Partial Discharge Testing

PROPAGATION OF HIGH FREQUENCY PARTIAL
DISCHARGE SIGNAL IN POWER CABLES
2009
HIO NAM JOHNSON, O
Supervisor: Assoc. Prof. Trevor R. Blackburn
School of Electrical Engineering and Telecommunications

The University of New South Wales, Australia
THE UNIVERSITY OF NEW SOUTH WALES
Thesis/Dissertation Sheet
Surname or Family name: O

First name: Hio Nam Other name/s: Johnson
Abbreviation for degree as given in the University calendar:
Doctor of Philosophy
School: Electrical Engineering & Faculty: Faculty of Engineering
Telecommunications
Title: Propagation of High Frequency Partial Discharge Signal in Power Cables
Abstract 350 words maximum:
The insulation lifetime of power cables is determined by several factors. One of the most important of these is the occurrence of partial discharge (PD)
at the dielectric. The ability to detect and locate a PD source is limited by attenuation of the high frequency PD pulses as they propagate through the
cable to the sensor. Therefore it is necessary to understand the high frequency response of such cables. The ultimate aim of this thesis is to develop an
accurate frequency-dependent cable model for detecting and locating degraded insulation regions on power cables, caused by partial discharge
activities.
Numerical methods can calculate field distribution in the vicinity of a cavity of non-standard shape which generates PDs, and is difficult to calculate by
analytical methods. The simulated results show the important influence of the shape of cavity on the electric stress within it. The cavity stress
enhancement increases as the permittivity of the dielectric increases. The increase is greater for cavities with large diameter to thickness ratios. A cavity
with its axis parallel to the applied field direction has a higher stress enhancement. In addition the stress distribution in the cavity is smaller for spherical
cavities than for cylindrical types.
The research results show that the semi-conducting layers response voltage increases as frequency increases. This indicates that the semi-conducting
layers can have high sensitivity for detection of partial discharge signals and this may be a useful feature to incorporate in the design of cables and in the
application of cable models. By using ATPDraw, FEM and EMTP-RV techniques, three different types of cable models are developed. The simulated
results give a good agreement with the measured results on the single and three phase power cable. The developed cable model can use for
reconstruction of PD source signal by using the receiving signal captured at the cable ends. It is important to use the true pulse shape because it is
characteristic of the PD type and location.
An investigation into the possibility of detecting different PD patterns and signals when conducting PD tests using different sensor bandwidths is also
presented in this thesis. The occurrence of discharge activity was created by an artificial defect manufactured in the single core cable insulation. The
artificial defect generated internal discharge and was used to investigate the PD signal propagation on cross-linked polyethylene (XLPE) cable.
Capacitance coupled external sensors have been applied for the PD detection measurements and the results show that these external sensors have a
number of advantages compared to high frequency current transformer (HF-CT) sensors for the detection of PD pulses.
In addition, development of a method to detect cross-coupling of PD signals between phases of a three core cable and location of the PD source on the
three phase cable. In order to visualize recorded data gained by PD measurement of three phase cable under test, the 3PARD diagram was used. Each
data pulse is assigned to a single dot in the (scatter plot) diagram. The measured results show that the 3PARD diagram allowed the user to identify the
fault between phases with PD location.
The model used for reconstruction which includes the effect of semicon material in the losses provides accurate reproduction of the propagation
characteristics of high frequency PD pulses and the thesis work had used such a model to reconstruct PD waveforms of site PDs from the measured
signal for the first time. The use of the original waveform is important for PD identification and location in the practical situation.
Declaration relating to disposition of project thesis/dissertation
I hereby grant to the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or in
part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all
property rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation.
I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstracts International (this is applicable to doctoral
theses only).
…………………………………………………………… ……………………………………..……………… ……….……………………...…….…

Signature Witness Date
The University recognises that there may be exceptional circumstances requiring restrictions on copying or conditions on use. Requests for
restriction for a period of up to 2 years must be made in writing. Requests for a longer period of restriction may be considered in exceptional
circumstances and require the approval of the Dean of Graduate Research.
FOR OFFICE USE ONLY Date of completion of requirements for Award:
THIS SHEET IS TO BE GLUED TO THE INSIDE FRONT COVER OF THE THESIS

