art in cable monitoring at the time of this
research. This section is not intended to be a comprehensive listing of all condition monitoring
techniques. It is acknowledged that there are other techniques available, and significant
advancements are being made in the refinement of the existing technologies. However, thus far
these techniques are subject to similar limitations as the techniques discussed herein.
2.1 General Criteria for an Effective Technique
Condition monitoring of an electric cable involves the observation, measurement, or trending of

one or more indicators, which can be correlated to the condition or functional performance of the
cable. The intention is to provide information that can be used to determine the current ability of
a cable to perform within specified acceptance criteria, as well as to make predictions about its
remaining useful life. Ideally, the technique should have the following attributes:
• Non-destructive and non-intrusive (i.e., it does not require that the cable be disconnected
or disturbed),
• Capable of measuring property changes or indicators that are trendable and can be
consistently correlated to functional performance during normal service,
• Applicable to cable types and materials commonly used in existing nuclear power plants,
• Capable of being performed in situ under field conditions,
• Capable of detecting age-related degradation at levels that are low enough to allow
sufficient time to take corrective actions prior to cable failure,
• Capable of locating localized areas of degradation on a cable due to hot-spots, and
• Provides information that can be used to predict the remaining useful life of the cable.
In practice, there are no condition monitoring techniques currently available that have all of the
above attributes. Each has its own uses, as well as limitations that make it unsuitable as a
general tool for monitoring installed cables.
The following sections provide an overview of several currently available techniques. The
advantages of each technique are discussed, along with its limitations.
2.2 Mechanical Techniques
7
Techniques classified as mechanical involve the measurement or monitoring of some physical
attribute of the cable. Examples of mechanical techniques include visual inspections,
elongation-at-break, and compressive modulus measurement. Each of these techniques is
discussed below.
2.2.1 Visual Inspection
Visual inspection is an effective technique for providing a qualitative assessment of a cable’s

condition. While no quantitative data is obtained, the results can be trended to provide an
assessment of how fast a cable is degrading under the operating conditions to which it is
exposed. This technique can be used to evaluate the condition of a cable and determine if more
extensive testing is required to characterize its condition. Past studies have found that, in most
cases, cables that appear to be in good physical condition through visual inspection show
acceptable electrical performance.
The major advantage of the visual inspection is that it is inexpensive to perform and it does not
require any expensive equipment. It is recommended that a standardized procedure be
developed to ensure that a consistent inspection approach is used and that all of the important
cable attributes are inspected.
The most serious limitation of this technique is that the cable to be inspected must be accessible
and visible. In some cases, cables may be installed in closed conduits or buried beneath other
cables in a cable tray. In these cases, visual inspection would not be directly useful. Also, even
for visually accessible cables, usually only the jacket can seen. Therefore, inspection of the
insulation would probably not be possible. However, visual inspection of representative cables
that are accessible could be used to provide an indirect measure of the condition of the
inaccessible cables.
2.2.2 Elongation-At-Break
Elongation-at-break (EAB) is a measure of a material's resistance to fracture under an applied

tensile stress. It is often termed the "ductility" of a material and is defined as the percent
increase in elongation at the time of fracture. EAB is a valuable parameter for monitoring aging
in cable materials because of its sensitivity to microstructural changes in polymers brought about
by service aging. It is a widely accepted measure of polymer aging and is used as a reference in
evaluating other techniques.
In general, EAB measurements provide a useful quantitative assessment of the remaining

integrity of a cable. As the EAB decreases to a low value, and crack initiation and propagation
become possible from in-service stresses, moisture intrusion and current leakage could become
a problem. At this time there does not appear to be any commonly accepted criterion for the
minimum EAB of a cable material that will define the end of its useful service life. A
conservative value of $50 percent has been used as an acceptance criterion, even though there is
usually some useful service life remaining.
8
The major concern with EAB testing is that it is a destructive test, and relatively large amounts
of cable are required. They can only be obtained if cable is removed from service, or if
sacrificial cable samples are available for periodic evaluation.
2.2.3 Compressive Modulus
Compressive modulus is a material property defined as the ratio of compressive stress to

compressive strain below the proportional limit. As cable insulation and jacket materials age
they tend to harden, which will cause the compressive modulus of the materials to increase. By
monitoring this change in compressive modulus, an estimate of the degradation rate of the
material can be made. Compressive modulus is measured using a device developed by EPRI
known as the polymer indenter.
In many instances, cables are not easily accessible in nuclear plants. Often they may be stacked
in cable trays or run through conduits, which severely limits access to perform CM testing. For
cables which are accessible, the indenter can be used to determine the modulus of the outer
jackets. For those cables that are not accessible, it is not desirable nor possible, in some
instances, to excessively handle the cables to permit indenter testing. To monitor the modulus of
individual jackets or insulation, access would be desirable at a termination point. This is not the
optimum situation, since the termination points may be physically located in a different plant
location and exposed to very different ambient conditions than the remainder of the cable.
Measurement of compressive modulus is a non-destructive test that does not affect the continued
operation of the cable. The indenter is computer controlled and the software prevents the probe
from penetrating the material being tested, from leaving residual marks, or from causing any
other type of damage. It is suitable for use on various materials (XLPE, EPR, Hypalon® and
Neoprene). However, not every material produces significant age-induced changes in
compressive modulus that can easily be correlated with thermal or radiation exposure. For low
levels of aging, relatively small modulus changes can be seen, which, in the absence of baseline
measurements, might be difficult to correlate with aging. Cable construction and manufacture
were also found to produce different modulus results for the same aging. Compressive modulus
data correlate well with EAB measurements, which is an indication of this technique’s
usefulness.
2.3 Chemical Techniques
Techniques classified as chemical involve the measurement or monitoring of some attribute of

the cable’s insulation related to changes in its chemistry or molecular makeup. Examples of
chemical techniques include oxidation induction time (OIT) and Fourier Transform Infrared
Spectroscopy (FTIR). Each of these techniques is discussed below.
2.3.1 Oxidation Induction Time
9
Polymeric materials used to insulate cable conductors typically include an anti-oxidant in their
formulation to mitigate oxidation, which will degrade the polymer over time. As the cables age,
the anti-oxidant additives are gradually lost from the polymers due to diffusion and volatilization
from the surface. As the additives are depleted, oxidation will gradually increase.
Oxidation induction time is a technique for measuring the time required for a polymer to begin to
oxidize under controlled conditions. When compared to values from new cables, the OIT
provides an indication of the amount of aging on the cable. The process involves the controlled
heating of a small scraping of insulation material from the cable. As the sample is heated under
an oxygen environment, the amount of energy liberated is measured. An increase in this energy
indicates the onset of oxidation, which is an exothermic reaction. The time to oxidation onset is
measured and represents the oxidation induction time.
OIT has been shown to provide useful correlations to EAB values for some cable insulation
materials. However, access to the cable is required to obtain the test specimen. Also, the test
must be performed in a laboratory setting, and the results only provide data on a localized
portion of the cable.
2.3.2 Fourier Transform Infrared (FTIR) Spectroscopy
FTIR Spectroscopy is a technique for analyzing the structure of molecules. The principle
involves the measurement of absorbance or transmittance of infrared radiation by molecular
structures, including those for polymers. As the radiation passes through a polymer, atoms
absorb radiation and begin to vibrate. For a particular chemical bond, maximum vibration
occurs for a specific wavelength of radiation. Therefore, by irradiating a specimen with a
continuous spectrum of infrared radiation, and measuring the peaks (wavelengths) at which
maximum absorbance or transmittance occurs, the chemical bonds that are vibrating may be
identified from standard wavelengths that are available from the open literature. The presence of
certain molecular bonds that develop as a result of oxidation can be used as an indicator of aging.
One of the potential problems with the FTIR spectroscopy technique is that it is a surface
examination procedure in which the infrared radiation passes into the surface of the specimen
and is refracted back into the crystal. By analyzing the intensity of the incident and reflected
rays, the transmittance is determined. Since accelerated thermal aging at elevated temperatures
will produce an oxidation gradient at the specimen surface, the spectroscope will detect a higher
amount of oxidation that the average bulk value. The technique would, therefore, be expected to
give a more accurate estimate of bulk aging for naturally-aged specimens, for which oxidation
gradients will be minimized at the lower service aging temperatures.
2.4 Electrical Techniques
10
Techniques classified as electrical involve the measurement or monitoring of some electrical
property of the cable. Examples of electrical techniques include insulation resistance (IR),
dielectric loss, voltage withstand testing, time-domain reflectometry, and partial discharge
testing. Each of these techniques is discussed below.
2.4.1 Insulation Resistance
Insulation resistance measurements are commonly performed to determine the current condition
of cable insulation. By applying a voltage from the conductor to ground, the resistance of the
insulation separating them can be measured. The advantages of this test are that it is relatively
easy to perform and requires inexpensive equipment. If the insulation resistance decreases in a
predictable manner as the insulation ages, trending of this parameter could be useful as a
condition monitoring technique for electric cables.
When a dc voltage is applied to a test specimen, the total current flowing in the insulation from
the conductor to ground is equal to the sum of the capacitive charging current, leakage current,
and dielectric absorption current. These three component currents change with time. The
capacitive charging current and the dielectric absorption current will initially be relatively high
when the test voltage is first applied to the test specimen. Once the insulation, which behaves
like a capacitor, is energized and charges have aligned across the insulation, these currents will
taper off and eventually approach zero. However, leakage current will typically start at zero and
gradually increase. In high integrity insulation, leakage current will reach and maintain a steady
value after a certain amount of time. If the insulation is badly deteriorated, wet, or contaminated,
the leakage current will be greater than that found in good insulation and it could continue to
increase over time. As a result, the total current flowing in a test specimen will start out high
when a test voltage is first applied, and taper off in different ways over the next several minutes
depending on the condition of the insulation.
In high integrity insulation the insulation resistance will gradually increase after the test voltage
is applied, then a steady value will be reached. Because of this behavior, insulation resistance
measurements are taken using an ohmmeter first at one minute and again at ten minutes. The
ratio of the insulation resistance at ten minutes to the value measured at one minute is called the
polarization index.
One disadvantage of making insulation resistance measurements is that the cable under test must
be disconnected in order to attach the test instrument. However, this can be controlled by test
procedures with independent verification steps, as are commonly used for surveillance and
maintenance procedures in nuclear power plants.
The major factor affecting insulation resistance measurements is temperature. The effect of
temperature is predictable and can be corrected by normalizing the raw insulation resistance
readings to a common temperature, such as 16ºC (60ºF), for purposes of comparison. Using the
ratio of the 10 minute insulation resistance to the 1 minute reading, the polarization index can
provide a measure of dielectric condition that eliminates temperature effects. Consequently, in
11
order to collect data that are trendable, accurate time and temperature data must accompany the
raw quantitative insulation resistance measurements. For this reason, plant personnel often look
upon the insulation resistance test as a simple pass/fail test for the dielectric integrity of electrical
equipment and cables, but too irregular for trendable condition monitoring.
2.4.2 Dielectric Dissipation or Loss Factor
When a steady-state ac test voltage is applied to an insulated cable, the resulting apparent total
current that flows consists of a charging current due to the capacitance of the cable insulation
and a leakage current. The phase angle between the applied test voltage and the total current is
known as the dielectric phase angle. The complement of the phase angle is called the dielectric
loss angle.
The leakage current for electric cables is normally much smaller than the charging current, but it
is more sensitive to the condition of the insulation. As insulation deteriorates it is expected that
the leakage current will increase, while the capacitive current remains approximately constant.
Thus, the ratio of the magnitudes will increase. The ratio is called the dielectric dissipation
factor and is commonly used as a measure of insulation condition. Another means of describing
insulation condition is the dielectric power factor, expressed as the cosine of the dielectric phase
angle. At very low power factors (<10 percent), the dielectric power factor is approximately
equal to the dielectric dissipation factor.
Dielectric loss measurement is a very simple and straightforward condition monitoring

technique. Using a signal analyzer, an internal source provides an applied ac voltage signal to
the test specimen over a range frequencies, while measuring and recording the dielectric phase
angle at each increment. The phase angle data can then be used to calculate the dielectric loss or
power factor. Studies have shown that dielectric loss can be correlated to cable degradation.
Another advantage of this test is that it can be used to interrogate an entire cable system. The
data obtained provide information on the global condition of the cable, and can provide insights
into the condition of the entire cable system.
Some of the factors that affect the dielectric loss measurement technique include cable length,
humidity or moisture within the cable and insulation, and electrical equipment operating in the
vicinity of the test cable. The effect of length is very uniform and predictable, resulting in a
relative increase in insulation power factor as the length of cable increases. This effect is most
easily accounted for by making in situ baseline measurements for each cable to be monitored to
serve as a standard for comparison with similar measurements in the future. The effect of other
operating electrical equipment or energized cables in the same tray is typically concentrated at
the frequency of the operating equipment. In most cases, this is the 60Hz power frequency and it
has a more pronounced effect on longer cables than short ones. This problem can be avoided by
making measurements at an applied ac test voltage with a frequency below 50Hz or above 70Hz.
12
The major advantage of the dielectric loss technique is that the cable being tested does not have
to be completely accessible. The test equipment can be connected to the ends of the cable, and
the test can be performed without physically touching the length of the cable. Also, no material
samples are taken from the cable.
A disadvantage of the dielectric loss technique is that the cable under test must be disconnected
in order to attach the test instrument. However, this can be controlled by test procedures with
independent verification steps, as are commonly used for surveillance and maintenance
procedures in nuclear power plants.
2.4.3 Voltage Withstand Test
The high potential voltage withstand test is a measure of a cable insulation’s ability to withstand
a high applied voltage stress. To perform this test, the cable is connected to a source of high
voltage. The test voltage is applied to each individual conductor of the cable specimen and held
for five minutes, during which time the leakage current between the conductor and electrical
ground is recorded. The leakage current provides a measure of the insulation’s condition.
As with any electrical test, an advantage of this test is that it can interrogate an entire cable
system to identify weak points.
A disadvantage of this test is that, due to the high voltages applied, the test itself can degrade the
insulation. If this test is repeated a number of times, the dielectric strength may become so
weakened that the cable could fail because of the testing. As such, this test would not normally
be used in situ by nuclear power plant technicians to assess and trend the dielectric integrity of a
cable’s insulation system. Since it is a high voltage test, it has the potential to destroy a cable if
a voltage breakdown should occur. Even at lower levels of applied test voltage, within the
voltage rating of the cable, the cable insulation is incrementally weakened by each high potential
test. Thus, the risks of causing either catastrophic or incipient damage to cable insulation make
this an unsuitable condition monitoring method.
2.4.4 Time-Domain Reflectometry
Time-domain reflectometry (TDR) works on the same principle as radar. A pulse of low-voltage
energy is transmitted down a cable and the reflected waveform is captured on the instrument
scope. When the pulse reaches the end of the cable, or a fault along the cable, part or all of the
pulse energy is reflected back to the instrument. The TDR measures the time it takes for the
signal to travel down the cable and reflect back. The TDR then converts this time to distance
and displays the information as a waveform and/or distance reading. Knowing the length of the
cable, the reflected pulse can be interpreted as the end of the cable, or a fault in the cable.
The primary use of the TDR is for locating defects in a cable. Little information is provided on
the condition of un-faulted cable insulation.
13
2.4.5 Partial Discharge Testing
Another type of electrical test is the partial discharge test. When an insulation defect such as a
void is present, the defect will display a localized corona or ionization during exposure to a
high-voltage stress. As voltage builds up across a void, once the partial discharge inception
voltage is reached, the gas in the void will ionize and a discharge will occur. The discharge
redistributes electrical charge within the dielectric and is known as partial discharge.
The inception voltage for these voids tends to be constant, so that the measurement of total
charge being redistributed within the dielectric is believed to be a good indicator of the size of
the voids and their likelihood of becoming an incipient failure. Partial discharge testing requires
the measurement of small current pulses (<5 pC) while applying a relatively high test voltage.
2.5 Reasons to Focus on an Electrical Technique
From the previous overview of currently available condition monitoring techniques it can be
seen that there are advantages and limitations to each. Of the three categories of techniques
presented, the electrical techniques appear to have the highest potential of meeting the criteria
for an effective in situ technique discussed previously. Advantages of the electrical techniques
are the following:
• An entire cable system can be interrogated with one test to detect degradation,
• The cable does not have to be accessible along its entire length,
• Results can be correlated to other monitoring parameters, such as EAB,
• The data obtained can be used to locate localized defects,
• Hardware can be installed to provide continuous on-line monitoring without connecting
test equipment, and
• Material samples are not required to perform the test
• Loads at either end of an installed cable system can remain connected
• Non-intrusive, non-destructive voltage signals can be applied for examination
Based on these advantages, an electrical technique was focused on for this research. In
particular, broadband impedance spectroscopy was selected for evaluation.
14
3. THEORY OF BROADBAND IMPEDANCE SPECTROSCOPY
This section presents an overview of the theory behind broadband impedance spectroscopy
(BIS), and how impedance measurements are used to determine the electrical properties of
cables.
3.1 Equivalent Circuit
Over the past 70 years or more electric impedance models enabled the analysis of the dielectric
properties for numerous materials (for example, see Vaughan, 1969). Figure 2 depicts the
system being analyzed, which consists of two wires of length l with a voltage potential V(T )
across them.
Figure 2 Wire system equivalent circuit
The core concept of transmission theory is that a set of wires can be viewed as a distributed
system that is endowed with a frequency dependent resistance, R(ω) per meter, inductance, L(ω)
per meter, conductance, G(ω) per meter, and capacitance, C(ω) per meter. Note, the
conductance and capacitance are related to the wire insulation dielectric function, ε(ω) via
G (ω ) + i ω C(ω ) = iΛ ω ε (ω ) (3.1-1)
Here “i” represents the imaginary operator (square root of -1), and “Λ”is a structure factor that is
independent of frequency and is a real number. For example, for two identical wires whose
diameter is “d” and the center-to-center separation between the conductors is “s” the structure
factor would be:
15
π (3.1-2a)
Λ=
cosh − 1 s d ( )
If the system is a coaxial cable, with inner and outer radii r0 and r1, respectively, then
2π
Λ=
ln ⎛⎜ r0 r ⎞⎟
⎝ 1⎠ (3.1-2b)
If a voltage is established between the wires, then the system’s electrodynamics obey the
following equations (Durney and Johnson, 1969):
∂ ∂ (3.1-3)
V ( x, t ) = − R I ( x, t ) − L I ( x, t )
∂x ∂t
∂ ∂
I ( x, t ) = −G V ( x, t ) − C V ( x, t )
∂x ∂t (3.1-4)
Here, V(x, t) is the voltage established between the two wires at the space-time point (x, t) and
I(x, t) is the current flowing in the wires. The solutions of these equations define a set of normal
modes A(x, t)
[ ]
A( x , t ) = exp γ ± (ω ) x − i(ω t ) (3.1-5)
The propagation function is defined as follows:
γ ± (ω ) = ± [ R(ω ) + i ω L(ω )][G(ω ) + i ω C(ω )]

(3.1-6a)
= ±[α (ω ) + i β (ω )]
Here α(ω) is the dissipation (per meter) and 2π/β(ω) is the wavelength of the normal mode.
Furthermore, we can rewrite the propagation function as
16
γ ± (ω ) = ± [ R(ω ) + iω L(ω )] i Λ ε (ω ) (3.1-6b)
The impedance for open-circuited and short-circuited wires is given by the following equations:
Zopen (ω ) = Z0 (ω ) Coth[γ (ω ) l ] (3.1-7a)
Z short (ω ) = Z0 (ω ) Tanh[γ (ω ) l ] (3.1-7b)
Here the wire length is l, and Z0(ω), the characteristic impedance, is given by the following
equation:
R(ω ) + i ω L(ω )
Z0 (ω ) = (3.1-8)
G (ω ) + i ω C(ω )
3.2 Wire Properties and Wire Impedance Spectra
In this section, the approach for extracting wire properties from impedance data is presented. An
example is then given describing how this approach is used for wire prognostics by extracting
the wire properties from the impedance data of healthy wiring and from damaged wiring.
Specifically, the cases of damage due to (1) humidity and (2) abrasion are examined. The data
presented in this section were obtained from a research program sponsored by the FAA, as well
as Boeing's IR&D on aircraft wiring and are used herein for demonstration purposes.
3.2.1 Approach to Extract Wire Properties from Impedance Data
First, the wire properties are extracted from the measured broadband impedance spectra for the
following wiring configurations: open-circuited and short-circuited. The product of the
short-circuited and open-circuited impedances and their ratio yield the characteristic impedance
and the propagation function, as follows:
Zo2 (ω ) = Z short (ω ) Zopen (ω ) (3.2-1a)
Z short (ω )
γ (ω ) l = Tanh − 1
Zopen (ω )
(3.2-1b)
17
Using equations (3.1.6b) and (3.1.8) for the propagation function and characteristic impedance,
respectively, the above equations can be written as:
γ (ω ) Z0 (ω ) = R(ω ) + i ω L(ω ) (3.2-2a)
γ (ω )
= i Λ ω ε (ω )
Z0 (ω ) (3.2-2b)
Therefore, the wire properties per length are related to the characteristic impedance and the
propagation function via the following equations:
R(ω ) = Re [γ (ω ) Z0 (ω )] (3.2-3a)
1
L(ω ) = Im [γ (ω ) Z0 (ω )]
ω (3.2-3b)
⎡ γ (ω ) ⎤
ε1 (ω ) = Re ⎢ ⎥ (3.2-3c)
⎢⎣ ω Λ Z0 (ω ) ⎥⎦
⎡ γ (ω ) ⎤
ε 2 (ω ) = − Im ⎢ ⎥ (3.2-3d)
⎣ ω Λ Z0 (ω ) ⎦
Here, ε1(ω) and ε2(ω) are the real and imaginary portion of the dielectric function, respectively.
Before an example of a wiring system is presented, the behavior of the characteristic impedance
and propagation function will be discussed, as well as the open-circuited and short-circuited
impedance spectra. At low frequencies, the characteristic impedance is dominated by the wire’s
capacitance and therefore is large. Conversely, the propagation function is very small such that
γ(ω) l << 1. Thus, the open-circuited and short-circuited impedance spectra approach the
following limiting forms at low frequencies:
1
Zopen (ω ) →
i ω Λ l ε (ω ) (3.2-4a)
18
Z short (ω ) → l [ R(ω ) + i ω L(ω )] (3.2-4b)
Thus, at low frequencies the open-circuited wire impedance is dominated by the wire’s
capacitance. The magnitude of the open-circuited impedance depends nearly inversely on the
frequency and, therefore is large, whereas its phase is determined by the ratio ε2(ω)/ε1(ω). On
the other hand, at low frequencies the short-circuited impedance is independent of frequency
(ω → 0), is dominated by the wire’s resistance, and is small. At higher frequencies, the short-
circuited impedance exhibits a linear dependence on frequency with a slope determined by the
wire’s inductance. Thus, the low frequency portion of these spectra is completely dominated by
the wire’s electrical properties.
As an example, Figures 3 and 4 depict the log magnitude and phase of the impedance spectra at
low frequencies for a ten meter poly-x twisted pair wire that was open-circuited and short-
circuited. Here, the frequency ranges from 1kHz to 1MHz.
Open-Circuited Short-Circuited
1.5
Log10 |Z(ω)|
6 1.25
Log10 |Z(ω)|
1
5 0.75
4 0.5
0.25
3 4 5 6 3 4 5 6
Log10 ω Log10 ω
Figure 3 Low frequency impedance spectra log magnitude for a 10 meter poly-x
twisted pair wire
19
-86 80
Φ(ω)
Φ(ω)
-86.5 60
-87 40
-87.5 20
3 4 5 6
3 4 5 6
Log10 ω Log10 ω
Figure 4 Low frequency impedance spectra phase angle for a 10 meter poly-x
twisted pair wire
In the region where γ(ω) l > π, the impedance spectra exhibit oscillations with a period of
2π/[γ(ω) l]. Electrical waves now propagate on the wiring and the impedance is dominated by
wave interference. Furthermore, noting that ε1(ω) >> ε2(ω), the characteristic impedance and the
propagation function approach the following limits:
L(ω )
Z0 → (3.2-5a)
Λ ε1 (ω )
γ (ω ) → i ω Λ ε1 (ω ) L(ω )
(3.2-5b)
Examination of the open-circuited and short-circuited cable impedances, described by

Equations (3.1.7a) and (3.1.7b), reveal that the period of these oscillations is determined by the
inductance and the real component of the dielectric function. Information on the location of
defects, such as “hot-spots,” is contained in this portion of the impedance spectrum.
As an example, Figures 5 and 6 depict the impedance magnitude and phase spectra at high
frequencies for a ten meter poly-x twisted pair wire that was open-circuited and short-circuited.
Here the frequency ranged from 1MHz to 40MHz.
20
800 800
|Z(ω)|
|Z(ω)|
600 600
400 400
200 200
1´ 107 2´ 10 7 3´ 107 4´ 107 2´ 10 7 4´ 107 6´ 107 8´ 107 1´ 108
ω ω
Figure 5 High frequency impedance spectra magnitude for a 10 meter poly-x
twisted pair wire
Open Circuited Short Circuited
75 75
50 50
25 25
Φ(ω)
Φ(ω)
7 7 7 7
1´ 10 2´ 10 3´ 10 4´ 10 1´ 107 2´ 107 3´ 107 4´ 107
-25
ω -25
-50 -50 ω
-75 -75
Figure 6 High frequency impedance spectra phase angle for a 10 meter poly-x
twisted pair wire
3.2.2 Wire Properties from Impedance Data for Undamaged Wiring
The wire electrical properties can be extracted from impedance data using equations (3.2-3a)
through Eq. (3.2-3d). The frequency dependent resistance and inductance are shown in Figure 7.
Examination of the resistance, R(ω), shows that it increases significantly at frequencies over
50kHz due to the skin effect. Analysis shows that the resistance varies as ω1/2, as expected. The
skin effect manifests itself when the conductor’s skin depth, δs(ω), is on the order or less than the
radius of the conductor, r (here r~10-4m), i.e. δs(ω) < r. At that point, the electrical currents are
confined near the surface of the conductor within the skin depth. This occurs when ω ≈ 50kHz.
21
0.3 3.7 ´ 10-6
0.25 3.6 ´ 10-6

R(ω)
0.2 L(ω)
3.5 ´ 10 -6
-6
3.4 ´ 10
0.15
4.5 5 5.5 6
3 4 5 6
Log10ω Log10ω
Figure 7 Frequency-dependent wire resistance and inductance
The inductance also exhibits frequency dependence. However, L(ω) does not follow the usual
skin effect frequency dependence due to the wire’s complex microstructure, depicted in Figure 8.
Specifically, there are nineteen strands, all of which conduct current, and their associated
magnetic fields interfere with each other. Furthermore, their individual radii are over twenty
times smaller than the wire conductor itself. Thus, the skin depth is on the order of 100µ−300µ.
As a result, the frequency dependence is very weak. Finally, the low frequency behavior
(100Hz-1kHz) of the inductance (shown in Figure 7) is probably not valid. In particular, the
contribution of the inductance to the wire’s impedance is very small compared to that of the
resistance and dielectric function. As a result, the low-frequency portion of the inductance
(100Hz-1kHz) suffers from noise.
Figure 8 Micro picture of a poly-x wire

cross section
Figure 9 depicts the real and imaginary components of the wire’s dielectric function (times the
structure factor, Λ) in the frequency range from 100kHz to 1MHz. Examination of this figure
22
reveals that the real component is about 500-1000 times larger than the imaginary component.
Note that both components exhibit significant frequency dependence in the vicinity of 100Hz.
These structures arise from a slow, Debye ion process on the order of 100ms, in which ions
move in response to the ac electric fields associated with the probe voltage, V(ω). Furthermore,
there is a broad background that arises from very fast electronic processes, on the order of 0.1µs.
Analysis of the dielectric data reveals that this process is nearly independent of frequency and is
the largest contribution to the dielectric function. This electronic dielectric process follows the
fraction power law, in which the following relationship applies:
A1 ε 0
ε electronic (ω ) ≈
[
1 + (i ω τ ) p ] (3.2-6)
Here A1 is a fitting constant representing the strength of the process with a typical value ranging
from 2 to 4, ε0 is the dielectric constant of vacuum, τ is the time scale of the electronic dielectric
process, and p is a fitting constant ≈ 0.05. Note that as the frequency increases, the wire
insulation becomes more lossy. Finally, there is a small structure in the vicinity of 30kHz that
mirrors another process that is midway between the electronic and ion dielectric processes. The
origin of this process is unclear.
-13
1.6 ´ 10
-11
ε (ω)
4.835 ´ 10 1.4 ´ 10-13