Abstract
________________________________________________________________________
Abstract
The insulation lifetime of power cables is determined by several factors. One of the most
important of these is the occurrence of partial discharge (PD) at the dielectric. The ability
to detect and locate a PD source is limited by attenuation of the high frequency PD pulses
as they propagate through the cable to the sensor. Therefore it is necessary to understand
the high frequency response of such cables. The ultimate aim of this thesis is to develop
an accurate frequency-dependent cable model for detecting and locating degraded
insulation regions on power cables, caused by partial discharge activities.
Numerical methods can calculate field distribution in the vicinity of a cavity of non-
standard shape, which generates PDs, and is difficult to calculate by analytical methods.
The simulated results show the important influence of the shape of cavity on the electric
stress within it. The cavity stress enhancement increases as the permittivity of the
dielectric increases. The increase is greater for cavities with large diameter to thickness
ratios. A cavity with its axis parallel to the applied field direction has a higher stress
enhancement. In addition the stress distribution in the cavity is smaller for spherical
cavities than for cylindrical types.
The research results show that the semi-conducting layers response voltage increases as
frequency increases. This indicates that the semi-conducting layers can have high
sensitivity for detection of partial discharge signals and this may be a useful feature to
incorporate in the design of cables and in the application of cable models. By using
ATPDraw, FEM and EMTP-RV techniques, three different types of cable models are
developed. The simulated results give a good agreement with the measured results on the
single and three phase power cable. The developed cable model can be used for
reconstruction of PD source signal by using the receiving signal captured at the cable
I
Abstract
________________________________________________________________________
ends. It is important to use the true pulse shape because it is characteristic of the PD type
and location.
An investigation into the possibility of detecting different PD patterns and signals when
conducting PD tests using different sensor bandwidths is also presented in this thesis. The
occurrence of discharge activity was created by an artificial defect manufactured in the
single core cable insulation. The artificial defect generated internal discharge and was
used to investigate the PD signal propagation on cross-linked polyethylene (XLPE) cable.
Capacitance coupled external sensors have been applied for the PD detection
measurements and the results show that these external sensors have a number of
advantages compared to high frequency current transformer (HF-CT) sensors for the
detection of PD pulses.
In addition, development of a method to detect cross-coupling of PD signals between

phases of a three core cable and location of the PD source on the three phase cable. In
order to visualize recorded data gained by PD measurement of three phase cable under
test, the 3PARD diagram was used. Each data pulse is assigned to a single dot in the
(scatter plot) diagram. The measured results show that the 3PARD diagram allowed the
user to identify the fault between phases with PD location.
The model used for reconstruction, which includes the effect of semi-conducting material
in the losses, provides accurate reproduction of the propagation characteristics of high
frequency PD pulses and the thesis work had used such a model to reconstruct PD
waveforms of site PDs from the measured signal for the first time. The use of the original
waveform is important for PD identification and location in the practical situation.
II
Certificate of Originality
________________________________________________________________________
Certificate of Originality
I hereby declare that this submission is my own work and to the best of my knowledge it
contains no matter previously published or written by another person, nor material which
to a substantial extent has been accepted for the award of any other degree or diploma at
UNSW or any other educational institution, except where due acknowledgement is made
in the thesis. Any contribution made to the research by others, with whom I have worked
at UNSW or elsewhere, is explicitly acknowledged in the thesis.
I also declare that the intellectual content of this thesis is the product of my own work,
except to the extent that assistance from others in the project’s design and conception or
in style, presentation and linguistic expression is acknowledged.
Signed…………………………………………………
Date………………………………………………….
III
Copyright Statement
________________________________________________________________________
Copyright Statement
‘I hereby grant the University of New South Wales or its agents the right to archive and
to make available my thesis or dissertation in whole or part in the University libraries in
all forms of media, now or here after known, subject to the provisions of the Copyright
Act 1968. I retain all proprietary rights, such as patent rights. I also retain the right to use
in future works (such as articles or books) all or part of this thesis or dissertation.
I also authorize University Microfilms to use the 350 word abstract of my thesis in
Dissertation Abstract International (this is applicable to doctoral theses only).
I have either used no substantial portions of copyright material in my thesis or I have