εg22((ω)
4.83 ´ 10-11
1.2 ´ 10-13
T)
g1(T1)
1
4.825 ´ 10-11 -13

1´ 10
-11
4.82 ´ 10
8´ 10-14
-11
4.815 ´ 10
6´ 10-14
3.5 4.5 5 5.5 6
3.5 4.5 5 5.5 6
Log10ω Log10ω
Figure 9 Real and imaginary components of the wire insulation dielectric
function (open circuited)
3.2.3 Wire Properties from Impedance Data for Damaged Wiring
An important element of a successful cable condition monitoring technique is its ability to detect
damage on the cable due to exposure to a stressor. As part of the initial feasibility study for this
project, impedance data were evaluated for wiring that was damaged by exposure to elevated
humidity levels. Humidity was chosen as the damage stressor since models were already
available for this damage mechanism from previous research on aircraft wires (Rogovin and
Kendig, 2002).
23
If the wiring is damaged by exposure to elevated levels of humidity, corrosion and other new
microscopic processes may occur that are reflected in the wire’s electrical properties. As an
example, the case of poly-x twisted pair wiring exposed to humidity is examined.
Figure 10 depicts the log magnitude and phase of the impedance over the low frequency portion
(100Hz-1MHz) of the spectra for the cases in which the wires are open-circuited. Three
conditions are evaluated: (1) healthy wiring, (2) wiring exposed to 85%/85°F for 24 hours, and
(3) wiring exposed to 85%/85°F for 168 hours.
1 Healthy wiring
2 24 hours in humidity
-76
2
Log10|Z(ω)|
6 1 -78
-80 3
5
2 Φ(ω)
-82
-84
4
3 -86
3 4 5 6
-88 1
3 4 5 6
Log10ω
Log10ω
Figure 10 Log magnitude and phase of the impedance spectra for moist and
dry wiring (open circuited)
Examination of Figure 10 reveals that humidity changes the impedance spectra significantly.
Specifically, at low frequencies the impedance phase for the healthy wire Φdry(ω) ≈-90°, which
implies that little energy is dissipated in the wiring. When the wiring is exposed to humidity for
a period of time, the impedance phase migrates to more dissipative states where its value varies
from -85° < Φhumidity (ω) < -75°. Furthermore, the magnitude of the impedance is reduced
drastically. This is shown in Figure 11 where the ratio of the magnitude of the impedances are
plotted as a function of frequency. Note that the longer the wiring is exposed to humidity the
more conductive it becomes. Clearly, the wire’s dielectric properties can be seriously
compromised by humidity.
24
0.5
2 24 hours in humidity 2
0.45
|Zhumidity(ω)|/ |Zdry(ω)|
0.4
0.35
0.3 3
0.25
3 4 5 6
0.15 Log10ω
Figure 11 Ratio of the impedance spectra magnitude for moist and dry wiring
(open circuited)
Next, the electrical properties for dry and moist wiring are compared. Figure 12 depicts the
resistance and inductance of dry and moist wiring in the frequency range from 100Hz to 1MHz.
1 Healthy wiring
0.24
0.22
5´ 10-6 2
R(ω)
L(ω)
0.2
0.18 4.5 ´ 10
-6
0.16
3
0.14 2 3 3 4 5 6
0.12 3.5 ´ 10-6
1 1
3 4 5 6
Log10ω
Log10ω
Figure 12 Ratio of the resistance and inductance for moist and dry wiring as
a function of frequency (R is S/m and L is H/m - open circuited)
25
Referring to Figure 11, it should be noted that both of these electric properties are changed by
humidity. For example, the resistance of moist wiring is about 20% larger that dry wiring. To
understand this, the following model describing the resistance for a pair of twisted wires is
examined:
⎡ s ⎤
2 Rs (ω ) ⎢ ⎥
R(ω ) = d
πd ⎢ ( s ) − 1 ⎥⎥
2 (3.2-7)
⎣⎢ d ⎦
Here “s” is the distance between the wire centers, Rs(ω) is the resistance for a single wire
(s → ∞) and “d” is the diameter of the wire insulation. Note that equation (3.2-7) demands that
the resistance will increase as s/d increases. Since humidity swells the wire’s insulation, the
resistance will increase, as shown in Figure 12.
Figure 13 compares the real and imaginary components of the wire insulation’s dielectric
function. Note that the imaginary component increases by a factor of several hundred to over one
thousand with exposure to humidity. This behavior implies that new and significant microscopic
processes occur. These processes occur throughout the spectrum, implying both ionic and
electronic dielectric processes. Note that the real component of the dielectric function has
increased as well.
More important, these processes are different than those that occur in a dry wire. This can be
seen by examining the ratio of the imaginary to the real component of the dielectric function,
ρ = ε 2 (ω ) / ε1 (ω ) (3.2-8)
Examination of Figure 13 reveals that the ratio increases from 10-3 for dry wire to the range of
0.15 < ρ(ω) < 0.25 for wire exposed to humidity. Clearly, the exponent (p), response time (τ) and
the strength (A1) in the fraction power law, as represented by Eq. (3.2-6), have changed.
Furthermore, the imaginary component of the dielectric function shows structure near ω ≈
104Hz. Note that the longer the wiring is exposed to humidity, the further this peak migrates to
higher frequencies.
26
0.25
2
0.2
ρ(ω) 3
0.15 Healthy wiring
24 hours in humidity
0.1
168 hours in healthy
0.05
1
3 4 5 6
1 Healthy wiring Log10ω
3 168 hours humidity
2 24 hours of humidity
Figure 13 Ratio of the real and imaginary components of the wiring
insulation’s dielectric function for moist and dry wiring as a function of
frequency
3.3 BIS Test Procedure
The process for using the BIS technique to evaluate a cable’s condition involves two steps;
1) performing impedance measurements on the cable, and 2) analysis and interpretation of the
impedance data obtained using analytical models. Each of these steps is discussed in the
following sections.
3.3.1 Procedure for Making Impedance Measurements
Impedance measurements were made on the cable being evaluated over a broad range of
frequencies in the open circuited configuration. In this study, frequencies ranging from 100 Hz
to 100 MHz were used. The measurements were broken up into two frequency ranges; low
frequency, which included frequencies from 100Hz to 1MHz, and high frequencies, which
included frequencies from 1MHz to 100MHz. The use of two frequency ranges was determined
to be the best approach for taking impedance measurements due to the limitations of the test
equipment available. In the low frequency range, a Hewlett-Packard instrument was used since
it provided better impedance data accuracy in this range of frequencies. However, at higher
frequencies, the impedance data obtained from the HP instrument were “noisy” and lacked
adequate resolution. Therefore, in the high frequency range, an Eclypse instrument was used
27
that was specially designed to provide accurate impedance measurements at high frequencies.
Figure 14 shows the Eclypse instrument (top) and the HP instrument (bottom) used to perform
the impedance measurements.
Figure 14 Equipment and test setup

for making impedance measurements
For the laboratory measurements, the cable being evaluated was prepared by stripping
approximately ½ inch of insulation from the ends of each conductor. The conductors were then
connected to terminals on the instrument being used for the impedance measurement, as
discussed above. The actual impedance measurements were performed automatically by a laptop
computer based software program, which controlled the actual measurements and downloading
of the data obtained. The measurements involved the input of a sinusoidal voltage signal into the
cable with an amplitude of 1-volt. The frequency of the input signal was varied over the desired
range of measurement. Data points were obtained at 2000 discrete frequencies in each
measurement range, and included the impedance magnitude and phase angle. The data points
were downloaded to the computer for subsequent retrieval and analysis.
3.3.2 Analysis and Interpretation of Impedance Data
After the impedance measurements were obtained, the data were analyzed using analytical
models. Different models were developed for each specific type of degradation being evaluated,
and were used to interpret the data for purposes of detecting and locating degradation on the
cable.
28
In practice, the application of BIS would involve starting with a known condition of the cable in
an un-degraded (baseline) state. The impedance measurements of the cable after being degraded
would be compared to the baseline condition. An appropriate model would then be selected to
explain any differences in the cable’s impedance. Based on the model that best fits the data, the
type of degradation would be identified.
The development of analytical models for various types of cable degradation was one of the
main objectives of this research effort, and is discussed in the following sections of this report.
It should be noted that only one cable type was evaluated in this study. Therefore, the models
developed may not be applicable to other types of cables. The suitability of the models
developed in this study to other cable types is the subject of a future research program.
29
30
4. PREPARATION AND BASELINE TESTING OF CABLE TEST SPECIMENS
This section of the report discusses the process used to prepare the cable test specimens that were
used in this study.
4.1 Selection of Cable Types
A number of different types of cables are used in nuclear power plants, and a great deal of
knowledge is available concerning their construction and performance (see for example Thue,
1999; Moore, 1997; and Grayson, 2000). For this research, cable samples were prepared using
Rockbestos Firewall® III, 2/C #12 AWG instrumentation and control cable, which is a
commonly used type of cable in nuclear power plants. The cables had a cross-linked
polyethylene (XLPE) insulation on each conductor, and an overall Neoprene jacket covering the
bundle of insulated conductors. The testing was limited to one type of cable to allow a full range
of testing to be performed within the budget and schedule allocated for this study. Future
research is desirable to investigate the effectiveness of the BIS technique on other types of
cables.
4.2 Preparation of Test Samples
The test specimens were prepared using a single reel of cable stored at BNL. The cable reel was
obtained from the warehouse stock of a decommissioned nuclear power plant as part of a
previous research program for the NRC. A single reel of cable was used as the source of all of
the test specimens for this study to avoid differences in cable properties that can occur between
different batches of cable.
Lengths of cable were cut from the reel to prepare the test specimens. The test specimens ranged
in length from 10 meters (32.8 feet) to 100 meters (328 feet). Approximately 15 cm (6 inches) of
the outer jacket was stripped from each end of the test specimen, along with approximately
2.5 cm (1 inch) of insulation from each conductor end.
4.3 Baseline Testing of Test Samples
Prior to aging, each test specimen was tested to determine its baseline performance. The tests
used were insulation resistance and dielectric loss. These tests were chosen since they are well
established, traditional tests that can be used for comparison purposes with the BIS test data. The
tests were repeated after aging of the test specimens was completed.
31
4.4 Accelerated Aging of Test Samples
Thermal aging was applied to the cable test specimens using convection ovens at BNL to
simulate exposure to a service temperature of 50ºC (122ºF) for durations of 20 years, 40 years,
and 60 years. The aging duration was determined using the Arrhenius equation with an
activation energy of 1.33 eV, which is the value used in the original qualification tests for these
cables. The thermal aging was applied globally to the entire length of the test specimen. Cable
specimens with no thermal aging were also tested to provide baseline data on the cable
performance.
In addition to global aging, selected specimens received thermal aging over a portion of their
length to simulate exposure to hot-spots at predetermined locations on the cable. The hot-spot
lengths simulated were 0.5 meters (1.64 feet) and 1.0 meters (3.3 feet) long. The hot-spots were
simulated by inserting only a portion of the test specimen in the aging chamber. Figure 15
shows a typical aging chamber loading used to perform the global and hot-spot aging. Table 1
provides a matrix of the test specimens prepared for this study.
Figure 15 Typical aging chamber loading for global and hot-spot

aging of cable test specimens
32
Table 1 Cable Test Specimens for Broadband Impedance Spectroscopy Research
Cable Type = Rockbestos I&C, 2 Conductor, #12 AWG with XLPE Insulation and Neoprene Outer Jacket
Global Accelerated Thermal Aging Applied 1 Localized Accelerated Thermal Aging Applied 1
Length
Specimen
No. meters
ID No. Service Life Hot-Spot Hot-Spot
(inches) Aging Parameters Aging Parameters 2 Size
Simulated Simulated Location 3
(hours @ Temp) (hours @ Temp.) (meters)
(years @ Temp.) (years @ Temp.) (meters)
1. PNI-79-RB-188-801 1.0 (39.4) None None None - None -
2, PNI-79-RB-188-802 2.0 (78.7) None None None - None -
3, PNI-79-RB-188-803 9.0 (354.3) None None None - None -
4. PNI-79-RB-188-804 1.0 (39.4) 31.6 @121ºC (250ºF) 20 @ 50ºC (122ºF) None - None -
33

Table 1 Cable Test Specimens for Broadband Impedance Spectroscopy Research (continued)
Length
Specimen
No. meters
ID No. Service Life Size Hot-Spot Hot-Spot
(inches) Aging Parameters Aging Parameters 2
Simulated meters Simulated Location 3
(hours @ Temp) (hours @ Temp.)
(years @ Temp.) (inches) (years @ Temp.) (meters)
17, PNI-79-RB-188-817 10.0 (393.7) 31.6 @121ºC (250ºF) 20 @ 50ºC (122ºF) 6.96 @150ºC (302ºF) 1.0 (39.4) 20 @ 60ºC (140ºF) 8.0 - 9.0
18, PNI-79-RB-188-818 10.0 (393.7) 31.6 @121ºC (250ºF) 20 @ 50ºC (122ºF) 6.96 @150ºC (302ºF) 0.5 (19.7) 20 @ 60ºC (140ºF) 8.0 - 8.5
19. PNI-79-RB-188-819 10.0 (393.7) 31.6 @121ºC (250ºF) 20 @ 50ºC (122ºF) 33.1 @150ºC (302ºF) 1.0 (39.4) 20 @ 70ºC (158ºF) 8.0 - 9.0
22. PNI-79-RB-188-822 10.0 (393.7) 63.1 @121ºC (250ºF) 40 @ 50ºC (122ºF) 13.9 @150ºC (302ºF) 1.0 (39.4) 40 @ 60ºC (140ºF) 1.0 -2.0
34
33. PNI-79-RB-188-833 10.0 (393.7) 63.1 @121ºC (250ºF) 40 @ 50ºC (122ºF) 837.6 @150ºC (302ºF) 1.0 (39.4) 40 @ 90ºC (194ºF) 1.0 -2.0
34. PNI-79-RB-188-834 10.0 (393.7) 63.1 @121ºC (250ºF) 40 @ 50ºC (122ºF) 837.6 @150ºC (302ºF) 0.5 (197) 40 @ 90ºC (194ºF) 1.5 -2.0
Table 1 Cable Test Specimens for Broadband Impedance Spectroscopy Research (continued)
Length
Specimen
No. meters
ID No. Service Life Size Hot-Spot Hot-Spot
(inches) Aging Parameters Aging Parameters 2
Simulated meters Simulated Location 3
(hours @ Temp) (hours @ Temp.)
(years @ Temp.) (inches) (years @ Temp.) (meters)
35
44. PNI-79-RB-188-847 100.0 (3937) None - 1256.4 @150ºC (302ºF) 1.0 (39.4) 60 @ 90ºC (194ºF) 96.0 -97.0
45. PNI-79-RB-188-848 100.0 (3937) None - None - None -

1. Calculations based on activation energy of 1.33 eV.
2. Hot-spot aging time is in addition to global aging time.
3. Measured from right end of cable, as determined by orienting wording on cable so it can be read normally from left to right.
36
5. RESULTS OF BIS TESTING OF CABLE SAMPLES
This section presents the results of the BIS tests performed on the cable test specimens with
various types of degradation. The models developed for each type of degradation are presented
and, comparisons with actual test data are made.
5.1 Application of BIS to Thermally Aged Cables
The initial research to evaluate BIS was focused on the detection of degradation on cables that
had been exposed to thermal aging. This is a fundamental requirement of an effective cable
condition monitoring technique for nuclear plant cables since thermal aging is one of the
predominant stressors in such plants. Section 5.1.1 presents the results of tests on cables that
received global thermal aging to simulate various service conditions. These cables were tested
to characterize their impedance spectra, and demonstrate the capability of the BIS method for
detecting thermal degradation. Subsequently, models were developed to represent the measured
impedance spectra and the electrical properties of the cables.
In Section 5.1.2, the results from tests on cables with localized thermal aging, or hot-spots, is
presented. These tests were performed to determine if the BIS technique is capable of locating
localized degradation on a cable. The models and process used to locate degradation are
presented and discussed.
5.1.1 Detection of Degradation on Cables with Global Thermal Aging
The BIS technique was evaluated for its ability to detect global thermal aging on cables using the
test specimens shown in Table 2. These cables were aged to simulate 20, 40 and 60 years of
service at service temperatures of 50ºC (122ºF), 60ºC (140ºF), and 70ºC (158ºF). For each test
specimen, broadband impedance measurements were made over the low frequency range (100
Hz to 1 MHz) and over the high frequency range (1 MHz to 100 MHz).
Figure 16 shows the measured impedance phase spectra in the low frequency range for the cables
aged to simulate 60 years of service (Specimens 26, 41, and 42). In performing these
measurements, the cables were placed in a humidity chamber, which was maintained at a
constant temperature of 25º (77ºF) and a pressure of 1 atmosphere to minimize atmospheric
effects on the measurements. The cables were tested with the conductors connected to the test
instrument on one end, and “open” (unconnected) at the opposite end.
Referring to Figure 16, it is seen that as the amount of degradation on the cable increases, which
is represented by the higher simulated service temperature, the greater the cable’s impedance
phase shifts away from -90 degrees. This suggests that as the degradation on the cable increases,
the cable becomes more dissipative, which is expected. Similar results were obtained for cables
aged to simulate 20 years and 40 years of service (Appendix A).
37
Table 2 Specimens tested to evaluate BIS for detecting global thermal aging
Length
No. Specimen ID meters Service Life Simulated Hot-Spot Simulated
(inches)
16. PNI-79-RB-188-816 10 (393.7) 20 @ 50ºC (122ºF) None
37. PNI-79-RB-188-837 10 (393.7) 20 @ 60ºC (140ºF) None
38. PNI-79-RB-188-838 10 (393.7) 20 @ 70ºC (158ºF) None
21. PNI-79-RB-188-821 10 (393.7) 40 @ 50ºC (122ºF) None
39. PNI-79-RB-188-839 10 (393.7) 40 @ 60ºC (140ºF) None
40. PNI-79-RB-188-840 10 (393.7) 40 @ 70ºC (158ºF) None
26. PNI-79-RB-188-826 10 (393.7) 60 @ 50ºC (122ºF) None
41. PNI-79-RB-188-841 10 (393.7) 60 @ 60ºC (140ºF) None
42. PNI-79-RB-188-842 10 (393.7) 60 @ 70ºC (158ºF) None
A
Φ(ω) [Deg.] 50 ºC (122 ºF)
B 60 ºC (140 ºF)
-88.6 C 70 ºC (158 ºF)
C
-88.8
B A
ω [Hz]
200000 400000 600000 800000 1x106
-89.2
-89.4
Figure 16 Measured low frequency impedance phase spectra for cables with
global thermal aging to simulate 60 years of service at various temperatures
38
Figure 17 shows the impedance phase spectra for these cables in the high frequency range.
Examination of this figure reveals that the impedance phase for these cables differs significantly
as degradation increases, particularly for frequencies above 60MHz where large differences in
the zero crossings of the impedance phase spectra are seen. An expanded view for the frequency
range from 78 MHz to 86 MHz is shown in Figure 18 for both the impedance phase and
magnitude. The shift in both phase and magnitude to lower frequencies with increasing
degradation is readily observed. Similar results were obtained for cables aged to simulate 20
years and 40 years of service (Appendix A).
A
Φ(ω) [Deg.] 50 ºC (122 ºF)
B 60 ºC (140 ºF)
75
C 70 ºC (158 ºF)
50
C A
B
25
ω [Hz]
7 7 7 7
2x10 4x10 6x10 8x10 1x108
-25
-50
-75
Figure 17 Measured high frequency impedance phase spectra for cables with
global thermal aging to simulate 60 years of service at various temperatures
These results demonstrate that thermal degradation has an impact on a cable’s electrical
impedance spectra. By measuring the impedance over a range of frequencies and comparing the
impedance spectra to a known spectra for an undamaged cable, changes in the impedance phase
and magnitude can be observed and used as an indicator of cable degradation.
In the next section, models are developed that can be used to describe the changes in the
impedance spectra. These models can be used to characterize a cable’s electrical properties after
being exposed to global thermal degradation. The models can also be used to locate localized
areas of degradation on the cable, such as would be caused by environmental “hot-spots” along
the cable.
39
Φ(ω) [Deg.] |Z(ω)|/l [S/m]
300
C 40 A
B 250
C
20 A 200 B
ω [Hz] 150
7 7
7.8x10 8.2x10 8.6x107
100
-20
ω [Hz]
7 7
-40 7.8x10 8.2x10 8.6x107
A 50 ºC (122 ºF)
B 60 ºC (140 ºF)
C 70 ºC (158 ºF)
Figure 18 Expanded view of impedance phase and magnitude per unit length
spectra over the frequency range from 78 MHz to 86 MHz for cables with global
thermal aging to simulate 60 years of service at various service temperatures
5.1.2 Development of Models for Cables with Thermal Degradation
Analytical models are used to determine the electrical properties of a cable, thereby providing an
assessment of the cable’s health. In addition, the models are required to determine the location
of localized degradation on a cable. The development of analytical models for cables exposed to
thermal aging is discussed in this section.
In order to develop an analytical model, it is important to understand the basic approach to

modeling a cable. If a cable is exposed to an aging stressor uniformly along its length, it will be
in a “single state” and can be represented by two frequency dependent functions; the
characteristic impedance, Z0 (T), and the propagation function, ( (T). However, if a cable is
exposed to different levels of an aging stressor along its length, it will have sections with
different amounts of aging degradation and the cable will be in multiple states. In this case, each
section will have its own characteristic impedance and propagation function, and the cable must
be modeled as a complex combination of these parameters.
At low frequencies the cable’s impedance is non-oscillatory and, as discussed in Section 3.2.1,
the characteristic impedance and propagation function can be directly extracted from the
measured impedance spectra using equations 3.2-1a and 3.2-1b. At high frequencies, where the
cable’s impedance oscillates, it is not possible to directly extract the characteristic impedance
and propagation function. Instead, one of the following two approaches can be used:
40
• Extrapolative approach, or
• Physical modeling approach
The extrapolative approach is used here to model cables with thermally induced aging
degradation. In this approach, the function representing the cable’s characteristic impedance in
the low frequency range is first determined using equation 3.2-1a. This function is then
extrapolated to high frequencies ranging from 1MHz to 15MHz. By substituting the
extrapolated characteristic impedance into equation 3.1-7a, the propagation function for the high
frequency range can then be determined using the following equation:
⎡ Zopen (ω ) ⎤ (5.1-1)
γ (ω ) = Coth − 1 ⎢ ⎥
⎢⎣ Z0, extrapolated (ω ) ⎥⎦
where Zopen is the measured cable impedance in the open configuration, and Z0, extrapolated (T) is the
extrapolated characteristic impedance for the cable.
5.1.2.1 Extraction of Cable Properties from Measured Impedance Spectra
As discussed in Section 3.2.1, a cable’s characteristic impedance and propagation function can
be extracted from measured impedance data. The cable’s impedance is measured in the open
configuration and the shorted configuration, then equations 3.2-1a and 3.2-1b can be used to
obtain the cable’s characteristic impedance and propagation function . Once these functions are
known, the cable’s electrical properties can be determined using equations 3.2-3a to 3.2-3d. This
was performed for the thermally aged cable test specimens in Table 2. Results for the cables
aged to simulate 60 years of service are presented below. Similar results were obtained for the
remaining cables in Table 2 and are included in Appendix A.
In Figure 19, the magnitude,*Z0(T)*, and phase, Q(T), of the characteristic impedance spectra
are presented for the cables thermally aged to simulate 60 years of service (Specimens 26, 41,
and 42). Examination of this figure shows that the thermal aging has only a small effect on the
cable’s characteristic impedance, as evidenced by the changes in magnitude of less than 5%.
Figure 20 compares the real component, (r, and imaginary component, (i, of the propagation
function spectra for the cables aged to simulate 60 years of service. As shown, the real
component of the propagation function increases as the aging degradation increases (higher
simulated service temperature). This reflects the fact that the thermal aging process produces
chemical changes in the cable’s insulation.
41
A 50 ºC (122 ºF)
|Z0(ω)|/l [S/m] B 60 ºC (140 ºF) Ψ(ω)
112 C 70 ºC (158 ºF) C
-1 A
110 B
-2
108 B
106 A -3
104 C
4.6 4.8 5.2 5.4 5.6 5.8 6
4.6 4.8 5.2 5.4 5.6 5.8 6
Log10(ω) Log10(ω)
Figure 19 Characteristic impedance magnitude per unit length and phase spectra for
cables with global thermal aging to simulate 60 years of service at various temperatures
γ i(ω) γr(ω)
0.035 Log10(ω)
0.030
B 4.6 4.8 5.2 5.4 5.6 5.8 6
0.025 0.0008
A A
0.020 C C
0.0006
0.015
B
4.6 4.8 5.2 5.4 5.6 5.8 6 0.0004
A 50 ºC (122 ºF)
Log10(ω) 0.0002
B 60 ºC (140 ºF)
C 70 ºC (158 ºF)
Figure 20 Imaginary component and real component of propagation function for cables
with global thermal aging to simulate 60 years of service at various temperatures
With the characteristic impedance and propagation function now known, the cable’s electrical
properties can be determined. First, equation 3.2-3a is used to obtain the resistance of the cables,
as shown in Figure 21. The left-hand side of the figure shows the resistance per unit length of
the cable in ohms/meter, while the right-hand side shows the percent change in resistance for an
increase in service temperature from 50ºC to 60ºC, and from 50ºC to 70ºC. The percent change
was calculated using the following equation:
42
Rk (ω ) − R50 (ω ) 5.1-2
δRk (ω ) =
R50 (ω )
where Rk(T) is the resistance for service temperatures of 60ºC (140ºF) (k=60) and 70ºC (158ºF)
(k=70), and R50(T) is the resistance for a service temperature of 50ºC (122ºF).
R(ω)/l [S/m]
δR(ω)/l [S/m]
0.14 B
0.12 A C 0.01
ω [Hz]
ω
0.10
0.08 200000 600000 1x106
0.06
-0.01
0.04 E
0.02 ω [Hz] -0.02 D
200000 600000 6
1x10
A 50 ºC (122 ºF)
D 50 to 60 ºC
B 60 ºC (140 ºF)
E 50 to 70 ºC
C 70 ºC (158 ºF)
Figure 21 Resistance per unit length and change in resistance per unit length for cables
with global thermal aging to simulate 60 years of service at various temperatures
Referring to the left-hand side of Figure 21, it is seen that an increase in thermal aging (i.e.,
higher service temperature) results in a small but perceptible decrease in the cable’s resistance.
This implies that shrinkage of the insulation is greatest with the higher simulated service
temperature. The right-hand side of Figure 21 shows that the resistance decreases as service
temperature increases, which results in a negative percent change. Also, the percent change in
resistance increases (i.e., becomes more negative) at the higher frequencies.
Using the extracted resistance, an analytical model can now be developed to represent this
resistance spectra. The analytical model is based upon the knowledge that resistance is primarily
governed by the skin effect, which is a function of T½. Therefore, an equation of the following
form was fitted to the resistance spectra presented in Figure 21:
⎛ ω ⎞
Rk (ω ) = R0,k ⎜1 + ⎟ 5.1-3
⎝ ωk ⎠
where Rk(T) is the resistance for a cable aged to simulate service temperature k, Tk is the skin
43
frequency for a cable aged to simulate service temperature k, and R0,k is an adjustable constant.
The values for the skin frequency and R0,k for the cables evaluated are shown in Table 3.
Table 3 Parameters for analytical model of resistance for cables thermally aged to simulate 60
years of service at various service temperatures 1
Simulated Service
Skin Frequency (Tk) Constant (R0,k)
Temperature
50 ºC (122 ºF) 12,589.3 sec-1 0.0150 S/m
60 ºC (140 ºF) 12,589.3 sec-1 0.0143 S/m
70 ºC (158 ºF) 14,125.4 sec-1 0.0150 S/m
1. Parameters determined by fitting equation 5.1-3 to resistance data in Figure 20
The cable resistance determined using the analytical model is compared to the extracted cable
resistance in Figure 22 for the cable thermally aged to simulate 60 years of service at 50ºC
(122ºF). As shown, the analytical model is within 5% of the extracted resistance over the
frequency range from 400kHz to 1 MHz, which demonstrates reasonably good accuracy for this
model. Similar results were obtained for the other cables in Table 2, which are included in
Appendix A. The error in the analytical model arises from the fact that the classical relationship
of the skin effect was approximated by equation 5.1-3. However, the error introduced by this
approximation is acceptable for the purposes of this study.
A similar process was followed to obtain an analytical model for the cable inductance. Equation
3.2-3b was used to extract the inductance from the measured characteristic impedance and
propagation function spectra. The results are presented in Figure 23. The left-hand side shows
the inductance per unit length, L(T) Henries/meter, for the cables aged to simulate 60 years of
service. The right-hand side shows the percent change in inductance, *L(T), for an increase in
service temperature from 50ºC to 60ºC, and from 50ºC to 70ºC. The percent change was
calculated using the following equation:
Lk (ω ) − L50 (ω ) 5.1-4
δLk (ω ) =
L50 (ω )
where Lk(T) is the inductance for cables aged to simulate service temperatures of 60ºC (140ºF)
(k=60) and 70ºC (158ºF) (k=70), and L50(T) is the inductance for the cable aged to simulate a
service temperature of 50ºC (122ºF).
44
0.14 R( ω)/l [S/m]
A
B
0.12
ω [Hz]
400000 600000 800000 1x106
0.08
A Extracted
B Model
0.06
Figure 22 Comparison of resistance per unit length predicted by analytical model to that
extracted from measured impedance for cable with global thermal aging to simulate 60
years of service at 70ºC
L(ω)/l [H/m] δL(ω)/l [H/m]

4.6 0.02 D
4.4
0.01
4.2 B ω [Hz] ω [Hz]
400000 800000
400000 800000 E
3.8 C -0.01
A
-0.02
A 50 ºC (122 ºF)
D 50 to 60 ºC
B 60 ºC (140 ºF)
E 50 to 70 ºC
C 70 ºC (158 ºF)
Figure 23 Inductance per unit length and change in inductance per unit length for cables
with global thermal aging to simulate 60 years of service at 70ºC
Examination of the left-hand side of Figure 23 shows that the inductance tends to increase with
45
increasing thermal degradation, as evidenced by comparing the inductance for the 50ºC (122ºF)
service temperature (curve A) with that for the 60ºC (140ºF) service temperature (curve B). This
supports the notion that shrinkage of the insulation is greater with increased amounts of thermal
aging. A further increase in service temperature to 70ºC (158ºF) (curve C) results in a decrease
in inductance for this case. The reason for this decrease is not readily apparent. It is also noted
that the inductance decreases with increasing frequency. This is consistent with the skin effect,
although the inductance decreases with frequency as T-0.2, whereas the skin effect predicts a T-0.5
dependence.
An analytical model of the cable’s inductance is now developed based on the extracted
inductance spectra. Since the inductance does not behave as the classical skin effect, the
following phenomenological relation is used.
⎛ (ω τ k ) mk ⎞ 5.1-5
Lk (ω ) = L0,k ⎜1 − a k ⎟
⎝ 1 + (ω τ k ) 2 mk ⎠
where L0,k is the very low frequency inductance per unit length of the cable, and ak, Jk, and mk
are adjustable constants. The values for these parameters were determined for the cables aged to
simulate 60 years of service by fitting equation 5.1-5 to the extracted inductance spectra
presented in Figure 23 over the frequency range from 429 kHz to 1MHz. The rms error between
the model prediction and extracted data was minimized to determine the parameter values, which
are presented in Table 4.
Table 4 Parameter values for analytical model of inductance for cables aged to simulate 60 years
of service at various service temperatures 1
Service Temperature Very Low Frequency

ak Jk mk
Simulated Inductance (L0,k)
50 ºC (122 ºF) 4.84 :H/m 0.612 0.10 :sec-1 0.18
60 ºC (140 ºF) 4.84 :H/m 0.612 0.10 :sec-1 0.18
70 ºC (158 ºF) 5.00 :H/m 0.675 0.10 :sec-1 0.18
1. Parameters determined by fitting equation 5.1-5 to the inductance spectra presented in Figure 22.
Figure 24 compares the inductance extracted from the measured impedance spectra with the
inductance predicted by the analytical model for the cable aged to simulate 60 years of service at
70ºC (158ºF). As shown, the predicted inductance is within 1% of the extracted inductance over
the frequency range from 429 kHz to 1 MHz, which demonstrates good accuracy for the
analytical model. Similar results were obtained for the other cables in Table 2, which are
presented in Appendix A.
46
L(ω)/l [H/m]
A Extracted
3.95x10-6
B Model
3.90x10-6
A
3.85x10-6 B
ω [Hz]
400000 600000 800000 1x106
3.75x10-6
Figure 24 Comparison of inductance per unit length predicted by analytical model

to that extracted from measured impedance for cable with global thermal aging to
simulate 60 years of service at 70ºC
The capacitance and conductance are linked by the cable’s dielectric function, which is given by
equation 3.1-1. Therefore, these two electrical properties will be considered together. Using
equations 3.2-3c and 3.2-3d, the real and imaginary components of the dielectric function are
extracted from the characteristic impedance and propagation function. The capacitance and
conductance are then determined using equation 3.1-1.
Figure 25 presents the capacitance per unit length, C(T) Farads/meter, for the cables aged to
simulate 60 years of service. The left-hand side presents the capacitance per unit length for each
of the different service temperatures simulated. The right-hand side presents the percent change
in capacitance for an increase in service temperature from 50ºC to 60ºC (curve D), and 50ºC to
70ºC (158ºF) (curve E). Examination of this figure shows that the capacitance, and therefore, the
real component of the dielectric function, g1(T), decreases slowly with frequency and has an
average value of 0.35 to 0.38 nF/m. The effect of higher simulated service temperatures is
mixed; the capacitance of the cable aged to simulate a service temperature of 60ºC (140ºF)
shows a decrease compared to the cable aged to simulate 50ºC (122ºF). However, an increase is
noted when the capacitance for the cable aged to simulate 70ºC (158ºF) is compared to the cable
aged to simulate 50ºC (122ºF). This behavior is non-intuitive and the explanation is not readily
obvious.
47
C(ω)/l [x10-10 F/m] δC(ω)/l [x10-10 F/m]
3.8 0.06
E
3.7 C 0.04
3.6
0.02
3.5 A ω [Hz] ω [Hz]
200000 600000 1x10 6
200000 600000 1x106
B D
A 50 ºC (122 ºF)
D 50 to 60 ºC
B 60 ºC (140 ºF)
E 50 to 70 ºC
C 70 ºC (158 ºF)
Figure 25 Capacitance per unit length and change in capacitance per unit length for
cables with global thermal aging to simulate 60 years of service at various temperatures
Figure 26 presents the conductance per unit length for the cables aged to simulate 60 years of
service. The left-hand side of the figure shows the conductance per unit length normalized by
the frequency, G(T)/T mho-sec/m. The right-hand side shows the change in conductance,
*G(T), for an increase in service temperature from 50º to 60ºC (curve D), and from 50º to 70ºC
(curve E). As shown, the conductance increases with frequency and has an average value on the
order of 6x10-3 mho-sec/m. Also, as the simulated service temperature increases, the
conductance increases. The right-hand side of the figure shows that the conductance for the
cable aged to simulate 70ºC (158ºF) increased by approximately 40% above that for the cable
aged to 50ºC (122ºF). This indicates that the cable’s insulation has probably passed through a
glass transition phase and become more dissipative.
Using the capacitance and conductance spectra, analytical models for the cable’s dielectric
function are now developed. These models are based upon the fraction power law, which has
been used extensively over the past two decades to model dielectric materials. The model is
defined in the following equation:
⎛ ⎞
ε = c0 ⎜⎜1 + ⎟
A1 A2
+ 5.1-6
⎝ [
1 + (i ω τ 1 ) 1
n
] [ 1 + (i ω τ 2 )
n2
] ⎟
⎠
48
-12
δG(ω)/ω-l [x10-12 mho-s/m]
G(ω)/ω-l [x10 mho-s/m]
0.7
8.0 C
B 0.5
7.0
E
6.0 0.3
A
5.0
ω [Hz] D
6
0.1 ω [Hz]
200000 600000 1x10
200000 600000 1x106
A 50 ºC (122 ºF)
D 50 to 60 ºC
B 60 ºC (140 ºF)
E 50 to 70 ºC
C 70 ºC (158 ºF)
Figure 26 Conductance-to-frequency ratio per unit length and change in conductance-to-
frequency ratio per unit length for cables with global thermal aging to simulate 60 years
of service at various temperatures
where the second and third terms in the equation refer to electronic processes and mixed
processes that govern the insulator’s dielectric response to electrical forces. The constants c0,
A1, J1, n1, A2, J2, and n2 are determined based upon physical considerations. For example,
2#A2#4 due to the fact that the electronic processes dominate the material’s dielectric response
to electrical forces throughout the frequency range. Furthermore, these processes are very fast
(J2=0.1 :s) and nearly independent of frequency (0#n2#0.1). On the other hand, ionic processes
are not included in equation 5.1-6 since they are too slow (0.01s#J2#1s) and cannot affect the
material’s dielectric response to electrical forces. Also, the ionic processes are Debye in nature
and are about 5% less than A2 even at low frequencies (i.e., 1kHz or less, whereas we are
interested in the frequency range from 429kHz and above). Finally, the mixed processes are
midway between the ionic and electronic processes. Typically, the following ranges are found
for these parameters: 0.1#A1#1.0, 0.1#n1#0.5, and 10:s#J1#100:s.
The parameter values for the dielectric function analytical model were determined by fitting
equation 5.1-6 to the extracted capacitance and conductance over the frequency range from
429 kHz to 1 MHz, and minimizing the rms error between the modeled values and the extracted
values. The parameter values are presented in Table 5.
49
Table 5 Parameter values for dielectric function analytical model for cables thermally aged to
simulate 60 years of service at various service temperatures 1
Service
Temperature c0 A1 J1 n1 A2 J2 n2
Simulated
50 ºC (122 ºF) 0.2920 1.00 31.6 0.2500 6.00 0.10 0.010
60 ºC (140 ºF) 0.2874 0.85 31.6 0.2489 6.80 0.10 0.018
70 ºC (158 ºF) 0.2880 1.50 31.6 0.2500 5.75 0.01 0.125
1. Parameter values determined by fitting equation 5.1-6 to the capacitance and conductance spectra presented in Figures 24 and 25,
respectively.
Figure 27 compares the capacitance predicted by the model to the capacitance extracted from the
measured impedance spectra for the cable aged to simulate 60 years of service at 70ºC (158ºF).
As shown, the predicted capacitance is within 1% of the extracted capacitance over the
frequency range from 429 kHz to 1MHz, which demonstrates good accuracy for the model.
C(ω)/l [x10-10 F/m]
A Extracted
3.64
B Model
3.63
B
3.62
3.61 A
ω [Hz]
300000 400000 500000 600000 700000 800000
3.59
3.58
Figure 27 Comparison of capacitance per unit length predicted by analytical

model to that extracted from measured impedance for cable with global thermal
aging to simulate 60 years of service at 70ºC
50
Similarly, Figure 28 compares the conductance-to-frequency ratio, G(T)/T, predicted by the
model to the conductance-to-frequency ratio extracted from the measured impedance spectra for
the cable aged to simulate 60 years of service at 70ºC (158ºF). As shown, the predicted spectra
is within 10% of the extracted spectra over the frequency range from 429 kHz to 1MHz, which
demonstrates good accuracy for the model.
G(ω)/ω-l [x10-12mho-s/m]
A Extracted
8.25
B Model
8.00
7.75
A
7.50
B
7.25
ω [Hz]
300000 400000 500000 600000 700000 800000
6.75
6.50
Figure 28 Comparison of conductance-to-frequency ratio per unit length

predicted by analytical model to that extracted from measured impedance for
cable with global thermal aging to simulate 60 years of service at 70ºC
With the electrical properties modeled, the characteristic impedance and propagation function
can now be modeled using equations 3.1-8 and 3.2-2, respectively. Figure 29 compares the
characteristic impedance predicted by the analytical model to that extracted from the measured
impedance spectra for the cable thermally aged to simulate 60 years of service at 70ºC (158ºF).
As shown, the characteristic impedance predicted by the analytical model is within 1% of the
extracted spectra over the frequency range from 429 kHz to 1 MHz, which demonstrates good
accuracy for the model. Similarly, Figure 30 compares the propagation function predicted by the
analytical model to that extracted from the measured impedance spectra for the cable thermally
aged to simulate 60 years of service at 70ºC (158ºF). Again, good accuracy is demonstrated for
the model, with predicted values within 1% for the imaginary part, and within 7% for the real
part of the propagation function.
51
|Z0(ω)|/l [S/m] Ψ(ω)
101.9 -0.50
101.8 -0.75
-1.00
A
101.7
101.6 A B -1.25
101.5 -1.50 B
-1.75
101.4
ω [Hz] ω [Hz]
800000 -2.25 400000 600000 800000
600000 700000
A Extracted
B Model
Figure 29 Comparison of characteristic impedance magnitude per unit length and phase
spectra predicted by analytical model to that extracted from measured impedance for
γ i(ω) γr(ω)
0.035
0.030 ω [Hz]
0.025 A 400000 600000 800000
0.020 0.0008 A B
0.015 B
0.010 0.0006
ω [Hz] 0.0004
200000 600000 1x106
A Extracted
B Model
Figure 30 Comparison of the imaginary component and real component of the

propagation function predicted by analytical model to that extracted from measured
impedance for cable with global thermal aging to simulate 60 years of service at 70ºC
5.1.2.2 Extrapolation of Cable Properties to High Frequency
52
Next, the high frequency cable properties are determined by using the extrapolative approach,
which was discussed previously. The characteristic impedance and propagation function for the
cables aged to simulate 60 years of service are presented and discussed below. Similar analyses
were performed for the cables aged to simulate 20 years and 40 years of service, and these
results are presented in Appendix B.
The high-frequency characteristic impedance spectra were determined by extrapolating the low-
frequency spectra. Figure 31 presents the extrapolated high-frequency characteristic impedance
magnitude, *Z0(T)*, and phase, Q(T), spectra for cable specimens 26, 41 and 42. The limiting
value of the characteristic impedance magnitude (i.e., as T64) for the cables are as follows:
Specimen 26 (50ºC): Z0(T64) = 77.42 S
Specimen 41 (60ºC): Z0(T64) = 79.46 S
Specimen 42 (70ºC): Z0(T64) = 75.12 S
A 50 ºC (122 ºF)
B 60 ºC (140 ºF)
C 70 ºC (158 ºF)
|Z0(T)|/l [S/m]
Ψ(ω)
106 C
0.008 B
104
0.007
102 A
0.006
A B
100 0.005
0.004
98 C ω (Hz)
ω (Hz) 2x10 7
6x10 7
1x108
7 7 8
2x10 6x10 1x10
Figure 31 High frequency characteristic impedance magnitude and phase spectra

extrapolated from low frequency spectra for cables with thermal aging to simulate 60
years of service at various temperatures
Using Equation 5.1-1, each cables’ propagation function at high frequency was determined, as
shown in Figure 32. Examination of this figure shows that both the real and imaginary
components of the propagation function exhibit a number of oscillations, starting in the vicinity
of 10 MHz. These oscillations are a consequence of the extrapolative approach used, and do not
represent reality. Specifically, the impedance spectra are strong functions of the frequency in the
vicinity of the phase zero crossings. Thus, small differences between the actual characteristic
53
impedance and the extrapolated value give rise to these fictitious oscillations. Furthermore, as
the frequency increases, the differences between the actual and extrapolated values also increase.
To mitigate the impact of these anomalies on the evaluations performed herein, the extrapolated
propagation function data used in the evaluations were restricted to values in the frequency range
from 1 MHz to 10 MHz. The results in this frequency range are reliable.
A 50 ºC (122 ºF)
B 60 ºC (140 ºF)
C 70 ºC (158 ºF)
γ i(ω)
γr(ω)
3.5
C 0.1 C
2.5 0.08 B
B 0.06
1.5
A 0.04
0.5 0.02 A
ω (Hz) ω (Hz)
2x107 6x107 1x108 2x107 6x107 1x108
Figure 32 Imaginary and real components of the high frequency propagation

function determined from the extrapolated characteristic impedance for cables
with thermal aging to simulate 60 years of service at various temperatures
5.1.3 Detection of Localized Degradation on Cables
In this section, the use of BIS for detecting a localized area of degradation on a cable is
discussed. For purposes of this study, the localized area of degradation was simulated by
exposing a small section of a cable to elevated temperature in addition to the global thermal
aging applied to the cable. The cable specimens tested for this portion of the study are identified
in Table 6.
First, the high frequency impedance spectra for cables with and without hot-spots are compared
to demonstrate that localized damage does impact the cable’s impedance. This difference in
impedance spectra can be used to detect and locate localized degradation. The data presented
were obtained from BIS measurements on cables in the open configuration. Next, three tests are
presented that can be used to detect localized degradation. These tests do not provide absolute
values for cable properties; rather, the tests are based upon the differences between cables with
and without localized degradation.
54
Table 6 Specimens tested to evaluate BIS for locating localized areas of thermal aging
Hot-Spot Simulated
Length
Service Life Simulated Conditions Simulated/
No. Specimen ID meters
(Global Aging) Location on Cable (meters
(inches)
from left end)
16. PNI-79-RB-188-816 10 (393.7) 20 @ 50ºC (122ºF) None
17. PNI-79-RB-188-817 10 (393.7) 20 @ 50ºC (122ºF) 20 yrs. @ 60ºC/8.0m to 9.0m
21. PNI-79-RB-188-821 10 (393.7) 40 @ 50ºC (122ºF) None
26. PNI-79-RB-188-826 10 (393.7) 60 @ 50ºC (122ºF) None
5.1.3.1 Comparison of Impedance Spectra With and Without Localized Degradation
In this section, the impact of localized degradation on a cable’s impedance spectrum is

examined. The effect of degradation severity, size, and location are evaluated.
55
The impact of hot-spot severity was examined by comparing cables with simulated hot-spots
representing 60 years of exposure to thermal environments of 60ºC and 70ºC. The impedance
spectra for these cables were compared to each other, and to a cable with no hot-spot. Figure 33
presents the impedance magnitude and phase spectra over the frequency range of 14MHz to
22MHz for three cables; one with no hot-spot (curve A), one with a 1-meter hot-spot
representing 60 years at 60ºC (curve B), and one with a 1-meter hot-spot representing 60 years at
70ºC (curve C). Examination of this figure shows that the double peak structure present in the
impedance magnitude spectrum for the cable with no hot-spot (curve A) disappears when a hot-
spot is introduced onto the cable. Furthermore, the zero crossing of the impedance phase spectra
are shifted to lower frequencies when a hot-spot is present. In this case, the increase in
degradation severity from 60º (curve B) to 70ºC (curve C) appears to have very little effect, as
shown by the very similar impedance spectra.
|Z(ω)|/l [S/m] Φ(ω) [Degrees]

700 80
C 600 60
500 A 40
B
400 20
ω [Hzs]
A 300
1.4x107 1.8x107 2.2x107
200 -20
100 -40 C
ω [Hzs] B -60
1.4x107 1.8x107 2.2x107
A No Hot Spot
B 60 ºC (140 ºF) 1.0m Hot Spot
C 70 ºC (158 ºF) 1.0m Hot Spot
to 22MHz for cables with global thermal aging to simulate 60 years of service at 50ºC,
plus additional thermal aging to simulate hot-spots of different severity
Figure 34 presents the impedance magnitude and phase spectra over the frequency range of
39MHz to 45MHz for the same three cables. As shown in this figure, a shift toward lower
frequencies is noted for both the magnitude and phase spectra when a hot-spot is present. In this
case, the increase in degradation severity from 60ºC (curve B) to 70ºC (curve C) is more
pronounced with the less severe degradation resulting in an impedance spectrum that is closer to
that of the baseline cable with no hot-spot (curve A). The hot-spot with more severe degradation
tends to shift the impedance spectrum to lower frequencies.
56
60
C
150 40
A A B
20
100 B ω [Hzs]
7 7 7
3.9x10 4.1x10 4.3x10 4.5x107
C
50
ω [Hzs] -40
3.9x107 4.1x107 4.3x107 4.5x107 -60
A No Hot Spot
plus additional thermal aging to simulate a hot-spot
These results demonstrate that the presence of localized degradation does impact the cable’s
impedance spectra. Further, the impedance spectrum appears to be sensitive to the severity of
the hot-spot degradation at higher frequencies. Thus, comparison of a cables’s broadband
impedance spectrum to that for a cable with a known condition may be useful as an indicator of
the presence and severity of a hot-spot.
The impact of hot-spot size was investigated by comparing cables with hot-spots of 0.5 meters
and 1.0 meters. Again, the cables are compared to each other, as well as to a cable with no hot-
spot. Figure 35 presents the impedance magnitude and phase spectra over the frequency range of
14MHz to 22MHz for three cables; one with no hot-spot (curve A), one with a 0.5-meter hot-
spot representing 60 years at 70ºC (curve B), and one with a 1-meter hot-spot representing 60
years at 70ºC (curve C). Examination of this figure shows that the increase in degradation size
from 0.5 meters (curve B) to 1.0 meters (curve C) appears to have very little effect, as shown by
the very similar impedance spectra.
57
80
B C 600 60
A 40
400 20
ω [Hzs]
1.4x107 1.8x107 2.2x107
200 -20 C
A
ω [Hzs] B -40
1.4x107 1.8x107 2.2x107
A No Hot Spot
14MHz to 22MHz for cables with global thermal aging to simulate 60 years of service at
50ºC, plus additional thermal aging to simulate hot-spots of different sizes
39MHz to 45MHz for the same three cables. As shown in this figure, the increase in degradation
size from 0.5 meters (curve B) to 1.0 meters (curve C) is more pronounced with the smaller hot-
spot resulting in an impedance spectrum that is closer to that of the baseline cable with no hot-
spot (curve A). The larger size hot-spot tends to shift the impedance spectrum to lower
frequencies.
These results demonstrate that a cable’s impedance spectrum appears to be sensitive to the size
of the hot-spot at higher frequencies, as reflected by a shift to lower frequencies as the hot-spot
size increases. Thus, comparison of a cables’s broadband impedance spectrum to that for a cable
with a known condition may be useful as an indicator of the size of a hot-spot.
Finally, the impact of hot-spot location on a cable’s impedance spectrum was examined. This
testing was performed using a cable with a simulated 1-meter long hot-spot representing 60 years
at 70ºC located from 8.0 to 9.0 meters from the left end of the cable. The cable’s impedance was
measured twice; once with the point of measurement (POM) at the left end of the cable, and once
with the POM at the right end of the cable. This effectively yielded impedance spectra for a hot-
spot located 8.0 to 9.0 meters from the POM, and for a hot-spot located 1.0 to 2.0 meters from
the POM. These measurements were compared to each other, as well as to the impedance
spectrum from a cable with no hot-spot.
58
175 60
B B
150 C 40
C
125
20
100 A ω [Hzs]
7 7 7
75
A 3.9x10 4.1x10 4.3x10 4.5x107
50 -20
ω [Hzs]
7 7 7 -40
3.9x10 4.1x10 4.3x10 4.5x107
A No Hot Spot
plus additional thermal aging to simulate hot-spots of different sizes
14MHz to 22MHz for three cases; one with no hot-spot on the cable (curve A), one with a
1.0-meter hot-spot representing 60 years at 70ºC located 8.0m to 9.0 m from the POM (curve B),
and one with a 1-meter hot-spot representing 60 years at 70ºC located 1.0m to 2.0m from the
POM (curve C). Examination of this figure shows that the location of the hot-spot does impact
the impedance spectrum for the cable. With the hot-spot located close to the POM (curve B), the
two peak structure observed for the cable with no hot-spot (curve A) is not present. Also, the
maxima for the impedance magnitude, and the zero crossing for the impedance phase are shifted
to a lower frequency. With the hot-spot farther from the POM (curve C), the two peak structure
is still present; however, the impedance magnitude and phase zero crossing are also shifted to a
lower frequency.
39MHz to 45MHz for the same three cables. As shown in this figure, a consistent trend is not
readily apparent since the impedance spectra for the case in which the hot-spot is farther from
the POM (curve C) is very similar to the base line cable with no hot-spot (curve A), while that
for the case in which the hot-spot is close to the POM (curve B) shows a shift toward lower
frequencies.
These results demonstrate that the location of a hot-spot also impacts the cable’s impedance
spectrum as evidenced by a shift in the spectrum to lower frequencies. The data suggest that the
closer the hot-spot is to the POM, the greater is the impact on the measured impedance.
59
80
B B
600 60
A 40
C 20
400
ω [Hzs]
1.4x107 1.8x107 2.2x107
-20
A 200
C -40
ω [Hzs]
1.4x107 1.8x107 2.2x107
A No Hot Spot
B 70 ºC (158 ºF) Hot Spot at 1.0m-2.0m from POM
C 70 ºC (158 ºF) Hot Spot at 8.0m-9.0m from POM
50ºC, plus additional thermal aging to simulate hot-spots at different locations

A
60
150
C
40
A B
20
100 C ω [Hzs]
B 3.9x107 4.1x107 4.3x107 4.5x107
50
ω [Hzs]
7 7 7
3.9x10 4.1x10 4.3x10 4.5x107 -40
A No Hot Spot
50ºC, plus additional thermal aging to simulate hot-spots at different locations
60
5.1.3.2 Numerical Tests for Detecting Localized Degradation
In this section three numerical tests that can be used to detect the presence of a hot-spot based on
measured impedance spectra are presented. These tests are 1) average dissipation, 2) zero
crossing location, and 3) zero crossing line spacing. Each of these tests are discussed below.
5.1.3.2.1 Average Dissipation Test
As demonstrated in the previous evaluation of impedance spectra for the thermally aged cables,
as the degradation on the cable increases, the cable’s conductance per Hertz also increases.
Therefore, cables with a higher level of degradation are expected to show more dissipation when
the cable is subjected to an ac voltage. Since the presence of a hot-spot on a cable implies that
some portion of the cable was degraded more than the remainder of the cable, a cable containing
a hot-spot is expected to show more dissipation than a cable without a hot-spot. The phase
spectrum for the cable should reflect this increased dissipation by forcing the phase spectrum
further away from ±90°. This should be the case throughout the frequency range used to probe
the cable. However, since cable physics in the oscillatory region is dominated by wave
phenomena, the impedance phase is not expected to decrease at every discrete frequency when a
hot-spot is present on the cable. Rather, the average value of the absolute phase magnitude,
<*M(T)*>, should exhibit the aforementioned behavior and can be used as an indicator of the
presence of a hot-spot. This is defined by the following equation
ω2
Φ(ω ) ≡
∫ω Φ(ω ) dω
1
ω
∫ω dω 5.1-7
2
In Equation 5.1-7, the integration is performed over the frequency range from T1=1MHz to
T2=100MHz since this is the range of frequencies over which the impedance is measured.
5.1.3.2.2 Phase Zero Crossing Frequency Test
Changes in the zero crossing frequencies of the impedance phase spectra are useful as indicators
of cable degradation. At high frequencies, the impedance phase is a strong function of the
propagation function phase since the characteristic impedance phase is essentially zero.
Consequently, the frequencies at which the impedance phase zero crossings occur can be
determined by the following equation:
1 nπ 5.1-8
γ i (ω ) ≈ ≈
L(ω ) C(ω )
61
From the low frequency impedance data presented previously, it was shown that the cable
inductance and capacitance increase with increasing degradation of the cable. Although that data
are restricted to the non-oscillatory region of the impedance spectrum, it is reasonable to assume
that this behavior extends to higher frequencies. Based on this assumption, it is expected that the
zero crossings of the impedance phase spectrum will exhibit a shift to lower frequencies when
the spectrum for a cable containing a hot-spot is compared to that for a cable without a hot-spot.
The impedance data presented previously showed that, in some cases, the cable’s impedance
spectra contain a double peak structure. Also, in some cases additional zero crossings occur. To
compare the various zero crossings between different spectra, it is desirable to average the
spectra over a range of frequencies as follows:
ω+ ∆ ω
∫ω Φ(ϖ ) dϖ
Φ(ω ) = − ∆ω
ω+ ∆ ω
∫ω − ∆ω
dϖ 5.1-9
where j=T±)T, and )T=1.30MHz.
5.1.3.2.3 Phase Zero Crossing Line Spacing Test
In the phase zero crossing line spacing test the sum of the spacing between zero crossings is
compared. The line spacing of the zero crossings is defined by the following equation:
∑ (ω n − ω n− 1 )
ϑn = n= 2
5.1-10
∆ω meas
where n is the zero crossing being evaluated, k is the total number of zero crossings in the
frequency interval of interest, Tn is the frequency of the nth zero crossing, and )Tmeas is the
frequency spacing between measured data points. For this study, measurements were taken at
2000 equally spaced data points over the frequency range from 100kHz to 100MHz; therefore,
)Tmeas = (100x106 - 0.1x106)/2000 = 49,950.
Figure 39 shows an example of the averaged impedance phase spectrum, <*M(T)*>, over the
frequency range from 100kHz to 100MHz. Examination of this figure shows that the frequency
increments between the individual zero crossings can be viewed as line spacings.
62
<|Φ(ω)|>
80
60
40
20
2x107 4x107 6x107 8x107 1x108

ω1
ω [Hz]
ω2
Figure 39 Example of average impedance phase spectrum over

the frequency range from 100kHz to 100MHz for cable with
global thermal aging
5.1.3.2.4 Results of the Numerical Tests
Each of the three tests discussed above were applied to the cable test samples to determine their
effectiveness for detecting a hot-spot on a cable. Results for the cables with global thermal
aging to simulate 60 years of service are discussed below. Results for the cables with thermal
aging to simulate 20 years and 40 years of service are included in Appendix B.
First, the average of the phase magnitude was determined for each of the cables. Table 7
presents the results of this test, which compares the values for each of the cables aged to simulate
60 years of service, both with and without hot-spots. For this test, the frequency range was
restricted and included data from 7MHz to 100MHz since data measurements below 7MHz were
found to be unreliable. In the table the change in average impedance phase magnitude,
*<*M(T)*>, is determined by the following equation:
φ (ω ) − φ (ω )
δ φ (ω ) = B 5.1-11
φ (ω )
B
63
where +*M(T)*,B represents the average impedance phase magnitude for the baseline cable and
+*M(T)*, represents the average impedance phase magnitude of the cable being tested.
Examination of Table 7 shows that 7 of the 8 cables with hot-spots had an average impedance
phase magnitude lower than that for the baseline cable without a hot-spot. These results support
the finding that the average dissipation of the cable decreases with increased degradation, thus
suggesting that the average dissipation test may be a useful indicator for detecting hot-spots.
Table 7 Average impedance phase magnitude for cables with global thermal aging to simulate
60 years of service at 50ºC, plus additional thermal aging to simulate hot-spots
Cable Hot-spot Hot-spot Hot-spot <*N(T)*> * <*N(T)*>

Temperature Size Location
26. None None None 37.0340 -
27A. 60ºC (140ºF) 1.0m 1.0m to 2.0m 36.7463 + 0.78%
27B. 60ºC (140ºF) 1.0m 8.0m to 9.0m 36.5958 + 1.18%
28A. 60ºC (140ºF) 0.5m 1.5m to 2.0m 36.1382 + 2.42%
28B. 60ºC (140ºF) 0.5m 8.5m to 9.0m 35.6967 + 3.61%
29A. 70ºC (158ºF) 1.0m 1.0m to 2.0m 36.7463 + 0.78%
29B. 70ºC (158ºF) 1.0m 8.0m to 9.0m 35.5958 + 3.88%
30A. 70ºC (158ºF) 0.5m 1.5m to 2.0m 36.8930 + 0.38%
30B. 70ºC (158ºF) 0.5m 8.5m to 9.0m 37.2820 - 0.67%
Next, the zero crossing frequencies are compared between the baseline cable with no hot-spot
(cable 26) and each of the cables with a hot-spot (cables 27 to 30). To make this comparison, a
matrix of zero crossing frequency differences is created using the following relationship:
ω zc , n , 26 − ω zc , n , x
∆ω zc , n , x = 5.1-12
∆ω meas
where Tzc, n, 26 is the zero crossing frequency for the nth impedance phase zero crossing for cable
26 (baseline), Tzc, n, x is the zero crossing frequency for the nth impedance phase zero crossing for
cable x (x = 27, 28, etc. for cables with hot-spots), and )Tmeas is the frequency spacing between
measured data points, which for this study is 49,950Hz, as discussed previously. Positive
64
differences would indicate a downshift in frequency for the zero crossing for the cables with hot-
spots. The results are presented in Table 8.
Table 8 Differences in zero crossing frequencies between cable with no hot-spot (cable 26) and
cables with hot-spots (cables 27 to 30) at odd number zero crossings
Frequency Difference for Zero Crossing No.