obtained permission to use copyright material; where permission has not been granted
have applied/will apply for a partial restriction of the digital copy of my thesis or
dissertation.’
Signed…………………………………………………
Date………………………………………………….
IV
Authenticity Statement
________________________________________________________________________
Authenticity Statement
‘I certify that the Library deposit digital copy is a direct equivalent of the final officially
approved version of my thesis. No emendation of content has occurred and if there are
any minor variations in formatting, they are the result of the conversion to digital format.’
Signed………………………………………………
Date………………………………………………
V
Acknowledgement
________________________________________________________________________
Acknowledgement
The completion of this thesis report was made possible by the co-operation of numerous
individuals. I would like to take this opportunity to express my greatest appreciation for
their valuable contributions.
First and foremost, I would like to show my deepest thanks to my supervisor, Associate
Professor Dr Trevor R. Blackburn for his valuable guidance and advice throughout the
completion of this report. Thank you also for your comments and keen eyes in the
preparation and proof reading of this thesis.
I gratefully acknowledge the support and suggestions of Professor Mehdi Vakilian

(Sharif University of Technology, Iran), Senior Lecturer Dr. B. T. Phung and Technical /
OH&S Officer, Zhenyu Liu. Special thanks must go to my fellow colleagues Hao Zhang,
Mohammad Salay Naderi (Sharif University of Technology, Iran) and Kai Xian Lai for
our joint efforts in some experiments.
Additionally, I wish to express my appreciation to the School of Electrical Engineering

and Telecommunications for its providing the entire academic facilities.
I would also like to thank my family members, friends and Eric Leung family for the love
and support they provided throughout especially my elder brother Hiu Fong (Anderson),
O and my elder sister Hiu Loi (Biddy), O.
VI
Acknowledgement
________________________________________________________________________
Finally and most important, I wish to express my deepest gratitude and appreciation to
my dearest parents I Kin, O and Wai Peng (Rita), Wong. Great thanks go to my parents
who gave me life and taught me the way to treat life.
Hio Nam Johnson, O

Sydney, Australia
VII
Contents
________________________________________________________________________
Contents
ABSTRACT......................................................................................................................... I
CERTIFICATE OF ORIGINALITY................................................................................ III
COPYRIGHT STATEMENT........................................................................................... IV
AUTHENTICITY STATEMENT ..................................................................................... V
ACKNOWLEDGEMENT ................................................................................................ VI
CONTENTS................................................................................................................... VIII
TABLE OF CONTENTS.................................................................................................. IX
LIST OF FIGURES ...................................................................................................... XVII
LIST OF TABLES.....................................................................................................XXXV
VIII
Table of Contents
________________________________________________________________________
Table of Contents
CHAPTER 1
Introduction
1.0 Background ...................................................................................................... 1
1.1 Research Goals................................................................................................. 2
1.2 Achievements................................................................................................... 3
1.3 Synopsis of this thesis...................................................................................... 4
CHAPTER 2
Power Cables and Partial Discharge
2.0 Underground Transmission............................................................................... 6
2.0.1 Cable versus Overhead transmission line .................................................. 6
2.1 Power cables .................................................................................................... 9
2.1.1 PILC cables............................................................................................. 10
2.1.2 XLPE cables............................................................................................ 11
2.1.3 Classification of cables ........................................................................... 12
2.2 Cable Structural Materials .................................................................................. 13
2.2.1 Conductors ................................................................................................. 13
2.2.2 Conductor Resistance................................................................................. 16
2.2.3 Cable insulation ........................................................................................ 16
XLPE................................................................................................................. 17
P.V.C................................................................................................................. 17
Paper ................................................................................................................. 17
Rubber............................................................................................................... 17
Mineral.............................................................................................................. 18
Gas insulation.................................................................................................... 18
2.2.4 Electrical properties of insulating materials ............................................. 18
IX
Table of Contents
________________________________________________________________________
Dielectric constant (permittivity)...................................................................... 19