Cable
1 3 5 7 9 11 13 15 17 19 21
27A 0 0 0 +2 -4 -4 -2 -2 -7 +4 0
27B +1 +1 +1 0 -2 -2 +1 +1 0 -7 + 10
28A +1 +4 +7 +4 0 +3 +2 +6 +4 + 10 + 20
28B +1 0 +1 +5 -1 0 +2 +1 -2 0 + 12
29A 0 0 0 +2 -4 -4 -2 -2 -7 +4 0
29B +1 +1 +1 0 -2 -1 +1 +1 0 -7 + 10
30A +1 +2 +3 +6 +2 +1 +7 +5 +7 +1 + 19
30B +2 +3 +6 +5 +7 +7 +8 + 11 +4 + 22 + 25
Examination of Table 8 shows that there is no clear trend in the zero crossing frequencies. For
zero crossings 1, 3, 5, and 7 on the cables with hot-spots, each is either the same or downshifted
to a lower frequency relative to the baseline cable. For the higher zero crossings the results are
mixed with some showing a downshift and others showing a shift toward higher frequencies.
These results suggest that this test may not be effective at detecting hot-spots.
Finally, the phase zero crossing line spacing test is applied to the cables with global thermal
aging to simulate 60 years of service. Again, each cable with a hot-spot is compared to the
baseline cable without a hot-spot. The results are presented in Table 9.
Examination of Table 9 shows that the line spacing for most of the cables with hot-spots is
smaller (positive change) than that for the baseline cable. These results support the finding that
the zero crossings occur closer together for cables with hot-spots. Therefore, the line spacing
test may be a useful indicator of the presence of hot-spots on cables.
65
Table 9 Impedance phase zero crossing line spacing for cables with global thermal aging to
simulate 60 years of service, with and without hot-spots
Cable Line Spacing Percent Change from

Baseline Cable
26 (Baseline) 93.60 -
27A 93.60 0
27B 93.15 + 0.48%
28A 92.65 + 1.01%
28B 93.05 + 0.59%
29A 93.60 0
29B 93.15 + 0.48%
30A 92.70 + 0.96%
30B 92.45 + 1.23%
5.1.4 Locating Localized Degradation on Cables
This section presents the results of research on the use of BIS to locate localized areas of
degradation, or hot-spots, on a cable. The cables shown in Table 6 were also used for this
portion of the research. First, the theory behind hot-spot location is discussed. Subsequently,
results from the application of this theory are presented.
5.1.4.1 Theory Behind Hot-spot Location in Thermally Aged Cables
The formation of a hot-spot in a thermally aged cable occurs when a segment of the cable is
exposed to higher temperatures than the other portions of the cable. This could occur, for
example, if the cable is routed near a hot steam line in the plant. Figure 40 presents a schematic
depicting a model for a two-conductor cable with a hot-spot. In this figure, “H” represents the
healthy portion of the cable exposed to normal operating conditions, and “D” represents the
damaged portion of the cable exposed to elevated temperature conditions (i.e., hot-spot). The
total length of the cable is denoted by l, and the length of the hot-spot is indicated by la to lb.
The cable has a frequency dependent voltage applied to the two conductors represented by V(T).
66
H D H
V(T)
x=0 x=la x=lb x=l
Figure 40 Schematic model of a cable with a hot-spot
In Section 5.1, it was shown that aging degradation will alter the cable’s electrical properties.
Furthermore, the cable’s characteristic impedance and propagation function depend upon the
severity of the degradation. To model this dependence, the following equation can be used:
⎧
Z MODEL (ω ) = Z0, H (ω )⎨
[ ] [
⎪ Z0, H (ω ) coth γ H ( − a ) + Z0, D (ω ) coth γ D ( b − a )] ⎫⎪
⎬ 5.1-13
[ ] [
⎩ Z0, D (ω ) + Z0, H (ω ) coth γ H ( − b ) tanh γ D ( b −
⎪ a )] ⎪⎭
where Z0,H(T) and (H(T) represent the cable’s characteristic impedance and propagation function
in the healthy portion of the cable, and Z0,D(T) and (D(T) represent the cable’s characteristic
impedance and propagation function in the damaged (hot-spot) portion of the cable.
In Section 5.1.2 of this report, models for the thermally aged cables were developed that can be
used to approximate the cable’s actual electrical properties. Specifically, models were developed
based on low frequency impedance data and were extrapolated to high frequencies. Equation
5.1-1 was then used to obtain the cable’s high frequency propagation function. Although the
model has the correct cable impedance, since it is built into the model, the extrapolated
characteristic impedance is an approximation of the real function; therefore, the propagation
function is also an approximation of the real function. To minimize the introduction of modeling
error into the evaluation, the modeling is limited to the frequency range from 1MHz to 10MHz.
This is sufficient since the first three or four zero crossings of the impedance phase occur within
this frequency range.
The most direct method of determining a hot-spot location would be to invert equation 5.1-13
and solve for la and lb. However, inverting the equation yields a transcendental equation that
includes la and lb in a non-linear way. An easier method of locating a hot-spot is to calculate the
cable’s impedance using equation 5.1-13 based on assumed values of la and lb and compare the
result to the measured impedance. Based on the root mean square (rms) error between the value
predicted by the model and the measured value, new values of la and lb are chosen and the
process is repeated. The final hot-spot location is determined by minimizing the rms error.
67
The following process is used to determine a hot-spot location:
1. Measure the cable impedance
2. Calculate the cable impedance using equation 5.1-13 using an assumed hot-spot location
3. Compare the impedance predicted by the model with the measured value and calculate a
rms error
4. Based on the rms error, assume a new hot-spot location and repeat steps 2 and 3.
5. Repeat steps 2 through 4 until the rms error is minimized
In comparing the modeled impedance to the measured impedance, four different parameters are
calculated. These are defined by the following equations:
φ MODEL (ω n ) − φ MEASURED (ω n )
δφ (ω ) ≡ ∑ω = 1 MHz
ωn = 10 MHz
5.1-14
n
φ MEASURED (ω n )
Z MODEL (ω N ) − Z MEASURED (ω n )
δ Z (ω ) ≡ ∑ω = 1 MHz
ωn = 10 MHz
5.1-15
n
Z MEASURED (ω n )
5.1-16
δΘ(ω ) ≡ δφ (ω ) + Z (ω )
2 2
ν n , MODEL − ν n , MEASURED
δΩ ≡ ∑ω = 1 MHz
ωn = 10 MHz
5.1-17
n
ν n , MEASURED
where NMODEL(T) and NMEASURED (T) are the predicted and measured values of the cable
impedance phase, ZMODEL (T) and ZMEASURED (T) are the predicted and measured cable impedance
magnitude, and <n, MODEL and <n, MEASURED are the predicted and measured frequency of the nth
impedance phase zero crossing, and 1 and S are parameters developed specifically to represent
the model accuracy.
The relationships defined by equations 5.1-14 to 5.1-17 are referred to as the “error functions”
since they represent the error between the cable’s properties predicted by the model to the actual
measured value. Each of the error functions is evaluated in making a determination as to the
correct location of a hot-spot.
The process for locating a hot-spot is demonstrated below for cable 29. This cable has a 1 meter
68
hot-spot representing 60 years of service at 70ºC (158ºF). The hot-spot is located 7.0 to 8.0
meters from the left end of the cable. In this demonstration, the cable impedance point of
measurement is first taken at the left end of the cable (29A), then at the right end of the cable
(29B). Thus, one cable is used to demonstrate two hot-spot locations; namely, 7.0 to 8.0 meters
from the POM and 2.0 to 3.0 meters from the POM. In each case, a hot-spot location is assumed
and the error functions are calculated. The results are plotted as a function of length along the
cable to facilitate the identification of maxima and minima.
Figures 41 presents the results of the impedance phase error function calculations, <*M(T)>, and
impedance magnitude, <**Z(T)*>, for cable 29A as a function of distance along the cable.
Examination of this figure shows that the impedance phase error has an absolute minimum at
6.0 meters, while the impedance magnitude also has an absolute minimum at 6.0 meters. These
minima are relatively close to the actual hot-spot location of 7.0 to 8.0 meters from the POM.
<δΦ> <δ|Ζ|>
0.030
0.020
0.025
0.015 0.020
0.010 X [m] 0.015

2 4 6 8 10 X [m]
2 4 6 8 10
Actual position of the hot spot: 7m to 8m from the POM

Figure 41 Impedance phase and magnitude error functions for cable with thermal aging
to simulate a 1 meter hot-spot located 7.0 to 8.0 meters from the cable end representing
60 years of service at 70ºC
Figures 42 presents the results of the error function calculations, <*1(T)>, and zero crossing
frequency change, <*S>, for cable 29A as a function of distance along the cable. Examination
of this figure shows that <*1(T)> has an absolute minimum at 6.0 meters, while the zero
crossing frequency change, <*S>, has an absolute minimum at 5.0 to 7.0 meters. Again, these
minima are relatively close to the actual hot-spot location of 7.0 to 8.0 meters from the POM.
69
<δΘ> <δS>
0.008
0.025
0.006
0.020
0.004
0.015
0.002
0.010 X [m]
2 4 6 8 10 X [m]
2 4 6 8 10

Figure 42 Impedance phase zero crossing error functions for cable with thermal aging to
simulate a 1.0 meter hot-spot located 7.0 to 8.0 from the cable end representing 60 years
of service at 70ºC
From an evaluation of the four error functions obtained for this cable, a hot-spot location of 5.0
to 7.0 meters from the POM would be predicted. This is in excellent agreement with the actual
hot-spot location of 7.0 to 8.0 meters from the POM.
Figures 43 presents the results of the impedance phase error function calculations, <*M(T)>, and
impedance magnitude, <**Z(T)*>, for cable 29B as a function of distance along the cable.
Examination of this figure shows that the impedance phase error has an absolute minimum at 4.0
meters, while the impedance magnitude has an absolute minimum at 3.0 meters. These minima
are relatively close to the actual hot-spot location of 2.0 to 3.0 meters from the POM.
<δΦ> <δ|Ζ|>
0.04 0.04
0.03
0.03
0.02 X [m]
0.02
2 4 6 8 10
X [m]
2 4 6 8 10
Actual position of the hot spot: 2m to 3m from POM

Figure 43 Impedance phase and magnitude error functions for cable with thermal aging
to simulate a 1 meter hot-spot 2.0 to 3.0 meters from the end of the cable representing 60
years of service at 70ºC
70
Figures 44 presents the results of the error function calculations, <*1(T)>, and zero crossing
frequency change, <*S>, for cable 29B as a function of distance along the cable. Examination
of this figure shows that <*1(T)> has an absolute minimum at 3.0 meters, while the zero
crossing frequency change, <*S>, has an absolute minimum at 2.0 meters. Again, these minima
are relatively close to the actual hot-spot location of 2.0 to 3.0 meters from the POM. It is also
noted that <*S> has a second minima at 7.0 meters; however, based on the other error function
minima, this one is seen to be inconsistent and can be eliminated.
<δΘ> <δS>
0.008
0.030
0.006
0.025
X [m] 0.004
2 4 6 8 10 0.002
0.015 X [m]
2 4 6 8 10

Figure 44 Impedance phase zero crossing error functions for cable with thermal aging to
simulate a 1 meter hot-spot located 2.0 to 3.0 meters from the end representing 60 years
of service at 70ºC
From an evaluation of the four error functions obtained for this cable, a hot-spot location of 2.0
to 4.0 meters from the POM would be predicted. Again, excellent agreement with the actual hot-
spot location of 2.0 to 3.0 meters from the POM is obtained. Additional examples of hot-spot
location are included in Appendix B.
5.2 Application of BIS to Cables with Attached Loads
When cables are monitored in a nuclear power plant environment it is very likely that they will
be connected to a piece of equipment. Rather than disconnect the cable from the equipment, it
would be desirable to monitor the cable as installed. To determine the impact of attached loads
on the BIS method, tests were performed with a load attached to the test cables to determine if
degradation can be detected and located without disconnecting the cable. The results are
reported in this section.
71
5.2.1 Test Specimens Used for Testing with Attached Loads
For this portion of the research the cable test specimens presented in Table 10 were used.
Table 10 Specimens tested to evaluate BIS for testing cables with attached loads
Length Service Life Hot-Spot

No. Specimen ID
meters (inches) Simulated Simulated
26. PNI-79-RB-188-826 10 (393.7) 60yrs. @ 50ºC None
27. PNI-79-RB-188-827 10 (393.7) 60yrs. @ 50ºC 60yrs. @ 60ºC
Two different loads were used for these tests; 1) a small induction motor, and 2) a solenoid coil.
Both loads have very broad impedance spectra and produced very similar results. Therefore,
most of the testing was focused on the motor and those results are reported herein. The results
apply equally well to the coil or any load that has a capacitance and inductance in parallel. The
combination of test cable with load attached is referred to as the “cable system” in the following
discussion.
5.2.2 Low Frequency BIS Tests for Cables with Attached Loads
Impedance measurements were made on the cables alone in the open circuit configuration, and
with the load attached to determine their frequency response. The load used was a 1 Hp, 3-phase
ac induction motor. Figure 45 compares the low frequency impedance magnitudes for the cable
specimen with no hot-spot (Specimen 26), the load, and the cable system (cable connected to the
load). As shown, the impedance magnitude of the motor is less than that of the cable for
frequencies less than the crossing frequency, Tc, which is the point at which the impedance
curves cross (approximately 126kHz). As a result, most of the current will flow through the
motor and not through the cable’s capacitance. As the cable’s length increases, the crossing
frequency decreases.
Figure 45 also shows that as the frequency increases, the magnitude of the cable’s impedance
magnitude continues to decrease as 1/T, whereas the cable system’s impedance magnitude
decreases as 1/(T-Tc)2. Once the load’s impedance is greater than the cable’s impedance, less
current flows through the motor and more flows through the cable’s capacitance. As a result, the
cable system’s impedance will be dominated by the physics of the cable.
72
Log10|Z(ω| A Cable aged to 60yrs@50°C
B Motor alone
5 C Cable aged to 60yrs@50°C
plus motor
4 A
B
3
C
Log10 ω
3.5 4 4.5 5 5.5 6
Figure 45 Comparison of measured impedance magnitude for cable with

and without an attached load
Figure 46 compares the impedance phase spectra for the cable, load, and cable system. As
shown, the motor’s impedance phase is nearly 90 degrees at 100Hz and gradually decreases until
100kHz, at which point it drops below 0 degrees. This behavior implies that the motor can be
viewed as a simple L-C circuit with the inductance and capacitance in parallel. It is noted that
the motor’s impedance phase has a zero crossing at 200kHz and that the cable system itself has a
zero crossing at 120kHz. Thus, at low frequencies the cable system’s impedance is heavily
dominated by the motor. However, the zero crossing of the impedance phase does react to the
presence of a hot-spot.
Figure 47 compares the impedance phase spectra of a cable system consisting of an aged cable
without a hot-spot connected to a load to that of a system consisting of an aged cable with a
hot-spot connected to a load. The severity of the hot-spot is varied. As shown, the presence of a
hot-spot does affect the system’s impedance in the region of the zero crossing, although the
effect is small. Specifically, the hot-spot decreases the frequency at which the cable system’s
impedance phase has a zero crossing. The more severe the hot-spot, the greater the downshift in
zero crossing frequency.
73
Φ(ω)
50
B
25
C
Log10 ω
3.5 4 4.5 5 5.5 6
-25
A Cable aged to 60yrs@50°C
-50
B Motor alone
C Cable aged to 60yrs@50°C
-75 plus motor
A
Figure 46 Comparison of the impedance phase spectra for cables with and
without a load attached
A
Φ(ω) System with no hot spot
B System with hot spot of
7.5 60yrs@60°C
C System with hot spot of
5 60yrs@70°C
C
2.5 A
Log10 ω
5.065 5.075 5.084 5.085 5.090
-2.5
B
-5
Figure 47 Comparison of impedance phase in the frequency range from 116kHz to

123kHz for cables with and without a load attached
74
Figure 48 presents the difference in the impedance phase between the cable system with the hot-
spot and the cable system without the hot-spot. As shown, the differences are localized to the
vicinity of the impedance phase zero crossing. Also, as the severity of the hot-spot increases, the
difference also increases. Therefore, an accurate model of the cable system impedance can be
used to examine the behavior of the impedance phase in the vicinity of the low frequency zero
crossings and this can be used as an indicator for the presence of a hot-spot on the cable’s
insulation.
δΦ(ω)
A System with hot spot of
1.2 60yrs@60°C
System with hot spot of B
B
1 60yrs@70°C
0.8
0.6
0.4
0.2 A
Log10 ω
3.5 4 4.5 5 5.5 6
Figure 48 Change in impedance phase for cable system with a hot-spot from
cable system without a hot-spot
5.2.3 High Frequency BIS Tests for Cables with Attached Loads
In evaluating the high frequency BIS measurements, the frequency averaged second derivative of
the impedance phase spectra was found to provide useful information for detecting hot-spots on
a cable connected to a load. Specifically, the zero crossings of this parameter were found be to
downshifted when the cable’s insulation contains a hot-spot.
The second derivative of the impedance phase, A(T), is developed as follows. At each
frequency Tn at which a measurement is taken, M(Tn) is the measured impedance phase. This
phase value is averaged over a frequency range of Tn ± n0, where n0 = 0.749625MHz (15
frequency steps) for this parameter. This averaged impedance phase is represented by the
following equation:
75
m = n0
∑ Φ (ω ) n+m
Φ (ω n ) =
m = -n0 5.2-1
2 n0
The second derivative of the impedance phase can then be developed as follows:
Ω (ω n ) = Φ (ω n +10 ) - 2 Φ (ω n ) + Φ (ω n -10 ) 5.2-2
This is then averaged over a frequency range of Tn ± s0, where s0 = 1.22429MHz (25 frequency
steps) for this parameter to form the second derivative parameter as follows:
m =s 0
∑ Ω(ω ) n+m
Π (ω n ) =
m = -s0 5.2-3
2 s0
The parameter A(T) is the second derivative in frequency of the cable’s phase impedance. It is
observed that attached loads downshift the zero crossings of A(T). These zero crossings are
referred to herein as the “Detection Zero Crossings” (DZC). For a 10 meter cable, there are 14
DZCs between 1MHz and 100MHz.
The A(T) parameter was compared for the cable systems with and without a hot-spot. An
examination of the A(T) spectra showed that a frequency downshift occurs at all of the 14 DZCs
when a hot-spot is present on the cable. Figure 49 presents an example in which the A(T)
parameter is compared for the cable systems with and without a hot-spot in the vicinity of the
first DZC, which occurs at 11.4MHz. It is observed that a frequency downshift occurs when the
hot-spot is present. Similar results were found for the remaining DZCs.
Figure 49 also shows that the location of the hot-spot relative to the point of measurement also
affects the zero crossing frequency. As shown, the farther the hot-spot is from the POM, the
greater the downshift in frequency of the zero crossing. Thus, in principle, a cable system’s
impedance phase spectra are sensitive to the location of the hot-spot.
76
Π(ω)
0.02
B
ω [Hz]
1.08x107 1.12x107 1.14x107 1.16x107
C
A
-0.02
A System with no hot spot

-0.04 B System with 1m hot spot at
1 to 2 m from POM
C System with 1m hot spot at
8 to 9 m from POM
Figure 49 Comparison of A(T) parameter in the vicinity of the first zero

crossing for cable systems with and without a hot-spot
5.2.4 Phenomenological Models of Cables with Attached Loads
In this section a model for predicting the properties of a thermally aged cable with an attached
load is developed. The model is based on the phenomenological approach and, therefore, is
valid in the frequency range from 10Hz to 1MHz. Models are developed for the following two
cases:
• Thermally aged cable with no hot-spots and a load attached

• Thermally aged cable with a 1 meter hot-spot and a load attached
The first scenario modeled is shown schematically in Figure 50. The cable is 10 meters long and
has been thermally aged to simulate 60 years of service at 50ºC (specimen 26). No hot-spots are
present on the cable, and a load is attached to the far end of the cable, represented by an
impedance, ZL(T). The cable system’s impedance, Zmodel(T), is given by the following equation:
Z L (ω ) + Z 0 (ω ) Tanh [γ (ω ) l ] 5.2-4
Z Model (ω ) = Z 0 (ω )
Z 0 (ω ) + Z L (ω ) Tanh [γ (ω ) l ]
77
Cable
V(ω) Load ZL(ω)
x=l
Figure 50 Schematic of model for thermally aged cable with a load

attached
In Equation 5.2-4 Z0(T) and ((T) are the cable’s characteristic impedance and propagation
function, and ZL(T) is the impedance of the load.
In Section 5.1 phenomenological models were developed for thermally aged cables and the
characteristic impedance and propagation function were extracted from measured impedance
spectra using Equations 3.2-1a and 3.2-1b. Using that information and the measured impedance
of the load, the modeled impedance of the cable system can be constructed. The results are
shown in Figures 51 and 52, which compare the modeled versus measured impedance magnitude
and phase, respectively.
Figure 51 shows that in the frequency range from 100Hz to 1MHz, the modeled impedance
magnitude is within approximately 3% of the measured impedance magnitude. The error is
concentrated in the vicinity of the spectrum’s peak at approximately 126MHz. The model’s peak
is slightly downshifted relative to the measured peak.
Figure 52 shows that in the frequency range from 100Hz to 1MHz, the modeled impedance
phase spectra is within approximately 6% of the measured impedance phase. The error is
concentrated in the vicinity of the spectrum’s zero crossing at approximately 126MHz. The
model’s zero crossing is slightly downshifted relative to the measured zero crossing.
Figure 53 provides a comparison of the modeled and measured impedance magnitude and phase
for the cable system without a hot-spot.
78
|Z(ω)|
A A
3000 Measured
B Predicted
2500
B
2000
1500
1000
500
Log10 ω
3.5 4 4.5 5 5.5 6
Figure 51 Comparison of low frequency measured and predicted

impedance magnitude for a cable system with no hot-spots and an
attached load
Φ(ω)
50
A
25 B
Log10 ω
3.5 4 4.5 5 5.5 6
-25 A
Measured
B
-50 Predicted
-75
Figure 52 Comparison of low frequency measured and predicted impedance phase

spectra for a cable system with no hot-spots and an attached load
79
⏐Z(ω)⏐ Φ(ω)
A A 20
3200 Log10 ω
3000
5.025 5.075 5.125 5.175
B
2800 -20
2600
Log10 ω B
-40
5.025 5.075 5.125 5.175
A
Measured
B
Predicted
Figure 53 Comparison of high frequency measured and predicted impedance magnitude

and phase for a cable system with a hot-spot and an attached load
Next, the model for a cable system with a hot-spot is developed. This case is shown
schematically in Figure 54. The cable is 10 meters long and has been thermally aged to simulate
60 years of service at 50ºC (specimen 29). A 1 meter hot-spot representing 60 years at 70ºC is
present on the cable, and a load is attached to the far end of the cable, represented by an
impedance, ZL(T).
Cable
V(ω) Hot-Spots Load ZL(ω)
x = la x = lb x=l
Figure 54 Schematic of a cable system with a hot-spot and an attached

load
80
Figure 55 compares the measured and modeled impedance magnitude for the cable system in the
frequency range from 1kHz to 1MHz. As shown, the model is within approximately 1% of the
measured impedance magnitude, indicating good agreement.
|Z(ω)|
A A
3000 Measured
B Predicted
2500
B
2000
1500
1000
500
Log10 ω
3.5 4 4.5 5 5.5 6
Figure 55 Comparison of measured and modeled impedance

magnitude for cable system with a hot-spot and an attached load
Figure 56 compares the measured and modeled impedance phase for the cable system in the
frequency range from 1kHz to 1MHz. As shown, the model is within approximately 1% of the
measured impedance phase, indicating good agreement. It is also observed that the modeled
impedance phase in the vicinity of the zero crossing is slightly larger than the measured peak and
is downshifted relative to the measured value.
Figure 57 provides a comparison of the modeled and measured impedance magnitude and phase
for the cable system with a hot-spot.
81
Φ(ω)
50
A
25 B
Log10 ω
3.5 4 4.5 5 5.5 6
-25
A
Measured
-50 B
Predicted
-75
Figure 56 Comparison of low frequency measured and modeled impedance

phase spectra for a cable system with a hot-spot and an attached load
⏐Z(ω)⏐ Φ(ω)
A A 20
3200 Log10 ω
3000
B 5.02 5.06 5.10 5.14
-20
2800
Log10 ω B
5.02 5.06 5.10 5.14 -40
A
Measured
B
Predicted
cable system with a hot-spot and an attached load
5.2.5 Locating a Hot-Spot on a Cable with an Attached Load
82
Here, an approach is developed to locate the position of a hot-spot on a cable with an attached
load using the low frequency impedance spectrum. This work is based on the models developed
in the previous sections for a cable with an attached load. As a demonstration of the approach,
specimen 29 is used, which is thermally aged to simulate 60 years of service @ 50ºC, and has a
1 meter hot-spot representing 60 years @ 70ºC.
For this demonstration, the impedance phase spectrum in the vicinity of the zero crossing at
120kHz is focused on. The cable system’s impedance in the frequency range from 110kHz to
142kHz is calculated based on an assumed location of the hot-spot, and the relative difference
between the modeled phase and measured phase is calculated. Using an iterative procedure, the
assumed hot-spot location is incremented and the difference between measured and modeled
phase values is again calculated. The predicted hot-spot location is obtained when this
difference is a minimum. Figure 58 presents the results of the difference calculations as a
function of assumed hot-spot location. As shown, the hot-spot location is correctly predicted.
<δΦ(x)>
Actual and
Predicted
0.20 Location
0.15
0.10
0.05
x (meters)
2 4 6 8
Figure 58 Prediction of hot-spot location for a cable system with a load

attached
These results demonstrate that the low frequency BIS can be used to predict the location of a
hot-spot on a cable that has a load attached. Relatively good accuracy was demonstrated in the
example presented.
5.3 Application of BIS to Cables with Abrasion Damage
This section presents the results of testing performed to evaluate the use of the BIS method for
83
detecting and locating abrasion damage on cables. First, a discussion of the cables tested is
presented. Subsequently, results of tests performed at low frequency are discussed followed by
the results of tests performed at high frequency.
5.3.1 Abrasion Test Specimens
In this portion of the research, the cable test specimens shown in Table 11 were used.
Table 11 Specimens tested to evaluate BIS for detecting abrasion

No. Specimen ID
15. PNI-79-RB-188-815 10 (393.7) None None
Specimen 15 was used as the baseline representing a healthy cable with no aging or abrasion.
On specimen 22, one of the conductors was abraded using a Dremel tool to grind the cable’s
insulation down to the bare metal conductor over a length of 2mm (0.08in.). On specimens 29
and 30, both conductors were abraded as previously described. A conductive contaminant was
applied to the abraded portion of specimen 30. In all cases, the abrasion was located in the
middle of the cable at 5m (196.85in.) from the end. The following four cases were studied:
• Case A: Unabraded cable

• Case B: Cable with one conductor abraded
• Case C: Cable with two conductors abraded
• Case D Cable with two conductors abraded and a conductive contaminant applied
to the abraded portion of the cable
Case D represents a typical scenario expected in a power plant since a number of contaminants
are commonly found on cables that have not been touched in many years. These contaminants
can absorb moisture and hold it near the cables resulting in a conductive pathway between cables
and ground. For this study, the contaminant used was a mixture of 85% Skydrol hydraulic fluid,
11% iron filings, and 4% carbon black powder. This mixture had as a jelly-like consistency and
was applied directly to the cables in the location of the abrasion. Figure 59 presents a schematic
of the four cases studied.
84
Figure 59 Schematic of four cases studied to evaluate BIS
for detecting and locating abrasion
BIS measurements were made by placing the test specimen in a humidity chamber so that the
temperature and humidity surrounding the cable could be controlled. Test leads were connected
to the cables by extending the cable ends through ports in the wall of the humidity chamber.
Measurements were taken at various humidity levels while holding the temperature at 30ºC
(86ºF).
The BIS measurements were performed over two frequency ranges: 1) the low frequency or non-
oscillatory range, which was 1kHz to 1MHz, and 2) the high frequency or oscillatory range,
which was 1MHz to 100MHz. In the non-oscillatory range, the measurements were performed
with the HP-5094 device, which provided data points at 401 discrete frequencies equally spaced
over the range. In the oscillatory region the measurements were made with the Eclypse
instrument, which provided data points at 2000 discrete frequencies equally spaced over the
range.
5.3.2 Evaluation of Low Frequency BIS Results on Cables with Abrasion
The impedance response of abraded cables in the low frequency range is evaluated in the
following sections.
5.3.2.1 Low Frequency Impedance Measurements on Abraded Cables
85
Figure 60 compares the measured impedance phase spectra for cables with and without abrasion
exposed to an environment of 20% relative humidity (RH). As shown, the abrasion does not
affect the cables’ impedance phase in the non-oscillatory frequency range.
Φ(ω)
Log10 T
3.5 4 4.5 5 5.5 6
-89.1 A Unabraded
B
-89.2
2mm abrasion on one conductor
C
2mm abrasion on both conductors
-89.3
C
-89.4 A
B
-89.5
-89.6
Figure 60 Low frequency impedance phase spectra for cables with and
without abrasion exposed to an environment of 20% Relative Humidity
Figure 61 compares the impedance phase spectra for cables with and without abrasion exposed
to an environment of 85% RH. As shown, when a higher level of humidity is present, all of the
phase spectra have a new structure; i.e., a peak in the vicinity of 5kHz to 10kHz. This structure
arises from the increased humidity level, which provides a conductive pathway for leakage
currents. It is also observed that the abraded cable’s peak is shifted to higher frequencies relative
to the unabraded cable. This frequency shift may be attributable to the fact that the averaged
dielectric function of the abraded cable is smaller than that for the unabraded cable. As a result,
any structures must occur at higher frequencies. It is further observed that there is no readily
observable difference between the phase spectra of the cables with one and two conductors
abraded.
It should be noted that, while the change in the low frequency impedance phase spectra can be
used to detect abrasion under high humidity levels, this may not be practical in a nuclear power
plant environment since the humidity levels may not be high enough to observe these changes.
For example, the peaks previously discussed were not observed when BIS measurements were
made under conditions of 70% RH.
86
A Unabraded
B
Φ(ω) C
-87.00 2mm abrasion on both conductors
-87.25 A
-87.50 B
C
-87.75
Log10 T
3.5 4 4.5 5 5.5 6
-88.25
-88.50
-88.75
Figure 61 Low frequency impedance phase spectra for cables with and
without abrasion exposed to an environment of 85% Relative Humidity
Next, the case in which a conductivity path is present in the vicinity of the abraded portion of the
cables is evaluated (Case D). Figure 62 compares the measured impedance magnitude for cables
with and without abrasion, including cables with abrasion and a conductive contaminant applied
to the cables. The environment was 85% RH for these measurements. As shown, at the low end
of the frequency range (1kHz) the impedance decreases from approximately 245S to 7000S
when the conductive contaminant is present. This is consistent with the loss of insulation, which
degrades the cable’s insulating capability.
Figure 63 presents the impedance phase spectra for the cables with and without abrasion and a
conductive contaminant. Again the environment is 85% RH. As shown, the impedance phase
near the low end of the frequency range increases from approximately -89.5 degrees to -16.7
degrees when the conductive contaminant is placed on the abraded portion of the cable. This
supports the concept that there is a conductivity path between the abraded conductors since the
impedance phase behaves more like a resistor (which has zero degree phase) than a capacitor
(which has a 90 degree phase).
87
Log10 Z(T) A Unabraded
B
C
5 2mm abrasion on both conductors
D
4.5 A plus conductive contaminant
B
4
C
D
3.5
Log10 T
3.5 4 4.5 5 5.5 6
2.5
Figure 62 Impedance magnitude for cables with and without abrasion plus a
conductive contaminant exposed to an environment of 85% RH
A Unabraded
B
M(T) C
D
-20
plus conductive contaminant
-30
-40
-50 D
-60
-70
-80
A B C Log10 T
3.5 4 4.5 5 5.5 6
Figure 63 Impedance phase spectra for cables with and without

abrasion plus a conductive contaminant exposed to an environment of
85% RH.
88
Further examination of the previous two figures shows that, as the frequency increases, the effect
of the conductivity path on the cable’s properties decreases to the point where all of the
impedance spectra converge. This implies that the micro-processes taking place are slow; i.e.,
on the order of milli-seconds, which is reasonable since the contaminant used contained iron
particles.
5.3.2.2 Low Frequency Models of Abraded Cables
A phenomenological approach was used to model the characteristic impedance and propagation
function for abraded cables. First, the characteristic impedance and propagation function for
cables with abrasion only were extracted from the low frequency impedance measurements made
at 85% RH. Figure 64 presents the magnitude of the characteristic impedance for each of the
cables with and without abrasion. The difference from the unabraded case is also presented for
each abraded cable. As shown, the differences are very small (on the order of 1S to 2S or less).
|Ζ0(ω)| δ|Ζ0(ω)|
200 2.5
180 2
160 1.5
140 1 B
A C
120 0.5
B C Log10 T Log10 T
3.5 4.0 4.5 5.0 5.5 6 3.5 4.0 4.5 5.0 5.5 6
A Unabraded
B
C
without abrasion exposed to an environment of 85% RH
Figure 65 presents the characteristic impedance phase spectra for cables with and without
abrasion exposed to an environment of 85% RH. The change from the baseline (unabraded) case
is also presented. Again, the changes are very small ( on the order of ±2 degrees or less).
89
Ψ(ω) δΨ(ω)
Log10 T
3.5 4.0 4.5 5.0 5.5 6 0.2
-5
-10 C 0.1
-15 B Log10 T
-20 A 3.5 4.0 4.5 5.0 5.5 6
-25 -0.1
-30 C
-35 -0.2
B
A Unabraded
B
C
Figures 66 and 67 present the imaginary and real components of the propagation function for
each of the cables with and without abrasion. The difference from the unabraded case is also
presented for each abraded cable. As shown, the differences are very small (on the order of 1%
to 2% or less).
It should be noted that, while the parameter changes are on the order of 1% to 2%, only 0.02% of
the cable’s insulation was removed in these tests. Thus, the measured changes are two orders of
magnitude greater than the amount of insulation removed.
Next, the cables’ electrical properties were extracted from the measured low frequency
impedance spectra. Figure 68 compares the cables’ capacitance with and without abrasion
exposed to an environment of 85% RH. The relative change from the baseline (unabraded) case
is also presented. It is observed that the changes are on the order of 2%.
Figure 69 compares the cables’ conductance with and without abrasion exposed to an
environment of 85% RH. The relative change from the baseline (unabraded) case is also
presented. It is observed that the changes are relatively large; on the order of 15%.
It is noted that abrasion of a small portion of a cable’s insulation cannot by itself change the
cable’s electrical and material properties. However, abrasion down to the metallic conductor of
a cable allows the environment to enter the cable’s insulation and degrade its insulating
capabilities. Specifically, humidity can moisten the insulation and increase its dielectric
function. Also, humidity can form a weak conductivity path between the conductor and ground.
This behavior would not occur when the humidity is low.
90
γ i(ω)
δγi(ω) Log10 T
0.35
0.3 3.5 4.0 4.5 5.0 5.5 6
0.25 -0.0002
C
0.2 -0.0004
0.15 B -0.0006 C
0.1 -0.0008
A B
0.05 Log10 T -0.001
3.5 4.0 4.5 5.0 5.5 6
A Unabraded
B
C
with and without abrasion exposed to an environment of 85% RH
γr (ω) δγ r (ω) Log10 T

Log10 T
4.0 3.5 4.0 4.5 5.0 5.5 6
3.5 4.5 5.0 5.5 6
0.008 -0.00005
C
0.006 -0.0001
B C
0.004 -0.00015
A B
0.002 -0.0002
A Unabraded
B
C
Figure 67 Real component of the low frequency propagation function for cables with and
91
δC(ω)
C(ω)/l [F/m]
0.02
4.1x10-9 Log10 T
0.015
4.0 5.0 6.0
3.8x10-9 0.01 B
C
3.6x10-9 A B 0.005
Log10 T
3.5 4.0 4.5 5.0 5.5 6.0
A Unabraded
B
C
impedance spectra for cables with and without abrasion exposed to an environment of
85% RH
G(ω)/ ω−l (mho-sec/m) δG(ω)/ ω−l

B
B 0.1 C
1.8x10-10
Log10 T
1.4x10-10
A 3.5 4.0 4.5 5.0 5.5 6.0
-10 -0.1
1.0x10 Log10 T
3.5 4.0 4.5 5.0 5.5 6.0 -0.2
A Unabraded
B
C
impedance spectra for cables with and without abrasion exposed to an environment of
85% RH
92
Next, models for cables with both abrasion and a conductive contaminant are developed. The
characteristic impedance, propagation function, and electrical properties can be calculated based
on two scenarios, as follows:
1. Assume the cable is degraded uniformly along its entire length
2. Assume an interaction between the abraded cable conductors, localized at the

point where the conductive contaminant was placed
Since the actual phenomena can best be described as a composite of the above two scenarios,
two models were developed; 1) a spatially averaged degradation model, and 2) a localized cable
interaction model. These models are presented below.
For the spatially averaged cable degradation model, a phenomenological approach is used to
extract the cable’s characteristic impedance, propagation function, and electrical properties from
measured impedance data. Here, the extracted cable parameters represent the quantities spatially
averaged over the length of the cable.
Figure 70 presents the magnitude of the characteristic impedance for each of the cables with and
without abrasion plus a conductive contaminant. The difference from the unabraded case is also
presented for each abraded cable. As shown, the addition of a conductive contaminant greatly
decreases the magnitude of the cable’s characteristic impedance at low frequencies (on the order
of 80%). As the frequency approaches 1MHz, the magnitude of the characteristic impedance for
the cables with and without abrasion merge, as expected.
A Unabraded
D
plus a conductive contaminant
|Ζ0(ω)| [S]
δ|Ζ0(ω)| [S] Log10 T
200
3.5 4.0 4.5 5.0 5.5 6.0
-0.2
150
A -0.4
100 D
D -0.6
Log10 T
3.5 4.0 4.5 5.0 5.5 6.0 -0.8
without abrasion plus a conductive contaminant exposed to an environment of 85% RH
93
Figure 71 presents the characteristic impedance phase spectra for cables with and without
abrasion plus a conductive contaminant exposed to an environment of 85% RH. The difference
from the baseline (unabraded) case is also presented. As shown, the characteristic impedance
phase is significantly changed by the presence of a conductivity path. Specifically, the change in
phase, *Q(T), approaches zero as the frequency approaches 1MHz.
Ψ(ω) δΨ(ω)
20 D 3.5 4.0 4.5 5.0 5.5 6.0
10 -10 Log10 T
3.5 4.0 4.5 5.0 5.5 6.0
-10 -20
Log10 T
-20
A -30 D
-30
A Unabraded
D
Figures 72 and 73 present the imaginary and real components of the propagation function for
each of the cables with and without abrasion plus a conductive contaminant. The difference
from the unabraded case is also presented for each abraded cable. As shown, the imaginary
component of the propagation function increases when a conductive path is present. Again, the
larger changes occur at the very low frequencies (1kHz) and as the frequency increases the
change approaches zero. Similarly, the real component increases by a factor as large as eight at
the low frequencies when the conductive path is present. These results clearly show that the
cable is very dissipative and is on the verge of shorting out.
An examination of the cables resistance and inductance show that these properties show little or
no change in the presence of a conductive path. However, the cable insulation’s dielectric
function (averaged over the length of the cable) is greatly changed by the conductive path.
Figure 74 compares the cables’ capacitance per unit length with and without abrasion plus a
conductive contaminant exposed to an environment of 85% RH. The relative change from the
baseline (unabraded) case is also presented. It is observed that at low frequencies, the cable’s
capacitance per unit length increases by a factor of approximately eight when the conductive
path is present. This difference approaches zero as the frequency approaches 1MHz.
94
A Unabraded
D
γ i(ω) δγ i(ω)
0.35
0.8
D
0.25
0.6
D
0.15 0.4
A
0.2
0.05 Log10 T Log10 T
3.5 4.0 4.5 5.0 5.5 6.0 3.5 4.0 4.5 5.0 5.5 6.0
with and without abrasion plus a conductive contaminant exposed to an environment of
85% RH
A Unabraded
D
γr(ω) δγ r(ω)
0.035 8
D 7
0.025
6 D
0.015 5
A
0.005 4
Log10 T Log10 T
3.5 4.0 4.5 5.0 5.5 6.0 3.5 4.0 4.5 5.0 5.5 6.0
Figure 73 Real component of the low frequency propagation function for cables with and
95
A Unabraded
D
δC (ω)
C (ω)/l [F/m] 8
3.0x10-8 6
2.0x10-8 D 4
D
-8 2
1.0x10
A Log10 T
Log10 T
3.5 4.0 4.5 5.0 5.5 6.0
3.5 4.0 4.5 5.0 5.5 6.0
contaminant exposed to an environment of 85% RH
Figure 75 compares the cables’ conductance per unit length with and without abrasion plus a
conductive contaminant exposed to an environment of 85% RH. The relative change from the
baseline (unabraded) case is also presented. It is observed that at low frequencies, the cable’s
conductance increases by a factor of approximately 1,000 or more when the conductive path is
present. This difference approaches a factor of eight as the frequency approaches 1MHz.
For the localized cable interaction model, the conductivity path is viewed as a point structure in
which current can flow from one conductor to the other at a specified point located in the area of
the abrasion. Thus, it is assumed that the cable is not degraded throughout its entire length; only
a small region is degraded and this is approximated by a point at the location of the conductive
contaminant. Furthermore, it is assumed that the cable interaction that arises from the
conductive contaminant can be represented by a localized impedance, Zc(T), located at a point in
the cable where the conductive contaminant was placed. This model is depicted in Figure 76.
96
A
Unabraded
D
G(ω)/ω−l [mho-sec/m] δG(ω)
700
1.4x10-7
1.0x10-7 500
6.0x10-8 D 300
D
2.0x10 -8
A Log10 T 100
Log10 T
3.5 4.0 4.5 5.0 5.5 6.0
3.5 4.0 4.5 5.0 5.5 6.0
contaminant exposed to an environment of 85% RH
Z0(ω), γ(ω) Z0(ω), γ(ω)
V(ω) Zc(ω)
Z0(ω), γ(ω) Z0(ω), γ(ω)

lc
l
Total current I(ω)
Current flowing in the conductivity path Ic(ω)
Current flowing in the cable after the conductivity
Z0(ω), γ(ω) Cable baseline parameters
Figure 76 Schematic of the localized cable interaction model for abraded

cables plus a conductive contaminant
97
In this model, the current flowing in the cable is separated into two different components, as
follows:
I (ω ) = I cable (ω ) + I C (ω ) 5.3-1
where I(T) is the total current flowing in the cable, Ic(T) is the current that flows through the
localized impedance Zc(T), and Icable is the remaining current that continues to flow in the cable
after the localized area of abrasion. Therefore, the impedances due to the localized interaction,
Zc(T), and the cable length after the abraded point, Zcable(T), are in parallel with each other. This
can be modeled by treating the portion of cable after the conductivity path as a load, as shown in
Figure 77. Here, lc is the distance from the cable end to the abraded portion of the cable, and the
load represents the portion of cable after the conductivity path. The impedance of the portion of
the cable represented by the load, Zcable(T), is determined by the following equation:
Zcable (ω ) = Z0 (ω ) Coth[γ (ω )( − c )] 5.3-2
where l represents the length of the whole cable. In this equation it is assumed that the system is
in the open configuration. It should be noted that, if the impedance of the conductive path is less
than that for the remainder of the cable, the current will short through the conductive path.
Figure 77 Schematic representation of abraded cable with a conductive

path between conductors
The total impedance for the load is represented by the following equation:
Z cable (ω ) Z C (ω )
Z Load (ω ) =
Z cable (ω ) + Z C (ω ) 5.3-3
98
The impedance for a system consisting of a cable connected to a load is represented by the
following equation:
Z Load (ω ) + Z 0 (ω ) Tanh [γ (ω ) l ]
Z Model (ω ) = Z 0 (ω )
Z 0 (ω ) + Z Load (ω ) Tanh [γ (ω ) l ] 5.3-4
Thus, using the measured impedance for Zmodel(T), and the above equations, the impedance of the
conductive path, Zc(T), can be determined. Figure 78 presents the frequency dependence of the
magnitude of Zc(T) in the frequency range from 1kHz to 1MHz for cables with abrasion and
various locations of a conductive path between conductors. As shown, there is no significant
difference in the impedance for the different locations. Therefore, in the non-oscillatory region,
the modeled impedance, Zmodel(T), is essentially independent of the point at which the conductive
path is located, and it decreases with increasing frequency.
It is noted from equations 5.3-2 that Zcable(T) depends on its location on the cable, (l - lc). This is
physically reasonable since, in this region of the frequency spectrum, the electrical waves that
are used to probe the cable’s impedance have wavelengths that are much larger than the cable’s
length. However, the currents that flow in the cable and the conductivity path depend upon the
relative size of the impedances of these two paths. Furthermore, the cable’s total impedance
should react to the location of the conductivity path. This is discussed below.
|ZC(ω)| 1 zc= 0.0m

2
zc= 2.5m
7000 3 zc= 5.0m
4
zc= 7.5m
6000 5 zc= 10.0m
1
2
5000 3 4
4000 5
3000
Log10 ω
3.5 4 4.5 5 5.5 6
Figure 78 Impedance magnitude of conductivity path for abraded cable

with conductive path between conductors at various locations
99
Figure 79 compares the impedances of the conductivity path and the cable as a function of
frequency for the case in which the location of the conductivity path varies. As shown, at very
low frequencies, the magnitude of Zcable(T) is two to three orders of magnitude larger that ZC(T)
for all locations on the cable. Thus, for frequencies below 1kHz, most of the current flows
through the conductivity path and the total impedance is the same as ZC(T). At higher
frequencies ZC(T) approaches Zcable(T), depending upon the frequency and location of the
conductivity path; therefore, more current flows through the cable and less through the
conductivity path. In this region, the total impedance approaches Zcable(T). At frequencies near
1MHz, ZC(T) becomes larger than Zcable(T), and little current flows through the conductivity
path; therefore, the total impedance merges with the impedance Zcable(T).
Figure 80 presents the frequency dependence of the impedance phase for the conductivity path,
>(T), in the frequency range from 1kHz to 1MHz, along with the impedance phase for the total
impedance, M(T), and the impedance phase of Zcable(T), 0(T). As shown, at low frequencies
where most of the current flows through the conductivity path, M(T) is approximately equal to
>(T). As the frequency increases, M(T) approaches 0(T). This is to be expected since the ion
particles cannot follow the probe frequency and, therefore, less current flows in the conductivity
path. Also, if ZC(To) # Zcable(To), where To is the operating frequency of the system, a significant
portion of the current will flow through the conductivity path and the cable will short.
A zc= 0.0m E
zc= 10.0m
Log10 Z(ω) B zc= 2.5m F Interaction
C Impedance
zc= 5.0m G
8 D Total
zc= 7.5m Impedance
7 E
D
C
5
A B
G F
3.5 4 4.5 5 5.5 6
Log10 ω
Figure 79 Comparison of magnitude for total impedance, interaction
impedance, and conductivity path impedance for abraded cable with a
conductivity path at various locations
100
3.5 4 4.5 5 5.5 6
Log10 T
-20
ξ(ω)
-40
Φ(ω)
-60
-80 η(ω)
Figure 80 Comparison of conductivity path impedance phase,

>(T), cable impedance phase, 0(T), and total impedance phase,
M(T), for abraded cable with a conductivity path
5.3.3 Evaluation of High Frequency BIS Results on Cables with Abrasion
This section presents the results of BIS testing in the high frequency range (1MHz to 100MHz)
for abraded cables. The same cable test specimens used for the low frequency testing were used
for this portion of the research. Phenomenological models of the abraded cables are presented
that can be used at high frequency, and are the basis for detecting abrasion on cables. An
approach for locating abrasion on cables is also presented.
5.3.3.1 High Frequency Impedance Measurements on Abraded Cables
The open configuration impedance was measured for each of the cables using the Eclypse
instrument, as discussed previously. Figure 81 presents the averaged high frequency impedance
magnitude and phase spectra for cables with and without abrasion exposed to an environment of
85% RH. The impedance spectra were averaged over a frequency range of ±800MHz to remove
small structures in the spectra associated with additional zero crossings. These small structures
would prevent an accurate analysis of the spectra. As shown, small but significant changes
occur in the impedance spectra when abrasion is present on the cable.
101
|Ζ(ω)| Φ(ω)
700 75
A 50
500 25
B 2x107 4x107 6x107 8x107
300 C -25 ω (Hz)
D -50
100
7
2x10 4x10 7
6x10 7
8x10 7 -75
ω (Hz)
AUnabraded
B
C
D
Figure 81 Comparison of high frequency impedance magnitude and phase
spectra for cables with and without abrasion
Figure 82 shows an expanded view of the high frequency impedance magnitude and phase for
the frequency range from 8.4 MHz to 9.6 MHz, in the vicinity of the first zero crossing. As
shown, the phase zero crossing and the first maxima of the impedance magnitude are shifted to a
higher frequency when the cable is abraded. The largest change is observed for the case in
which the abrasion is covered by a conductive contaminant, thus establishing a conductivity
path.
Figure 83 shows an expanded view of the high frequency impedance magnitude and phase for
the frequency range from 64 MHz to 65.5 MHz. As shown, the phase zero crossing and the
maxima of the impedance magnitude are again shifted to a higher frequency when the cable is
abraded. The changes in this region of the spectra are larger than those in the vicinity of the first
zero crossing. This implies that a small portion of the process remains that is due to the iron
particles in the contaminant. It is noted that the frequency shift in the zero crossing of the cable
with one conductor abraded is larger than that for the cable with both conductors abraded. Thus,
there is no simple pattern in these spectra.
102
|Ζ(ω)| Φ(ω)
60
750 D
D 40
A B 20
700 ω (Hz)
8.4x106 8.8x106 9.2x106
C
650 -20
8.4x106 8.8x106 9.2x106 -40
ω (Hz)
A Unabraded
B
C
D
Figure 82 Comparison of high frequency impedance magnitude and phase spectra for the
frequency range from 8.4 MHz to 9.6MHz for cables with and without abrasion
|Ζ(ω)| Φ(ω)
D D
280
30
B B
270 C 20
C A 10
260 A ω (Hz)
7 7 7
6.46x10 6.48x10 6.52x10 6.54x107
ω (Hz)
6.46x107 6.48x107 6.52x107 6.54x10
7 -10
A
Unabraded
B
C
D
plus contaminant
Figure 83 Comparison of high frequency impedance magnitude and phase spectra in the
frequency range from 64MHz to 65.5MHz for cables with and without abrasion
103
5.3.3.2 High Frequency Phenomenological Models of Abraded Cables
In Section 5.2 the propagation function for thermally degraded cables was extracted from the
high frequency portion of the cable’s measured impedance by extrapolating the low frequency
characteristic impedance to higher frequencies. However, the extrapolated characteristic
impedance exhibited incorrect behavior due to its strong dependence on frequency in the vicinity
of 1MHz. While the extracted characteristic impedance is correct, its form does not lend itself to
extrapolation.
To model the high frequency response of abraded cables, the high frequency portion of the
characteristic impedance is approximated by a constant, z0, which is equal to the value of the
characteristic impedance at a frequency of 1MHz, Z0(T=106). The resulting propagation
function is given by the following equation:
1 ⎡ Zopen (ω ) ⎤
γ (ω ) = Coth − 1 ⎢ ⎥ 5.3-5
l ⎣ z0 ⎦
Figure 84 presents the extracted imaginary component of the propagation function for the cables
with and without abrasion. As shown, only small differences are noted when abrasion is present.
Examination of the real component of the propagation function reveals that, similar to the results
for cables with thermal degradation discussed in Section 5.2, the real component of the
propagation function exhibits large oscillations that are unphysical in nature and arise from the
approximation of the characteristic impedance. To overcome this anomalous behavior, the real
component of the propagation function is fitted to the following function:
γ r (ω ) = β + δ ω 5.3-6
where $ and 6 are fitting constants with the values shown in Table 12.
Figure 85 presents the real component of the propagation function for the cables based on
equation 5.3-6. As shown, the abraded cables are similar to the unabraded cables in terms of
dissipation; however, when the conductive contaminant is applied, the cable becomes much
more dissipative.
104
A Unabraded
B
C
γ i(ω) D
30
25
A
20
B
15 C
D
10
5
ω (Hz)
7 7 7 7
2x10 4x10 6x10 8x10 1x108
Figure 84 Imaginary component of the high frequency

propagation function for cables with abrasion plus a conductive
contaminant
Table 12 Fitting constants for equation representing the real component of

the high frequency propagation function for cables with abrasion plus a
conductive contaminant
Case $ 6
A: Unabraded -0.00380219 0.0115745
B: One Conductor Abraded -0.00518014 0.0115005
C: Two Conductors Abraded -0.00552897 0.0116347
D: Two Conductors Abraded Plus -0.00398439 0.0123597
Conductive Contaminant
105
γr(ω)
0.5
C A
0.4 D B
0.3
0.2
0.1
ω (Hz)
7 7 7 7
2x10 4x10 6x10 8x10 1x108
A Unabraded
B 2mm abrasion on one conductor
C 2mm abrasion on both conductors
D 2mm abrasion on both conductors
Figure 85 Real component of the high frequency propagation function
for cables with abrasion plus a conductive contaminant
A similar approach is used to model the electrical properties of the abraded cables at high
frequencies. The following equations are used:
Resistance: R(ω ) = β + δ ω 5.3-7
δ
Inductance: L(ω ) = β +
ω 0.05 5.3-8
δ α
Capacitance: C(ω ) = β + + 0.5
ω 0.05
ω 5.3-9
δ α
Conductance: G(ω ) = β + + 0.5
ω 0.05
ω 5.3-10
The values of the fitting constants are presented in Table 13.
106
Table 13 Fitting constants for equations representing the high frequency electrical properties of
abraded cables
Case Fitting Constant
$ 6 "
Resistance
Case A: Unabraded Cable 0.2 0.005955662 0
Case B: One Conductor Abraded 0.2 0.005955662 0
Case C: Two Conductors Abraded 0.2 0.005955662 0
Inductance
Case A: Unabraded Cable 4.0 x 10-6 1.11936 0
Case B: One Conductor Abraded 4.0 x 10-6 1.11936 0
Case C: Two Conductors Abraded 4.0 x 10-6 1.11936 0
Capacitance
Case A: Unabraded Cable 1.86 x 10-10 2.97289 x 10-8 1.04147 x 10-9
Case B: One Conductor Abraded 1.86 x 10-10 1.48645 x 10-8 0.52073 x 10-9
Case C: Two Conductors Abraded 1.86 x 10-10 0 0
Conductance
Case A: Unabraded Cable 0 2.97289 x 10-8 3.60049 x 10-11
Case B: One Conductor Abraded 0 1.48645 x 10-8 1.80024 x 10-11
Case C: Two Conductors Abraded 0 0 0
The following figures present the electrical properties of the abraded cables as represented by
Equations 5.3-7 to 5.3-10.
107
R(ω)/l [S/m]
80
A
C
B
60 D
40
20
ω (Hz)
7 7 7 7
2x10 4x10 6x10 8x10 1x108
A Unabraded
C
Figure 86 Resistance spectra for abraded cables in the high frequency
range from phenomenological model
L(ω)/l [H/m]
3.7x10-5 A Unabraded
3.6x10-5
3.5x10-5
20 C
A B
3.4x10-5
D
-5
3.3x10
ω (Hz)
7 7 7 7
2x10 4x10 6x10 8x10 1x108
Figure 87 Inductance spectra for abraded cables in the high frequency range
from phenomenological models
108
C(ω)/l [F/m]
A Unabraded
3.8x10 -9 B 2mm abrasion on one conductor
3.7x10-9
3.6x10-9
D
3.5x10-9 C
A B
7 ω (Hz)
2x10 4x107 6x107 8x107 8
1x10
Figure 88 Capacitance spectra for abraded cables in the high frequency range from
phenomenological models
G(ω)/ω−l [mho-sec/m]
A Unabraded
1.25x10-10 C 2mm abrasion on both conductors
1.00x10-10 plus conductive contaminant
0.75x10-10 D
A
0.50x10-10
C
-10
B
0.25x10
ω (Hz)
7 7 7 7
2x10 4x10 6x10 8x10 1x108
Figure 89 Conductance spectra for abraded cables in the high frequency range from
phenomenological models
5.3.4 Detection of Abrasion Damage Using BIS
109
Two different approaches were developed for the detection of abrasion on cables using BIS. The
first involves the properties of the propagation function; specifically the inverse of the velocity
of propagation of electrical signals that propagate on the cable, vp-1(T), which is defined by the
following equation:
dγ i (ω )
v -p1 (ω ) ≡ 5.3-11
dω
The second approach is phenomenological in nature and involves the zero crossings of the
second derivative of the impedance phase, A(T), which is defined by the following equation:
d2 Φ
Π (ω ) ≡
dω 2 5.3-12
In Equation 5.3-12, <M> implies an average of the impedance phase over a frequency range of
±600kHz.
It can be shown that the slope of vp-1(T) increases whenever the cable’s insulation is abraded.
From Equation 5.3-11 the imaginary component of the propagation function is required in the
oscillatory region of the frequency spectrum. However, the slope of the extracted parameter
exhibits large oscillations that are unphysical in nature. This behavior arises from the
approximation used for the cable’s characteristic impedance. To overcome this anomalous
behavior, the imaginary component of the propagation function is fitted to a linear function, as
follows:
γ i (x ) = α + x ς 5.3-13
In this equation, discrete x values from 1 to 2000 replace the continuous frequency spectra, T,
over the range from 1MHz to 100MHz. Recall that the impedance measurements were made at
2000 discrete frequencies in the high frequency range. These individual data points correspond
to the x values from 1 to 2000. This approximation ignores the propagation velocity’s
dispersion. Figure 90 presents the slope of this equation, H. As shown, the slope for the
unabraded cable is smaller than that for any of the abraded cables. As a result the slope for each
abraded cable can be compared to the baseline (unabraded) cable as an indicator for the presence
of abrasion on the cable. These results imply that the velocity of propagation for electrical
signals traveling on the cable decreases when the insulation is abraded.
110
ς
C
B
D 2x107 4x107 6x107 8x107 1x108
ω (Hz)
0.01678
A Unabraded
0.01676 C 2mm abrasion on both conductors
0.01674 A
Figure 90 Slope of the equation representing the imaginary component of the high
frequency propagation function for abraded cables
The approach using the zero crossings of the second derivative of the impedance phase, A(T), to
detect abrasion is now discussed. Figure 91 compares the zero crossings of A(T) for the cables
with and without abrasion. It is observed that the zero crossings are downshifted to lower
frequencies when abrasion is present on the cables. The reason for this behavior is not
understood. This behavior is not observed for the impedance phase spectra. In general, the zero
crossing of the impedance phase depends upon the product [L(T) C(T)]-½, and this is true for
A(T) as well. However, in the oscillatory region of the frequency spectrum the impedance is
dominated by wave phenomena, especially interference. As a result, the behavior of the zero
crossings when the cable’s properties are changed is complex and cannot easily be explained.
Figure 92 presents A(T) in the frequency range from 4.2MHz to 5.2MHz. As shown the zero
crossing of the abraded cables is downshifted relative to the unabraded cable. It is noted that the
zero crossing of the abraded cables are the same, despite the differences in the cable’s insulation.
111
Π (ω)
ω (H z)
4 .5 x1 0 6 4 .6 x1 0 6 4 .7 x1 0 6 4 .8 x1 0 6
-1 A
D
B
C
-2
A U n a b ra d ed
-3 B 2 m m a b ra sio n o n o n e co n d u cto r
C 2 m m a b ra sio n o n b o th co n d u cto rs
D 2 m m a b ra sio n o n b o th co n d u cto rs
p lu s co n d u ctiv e co n ta m in a n t
Figure 91 Comparison of zero crossings in the frequency range 4.4MHz
to 4.8MHz for the second derivative of the impedance phase, A(T), for
cables with and without abrasion
Π(ω) 4
A
2
D C B
ω (Hz)
2.26x10 7 2.28x10 7 2.32x10 7 2.34x10 7
-2
A Unabraded
B
C
D 2mm abrasion on both conductors plus
Figure 92 Comparison of the zero crossings of the second derivative of the
impedance phase, A(T), in the frequency range from 22.4MHz to 23.8MHz for
cables with and without abrasion
112
Similar results were found for other zero crossings examined. These results show that a select
number of zero crossings of A(T) can be used to determine if the cable’s insulation is abraded.
When abrasion is present, the zero crossings for A(T) are downshifted in frequency. For the
tests performed in this study, 28 out of 29 zero crossings examined followed this rule.
5.3.5 Location of Abrasion Damage Using BIS
The high frequency phenomenological models developed previously for abraded cables are not
useful for locating this damage. This is due to the following observation. The cable insulation’s
dielectric function follows the fractional power law for dielectric materials, which is:
⎛ ⎞
ε (ω ) = ε 0 ⎜⎜ 1 +
A1 A2 A3
+ + ⎟⎟
⎝ (
1 + (i ω τ 1 )) (
n1
1 + (i ω τ 2 ) ) (
n2
1 + (i ω τ 3 ) 3
n
) ⎠
5.3-14
The first term in the above equation refers to ionic processes, while the second and third terms
refer to electronic processes that contribute to the cable insulation dielectric function. Here, g0 is
the dielectric function of vacuum, and A1, A2 and A3 refer to various micro processes that
dominate the insulation’s dielectric function. When a portion of the insulation is abraded down
to the metallic conductor, the parameters A1 = A2 = A3 = 0, and g(T) = g0. The
phenomenological models developed previously deal with the capacitance between two
conductors in a cable via the structure factor, 7, and knowledge of this structure factor is
imprecise. Therefore, a model is required that includes knowledge of all the parameters
contained in the cable’s capacitance. Also, if only one of the wires is abraded, then the
capacitance per unit length and conductance per unit length are given by the following equation:
⎛ 1 1 1 ⎞
⎜ A1 A2 A3 ⎟
G (ω ) + i ω C(ω ) = Λ ε 0 ⎜1 + 2 + 2 + 2 ⎟
⎜⎜ (
1 + (i ω τ 1 ) 1
n
) (
1 + (i ω τ 2 ) 2
n
) (
1 + (i ω τ 3 ) 3
n
) ⎟⎟ 5.3-15
⎝ ⎠
If both of the conductors are abraded to the metallic conductor (thus there is no insulation
present), the capacitance per unit length and conductance per unit length are given by the
following equation:
G (ω ) + i ω C (ω ) = Λ ε 0 5.3-16
The phenomenological models developed previously cannot accurately represent this scenario.
As such, at high frequencies the following models for resistance per unit length and inductance
per unit length can be used:
113
⎛ ω ⎞⎟
R (ω ) = R 0 ⎜⎜ 1 +
ω s ⎟⎠
5.3-17
⎝
⎛
L(ω ) = L 0 ⎜⎜1 - 0.5
(ω/ω l )nl ⎞
⎟
1 + (ω/ω l ) l
2n ⎟ 5.3-18
⎝ ⎠
At high frequencies the inoic portion of g(T) can be ignored since the ions are too slow to react
to the high frequency fields, and the fractional power law for the insulation’s dielectric function
reduces to the following equation:
⎛ ⎞
ε (ω ) = Λ ε 0 ⎜⎜ 1 +
A2 A3
+ ⎟
⎝ (1 + (i ω τ 2 )
n2
) (1 + (i ω τ ) )
3
n3 ⎟
⎠
5.3-19
For a cable with a twisted pair of conductors, the structure factor becomes:
π
Λ=
Arc Coth s d ( ) 5.3-20
Here “s” is the distance between the centers of the conductors and “d” is their diameter. The
various constants in the above equations arise from various physical processes. At high
frequencies the cable’s resistance is dominated by the skin effect and Equation 5.3-17 is the
asymptotic form of the skin effect, with Ts being the skin frequency. The various constants can
be determined by obtaining the cable’s characteristic impedance and propagation function from
the measured impedance spectra, and then calculating the open configuration impedance using
assumed values for the constants. The measured impedance can then be compared to the
calculated impedance and the error calculated. Using an iterative process, the final value of the
constants can be determined by minimizing the rms error between the calculated impedance and
the measured impedance.
For the unabraded cable (Specimen 15) the best fit values for the constants are the following:
Resistance per unit length: R0 = 0.02S/m
Ts = 28.183kHz
Inductance per unit length L0 = 4.1:H/m