Dissipation Factor ............................................................................................. 19
Dielectric Strength ............................................................................................ 20
Electrical Trees ................................................................................................. 20
Water Trees....................................................................................................... 21
Thermal Breakdown.......................................................................................... 23
2.2.5 Semi-conducting Screen Materials ........................................................... 23
2.2.6 Carbon–polyethylene compound .............................................................. 23
2.2.7 Conductive paper layer ............................................................................. 25
2.2.8 Screens and Jackets................................................................................... 26
2.3 Power cable diagnostics................................................................................. 26
2.3.1 Off-line diagnostics................................................................................... 27
Loss factor......................................................................................................... 27
Dielectric polarization measurements............................................................... 27
Polarization / depolarization current................................................................. 27
Return voltage................................................................................................... 27
Partial discharge diagnostics............................................................................. 28
Time Domain Reflectometry ............................................................................ 28
2.3.2 On-line diagnostics ................................................................................... 28
DC current measurement .................................................................................. 28
Partial discharge diagnostics............................................................................. 28
2.4 Partial Discharge............................................................................................. 29
2.4.1 Partial Discharge definition and terminology ........................................... 29
2.5 Classification of Partial Discharge................................................................... 31
2.5.1 Internal Discharge...................................................................................... 34
2.5.2 Corona Discharge....................................................................................... 35
2.5.3 Surface Discharge ...................................................................................... 36
2.5.4 Partial Discharge Patterns ......................................................................... 37
2.6 Partial Discharge Detection ............................................................................. 39
2.6.1 Non-Electrical Methods ............................................................................ 40
2.6.2 Electrical Methods .................................................................................... 41
X
Table of Contents
________________________________________________________________________
2.6.3 Calibration................................................................................................ 41
2.6.4 Direct Coupling to HV Connection ......................................................... 41
2.6.5 PD detection using High Frequency Current Transformer or other External
Sensors .................................................................................................................. 44
2.6.6 VHF/UHF PD Detection System .............................................................. 45
2.7 Partial Discharges in Cables ......................................................................... 47
2.7.1 Partial Discharge Signal Characteristics................................................... 47
2.7.2 Attenuation and Velocity Characteristics of Cables ................................ 48
2.7.3 Time and Frequency Domain PD Testing ................................................ 51
CHAPTER 3
Cable Modelling
3.0 Introduction..................................................................................................... 53
3.1 Theory of the Transmission line ..................................................................... 54
3.2 Historical Development in Cable Modeling .................................................. 58
3.2.1 Influences from the Cable Design.............................................................. 58
3.2.2 Calculation of Propagation Characteristics................................................ 59
3.3 Methodology ................................................................................................... 60
3.3.1 Analytical method...................................................................................... 61
3.3.2 Background of Five EMTP Programs........................................................ 62
3.3.3 Numerical Integrator Substitution.............................................................. 64
3.4 Transmission lines model ............................................................................... 64
3.4.1 Frequency independent model ................................................................... 66
PI Section Model............................................................................................... 66
Bergeron’s model.............................................................................................. 68
Constant distributed parameter models............................................................. 70
3.4.2 Frequency-dependent model...................................................................... 70
Semlyen's Model............................................................................................... 71
José Martí's Model ............................................................................................ 71
L Martí's Model ................................................................................................ 74
Taku Noda's Model........................................................................................... 74
XI
Table of Contents
________________________________________________________________________
3.4.3 Underground Cable Parameters ................................................................. 77

3.5 Issue of semi-conducting layers in ATP programs ......................................... 78
3.5.1 Importance of including semi-conducting layers in ATP models ........ 79
3.6 Numerical Method .......................................................................................... 83
3.6.1 Finite Element Method ............................................................................. 85
3.6.2 Advantage of FEM..................................................................................... 87
3.7 Comparison between Both Methods............................................................... 89
3.8 Conclusion and Summary ................................................................................... 90
3.8.1 ATP Program versus FEM Technique..................................................... 91
3.8.2 Factors that will affect the results of cable simulation: ............................. 92
CHAPTER 4
Field Analysis
4.0 Introduction..................................................................................................... 93
4.0.1 Partial Discharge in Cavities...................................................................... 94
4.1 Analytical method............................................................................................. 94
4.2 Numerical method............................................................................................. 97
4.2.1 Finite Element Method .............................................................................. 97
4.2.2 Model Overview .......................................................................................... 99
4.3 Results and Discussion ................................................................................... 102
a) Influence of cavity size and thickness of insulation ................................... 102
b) Influence of cavity position ........................................................................ 105
c) Cavity orientation ....................................................................................... 106
d) Field variation inside the cavity.................................................................. 107
e) Influence of nearby cavities........................................................................ 109
f) Comparison of cavity shapes....................................................................... 110
4.4 Influence of different parameters of conducting, semi-conducting and insulating
layers ....................................................................................................................... 111
4.4.1 Influence of the cable insulation ................................................................ 111
A). Thickness of the Insulation....................................................................... 111
B) Dielectric loss tangent................................................................................ 113
XII
Table of Contents
________________________________________________________________________
4.4.2 Influence of skin effect .............................................................................. 119

4.5 Influence of Semi-conducting Layers............................................................... 124
4.5.1 The Parameter of Semi-Conducting Layers............................................... 124
4.5.2 The Thickness of Semi-Conducting Layers............................................... 139
4.6 Conclusion ...................................................................................................... 143
CHAPTER 5
High Frequency Pulse Measurements and Analysis
5.0 Introduction....................................................................................................... 145
5.1 Characteristics of Power Cable......................................................................... 145
5.2 Purpose of the test............................................................................................. 146
5.3 Test Setup for Cable Measurements ................................................................. 146
5.4 Sensor Selection.............................................................................................. 148
5.4.1 Sensor Types............................................................................................ 148
Capacitive sensor ............................................................................................ 149
Sheath champing sensor.................................................................................. 150
Axial magnetic field sensors........................................................................... 151
High frequency current transformer................................................................ 152
Rogowski coil ................................................................................................. 154
5.5 Measurement of Semi-Conducting Material................................................... 155
5.5.1 Cable Sample ............................................................................................. 155
5.5.2 Preparation of the semi-conducting sample............................................. 156
5.5.3 Measurement Setup and Results ................................................................ 156
5.6 Verification of the High Frequency Cable Model .......................................... 158
5.6.1 Verification of the results in frequency domain ...................................... 160
a) High frequency sinusoidal source measurement ........................................ 160
b) High frequency sinusoidal signals on the semi-conducting layers............. 161
5.62 Verification of the results in frequency domain ......................................... 164
a) Verification of paper insulated cable model ............................................... 167
b) Simulation of PD propagation by ATP....................................................... 169
C) Verification of XLPE cable model ............................................................ 170
XIII
Table of Contents
________________________________________________________________________
d) Simulation of PD propagation in XLPE cables by ATP ............................ 172