T1 = 10MHz
n1 = 0.275
Dielectric Function 7g0 = 0.185nF/m
114
A2 = 1.2, J2 = 2.82:s, n2 = 0.5
A3 = 3.94 J3 = 0.1:s, n3 = 0.022
The characteristic impedance and propagation function are then calculated using Equations
3.2-1a and 3.2-1b. The open configuration impedance is then calculated using Equation 3.2-7a.
Φ(ω) [Degrees]
75
B
50
A
25
ω [Hz]
6 6 6 6
2x10 4x10 6x10 8x10 1x107
-25
-50
-75 A
Measured
B Predicted
Figure 93 Comparison of predicted and measured impedance phase in the
range from 2.55MHz to 10.04MHz for unabraded cable (Specimen 15)
Figure 94 compares the cable impedance phase predicted by the model to the measured
impedance. The rms error of the impedance phase is 4.76%, indicating good agreement between
the predicted and measured values. As shown, the impedance phase predicted by the model
deviates slightly from the measured impedance after the first zero crossing. Also, there is a
small deviation between the model and the measured values in the vicinity of 1MHz to 2MHz.
Since the model was developed in the frequency range from 2.55MHz to 10.04MHz it is not
surprising that deviations appear outside this range.
Figure 94 compares the cable impedance magnitude predicted by the model to the measured
impedance. Again, good agreement is observed between the predicted and measured values.
115
|Z(ω) |/l [S /m]
2500
A
Measured
2000
B Predicted
1500
1000
A
500
B
ω [Hz]
2x106 4x106 6x106 8x106 1x107
Figure 94 Comparison of predicted and measured impedance magnitude in

the range from 2.55MHz to 10.04MHz for unabraded cable (Specimen 15)
Next, the characteristic impedance and propagation function for cables with one or both
conductors abraded is determined. This is useful since the abraded cable is modeled as if it is in
two states; 1) the conductor has all of its insulation (i.e., unabraded), and 2) the conductor is
bare.
Figures 95 and 96 present the characteristic impedance magnitude and phase in the frequency
range from 2.55MHz to 10.04MHz for the cables with and without abrasion. As shown, the
characteristic impedance magnitude increases when the insulation is abraded. The characteristic
impedance phase decreases when the insulation is abraded.
Figures 97 and 98 present the imaginary and real components of the propagation function in the
frequency range from 2.55MHz to 10.04MHz for the cables with and without abrasion. In all
cases the behavior of the propagation function is as expected in light of the fact that the cable’s
capacitance and conductance decrease when the insulation is abraded.
116
|Z0 (T)|
C
100
A Unabraded
90 B 2mm abrasion on one conductor
80
B
70
A ω (Hz)
4x10 6 6x10 6 8x10 6 1x10 7
Figure 95 Characteristic impedance magnitude in the frequency range

from 2.55MHz to 10.04MHz for cables with and without abrasion
Ψ(ω)
A
0.2
B
ω (Hz)
4x106 6x106 8x106 1x107
-0.2
C
-0.4
A Unabraded
-0.6
C
Figure 96 Characteristic impedance phase spectra in the frequency range
from 2.55MHz to 10.04MHz for cables with and without abrasion
117
A Unabraded
γ i(ω) B
0.35 2mm abrasion on one conductor
0.3
0.25 A
0.2 B
0.15
C
0.1
0.05
ω (Hz)
6 6 6
4x10 6x10 8x10 1x10 7
Figure 97 Imaginary component of the propagation function in the frequency

range from 2.55MHz to 10.04MHz for cables with and without abrasion
A Unabraded
γr(ω) B 2mm abrasion on one conductor
0.005
A
0.004
0.003 B
0.002
0.001 C
ω (Hz)
4x106 6x106 8x106 1x107
Figure 98 Real component of the propagation function in the frequency

range from 2.55MHz to 10.04MHz for cables with and without abrasion
118
In this section, an approach for locating abrasion damage is presented using the cable models
developed in the previous section. Specifically, two cases are evaluated; 1) a cable with the
insulation abraded on one conductor, and 2) a cable with the insulation abraded on both
conductors. In both cases, the models are able to predict the location of the abrasion within 10%.
The approach used is the same as that used to locate insulation damage when the cable is
thermally aged, which is discussed in Section 5.2. As such, different models are developed for
each of the two cases being studied.
5.3.5.1 Model for Cable with One Conductor Abraded
The scenario evaluated here is depicted in Figure 44 as Case B. A 2mm (0.08in.) length of
insulation is abraded from one conductor in the middle of the cable. The total cable length is
10m (393.7in.) and the start and end points of the abraded section are labeled la and lb,
respectively. The cable is not shorted at the point of abrasion.
For this case, the abraded cable is in two states; the abraded portion is in a “damaged” state, and
the unabraded portion is in a “healthy” state. The model for the impedance of the cable is
represented by the following equation:
Z0 , h ( ω ) Coth [ γ h ( ω ) ( − a )] + Z 0 , d ( ω ) Tanh [ γ d (ω ) ( b − a )]
Zmodel ( ω ) = Z 0 ,h ( ω )
Z 0 , d ( ω ) + Z 0 ,h ( ω ) Coth[ γ h (ω ) ( − b )] Tanh [ γ d (ω ) ( b − a )] 5.3-21
In Equation 5.3-21, the subscript “d” refers to the cable property in the damaged state, and the
subscript “h” refers to the cable property in the healthy state. The parameter “l” is the total
length of the cable. This model will be used together with the measured impedance to locate the
abrasion damage on the cable.
Figure 99 presents the measured impedance phase spectra in the frequency range from
4.048MHz to 15.04MHz for the cables with and without abrasion. As shown, the cables are very
similar with small differences in the spectra after the first zero crossing. These differences are
on the order of the deviations observed between the models presented in the previous section.
119
Φ (ω) A U nabraded
B 2m m ab rasion on one conductor
75
50
A
25
B
ω (H z)
0.8x10 7 1.0x10 7 1.2x10 7 1.4x10 7 1.6x10 7
-25
-50
-75
Figure 99 Measured impedance phase spectra in the frequency range from

4.048MHz to 15.04MHz for cables with and without abrasion on one conductor
To locate the abrasion damage on the cable, the following procedure is used:
• Calculate the impedance of the cable using Equation 5.3-21 and an assumed location for
the abrasion
• Compare the calculated impedance to the measured impedance and compute the rms error
in the values at each data point
• Using an iterative process, assume new locations for the abrasion along the cable and
compute the rms error between the predicted and measured impedance for each location
• The location that minimizes the rms error is the predicted location of the abrasion
For this example, abrasion locations were assumed every 1 meter (39.37in.) along the cable. The
impedance phase values were compared at discrete frequencies in the range from 4.048MHz to
15.04MHz. The error in the predicted phase values was computed using the following equation:
ω n n = 15.04MHz
∑Φ Model(ω , l ) - Φ Measured (ω )
n b
5.3-22
ω n = 4.048MHz
δΦ = ω n n = 15.04MHz
∑Φ Measured (ω )
ω n = 4.048MHz
120
In this example, the model predicts that the abrasion is located at 6 meters from the end of the
cable and the correct location is 5 meters. Figure 100 presents the results of the error
calculations showing the minima at 6 meters, which represents the predicted damage location. It
is noted that the error at this predicted location is 8.55%.
<δΦ(x)>
0.088
0.0875
0.087
Actual Location
0.0865
x (meters)
2 4 6 8 10
0.0855
Predicted
Location
Figure 100 Predicted location of abrasion damage for cable with one conductor
abraded
5.3.5.2 Model for Cable with Two Conductors Abraded
The scenario evaluated here is depicted in Figure 44 as Case C. A 2mm (0.08in.) length of
insulation is abraded from both conductors in the middle of the cable. The total cable length is
10m (393.7in.) and the start and end points of the abraded section are labeled la and lb,
respectively. The cable is not shorted at the point of abrasion.
Figure 101 presents the measured impedance phase spectra in the frequency range from
4.048MHz to 15.04MHz for the cables with and without abrasion. As shown, the cables are very
similar with small differences in the spectra before the first zero crossing and in the region
before and after the third zero crossing. Again, these differences are on the order of the
deviations observed between the models presented previously.
121
Φ(ω) A Unabraded
B 2mm abrasion on two conductors
75
50
A
25
B
ω (Hz)
7 7 7 7
0.8x10 1.0x10 1.2x10 1.4x10 1.6x107
-25
-50
-75
Figure 101 Measured impedance phase spectra in the frequency range from
4.048MHz to 15.04MHz for cables with and without abrasion on both
conductors
Using the appropriate characteristic impedance and propagation function, the iterative process
described previously in this section was used to predict the location of the abrasion. The results
predict a location of 6 meters, while the correct location is 5 meters. Figure 102 presents the
error calculations and the predicted versus actual location of the abrasion.
These results demonstrate the BIS method can be used to predict the location of abrasion damage
with relatively good accuracy.
5.4 Application of BIS to Cables with Cracking Damage
Cracking of the insulation on aged cables is a potential problem in nuclear power plants since it
can lead to leakage currents that may cause the system to fail. This is particularly true under
humid or wet conditions, during which moisture can enter the crack and create a conductive path
between the cable conductor and ground. Therefore, it is desirable for a condition monitoring
technique to be able to detect and locate cracking in cables.
In this section, the BIS method is evaluated for its ability to detect and locate cracking damage
on cables.
122
<δΦ(x)>
Actual Location
0.0931
x (meters)
2 4 6 8 10
0.0929
0.0928
0.0927 Predicted
Location
Figure 102 Predicted location of abrasion damage on cable with both conductors
abraded
5.4.1 Test Specimens Used for Testing of Cables with Cracking
For this portion of the research, unaged cables were used. To simulate the cracking damage a
knife was used to slice the insulation on both conductors of the cable. A total of 12 simulated
cracks were placed on the cable over a length of 2.54 cm (1in.). The crack location was 4 meters
from the left end of the cable. Figure 103 shows a schematic of the scenario evaluated.
Baseline impedance measurements were obtained using the Eclypse instrument and the HP
instrument. Measurement parameters used were a 1-Volt oscillation level, medium integration
time, and 4-samples of averaging. High frequency measurements were collected at 2000 discrete
frequency points between 1MHz and 100MHz. Low frequency measurements were collected at
logarithmically spaced points over the frequency range from 1kHz to 1MHz. Measurements
were made with the conductors in both the open and the shorted configuration, and from both
ends of the cable. The cable was coiled and placed in a humidity chamber that was maintained at
60%RH and 30ºC for all measurements.
123
4m
A V(T) B
9m
12 cracks over 2.54 cm
Figure 103 Schematic of cable with cracking damage
After baseline impedance measurements were made, the cable was removed from the
environmental chamber and 12 cracks were simulated on each conductor of the cable. Each
simulated crack was equally spaced at 0.21mm, with a width of 0.1mm. The cracks were first
made to a depth of 50% of the insulation thickness and measurements were made. A conductive
contaminant consisting of 85% Skydrol hydraulic fluid, 11% iron particles, and 4% carbon black
powder was applied to the cracked portion of the cable and the impedance measurements were
repeated. The cracks were then extended completely through the insulation (100%) and
additional conductive contaminant was applied, after which the impedance measurements were
again performed. Thus, four sets of impedance measurements were made on the cable.
5.4.2 Low Frequency BIS Measurements of Cables with Cracking
Figure 104 compares the impedance phase spectra in the low frequency range from 1kHz to
1MHz for cables with and without cracking. As shown, there is only a small difference between
the uncracked cable and the cable with cracks. Even the conductive path does not appear to
impact the impedance phase. This can be attributed to the lack of a conductivity path between
the conductors, even with the conductive contaminant applied. A possible reason for this is the
size of the iron filings used, which were on the order of 100:. This is approximately the same
dimension as the crack width. Thus, it is believed that the iron particles could not enter the crack
and make a connection with the cable conductor.
124
Φ(ω)
Φ(ω) [Hz] ω [Hz]
200000 400000 600000 80000 1x106

0
-89.1
ω
-89.2
B
C A
-89.3
-89.4 A
No
B
50% Cracking
-89.5 C 100% Cracking plus a
Figure 104 Comparison of impedance phase for cables with and without
cracking
Although the differences in the impedance spectra are very small, the data were analyzed to
determine the spatially averaged electrical properties of the cable. Figures 105 and 106 present
the characteristic impedance magnitude and phase spectra, and the propagation function,
respectively, for the cable with and without cracking. As shown, only small differences are
observed between the cable with no cracking and the cable with cracking.
Figure 107 compares the resistance per unit length for the cable with and without cracking. As
shown the resistance per unit length does not change when the cable contains cracks, even when
a conductive contaminant is applied to the cracks.
Figure 107 compares the inductance per unit length for the cable with and without cracking. As
shown the inductance per unit length increases slightly when the cable contains cracks and a
conductive contaminant is applied to the cracks. This change is more apparent near the upper
end of the low frequency region.
Figure 108 compares the capacitance per unit length for the cable with and without cracking. As
shown the capacitance per unit length increases slightly when the cable contains cracks and a
conductive contaminant is applied to the cracks. This change is also more apparent near the
upper end of the low frequency region. It is observed that the capacitance per unit length for the
cable with cracking but no conductive contaminant was higher than the baseline cable with no
cracking, and the cable with cracking plus a conductive contaminant.
125
Log10|Z(T)| Ψ(ω)
2.35
-1
2.3
2.25 -2
2.2
-3
2.15
2.1 -4
2.05
ω [Hz] ω [Hz]
3.5 4 4.5 5 5.5 6
200000 600000 1x106
A ω
No
B 1
50% Cracking
C 100% Cracking plus a
Figure 105 Characteristic impedance log magnitude and phase for cables with and
without cracking
γ i (ω)
γr (ω)
0.035 0.0012
0.03 ω [Hz]
0.025 C 20000 60000 1x106
0.0008
0.02 B C
0.015 A 0.0006
0.01 0.0004 A B
0.005 ω [Hz]
0.0002
20000 60000 1x106
A
No Cracking
B
50% Cracking
C
100% Cracking plus a
Figure 106 Imaginary and real components of the propagation function for
cables with and without cracking
126
R(ω)/l [Ω/m]
0.175
0.15
C
0.125 B
A
ω [Hz]
200000 600000 1x106
0.075 A
No Cracking
B
0.05 50% Cracking
C
0.025
Figure 107 Resistance per unit length for cables with and without cracking
L(ω)/l [H/m]
A
No Cracking
B
4.4x10-6 50% Cracking
C
4.2x10-6
ω [Hz]
100000 200000 300000 400000
A
3.8x10-6 B
C
Figure 108 Inductance per unit length for cables with and without cracking
127
C(ω)/l [F/m]
A
3.58x10 -10 No Cracking
B
50% Cracking
C
3.56x10-10 100% Cracking plus a
3.54x10-10
3.52x10-10
B
A C ω [Hz]
50000 100000 150000 200000
Figure 109 Comparison of capacitance per unit length for cable with and
without cracks
Figure 110 compares the conductance per frequency per unit length for the cable with and
without cracking. As shown the conductance per unit length increases slightly when the cable
contains cracks and a conductive contaminant is applied to the cracks. This change is also more
apparent near the upper end of the low frequency region. It is observed that the capacitance per
unit length for the cable with cracking but no conductive contaminant was higher than the
baseline cable with no cracking, and the cable with cracking plus a conductive contaminant.
5.4.3 High Frequency BIS Measurements of Cables with Cracking
In this section the high frequency impedance spectra of the cable test specimens is analyzed to
develop an approach for detecting cracking on cables. Specifically, a similar approach to that
used in Section 5.2 for evaluating attached loads is used here in which the zero crossings of the
second derivative of the impedance phase spectra, A(T), are examined for differences. These
differences are then used as indicators for the presence of cracking. The development and
definition of the parameter A(T) was presented previously in Equations 5.2-1 through 5.2-3.
It is observed from the test results that, for a 9-meter cable, there are eight detection zero
crossings (DZCs) in the high frequency range used for the cracking tests (i.e., 1MHz to
100MHz).
Figure 111 compares the first DZC for A(T) for the cables with and without cracking. As
shown, the DZC is downshifted to lower frequencies when cracking is present on the cables.
128
Similar results were observed for the other seven DZCs. These results suggest that monitoring
the zeros crossings of the parameter A(T) may be an effective indicator for the presence of
cracking on a cable.
G(ω)/ω−l [mho-sec/m]
A
No Cracking
4.25x10-12 B
50% Cracking
4.00x10-12 C
3.75x10-12 C
-12
3.50x10 B
3.25x10-12 A ω [Hz]
50000 100000 150000 200000

-12
2.75x10
Figure 110 Conductance per frequency per unit length for cable with and
without cracks
129
Π(ω) A
No Cracking
4 B
50% Cracking
C 50% Cracking plus a
2 conductive contaminant
D 100% Cracking plus a
ω [Hz]
1.48x107 1.52x107 1.54x107 1.56x107 1.58x107
B A
-2
C D
-4
-6
Figure 111 Comparison of second derivative of the impedance spectra,

A(T), for cables with and without cracking
It should be noted that cracking of a cable’s insulation arises when a cable that has been
embrittled by long-term exposure to environmental stresses is physically stressed. For example,
bending or crushing of an embrittled cable can lead to cracking of the insulation. In the
experiments performed as part of this research, cracking was simulated by slicing the cable’s
insulation using a knife. This approach was selected since it was an efficient method of
simulating the physical characteristics of a crack; however, the approach does not accurately
capture the physics or the chemistry of the cracking. As a result, the results demonstrate the
potential for detecting cracking using the BIS method; however, accurate models for real world
cracking are not currently available, thus, a model for locating cracking could not be developed
as part of this study. Additional research using more realistically simulated cracking is
recommended to further evaluate the BIS method for detecting this type of degradation.
Continuous BIS measurement during the aging process would be beneficial to understanding and
modeling the cracking process. Chemical analysis of the cable’s insulation would also be useful.
5.5 Application of BIS to Cables with Varying Environments
Electric cables in nuclear power plants can be very long and can traverse several different areas
of the plant. Thus, the environment along the length of the cable may vary. In order to apply a
condition monitoring technique to cables in their installed configuration, the impact of this
varying environment on the measurements must be evaluated. To investigate the impact of
130
varying environments, long runs of cable were tested. Several different environments were
imposed on sections of the cable during the tests to determine the impact on the test results.
5.5.1 Specimens Tested to Evaluate BIS for Cables with a Varying Environment
For this portion of the research, the cable test specimens shown in Table 14 were tested. The 10
meter cables were used to provide data for developing the phenomenological models for the 100
meter cables. The 100 meter cables were tested with constant temperature and humidity on the
external surface of the cables, and with a varying environment along their length. Comparisons
were made between the test results not only for the different environments, but also for the
cables with and without the hot-spot.
BIS tests were performed over the frequency range from 1kHz to 1MHz using the HP
instrument. Due to the length of test specimens 47 and 48, these cables were tested with the first
10 meters inside the test facility building, and the remainder outside the building. As a result,
the cables were exposed to two different environments. Inside the building, the temperature was
maintained at approximately 20ºC (68ºF) with low humidity. The portion of cable inside the
building was also placed in a humidity chamber for several tests. The temperature outside the
building varied several degrees as a function of location, and as a function of time of day.
Table 14 Specimens tested to evaluate BIS for testing cables with a varying environment

No. Specimen ID
26. PNI-79-RB-188-826 10 (393.7) 60yrs. @ 50ºC None
43. PNI-79-RB-188-843 10 (393.7) 60yrs. @ 90ºC None
47. PNI-79-RB-188-847 100 (3,937) None 60yrs. @ 90ºC
48. PNI-79-RB-188-848 100 (3,937) None None
Figure 112 depicts the test setup schematically.
131
Inside the building
50m
Humidity
Chamber
Hot spot at
96m to 97m
Figure 112 Schematic of test setup for cables with varying environment
5.5.2 Low Frequency BIS Results
Figure 113 compares the measured impedance magnitude for the 100 meter cable without any
aging or hot-spot (specimen 47) using external temperatures of 24ºC (75ºF), and 42ºC (107.6ºF)
along the entire length of the cable. The results presented cover a frequency range of 400kHz to
1MHz. As shown, the impedance magnitude peak value is downshifted slightly for the case in
which the higher external temperature was used.
132
|Ζ ( ω)|
1200 B
A
1000
A 2 4 ºC (7 5ºF )
800 B
4 2 ºC (1 0 8ºF )
600
400
200
ω [H z]
400000 600000 800000 1 x1 0 6
Figure 113 Impedance magnitude for 100 meter cable with no aging or
hot-spot and external temperatures of 24ºC (75ºF), and 42ºC (107.6ºF)
Figure 114 compares the impedance phase spectra for the 100 meter cable with no aging or hot-
spots (specimen 48) and external temperatures of 24ºC (75ºF), and 42ºC (107.6ºF). As shown,
the higher external temperature causes a downshift in the first zero crossing of the impedance
phase.
Φ(ω)
20
A 24ºC (75ºF)
B 42ºC (108ºF)
10
A
432000 434000 436000 438000 440000 442000 444000
-10
B
-20
-30
Figure 114 Impedance phase spectra for 100 meter cable with no aging or
133
Figure 115 compares the impedance phase spectra for the 100 meter cable with no aging or hot-
spot and varying humidity levels over the first 10 meters of cable. For this test, the first 10
meters of the cable were placed in a humidity chamber and the conditions were maintained at
29ºC (85ºF) and 85% RH. As shown, the higher external humidity causes a downshift in the first
zero crossing of the impedance phase, similar to that observed for the elevated temperature case.
Φ(ω)
A
20 No humidity
B
Humidity
10 A
ω [Hz]
432000 434000 436000 438000 440000 442000 444000
-10 B
-20
-30
Figure 115 Impedance phase spectra for 100 meter cable with no aging or
hot-spot and external temperatures of 24ºC (75ºF) low humidity, and 29ºC
(85ºF) and 85% RH.
Next, the impact of a varying external temperature on a cable with a 1 meter hot-spot (specimen
47) was examined. Figure 116 compares the measured impedance phase spectra for the 100
meter cable with a 1 meter hot-spot and external temperatures of 24ºC (75ºF), and 42ºC
(107.6ºF). As shown, the higher external temperature causes an upshift in the zero crossing of
the impedance phase. This is contrary to previous observations and the reason for this behavior
is not readily apparent.
Finally, a comparison of the two 100 meter cables was made to examine the impact of the hot-
spot on the cables properties, particularly in the vicinity of the zero crossings. Figure 117
compares the measured impedance phase spectra for the two 100 meter cables. For this
measurement, the external temperature was maintained at 24ºC (75ºF). As shown, the zero
crossing of the impedance phase is downshifted by the presence of the 1 meter hot-spot. It
should be noted that for this figure, the impedance phase is limited to the range from -90 to +90,
since this is the physical range of the impedance phase. Measured values outside this range are
134
unphysical in nature and arise from instrument error, therefore, they are eliminated from the
figure. However, the zero crossing frequencies presented are correct.
Φ(ω)
30
A 24ºC (75ºF)
20 B 42ºC (108ºF) A
10
ω [H z]
430000 435000 440000 445000
-10
B
-20
-30
-40
Figure 116 Impedance phase spectra for 100 meter cable with 1 meter
Φ(ω) B
75
50 A
25
ω [Hz]
400000 600000 800000 1x10 6
-25
A No Hot-Spot
-50 B
1m Hot-Spot
-75
Figure 117 Impedance phase spectra for 100 meter cables with and
without a 1 meter hot-spot, and an external temperature of 24ºC (75ºF)
135
Figure 118 shows an expanded view of the measured impedance phase spectra for the two 100
meter cables in the vicinity of the first zero crossing with an external temperature of 24ºC (75ºF).
The downshift in the zero crossing of the impedance phase is clearly visible with the presence of
the 1 meter hot-spot.
Φ(ω)
A 24ºC (75ºF)
20
B B
42ºC (108ºF)
10
ω [Hz]
430000 435000 440000 445000
-10
A
-20
-30
-40
Figure 118 Expanded view of the impedance phase spectra in the

frequency range from 425kHz to 445kHz for 100 meter cables with
and without a 1 meter hot-spot, and an external temperature of 24ºC
(75ºF)
Figure 119 shows an expanded view of the measured impedance phase spectra for the two 100
meter cables in the vicinity of the first zero crossing with an external temperature of 42ºC
(107.6ºF). Again, a downshift in the zero crossing of the impedance phase is observed with the
presence of the 1 meter hot-spot.
Figure 120 summarizes the comparison of the impedance phase spectra for the various cases
examined above; namely, 100 meter cables with and without a hot-spot, and external temperature
of 24ºC (75ºF) and 42ºC (107.6ºF). This figure shows the different behavior that these two
cables exhibit when the mean outside temperature changes. For the cable without a hot-spot,
increasing the external temperature results in a downshift in the impedance phase zero crossing,
while the reverse is true for the cable with the 1-meter hot-spot. This reverse behavior is
attributed to the presence of the hot-spot.
136
Φ(ω)
20
B
10
ω [Hz]
430000 435000 440000 445000
-10
-20 A No Hot-Spot
A B
-30 1m Hot-Spot
-40
Figure 119 Expanded view of the impedance phase spectra in the

frequency range from 425kHz to 445kHz for 100 meter cables with and
without a 1 meter hot-spot, and an external temperature of 42ºC (107.6ºF)
Φ(ω)
20
10
ω [Hz]
430000 435000 440000 445000
-10
B A
-20 C A 24ºC No Hot-Spot
-30 D B
42ºC No Hot-Spot
C 24ºC 1m Hot-Spot
-40
D
42ºC 1m Hot-Spot
Figure 120 Summary comparison of impedance phase spectra for 100
meter cables with and without a 1-meter hot-spot and external
temperatures of 24ºC (75ºF) and 42ºC (107.6ºF)
137
5.5.3 Phenomenological Models of Cables with Varying Environment
In this section a phenomenological model for a cable containing a hot-spot with a varying
external environment is developed. The approach taken is to consider the cable to be in two
states; namely a healthy state with no hot-spots, and a degraded state with the hot-spot. Each
state is then modeled separately accounting for a varying environment, then the models are
combined to represent the cable with a hot-spot. To develop the separate models, impedance
results from test specimen 43, which was thermally aged to simulate 60 yrs @ 90ºC , and
specimen 26, which was thermally aged to simulate 60 yrs @ 50ºC were used.
Figure 121 compares the measured impedance magnitude for the two 10 meter cables in the
frequency range from 1kHz to 1MHz. As shown, the impedance magnitude for the cable with
the more severe aging is lower. Thus, the more severe aging has made the cable’s insulation
more conductive.
Log10 ⏐Z(ω)⏐
5.5 A
Aged to Simulate 60 yrs. @ 50°C
B
5
4.5
A
B
3.5
Log10 ω
3.5 4 4.5 5 5.5 6
2.5
Figure 121 Comparison of impedance magnitude in the frequency

range from 1kHz to 1MHz for cables aged to simulate 60 yrs @ 50ºC
and 60 yrs @ 90ºC
Figure 122 compares the measured impedance phase spectra for the two 10 meter cables in the
frequency range from 1kHz to 1MHz. As shown, the impedance phase for the cable with the
more severe aging is higher than that for the cable with less aging for frequencies less than
approximately 400kHz. Again, this indicates that the more severe aging has made the cable’s
138
insulation more conductive. It is observed that the impedance phase for specimen 43 drops
significantly over 0.5MHz. The reason for this is not clear.
Φ(ω)
A
-88.7 Aged to Simulate 60 yrs. @ 50°C
B
-88.8
-88.9
Log10 ω
3.5 4 4.5 5 5.5 6
-89.1
B
-89.2
A
-89.3
-89.4
Figure 122 Comparison of impedance phase spectra for 10 meter

cables aged to simulate 60 yrs @ 50ºC and 60 yrs @ 90ºC
The cables characteristic impedance and propagation function are now extracted from the
measured impedance in the frequency range from 1kHz to 1MHz. Figure 123 compares the
extracted characteristic impedance magnitude for the 10-meter cables. As shown, the more
severe aging results in a decrease in the characteristic impedance magnitude.
Figure 124 compares the extracted characteristic impedance phase spectra for the 10-meter
cables. As shown, the more severe aging results in a decrease in the characteristic impedance
phase. The largest difference occurs at the lowest frequencies in this range, i.e., 1kHz. This
behavior arises from the fact that more severe aging degrades the cable’s electrical properties.
Figures 125 and 126 compare the extracted imaginary and real components, respectively, of the
propagation function for the 10-meter cables. As shown, the more severe aging results in an
increase in these parameters.
139
Z0(ω)
220 A
200 B
180
160
140
A
120
100 B Log10 ω
3.5 4 4.5 5 5.5 6
Figure 123 Comparison of extracted characteristic impedance magnitude

for 10 meter cables aged to simulate 60 yrs. @ 50ºC and 60 yrs @ 90ºC
140
Ψ(ω) Log 10 ω
3.5 4 4.5 5 5.5 6
-10
A
-20 A
B
B Aged to Simulate 60 yrs. @ 90°C
-30
Figure 124 Comparison of extracted characteristic impedance phase

spectra for 10 m cables aged to simulate 60 yrs.@50ºC and 60 yrs@90ºC
A
γ i (ω) Aged to Simulate 60 yrs. @ 50°C
B
0.04 Aged to Simulate 60 yrs. @ 90°C
0.03
B
0.02
A
0.01
ω [H z]
200000 400000 600000 800000 1x10 6
Figure 125 Comparison of imaginary component of the propagation

function for 10 m cables aged to simulate 60 yrs.@50ºC and 60 yrs@90ºC
141
γr (ω)
0.0012
ω [Hz]
200000 400000 600000 800000 1x106
0.0008
B
A
0.0006
0.0004
A
0.0002 B
Figure 126 Comparison of the real component of the propagation function

for 10 meter cables aged to simulate 60 yrs. @ 50ºC and 60 yrs @ 90ºC
The electrical properties of the cables are now determined for the 10-meter cables aged to
simulate 60 yrs. @ 50ºC and 60 yrs @ 90ºC. Figure 127 compares the resistance per unit length
for the cables, along with the relative difference in resistance per unit length. It is observed that
both resistance curves follow the square root frequency dependence due to the skin effect. As
shown, at very low frequencies, the resistance for the cable with the more severe aging is larger.
For direct currents, the resistance per unit length depends on the geometry of the twisted
conductor pair according to the following equation:
2 Rs ⎡ s/d ⎤
R dc = ⎢ ⎥ 5.5-1
πd ⎢ (s/d )
2
− 1 ⎥⎦
⎣
where “s” is the distance between the cable conductor centers, “d” is the wire insulation’s
diameter, and Rs is the skin resistance of a single conductor. Thus, the smaller the diameter of
the conductor insulation, the larger the resistance at direct current. These results imply that the
insulation shrinkage increases when the aging becomes more severe, which is consistent with
intuition. It is observed that for frequencies above 100kHz, the relative difference in resistance
per unit length is negative, which implies that the resistance for the cable with the more severe
aging becomes smaller that the cable with less aging. This behavior is an anomaly for which no
readily apparent explanation is available.
142
R(ω)/l [Ω/m] δR(ω)
0.14
A 0.15
0.12 B
ω [Hz] 0.10
200000 600000 1x107
0.08 0.05
0.06 B ω [Hz]
0.04 200000 600000 1x107
0.02 -0.05
A
B
Figure 127 Comparison of resistance per unit length and relative difference in resistance
per unit length for 10-meter cables aged to simulate 60 yrs. @ 50ºC and 60 yrs @ 90ºC
Figure 128 compares the inductance per unit length for the cables, along with the relative
difference in inductance per unit length. As shown, the inductance is smaller when the aging
severity is increased. This behavior does not reflect the standard formula for the dc inductance
per unit length dependence on the cable geometry, which is defined by the following equation:
µ ⎛s⎞
L= Cosh -1 ⎜ ⎟
π ⎝d⎠ 5.5-2
This behavior is not currently understood; however, cables have a very complicated structure
that involves a number of wire strands that form the conductor. Each strand creates its own
magnetic field and it is possible that the individual contributions from each strand to the cables’s
magnetic field may interfere with each other. In that case, Equation 5.5-2 would not describe the
cable’s inductance.
143
L(ω)/l [H/m] δL(ω)
4.4x10-6 -0.055
A
-0.06
4.2x10-6 B
Log10 ω -0.065
3.5 4 4.5 5 5.5 6
3.8x10-6 -0.07
3.6x10-6 B -0.075
ω [Hz]
-6
3.4x10 200000 600000 1x107
A
B
Figure 128 Comparison of inductance per unit length and relative difference in
inductance per unit length for 10m cables aged to simulate 60yrs.@50ºC and
60yrs@90ºC
Figure 129 compares the capacitance per unit length for the cables, along with the relative
difference in capacitance per unit length. As shown, the capacitance is increased when the aging
severity is increased. This behavior does not reflect the dc capacitance per unit length
dependence on the cable’s geometry, as defined by the following equation:
πε 1 (ω )
C(ω ) =
⎛s⎞ 5.5-3
Cosh -1 ⎜ ⎟
⎝d⎠
In the above equation, g1(T) refers to the real component of the insulation’s dielectric function.
However, the thermal aging can change the insulation’s chemistry. Throughout this frequency
range, the thermally degraded cable’s insulation has a capacitance that is larger than for the less
degraded cable. Furthermore, the capacitance exhibits a weak dependence on frequency. That
implies that the more severe aging degradation affected all of the microprocesses that control the
insulation’s dielectric function.
144
C(ω)/l [F/m] δC(ω)
5.5x10-10 B
0.605
-10
ω [Hz]
5.0x10
200000 600000 1x107
-10
4.5x10 0.595
Log10 ω
3.5 4 4.5 5 5.5 6 0.59 B
-10
4.0x10 A 0.585
A
B
Figure 129 Comparison of capacitance per unit length and relative difference in
capacitance per unit length for 10m cables aged to simulate 60 yrs@50ºC and
60 yrs@90ºC
Figure 130 compares the conductance per unit length for the cables, along with the relative
difference in conductance per unit length. As shown, the conductance is increased when the
aging severity is increased. This behavior does not reflect the dc conductance per unit length
dependence on the cable’s geometry, as defined by the following equation:
π ω ε 2 (ω )
G (ω ) =
⎛s⎞
Cosh -1 ⎜ ⎟ 5.5-4
⎝d⎠
In the above equation, g2(T) refers to the imaginary component of the insulation’s dielectric
function. However, the thermal aging can change the insulation’s chemistry. Throughout this
frequency range, the degraded cable’s insulation has a capacitance that is larger than for the
cable that is less severely degraded.
145
G(ω)/ω−l [mho-sec/m] δG(ω)/ω
1.2
B 1.0
1x10-11
9x10-10 0.8
8x10-10 B
0.6
7x10-10 0.4
6x10-10
0.2
5x10-10
A Log10 ω ω [Hz]
3.5 4 4.5 5 5.5 6 200000 600000 1x107
A
B
Figure 130 Comparison of conductance per unit length and relative difference in
conductance per unit length for 10m cables aged to simulate 60 yrs@50ºC and 60 yrs@90ºC
With the electrical properties of the individual 10-meter cables modeled, a model can be
developed for the 100 meter cable with a hot-spot. The model is then applied to the impedance
measurements made on the cables shown schematically in Figure 1. The geometry of the model
is shown schematically in Figure 131.
Sh Sd Sh
V(ω)
x=0 la lb x=l
Figure 131 Schematic of model for 100-meter cable containing a hot-spot
Referring to the above figure, Sh refers to the healthy section of cable with no aging, and Sd
refers to the damaged section of cable thermally aged to simulate a hot-spot. Also, la and lb refer
to the start and end location, respectively, of the damaged hot-spot section of cable.
The modeled impedance for this configuration is represented by the following equation:
146
⎛ Z (ω ) coth [γ h (l - la )] + Z 0,d (ω ) coth [γ d (lb - la )] ⎞
Z model (ω ) = Z 0, h (ω ) ⎜⎜ 0,h ⎟
⎝ Z 0,d (ω ) + Z 0, h (ω ) coth [γ h (l - lb )] tanh [γ d (lb - la )] ⎠
⎟ 5.5-5
where Z0,h(T) and Z0,d(T) are the characteristic impedances, and (h(T) and (d(T) are the
propagation functions for the healthy and damaged sections of cable, respectively. In applying
this model, the impedance data from the 10-meter cable aged to simulate 60yrs @ 50ºC are used
to represent the healthy section, and the impedance data from the 10-meter cable aged to
simulate 60yrs @ 90ºC are used to represent the damaged section. It should be noted that the
impedance measurements for the 10-meter cables were performed with the cables in a humidity
chamber and the conditions maintained at 30ºC (85ºF) and 20% RH. Thus, there is some error
introduced in using impedance data from these cables to model the 100 meter cables, which were
measured under slightly different conditions of 24ºC (75ºF) and low humidity.
To determine the hot-spot location relative to the point of measurement, it is first assumed that la
is zero. The model is used to calculate the impedance and the results are compared to the
measured impedance. In the frequency range from 1kHz to 1MHz. The assumed hot-spot
location is then changed and the impedance calculation is repeated. This process is continued
until the hot-spot location is varied over the entire length of the cable. For each calculation, the
error in the impedance phase is calculated using the following equation:
m = n0 + n
∑Φ Measuremen t (ω m ) - Φ Model (ω m , la )
δΘ(la ) ≡
m = n0 −n
m = n0 + n
∑ Φ (ω )
m = n0 −n
m 5.5-6
where MMeasurement (Tm) is the measured impedance phase at the frequency Tm and MModel (Tm, la)
is the model’s impedance phase at the frequency Tm based on the assumption that the hot-spot
location is la. In addition, Tn0=437,419Hz and the lower and upper frequencies were 429, 949Hz
and 445, 020Hz, respectively. The results are shown in Figure 132.
147
δΘ (x)
0.5
0.4
0.3
0.2
0.1
x [m]
20 40 60 80
Figure 132 Error in predicted impedance phase as a function of

assumed hot-spot location
Examination of the above figure shows that the model predicted the hot-spot location to be
within 90 meters to 99 meters from the point of measurement. The error varies from
approximately 56% down to approximately 5% near the correct location. To more accurately
locate the hot-spot, the calculations were repeated in the region from 90 meters to 99 meters
using refined assumptions for the hot-spot location. The predicted location is then found to be
99 meters from the point of measurement. This is in good agreement with the correct location of
96 meters. Thus, the model’s prediction is within 3% of the correct location. Figure 133
presents the results of this calculation. As shown, the model is within 4.2% of the correct
location.
These results demonstrate that BIS can be used to locate hot-spots on cables even when the
environment is variable along the external surface of the cable. For the example presented
herein, the model’s prediction of hot-spot location is within a few percent of the correct position.
This is particularly interesting in light of the fact that parameters from the 10 meter cables aged
at slightly different conditions were used to represent the 100 meter cable being modeled. In
addition, 90% of the 100 meter cable was outside the building during testing, which resulted in a
dynamic external environment. In spite of these adverse factors, the model was able to predict
the hot-spot location with good accuracy.
148
δΘ (x) Correct
Location
0.050
x [m]
92 94 96 98
0.048
0.046
0.044 Predicted
Location
0.042
Figure 133 Error in predicted impedance phase as a function of

assumed hot-spot location over the cable length from 90 meters to 99
meters
149
150
6. EVALUATION OF THE BIS METHOD FOR NUCLEAR PLANT CABLES
The goal of this research program was to evaluate the effectiveness of the BIS method for
detecting and locating degradation in electric cables in nuclear power plant applications. To
achieve this goal, a commonly used low-voltage instrumentation and control (I&C) cable was
selected to prepare cable test specimens, which were exposed to accelerated aging to simulate
various types of service conditions in a typical nuclear power plant application. Specific types of
degradation were inflicted on the cable after which the cable test specimens were tested using the
BIS method. The types of degradation evaluated are the following;
• Global thermal aging of the cable

• Global plus localized thermal aging of the cable (i.e., hot-spots)
• Thermal degradation with cracking, and
• Abrasion damage to the cable
In testing the cables, the following three different cable configurations were evaluated to provide
a representative sampling of the applications commonly expected in a nuclear power plant
setting:
• Constant temperature and humidity along the cable with no load attached
• Constant temperature and humidity along the cable with a load attached
• Varying environments along the length of the cable with no load attached
For each configuration, BIS measurements were made and the data were analyzed to determine if
the degradation could be detected and located. Detection and location of the degradation is
based on the use of the measured impedance data, the governing equations for the electrical
properties of an electric cable, and models developed to analytically represent the cable
properties. The results for each of the different configurations are discussed below.
6.1 Results for Global Thermal Aging
The test results on the cables with global thermal aging clearly show that the BIS method is
capable of detecting thermal degradation of a cable. In the low frequency range, comparison of
the measured impedance phase spectra show an increase in the impedance phase angle when
thermal aging is present on the cable. This increase in impedance phase indicates that the cable
is more dissipative, and can be used as an indicator of aging degradation on the cable.
The high frequency BIS data show that, as thermal aging is applied to the cable, the impedance
magnitude and phase spectra are downshifted to lower frequencies. Therefore, by measuring the
151
impedance of a cable and comparing it to a baseline for a known cable in good condition, this
downshift in frequency can also be used as an indicator of thermal degradation on the cable.
Analytical models were developed to predict the electrical properties of a degraded cable based
on measured impedance data. The approach used is based upon the fact that, at low frequencies
the cable’s impedance is non-oscillatory and the characteristic impedance and propagation
function can be directly extracted from the measured impedance spectra. At high frequencies,
where the cable’s impedance oscillates, it is not possible to directly extract the characteristic
impedance and propagation function. Instead, the cable’s characteristic impedance in the low
frequency range is first determined. This function is then extrapolated to the high frequency
range. The propagation function for the high frequency range can then be determined. Once the
characteristic impedance and propagation function are known, the models developed for the
cable’s electrical properties can be used to predict the electrical properties of the cable. Results
showed relatively good accuracy for the models developed.
These results demonstrate that the BIS method can detect thermal degradation and predict the
electrical properties of the degraded cable with relatively good accuracy for the type of cable
tested in this study. It should be noted that this testing was performed in a laboratory
environment. Testing in a actual plant environment is recommended to verify the effectiveness
of the BIS method for detecting this type of degradation.
6.2 Results for Global Plus Localized Thermal Aging (Hot-Spot)
The results from the testing of cables with global thermal aging were used as the basis for
detecting and locating localized thermal aging (i.e., hot-spots) in cables. The high frequency
impedance spectra for cables with and without hot-spots were compared and they demonstrated
that localized damage does impact the cables impedance characteristics. The presence of a hot-
spot tends to downshift the frequency of the impedance spectra. This difference in impedance
spectra can be used to detect and locate localized degradation.
The impact of hot-spot severity, size, and location were also investigated in this study. Tests
were performed to compare the impedance response of cables with a hot-spot representing
60 years of service at 60ºC (140ºF) and 60 years of service at 70ºC (158ºF). The results indicate
that the difference in hot-spot severity was not significant enough for the low frequency
impedance spectra to show a measurable difference. However, the high frequency spectra did
show a downshift in frequency. This change is useful for estimating the size of a hot-spot.
Cables with hot-spots 1.0 meters in length and 0.5 meters in length were also compared to
evaluate the impact of hot-spot size, and similar results were obtained.
To determine the sensitivity of the BIS method to hot-spot location, test cables were measured
from both ends. This effectively provided data for a hot-spot at two different distances from the
point of measurement. Results demonstrated that the location of the hot-spot does impact the
152
impedance spectra, and this difference can be used to locate the degradation. The closer the hot-
spot is to the point of measurement, the greater is the change in the impedance spectra.
A method for locating a hot-spot was developed based on the models for the cable’s properties.
Specifically, the cable’s impedance is calculated using equation 5.1-13 based on assumed values
of the hot-spot location. This calculated value is compared to the measured impedance. Based
on the root mean square (rms) error between the value predicted by the model and the measured
value, new values of hot-spot location are chosen and the process is repeated. The final hot-spot
location is determined by minimizing the rms error.
In comparing the modeled impedance to the measured impedance, four different parameters are
used, which are based on the impedance magnitude and the impedance phase zero crossings.
Results show that this approach can be used to predict hot-spot location with an accuracy of
10%.
It should be noted that this testing was performed in a laboratory environment with a known,
well defined hot-spot on the test specimen. Also, only one hot-spot was present on the cable.
Therefore, generalization of these results to cables with multiple hot-spots would not be
appropriate. However, these results do show promise that the BIS method is a powerful tool for
locating localized damage in an installed cable system.
6.3 Results for Global Thermal Aging with an Attached Load
Impedance measurements were made on cables with localized thermal aging both with and
without a load attached to the cable. The goal was to determine the impact of the load on the
ability of the BIS method to detect and locate a hot-spot. The results showed that, at low
frequencies, the cable system’s impedance is heavily dominated by the load. However, the zero
crossing of the impedance phase does react to the presence of a hot-spot. Therefore, an accurate
model of the cable system impedance can be used to examine the behavior of the impedance
phase in the vicinity of the low frequency zero crossings. This can be used as an indicator for
the presence of a hot-spot on the cable’s insulation even with a load attached to the cable.
In evaluating the high frequency BIS measurements for cables with an attached load, the
frequency averaged second derivative of the impedance phase spectra was found to provide
useful information for detecting hot-spots. Specifically, the zero crossings of this parameter
were found to be downshifted when the cable’s insulation contains a hot-spot. An examination
of the A(T) spectra showed that a frequency downshift occurs at all of the 14 DZCs when a hot-
spot is present on the cable. Thus, detection of a hot-spot appears to be feasible for cables with a
load attached.
Models were also developed to predict the electrical properties for cables with hot-spots and an
attached load. Using these models and an approach similar to that used for the detection of hot-
spots without an attached load, the location of a hot-spot was correctly predicted. These results
153
demonstrate that the BIS method is capable of detecting and locating localized degradation on
cables even if there is a load attached to the cable. Some knowledge of the impedance response
of the load is useful.
6.4 Results for Global Plus Localized Abrasion Damage
The testing on cables with simulated abrasion damage showed that the low frequency impedance
spectra do not respond to the presence of abrasion unless a high humidity level is present to
create a conductivity path. When a higher level of humidity is present, all of the phase spectra
have a new structure (i.e., a peak) in the vicinity of 5kHz to 10Khz. This structure arises from
the increased humidity level, which provides a conductive pathway for leakage currents.
It is also observed that the abraded cable’s impedance peak is shifted to higher frequencies
relative to the unabraded cable. This frequency shift may be attributable to the fact that the
averaged dielectric function of the abraded cable is smaller than that for the unabraded cable. As
a result, any changes to the spectra must occur at higher frequencies. It is further observed that
there is no readily observable difference between the phase spectra of the cables with one and
two conductors abraded.
It should be noted that, while the change in the low frequency impedance phase spectra can be
used to detect abrasion under high humidity levels, this may not be practical in a nuclear power
plant environment since the humidity levels may not be high enough to observe these changes.
For example, the peaks previously discussed were not observed when BIS measurements were
made under conditions of 70% RH.
An evaluation of the high frequency data shows that the phase zero crossing and the first maxima
of the impedance magnitude are shifted to a higher frequency when the cable is abraded. The
largest change is observed for the case in which the abrasion is covered by a conductive
contaminant, thus establishing a conductivity path.
Models were developed to represent the electrical properties of the abraded cable. Using these
models together with the measured impedance spectra, abrasion damage was located on a test
specimen with an accuracy of 15%. An iterative approach similar to that used for detecting hot-
spots was used in which an assumed damage location is used to calculate the cable impedance.
This is compared to a measured value and an iterative process is used to minimize the rms error
between predicted and measured values. These results demonstrate that the BIS method is
potentially useful for detecting and locating abrasion damage on a cable.
6.5 Results for Global Thermal Aging with Cracking Damage
154
Testing of cables with simulated cracking was performed to evaluate the effectiveness of the BIS
method for detecting and locating this type of degradation. The impedance phase spectra in the
low frequency range from 1kHz to 1MHz for cables with and without cracking show that there is
only a small difference between the uncracked cable and the cable with cracks. Even a
conductive contaminant applied to the cable does not appear to impact the impedance phase.
This can be attributed to the lack of a conductivity path between the conductors, even with the
conductive contaminant applied. A possible reason for this is the size of the iron filings used,
which were on the order of 100:. This is approximately the same dimension as the crack width.
Thus, it is believed that the iron particles could not enter the crack and make a connection with
the cable conductor.
The high frequency impedance spectra of the cable test specimens showed that the zero crossings
of the second derivative of the impedance phase spectra, A(T), did show differences that may be
useful as indicators for the presence of cracking. Monitoring the zero crossings of the parameter
A(T) may be an effective indicator for the presence of cracking on a cable.
In the experiments performed herein, cracking was simulated by slicing the cable’s insulation
using a knife. This approach was selected since it was an efficient method of simulating the
physical characteristics of a crack; however, the approach does not accurately capture the
physics or the chemistry of cracking that would occur in real plant environments. As a result,
the results demonstrate the potential for detecting cracking using the BIS method; however,
accurate models for real world cracking are not currently available. Additional research using
more realistically simulated cracking is recommended to further evaluate the BIS method for
detecting this type of degradation. Continuous BIS measurement during the aging process would
be beneficial to understanding and modeling the cracking process. Chemical analysis of the
cable’s insulation would also be useful.
6.6 Results for Global Thermal Aging with a Varying Environment
Testing of long cables (100 meters) was performed to evaluate the impact of varying
environments along the external surface of the cables on BIS measurements. One of the cables
contained a 1-meter hot-spot that simulated 60 years of service at 90ºC. The hot-spot was
located at 96 meters from the end of the cable. Two different environments were simulated
during the testing process.
The predicted location using BIS was 99 meters from the point of measurement. This is in good
agreement with the correct location of 96 meters. Thus, the model’s prediction is within 3% of
the correct location. The results demonstrate that the BIS method can be used to locate hot-spots
on cables even with varied environments along the external surface of the cable. For the
example presented herein, the model’s prediction of hot-spot location is within a few percent of
the correct position.
155
156
7. CONCLUSIONS AND CONSIDERATIONS FOR FOLLOW-ON RESEARCH
7.1 Conclusions on the Effectiveness of the BIS Method
The research program reported herein evaluated the BIS method as a prognostic and diagnostic
tool for monitoring installed electric cable systems in nuclear power plants. The results
demonstrated that this method is effective for nuclear plant cables, and that it is capable of
detecting and locating localized degradation within a cable in its incipient stage prior to failure.
Various degradation scenarios were evaluated, including the following:
• global thermal aging damage,

• localized thermal aging damage,
• abrasion damage, and
• cracking damage
The following conclusions are drawn based on the results of this research:
• The BIS method was clearly able to detect the presence of thermal degradation on
the cables used in this study. Specifically, the impedance phase spectra of the
cables tested were observed to increase as the amount of thermal degradation on
the cable increased. This increase can be used as an indicator of global thermal
degradation.
or hot-spots on the cables. Specifically, a shift in the zero crossings of the
impedance phase spectra were observed when a hot-spot was present on the cable.
• An approach was developed for locating hot-spots on a cable using models of the
cables electrical properties. The models were able to predict the hot-spot location
with an accuracy of 10%.
• The BIS method was not sensitive enough to distinguish between the different
severities and sizes of hot-spots simulated in the low frequency range; however,
measurements in the high frequency range did detect the differences. Additional
research is warranted to establish the sensitivity limits for this technique
• The BIS method was able to detect and locate the presence of abrasion damage on
a cable. The models and approach used are similar to that for detecting and
locating thermal hot-spots.
• The BIS method was demonstrated to be effective for detecting and locating
degradation on cables with an attached load. This is important since it is
desirable to have a technique that can test cables in their installed configuration,
without having to disconnect them from attached equipment.
157
• The BIS method was able to detect the simulated cracking damage on cables.
However, the simulation method used in this study was determined not to
accurately represent the cracking phenomena. Additional research should be
considered to more accurately evaluate the BIS method on actual cable cracking.
or hot-spots on the cables even with a varying environment along the external
surface of the cables. For the example presented herein, the model’s prediction of
hot-spot location is within a few percent of the correct position.
The results from this study suggest that the technology that makes use of BIS may represent a
breakthrough in the condition monitoring of installed electric cables. In a laboratory setting, this
technique was able to detect and locate the various forms of aging degradation commonly found
in nuclear power plant applications. This technique has the capability to provide a means of
scanning an entire cable system to determine its acceptability for continued service. The
technique can be applied in situ with low voltage signals that will not damage the cable under
test, and can detect aging damage in an incipient state prior to a failure occurring.
7.2 Considerations for Follow-On Research to Demonstrate the Effectiveness of the BIS
Method in Nuclear Plant Environment
While the BIS method shows great promise as a prognostic and diagnostic technique for installed
cable systems, additional research is desirable before it can be applied in the field. Specifically,
the following additional items should be considered during future research, preferably in a
collaborative manner:
• Research on the detection and location of cracking damage in cables using more
realistic simulation of the cracking damage,
• The technique should be demonstrated on additional types of cable to determine
its usefulness for other materials and cable configurations
• The technique should be evaluated with different types of loads attached to the
cables to determine the impact of loads encountered in a plant environment
• The technique should be demonstrated in an actual plant environment to
determine the impact of the various environmental factors that may impact the
BIS measurements
• The technique should be demonstrated on blind test samples in which the type,
severity, size, and location of the degradation are unknown
Based upon the promising results obtained from this phase of the research effort, the Office of
Nuclear Regulatory Research (RES) is considering a follow-on research effort. The primary
focus of the next phase of the research should be on the demonstration of the BIS method for
installed cable systems in operating nuclear power plant environment.
158
8. REFERENCES
Durney, C. and Johnson, C., “Introduction to Modern Electromagnetics,” McGraw-Hill, 1969.
Grayson, S.J., et al., “Fire Performance of Electric Cables,” Interscience Communications, Ltd.,
United Kingdom, 2000.
Lofaro, R.J., “Proceedings of the International Conference on Wire System Aging,”

NUREG/CP-0179, Brookhaven National Laboratory, November 2002.
Lofaro, R.J., et al., “Assessment of Environmental Qualification Practices and Condition

Monitoring Techniques for Low-voltage Electric Cables,” NUREG/CR-6704, Brookhaven
National Laboratory, Upton, NY, USA, February 2001.
Moore, G.F., “Electric Cables Handbook,” Blackwell Science, Inc., Malden, MA, USA, 3rd
Edition, 1997.
National Science and Technology Council, Committee on Technology, United States White
House, “Review of Federal Programs for Wire System Safety,” Wire System Safety Interagency
Working Group, November 2000.
Rogovin, D. and Kendig, M., “Diagnostic/Prognostics of Aged Wiring Using Broadband

Frequency Impedance,” 6th Joint FAA/DOD NASA Conference on Aging Aircraft, San
Francisco, 2002.
Thue, W.A., “Electric Power Cable Engineering,” Marcel Dekker, Inc., New York, NY, USA,
1999.
Vaughan, W., “Dielectric Properties and Molecular Behavior,” Chap. 2, Van Nostrand, 1969.
Villaran, M. and Lofaro, R.J., “Evaluation of Aging and Environmental Qualification Practices
for Power Cables Used in Nuclear Power Plants” NUREG/CR-6794, Brookhaven National
Laboratory, Upton, NY, USA, January 2003.
159
160
Appendix A
BIS Test Results for Cables with Global Thermal Aging Simulating 20 years and 40 years
of Service
161
162
List of Figures
Page No.
Figure A-1 Low frequency impedance phase spectra for cables globally aged to simulate 20
years of service at various service temperatures . . . . . . . . . . . . . . . . . . . . . . . . 168
Figure A-2 High frequency impedance phase spectra for cables with global thermal aging to
simulate 20 years of service at various service temperatures . . . . . . . . . . . . . . 169
Figure A-3 Expanded view of impedance phase and magnitude per unit length spectra for the
frequency range from 78 MHz to 86 MHz for cables with global thermal aging to
Figure A-4 Characteristic impedance magnitude per unit length and phase spectra for cables
with global thermal aging to simulate 20 years of service at various service
temperatures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
Figure A-5 Imaginary and real components of the propagation function spectra for cables
temperatures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
Figure A-6 Resistance per unit length and change in resistance per unit length for cables with
global thermal aging to simulate 20 years of service at various service
temperatures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
Figure A-7 Comparison of resistance per unit length predicted by analytical model to
resistance extracted from measured impedance for cable with global thermal
aging to simulate 20 years of service at 60ºC . . . . . . . . . . . . . . . . . . . . . . . . . . 171
Figure A-10 Inductance per unit length and change in inductance for cables with global
thermal aging to simulate 20 years of service at various service temperatures . 173
Figure A-11 Comparison of inductance per unit length predicted by analytical model to
inductance per unit length extracted from measured impedance for cable with
global thermal aging to simulate 20 years of service at 50ºC . . . . . . . . . . . . . . 173
163
Figure A-14 Capacitance per unit length and change in capacitance per unit length for cables
temperatures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
Figure A-15 Comparison of capacitance per unit length predicted by analytical model to
capacitance per unit length extracted from measured impedance for cable with
Figure A-17 Conductance-to-frequency ratio per unit length and change in conductance-to-
frequency ratio per unit length for cables with global thermal aging to simulate
20 years of service at various service temperatures . . . . . . . . . . . . . . . . . . . . . . 176
Figure A-19 Comparison of conductance-to-frequency ratio per unit length predicted by

analytical model to that extracted from measured impedance for cable with global
thermal aging to simulate 20 years of service at 50ºC . . . . . . . . . . . . . . . . . . . . 177


Figure A-22 Comparison of characteristic impedance magnitude per unit length and phase
spectra predicted by analytical model to that extracted from measured impedance
for cables with global thermal aging to simulate 20 years of service at 50ºC . . 179
164
Figure A-23 Comparison of imaginary and real components of the propagation function
cable with global thermal aging to simulate 20 years of service at 50ºC . . . . . 179
Figure A-24 Comparison of characteristic impedance magnitude and phase spectra predicted
by analytical model to that extracted from measured impedance for cable with
cable with global thermal aging to simulate 20 years of service at 60ºC . . . . . 180
for cable with global thermal aging to simulate 20 years of service at 70ºC . . 181
Figure A-27 Comparison of imaginary component and real component of the propagation
function for cable with global thermal aging to simulate 20 years of service at
70ºC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
Figure A-28 Low frequency impedance phase spectra for cables with global thermal aging to
Figure A-29 High frequency impedance phase spectra for cables with global thermal aging to
Figure A-30 Expanded view of impedance phase and magnitude per unit length spectra over
the frequency range from 78 MHz to 86 MHz for cables with global thermal
aging to simulate 40 years of service at various service temperatures . . . . . . . 183
Figure A-31 Characteristic impedance magnitude and phase spectra for cables with global
Figure A-32 Imaginary and real components of the propagation function for cables with global
Figure A-33 Resistance per unit length and change in resistance per unit length for cables with
global thermal aging to simulate 40 years of service at various service
temperatures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
Figure A-34 Comparison of resistance per unit length predicted by analytical model to that
simulate 40 years of service at 50ºC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
165
Figure A-37 Inductance per unit length and change in inductance per unit length for cables
temperatures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
Figure A-38 Comparison of inductance per unit length predicted by analytical model to that
Figure A-41 Capacitance per unit length and change in capacitance per unit length for cables
temperatures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
Figure A-42 Comparison of capacitance per unit length predicted by analytical model to that
simulate 40 years at 50ºC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
Figure A-45 Conductance-to-frequency ratio per unit length and change in conductance-to-
frequency ratio per unit length for cables with global thermal aging to simulate
40 years of service at various service temperatures . . . . . . . . . . . . . . . . . . . . . . 191
166


Figure A-52 Comparison of the imaginary component and real component of the propagation
function predicted by analytical model to that extracted from measured
impedance for cable with global thermal aging to simulate 40 years of service at
50ºC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
60ºC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
70ºC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
167


50ºC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
168
60ºC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
169
A.1 Test Results for Cables with Global Thermal Aging to Simulate 20 Years of Service
Φ(ω) [Deg.]
-88.6
-88.8
C
B A
ω [Hz]
200000 400000 600000 800000 1x106
-89.2
A 50 ºC (122 ºF)
B 60 ºC (140 ºF)
-89.4
C 70 ºC (158 ºF)
Figure A-1 Low frequency impedance phase spectra for cables globally
aged to simulate 20 years of service at various service temperatures
The following figures present the results of broadband impedance tests performed on cables with
global thermal aging to simulate 20 years of service in a nuclear power plant at service
temperatures of 50ºC (122ºF), 60ºC (140ºF), and 70ºC (158ºF). These results supplement the
data presented in Section 5 of this report.
170
Φ(ω) [Deg.]
75
50 C A
25
B
ω [Hz]
2x10 7
4x10 7
6x107
8x107 1x108
-25
-50
A 50 ºC (122 ºF)
-75 B 60 ºC (140 ºF)
C 70 ºC (158 ºF)
Figure A-2 High frequency impedance phase spectra for cables with global
thermal aging to simulate 20 years of service at various service
temperatures
A 50 ºC (122 ºF)
B 60 ºC (140 ºF)
C 70 ºC (158 ºF)
Φ(ω) [Deg.] |Z(ω)|/l [S/m]
300
40 C
A 250 B
20 B 200
ω [Hz]
150
7.8x107 8.2x107 8.4x107 8.6x107 C
A
-20 100
ω [Hz]
-40 7.8x107 8.2x107 8.4x107 8.6x107
Figure A-3 Expanded view of impedance phase and magnitude per unit length
spectra for the frequency range from 78 MHz to 86 MHz for cables with global
171
A 50 ºC (122 ºF)
B 60 ºC (140 ºF)
|Z0(ω)|/l [S/m] Ψ(ω)
C 70 ºC (158 ºF)
110 -1
109 B
108 -2 C A
107 A
106 -3
B
105 C
-4
104
4.6 4.8 5.2 5.4 5.6 5.8 6 4.6 4.8 5.2 5.4 5.6 5.8 6
Log10(ω) Log10(ω)
Figure A-4 Characteristic impedance magnitude per unit length and phase spectra for
cables with global thermal aging to simulate 20 years of service at various service
temperatures
γi(ω)
γr(ω)
0.035 Log10(ω)
0.030 4.6 4.8 5.2 5.4 5.6 5.8 6
0.025 A 0.0008
0.020
C A
B 0.0006
0.015 C B
0.0004
4.6 4.8 5.2 5.4 5.6 5.8 6
0.0002
Log10(ω)
A 50 ºC (122 ºF)
B 60 ºC (140 ºF)
C 70 ºC (158 ºF)
Figure A-5 Imaginary and real components of the propagation function spectra for
temperatures
172
A 50 ºC (122 ºF)
D 50 to 60 ºC
B 60 ºC (140 ºF)
E 50 to 70 ºC
C 70 ºC (158 ºF)
δR(ω)/l [S/m]
R(ω)/l [S/m]
0.14
B C -0.005 D
A
0.12 -0.010
ω [Hz]
0.10 400000 600000 800000 1x106 -0.015
E
0.08
ω [Hz]
0.06 400000 600000 800000
Figure A-6 Resistance per unit length and change in resistance per unit length for
temperatures
R(ω)/l [S/m]
A Extracted
B Model
0.14
A B
0.13
0.12
0.11
ω [Hz]
600000 700000 800000 900000 1x106
Figure A-7 Comparison of resistance per unit length predicted by

analytical model to resistance extracted from measured impedance for
173
R(ω)/l [S /m]
A Extracted
0.14 B M odel
A B
0.13
0.12
0.11
ω [H z]
600000 700000 800000 900000 1x10 6
Figure A-8 Comparison of resistance per unit length predicted by analytical

model to resistance extracted from measured impedance for cable with global
thermal aging to simulate 20 years of service at 70ºC
R (ω) /l [S /m ] A E xtracted
B M od el
0 .1 4
A B
0 .1 3
0.1 2
0 .1 1
ω [H z]
60 0 0 00 70 0 0 00 80 0 0 00 90 0 00 0 1 x1 0 6

model to resistance extracted from measured impedance for cable with global
174
A 50 ºC (122 ºF)
D 50 to 60 ºC
B 60 ºC (140 ºF)
E 50 to 70 ºC
C 70 ºC (158 ºF)
L(ω)/l [H/m] δL(ω)/l [H/m]
E
3.9
0.020
3.8 0.016
B C
3.7
A ω [Hz]
400000 600000 800000
0.012 ω [Hz]
400000 600000 800000
0.008
D
Figure A-10 Inductance per unit length and change in inductance for cables with global
A Extracted
L(ω)/l [H/m]
3.72x10-6 B Model
ω [Hz]
600000 700000 800000 900000 1x10-6
3.68x10-6
3.66x10-6
A
3.64x10-6
B
3.62x10-6
Figure A-11 Comparison of inductance per unit length predicted by

analytical model to inductance per unit length extracted from measured
impedance for cable with global thermal aging to simulate 20 years of service
at 50ºC
175
L(ω)/l [H/m] A Extracted
B Model
3.74x10-6
3.72x10-6
ω [Hz]
600000 700000 800000 900000 1x106
3.68x10-6
A
3.66x10-6 B
3.64x10-6

impedance for cable with global thermal aging to simulate 20 years of
service at 60ºC
L(ω)/l [H/m]
A Extracted
3.80x10 -6
B M odel
3.78x10 -6
A B
-6
3.76x10
3.74x10 -6
3.72x10 -6
ω [H z]
600000 700000 800000 900000 1x10 6

impedance for cable with global thermal aging to simulate 20 years of service
at 70ºC
176
C(ω)/l [x10-10 F/m] δC(ω)/l [x10-10 F/m]
0.025 E
3.54
C 0.020
3.52
ω [Hz]
400000 600000 800000 0.015
3.48
3.46 0.010
3.44 B 0.005 D
3.42 A ω [Hz]
400000 600000 800000
A 50 ºC (122 ºF)
D 50 to 60 ºC
B 60 ºC (140 ºF)
E 50 to 70 ºC
C 70 ºC (158 ºF)
Figure A-14 Capacitance per unit length and change in capacitance per unit length for
temperatures
C(ω)/l [x10-10 F/m]

3.450
A Extracted
3.445 B Model
3.440
3.435
B
3.430
A
3.425
ω [Hz]
600000 700000 800000 900000 1x106
3.415
Figure A-15 Comparison of capacitance per unit length predicted by

analytical model to capacitance per unit length extracted from
177
C(ω)/l [x10-10 F/m]
A Extracted
B Model
3.51
A
ω [Hz]
600000 700000 800000 900000 1x106
3.49
3.48
B
3.47
3.46
Figure A-16 Comparison of capacitance per unit length predicted by

analytical model to capacitance per unit length extracted from
-12
δG(ω)/ω-l [x10-12mho-s/m]
G(ω)/ω-l [x10 mho-s/m]
0.4
7.5 E
7.0 C 0.3
6.5
6.0 0.2
B
5.5 A
ω [Hz] 0.1
5.0 400000 600000 800000 D ω [Hz]
400000 600000 800000
A 50 ºC (122 ºF)
D 50 to 60 ºC
B 60 ºC (140 ºF)
E 50 to 70 ºC
C 70 ºC (158 ºF)
Figure A-17 Conductance-to-frequency ratio per unit length and change in
conductance-to-frequency ratio per unit length for cables with global
temperatures
178
C(ω)/l [x10-10 F/m]
A Extracted
3.445 B Model
3.440
A
3.435
B
3.430
3.425
ω [Hz]
600000 700000 800000 900000 1x106
Figure A-18 Comparison of capacitance per unit length predicted by analytical

model to capacitance per unit length extracted from measured impedance for
G (ω)/ω-l [x10 -12 m ho-s/m ]

ω [H z]
600000 700000 800000 900000 1x10 6
5.8
A
5.6 B
5.4 A Extracted
B M odel
5.2
Figure A-19 Comparison of conductance-to-frequency ratio per unit length

179
G(ω)/ω-l [x10 -12 mho-s/m]
6.2
A
ω [Hz]
600000 700000 800000 900000 1x10 6
5.8
B
5.6
A Extracted
5.4 B Model

G(ω)/ω-l [x10 -12mho-s/m]

A Extracted
B Model
7.6
A
7.4
7.2
ω [Hz]
600000 700000 800000 900000 1x10 6
6.8
B
6.6

predicted by analytical model to that extracted from measured impedance for cable
with global thermal aging to simulate 20 years of service at 70ºC
180
|Z0(ω)|/l [S/m] Ψ(ω)
-0.75
105 -1.00
-1.25
104 A -1.50 A
-1.75 B ω [Hz]
B ω [Hz]
103 400000 600000 800000
400000 600000 800000 -2.25
A Extracted
B Model
Figure A-22 Comparison of characteristic impedance magnitude per unit length and
phase spectra predicted by analytical model to that extracted from measured impedance
for cables with global thermal aging to simulate 20 years of service at 50ºC
γ i(ω) γr(ω)
0.035 ω [Hz]
600000 800000 1x106
0.030 A 0.0009 A B
0.025 B 0.0008
ω [Hz] 0.0007
600000 800000 1x106
0.0006
0.015
A Extracted
B Model
predicted by analytical model to that extracted from measured impedance for cable with
global thermal aging to simulate 20 years of service at 50ºC
181
|Z0(ω)|/l [S/m] Ψ(ω)
-0.75
105 -1.00
-1.25
A A
104 -1.50
B ω [Hz] B
-1.75 ω [Hz]
103 400000 600000 800000 400000 600000 800000
-2.25
A Extracted
B Model
Figure A-24 Comparison of characteristic impedance magnitude and phase spectra

γ i(ω) γr(ω)
0.035 ω
600000 800000 1X106
0.030 A B 0.0009 A B
0.025 0.0008
ω 0.0007
600000 800000 1X106
0.0006
0.015
A Extracted
B Model
182
|Z0(ω)|/l [S/m] Ψ(ω)
-0.75
105 -1.00
-1.25
A A
104 -1.50
B -1.75 B
103 ω [Hz] ω [Hz]
400000 600000 800000 -2.25 400000 600000 800000
A Extracted
B Model
for cable with global thermal aging to simulate 20 years of service at 70ºC
γ i(ω) γr(ω)
.035 ω [Hz]
A B 600000 800000 1x106
0.0009
0.030 A B
0.0008
0.025
0.0007
ω [Hz]
600000 800000 1x106 0.0006
.015 A Extracted
B Model
Figure A-27 Comparison of imaginary component and real component of the

propagation function for cable with global thermal aging to simulate 20 years of service
at 70ºC
183
A 50 ºC (122 ºF)
Φ(ω) [Deg.]
B 60 ºC (140 ºF)
C 70 ºC (158 ºF)
-88.6
C
-88.8
B A
ω [Hz]
200000 400000 600000 800000 1x106
-89.2
-89.4
Figure A-28 Low frequency impedance phase spectra for cables with global
temperatures
184
Φ(ω) [Deg.]
75
50
A
C
25 B
ω [Hz]
7 7 7 7
2x10 4x10 6x10 8x10 1x108
-25
-50
A 50 ºC (122 ºF)
-75 B 60 ºC (140 ºF)
C 70 ºC (158 ºF)
Figure A-29 High frequency impedance phase spectra for cables with global
temperatures
A 50 ºC (122 ºF)
B 60 ºC (140 ºF)
C 70 ºC (158 ºF)
Φ(ω) [Deg.] |Z(ω)|/l [S/m]
225
40 200 C
C A A
175
20
B
150 B
ω [Hz] 125
7.8x107 8.2x107 8.6x107 100
-20 75
ω [Hz]
-40 7.8x107 8.2x10 7
8.6x107
Figure A-30 Expanded view of impedance phase and magnitude per unit length spectra
over the frequency range from 78 MHz to 86 MHz for cables with global thermal aging to
simulate 40 years of service at various service temperatures
185
A 50 ºC (122 ºF)
|Z0(ω)|/l [S/m] B 60 ºC (140 ºF) Ψ(ω)
112 C
C 70 ºC (158 ºF) -1
110 B
-2 A
108 B
106 -3
A
104 -4
C
4.6 4.8 5.2 5.4 5.6 5.8 6 4.6 4.8 5.2 5.4 5.6 5.8 6
Log10(ω) Log10(ω)
Figure A-31 Characteristic impedance magnitude and phase spectra for cables with
global thermal aging to simulate 40 years of service at various service temperatures
A 50 ºC (122 ºF)
B 60 ºC (140 ºF)
γ i(ω) C 70 ºC (158 ºF) γr(ω)
0.035 Log10(ω)
0.030 4.6 4.8 5.2 5.4 5.6 5.8 6
B
0.025 0.0008
0.020 A C A
0.0006
0.015 C
B
0.0004
4.6 4.8 5.2 5.4 5.6 5.8 6
0.005 Log10(ω) 0.0002
Figure A-32 Imaginary and real components of the propagation function for cables with
global thermal aging to simulate 40 years of service at various service temperatures
186
R(ω)/l [S/m] δR(ω)/l [S/m]
-0.010
0.14
B C
A
0.12 -0.014
ω [Hz] E
-0.018 ω [Hz]
6
400000 600000 800000 1x10
0.08 400000 600000 800000
-0.022
D
0.06
A 50 ºC (122 ºF)
D 50 to 60 ºC
B 60 ºC (140 ºF)
E 50 to 70 ºC
C 70 ºC (158 ºF)
Figure A-33 Resistance per unit length and change in resistance per unit length for
temperatures
R( ω)/l [S /m]
0.14 A
B
0.12
ω [Hz]
400000 600000 800000 1x10 6
0.08
A Extracted
B Model
0.06

model to that extracted from measured impedance for cable with global
187
R(ω)/l [S/m]
A Extracted
0.14 B A
M odel
0.13
B
0.12
0.11
ω [H z]
600000 700000 800000 900000 1x10 6

R( ω)/l [S/m]
A Extracted
0.14 B A
M odel
0.13
B
0.12
0.11
ω [H z]
600000 700000 800000 900000 1x10 6

188
L(ω)/l [x10-6H/m] δL(ω)/l [x10-6 H/m]
3.95 0.035 D
3.90
0.030
3.85
3.80 B
0.025
3.75
A ω [Hz]
3.70 C 400000 600000 800000
3.65
ω [Hz] 0.015
400000 600000 800000 6
1x10 E
A 50 ºC (122 ºF)
D 50 to 60 ºC
B 60 ºC (140 ºF)
E 50 to 70 ºC
C 70 ºC (158 ºF)
Figure A-37 Inductance per unit length and change in inductance per unit length for
temperatures
L(ω)/m (H/m)
A Extracted
-6
3.7x10
B Model
3.68x10-6
3.66x10-6
3.64x10-6
A
3.62x10-6 B
ω (Hz)
600000 700000 800000 900000 1x106

189
L(ω)/m (H/m) A Extracted
3.82x10 -6 B Model
ω (Hz)
600000 700000 800000 900000 1x10 6
3.78x10 -6
A
3.76x10 -6
B
3.74x10 -6
3.72x10 -6

L(ω)/m (H/m)
A Extracted
3.74x10 -6
B M odel
3.72x10 -6
ω (H z)
600000 700000 800000 900000 1x10 6
3.68x10 -6
A
B
3.66x10 -6
3.64x10 -6
Figure A-40 Comparison of inductance per unit length predicted by analytical

190
C(ω)/l [x10-10 F/m] δC(ω)/l [x10-10 F/m]
3.60
0.02
E
3.55 C
0.01
3.50
ω [Hz]
A 400000 600000 800000
3.45 -0.01
ω [Hz]
400000 -0.02
800000 D
3.35 B -0.03
A 50 ºC (122 ºF)
D 50 to 60 ºC
B 60 ºC (140 ºF)
E 50 to 70 ºC
C 70 ºC (158 ºF)
Figure A-41 Capacitance per unit length and change in capacitance per unit length for
temperatures
C(ω)/l [x10-10 F/m] A Extracted

3.52
B Model
3.51 B
ω [Hz]
400000 600000 800000 1x106
3.49
A
3.48
3.47
3.46

aging to simulate 40 years at 50ºC
191
C(ω)/l [x10 -10 F/m]
3.44 A Extracted
B Model
3.42
B
ω [Hz]
400000 600000 800000 1x10 6
3.38 A
3.36

C(ω)/l [x10-10 F/m] A Extracted

B Model
3.55
3.54 B
A
3.53
ω [Hz]
600000 800000 1x10 6

192
G(ω)/ω-l [x10-10 mho-s/m] δG(ω)-l [x10-10 mho-s/m]
8.5 E
8.0 C 0.3
7.5
7.0 B 0.2
6.5
6.0 0.1
A D
5.5 ω [Hz] ω [Hz]
400000 600000 800000 1x106 400000 600000 800000
A 50 ºC (122 ºF)
D 50 to 60 ºC
B 60 ºC (140 ºF)
E 50 to 70 ºC
C 70 ºC (158 ºF)
Figure A-45 Conductance-to-frequency ratio per unit length and change in conductance-
to-frequency ratio per unit length for cables with global thermal aging to simulate 40
years of service at various service temperatures
A
G(ω)/ω-l [x10-10 mho-s/m] Extracted
B Model
6.50
6.25
6.00
B
5.75
5.50 A
5.25
ω [Hz]
400000 600000 800000 1x106

193
G(ω)/ω-l [x10 -10 mho-s/m]
A Extracted
3.44 B Model
3.42
B
ω [Hz]
400000 600000 800000 1x10 6
A
3.38
3.36

A
G(ω)/ω-l [x10 -10 mho-s/m] Extracted
8.50
B Model
8.25
A
8.00
7.75
7.50 B
7.25
ω [Hz]
600000 700000 800000 900000 1x10 6

predicted by analytical model to that extracted from measured impedance
194
|Z0(ω)|/l [S/m]
Ψ(ω)
104.5
-0.75
104.0 -1.00
103.5 -1.25
-1.50 A
103.0
-1.75 B ω [Hz]
102.5
B A
400000 600000 800000
ω [Hz] -2.25
400000 600000 800000
A Extracted
B Model
|Z0(ω)|/l [S/m] Ψ(ω)
107.5 -0.50
-0.75
107.0
-1.00
106.5 -1.25 A
106.0 A -1.50
105.5
-1.75 B ω [Hz]
B ω [Hz] 400000 600000 800000
400000 600000 800000 -2.25
A Extracted
B Model
195
|Z0(ω)|/l [S/m] Ψ(ω)
-0.50
-0.75
103.5
-1.00
103.0 B -1.25 A
-1.50
102.5 A B
ω [Hz] -1.75
ω [Hz]
400000 600000 800000 400000 600000 800000
A Extracted
B Model
γ i(ω) γr(ω)
0.035 ω [Hz]
0.030 600000 800000 1X106
A 0.0009
0.025 0.0008 A B
0.020 B 0.0007
0.0006
0.015
ω [Hz] 0.0005
600000 800000 1X106 0.0004
A Extracted
B Model
Figure A-52 Comparison of the imaginary component and real component of the
196
γ i(ω) γr(ω)
0.035 ω [Hz]
0.030 600000 800000 1X106
A 0.0009
0.025 0.0008 A B
0.020 B 0.0007
0.0006
0.015
ω [Hz] 0.0005
600000 800000 1X106 0.0004
A Extracted
B Model
γ i(ω) γr(ω)
0.035
0.030
ω [Hz]
A 600000 800000 ´ 6
1X10
0.025
B 0.0008 A
0.020 B
0.015 0.0006
ω [Hz]
600000 800000 1X106 0.0004
A Extracted
B Model
197
R(ω)/l [S/m]
0.14 A
B
0.12
ω [Hz]
400000 600000 800000 1x106
0.08
A Extracted
B Model
0.06
that extracted from measured impedance for cable with global thermal aging to simulate
198
R(ω)/l [S/m]
0.14
0.12
A
B ω [Hz]
400000 600000 800000 1x10 6
0.08
A Extracted
B Model
0.06

L(ω)/m (H/m)
A Extracted
3.85 x10 -6 B Model
3.80 x10 -6
A
B
3.75 x10 -6
ω (Hz)
400000 600000 800000 1x10 6

199
L(ω)/m (H/m)
A Extracted
-6
3.95x10
B Model
3.90 x10 -6
A
3.85 x10 -6
B
ω (Hz)
400000 600000 800000 1x10 6
3.75 x10 -6

C(ω)/l [x10 -10 F/m]

3.45
A Extracted
3.44 B
B Model
3.43
3.42 A
3.41
ω [Hz]
6
400000 600000 800000 1x10
3.39

200
that extracted from measured impedance for cable with global thermal aging to simulate
G(ω)/ω-l [x10 -10 mho-s/m]

A Extracted
B Model
6.75
6.50
6.25
6.00
A
B
5.75
5.50
5.25
ω [Hz]
400000 600000 800000 1x10 6

201
G(ω)/ω-l [x10-10 mho-s/m]
A Extracted
8.25
B Model
8.00
7.75
A
7.50
B
7.25
ω [Hz]
300000 400000 500000 600000 700000 800000
6.75
6.50

|Z0(ω)|/l [S/m] Ψ(ω)
106.0 -0.75
-1.00
105.5 -1.25 A
105.0
B -1.50
-1.75 B
104.5
A ω [Hz]
ω [Hz] -2.25
400000 600000 800000
400000 600000 800000
A Extracted
B Model
202
|Z0(ω)|/l [S/m] Ψ(ω)
-0.50
108.0
-0.75
107.5 -1.00
-1.25
A
107.0 A B -1.50
-1.75 B
106.5 ω [Hz]
ω [Hz] 400000 600000 800000
400000 600000 800000 -2.25
A Extracted
B Model
γ i(ω) γr(ω)
0.035 ω [Hz]
0.030 400000 600000 800000 1x106
A 0.0009
0.025 0.0008 A B
0.020 B 0.0007
0.015 0.0006
ω [Hz] 0.0005
400000 600000 800000 1x106 0.0004
A Extracted
B Model
203
γ i(ω) γr(ω)
0.035 ω [Hz]
0.030 400000 600000 800000
A 0.0009
0.025 0.0008 B
B 0.0007
0.020
0.0006
0.015 ω [Hz] 0.0005 A
400000 600000 800000 1x106 0.0004
A Extracted
B Model
204
Appendix B
BIS Test Results for Cables with Localized Thermal Aging (Hot Spots)
205
206
List of Figures
Page No.
Figure B-1 Characteristic impedance magnitude per unit length and phase spectra over the
frequency range from 1MHz to 100MHz for cables with global thermal aging to
simulate 20 years of service at various temperatures . . . . . . . . . . . . . . . . . . . . 207
Figure B-2 Imaginary and real components of the propagation function over the frequency
range from 1MHz to 100MHz for cables with global thermal aging to simulate
20 years of service at various temperatures . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
Figure B-3 Impedance magnitude per unit length and phase spectra over the frequency range
from 39MHz to 45 MHz for cables with global thermal aging to simulate 20 years
of service at 50ºC, plus additional thermal aging to simulate hot spots of different
severity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
Figure B-4 Imaginary and real components of the propagation function over the frequency
range from 1MHz to 100MHz for cables with global thermal aging to simulate
40 years of service at various temperatures . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
Figure B-5 Characteristic impedance magnitude per unit length and phase spectra over the
frequency range from 1MHz to 100MHz for cables with global thermal aging to
simulate 40 years of service at various temperatures . . . . . . . . . . . . . . . . . . . . 209
severity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
of service at 50ºC, plus additional thermal aging to simulate hot spots at different
locations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
207
locations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
severity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
severity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
locations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
208
A 50 ºC (122 ºF)
B 60 ºC (140 ºF)
C 70 ºC (158 ºF)
|Z0(T)|/l [S/m] Ψ(ω)

102 C
0.4 B
100 0.3 A
98 B 0.2
A C
ω (Hz) ω (Hz)
7
2x10 6x10 7
1x108 2x10 7
6x10 7
1x108
Figure B-1 Characteristic impedance magnitude per unit length and phase spectra over
the frequency range from 1MHz to 100MHz for cables with global thermal aging to
simulate 20 years of service at various temperatures
A 50 ºC (122 ºF)
B 60 ºC (140 ºF)
C 70 ºC (158 ºF)
γ i(ω) γr(ω)
C
C 0.05 B
3
B 0.04 A
2 A 0.03
0.02
1
0.01
ω (Hz) ω (Hz)
7 8 7
2x10 6x10 7
1x10 2x10 6x10 7
1x108
Figure B-2 Imaginary and real components of the propagation function over the
frequency range from 1MHz to 100MHz for cables with global thermal aging to simulate
20 years of service at various temperatures
209
A
60 C
150 A 40 B
C 20
100
ω [Hzs]
B 3.9x107 4.1x107 4.3x107 4.5x107
-20
50
ω [Hzs] -40
7 7 7 7
3.9x10 4.1x10 4.3x10 4.5x10 -60
A No Hot Spot
Figure B-3 Impedance magnitude per unit length and phase spectra over the frequency
range from 39MHz to 45 MHz for cables with global thermal aging to simulate 20 years
severity
A 50 ºC (122 ºF)
B 60 ºC (140 ºF)
C 70 ºC (158 ºF)
γ (ω)
i
γr(ω)
0.07 C
3 C
0.06 B
0.05
2 B
0.04 A
A 0.03
1
0.02
ω (Hz) 0.01 ω (Hz)
7
2x10 7
6x10 7
1x10 8
2x10 7
4x10 6x10 7
8x107
Figure B-4 Imaginary and real components of the propagation function over the
frequency range from 1MHz to 100MHz for cables with global thermal aging to simulate
40 years of service at various temperatures
210
A 50 ºC (122 ºF)
B 60 ºC (140 ºF)
C 70 ºC (158 ºF)
|Z0(T)|/l [S/m] Ψ(ω)

108 C
106 0.5 B
104
0.4
102 A
B
100 0.3
98 A C
ω (Hz) ω (Hz)
2x107 6x107 1x108 2x107 6x107 1x108
Figure B-5 Characteristic impedance magnitude per unit length and phase spectra over
the frequency range from 1MHz to 100MHz for cables with global thermal aging to
simulate 40 years of service at various temperatures

A 80
600 60
C A 40
400 20
ω [Hzs]
1.4x10 7
1.8x107 2.2x107
B 200 -20
-40 C
ω [Hzs] -60 B
1.4x107 1.8x107 2.2x107
A No Hot Spot
severity
211
A 80
600 60
B A 40
C
400 20 ω [Hzs]
7 7
1.4x10 1.8x10 2.2x107
200 -20
ω [Hzs] -40 C
B
1.4x107 1.8x107 2.2x107
A No Hot Spot
Figure B-7 Impedance magnitude per unit length and phase spectra over the
frequency range from 14MHz to 22 MHz for cables with global thermal aging to
simulate 20 years of service at 50ºC, plus additional thermal aging to simulate hot
spots of different size

36
40
34 30 A
C
32 B 20 C B
30
10 ω [Hzs]
A 4.12x10 7
4.16x10 7
4.20x107
28 ω [Hzs] -10
-20
4.12x107 4.16x107 4.20x107
A No Hot Spot
spots of different size
212
A
60
C
150 40
A B
20
100 C ω [Hzs]
7 7 7
3.9x10 4.1x10 4.3x10 4.5x107
B -20
50
-40
ω [Hzs]
3.9x10 7
4.1x10 7
4.3x10 7
4.5x107 -60
A No Hot Spot
locations
A 80
A 60
600
B 40
B
400
C 20 ω [Hzs]
7 7
1.4x10 1.8x10 2.2x107
-20
C 200
-40
ω [Hzs] -60
1.4x107 1.8x107 2.2x107
A No Hot Spot
frequency range from 14MHz to 22 MHz for cables with global thermal aging to simulate
20 years of service at 50ºC, plus additional thermal aging to simulate hot spots at
different locations
213
60 B
175 B
150 40
125 20 C
C 100 ω [Hzs]
7 7 7
3.9x10 4.1x10 4.3x10 4.5x107
75 -20
50 A ω [Hzs] -40
A
7 7 7 7
3.9x10 4.1x10 4.3x10 4.5x10 -60
A No Hot Spot
spots of different severity
B 400 60
40
300 A
A 20 ω [Hzs]
200 1.4x107 1.8x107 2.2x107
C 100 -40 C
ω [Hzs] -60
1.4x107 1.8x107 2.2x107 B
A No Hot Spot
40 years of service at 50ºC, plus additional thermal aging to simulate hot spots of
different severity
214
B 500
400 60
C A 40
300 B
20 ω [Hzs]
A 200 7 7
1.4x10 1.8x10 2.2x107
100 ω [Hzs]
-40 C
1.4x107 1.8x107 2.2x107
-60
A No Hot Spot
simulate 40 years of service at 50ºC, plus additional thermal aging to simulate
hot spots of different size
Φ(ω) [Degrees]
|Z(ω)|/l [S/m]
60
175
150 40
C B
125 C B 20
ω [Hzs]
100 7 7 7
3.9x10 4.1x10 4.3x10 4.5x107
75 A -20
50 A
ω [Hzs] -40
3.9x107 4.1x107 4.3x107 4.5x107 -60
A No Hot Spot
40 years of service at 50ºC, plus additional thermal aging to simulate hot spots of
different size
215
175 C
60
150 40
125 C B
20 ω [Hzs]
100 7 7 7
3.9x10 4.1x10 4.3x10 4.5x107
75
50 B A ω [Hzs] -40 A
3.9x107 4.1x107 4.3x107 4.5x107 -60
A No Hot Spot
B 1.0m Hot Spot 1-2m from POM
C 1.0m Hot Spot 8-9m from POM
40 years of service at 50ºC, plus additional thermal aging to simulate hot spots at
different locations
216

CR 6904

Uploaded by

Copyright:

Available Formats

CR 6904

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

CR 6904

Uploaded by

Copyright:

Available Formats

NUREG/CR-6904

Evaluation of the Broadband

U.S. Nuclear Regulatory Commission

Evaluation of the Broadband

J. Vora, NRC Project Manager

TABLE OF CONTENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii

LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxiii

EXECUTIVE SUMMARY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxv

LIST OF SYMBOLS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxxiii

2. REVIEW OF CONDITION MONITORING TECHNIQUES . . . . . . . . . . . . . . . . . . . . . . . . . 7

3. THEORY OF BROADBAND IMPEDANCE SPECTROSCOPY . . . . . . . . . . . . . . . . . . . . 15

4. PREPARATION AND BASELINE TESTING OF CABLE TEST SPECIMENS . . . . . . . . 31

5. RESULTS OF BIS TESTING OF CABLE SAMPLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

6. EVALUATION OF THE BIS METHOD FOR NUCLEAR PLANT CABLES . . . . . . . . . 149

7. CONCLUSIONS AND CONSIDERATIONS FOR FOLLOW-ON RESEARCH . . . . . . . 155

Figure 6 Frequency-dependent wire resistance and inductance . . . . . . . . . . . . . . . . . . . . . . 22

Figure 7 Micro picture of a poly-x wire cross section . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

Figure 8 Real and imaginary components of the wire insulation dielectric

Figure 13 Equipment and test setup for making impedance measurements . . . . . . . . . . . . . . 28

Figure 19 Imaginary component and real component of propagation function for

Figure 21 Comparison of resistance per unit length predicted by analytical model to

Figure 23 Comparison of inductance per unit length predicted by analytical model to

Figure 25 Conductance-to-frequency ratio per unit length and change in conductance-to-

Figure 29 Comparison of the imaginary component and real component of the

Figure 30 High frequency characteristic impedance magnitude and phase spectra

Figure 39 Schematic model of a cable with a hot-spot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

Figure 46 Comparison of impedance phase in the frequency range from 116kHz to

Figure 50 Comparison of low frequency measured and predicted impedance magnitude

Figure 51 Comparison of low frequency measured and predicted impedance phase

Figure 52 Comparison of high frequency measured and predicted impedance magnitude

Figure 53 Schematic of a cable system with a hot-spot and an attached load . . . . . . . . . . . . 80

Figure 54 Comparison of measured and modeled impedance magnitude for cable

Figure 55 Comparison of low frequency measured and modeled impedance phase

Figure 76 Schematic representation of abraded cable with a conductive path between

Figure 78 Comparison of magnitude for total impedance, interaction impedance, and

Figure 79 Comparison of conductivity path impedance phase, >(T), cable impedance

Figure 82 Comparison of high frequency impedance magnitude and phase spectra in

Figure 83 Imaginary component of the high frequency propagation function for

Figure 90 Comparison of zero crossings in the frequency range 4.4MHz to 4.8MHz

Figure 93 Comparison of predicted and measured impedance magnitude in the range

Figure 94 Characteristic impedance magnitude in the frequency range from 2.55MHz

Figure 95 Characteristic impedance phase spectra in the frequency range from

Figure 96 Imaginary component of the propagation function in the frequency

Figure 102 Schematic of cable with cracking damage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124

Figure 122 Comparison of extracted characteristic impedance magnitude for 10 meter

Figure 123 Comparison of extracted characteristic impedance phase spectra for 10 m

Figure 124 Comparison of imaginary component of the propagation function for

Figure 131 Error in predicted impedance phase as a function of assumed hot-spot

Table 8 Differences in zero crossing frequencies between cable with no hot-spot

Table 11 Specimens tested to evaluate BIS for detecting abrasion . . . . . . . . . . . . . . . . . . . . 84

Existing regulatory requirements relevant to electric cables:

Electric power systems important to safety shall be designed to permit appropriate

• 10CFR, part 50, Appendix B, Criterion XI

• Global thermal degradation,

Specific conclusions from this research are:

AWG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . American Wire Gauge

d . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Diameter of cable conductors (cm/in.)

* . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Difference between parameters

Broadband impedance spectroscopy (BIS) is a technique being developed by the Boeing