5.7 Propagation Constant vs. Cable Length.......................................................... 176
5.8 Reconstruction of Signal Source..................................................................... 178
5.9 Conclusion ...................................................................................................... 180
CHAPTER 6
Partial Discharge Detection
6.0 Introduction................................................................................................... 182
6.0.1 Characteristics and structures of XLPE cable.......................................... 183
6.0.2 High frequency attenuation...................................................................... 184
6.1 Measurement................................................................................................... 184
6.1.1 Measurement Set up................................................................................. 185
6.2 Selection of detection sensors......................................................................... 189
6.2.1 Conventional detection sensor ................................................................. 189
6.2.2 Comparison result .................................................................................... 189
6.3 Measurement Results ...................................................................................... 191
6.3.1. PD signal propagation............................................................................. 191
6.3.2. PDs pattern of the measurement ............................................................. 192
6.3.3. PDs pattern from the external sensors .................................................... 193
6.4 Evaluation of Partial Discharge ...................................................................... 195
6.4.1. Introduction of three different methods .................................................. 195
6.4.2. Measurement results of different location of sensors ............................ 197
6.4.3. Simulated results..................................................................................... 199
6.5 Conclusion ...................................................................................................... 201
CHAPTER 7
PD Characteristics and Defect Patterns in Three Phase Power Cables
7.0 Introduction................................................................................................... 202
7.1 Structures and Performance of XLPE Cables................................................. 203
7.2 Three phase cable parameters ......................................................................... 204
7.2.1 Method I: Calculating parameters with Ohm’s law................................. 206
XIV
Table of Contents
________________________________________________________________________
7.2.2 Method II: Calculating parameters using the energy method.................. 206
7.3 Experiment Setup............................................................................................ 208
7.4 PD Evaluation Technique ............................................................................... 210
7.5 Experimental Results ..................................................................................... 212
7.5.1 100pC calibrating signal ..................................................................... 212
7.5.2 Measurement under the noisy environment............................................... 214
7.5.3 Measurement under the laboratory ............................................................ 217
7.6 Conclusion ...................................................................................................... 220
CHAPTER 8
High Frequency PD Signal Propagation in Three Phase Power Cables
8.0 Introduction..................................................................................................... 221
8.1 Historical Development in Three Phase Power Cable Modelling ................... 222
8.2 Propagation Measurements............................................................................. 224
8.2.1 Meaurement Setup ................................................................................... 224
8.2.2 Structure of Three Phase Cables .............................................................. 225
A. Three core belted paper cable (Figure 8.3) ............................................... 226
B. Three phase XLPE cable ............................................................................ 226
8.2.3 Frequency Response Measurements .......................................................... 227
8.3 Measurement of Voltage Response to PD calibration signals using the Semi-
conducting Layers................................................................................................... 230
8.4 Cable Model...................................................................................................... 237
8.5 Use of PD calibrator signal for measurements ................................................. 240
8.5.1 Three core paper belted type cable ............................................................ 241
8.5.2. Three phase three phase XLPE cable........................................................ 244
8.6 Conclusion .................................................................................................. 248
CHAPTER 9
Conclusion and Future Study
9.0 Partial Discharges in Cavity........................................................................ 252
9.1 Influence from Semi-Conducting layers..................................................... 253
XV
Table of Contents
________________________________________________________________________
9.2 Sensor Selection.......................................................................................... 254

9.3 Cable Modeling........................................................................................... 255
9.4 HF Measurement and Simulation ............................................................... 255
9.5 PD Detection on Cables .............................................................................. 257
9.6 Suggestions for Further Work..................................................................... 258
References ............................................................................................... 260
Appendix I
The Parameter of Semi-Conducting Layers............................................................ 273
Appendix II
The Result of Frequency Response Measurements in Chapter 8.2.3 ..................... 290
Appendix III
Measured Voltage on Semi-conducting Layer of Phase A, B and C (-100pC) ...... 302
Appendix IV
Measured Voltage on Semi-conducting Layer of Phase A, B and C (100pC) ....... 307
XVI
Table of Figures
________________________________________________________________________
List of Figures
Figure 2.1: Schematic design of overhead line................................................................... 7

Figure 2.2: The ICNIRP guidelines set fundamental limits on electrical currents induced
in the body by EMF ............................................................................................................ 8
Figure 2.3: Structure of single core PILC cable with round conductor. ........................... 10
Figure 2.4: Structure of three core belted type cable with oval conductors. .................... 11
Figure 2.5: Structure of the 11kV single core XLPE cable. ............................................. 12
Figure 2.6: conductor shape ............................................................................................. 15
Figure 2.7: Electrical Treeing and paper degradation below a dry cable termination...... 21
Figure 2.8: Water trees in power cable insulation. ........................................................... 22
Figure 2.9: Surface of carbon- polyethylene compound ................................................. 24
Figure 2.10 (a): Real (1) part and imaginary (2) part of the complex conductivity.
Figure 2.10 (b): permittivity of semi-conducting carbon-polyethylene compound in a
frequency range up to 30 MHz . ....................................................................................... 24
Figure 2.11 (a): Real (1) part and imaginary (2) part of the complex conductivity:
Figure 2.11 (b): Real (r1) part and imaginary (r2) part of permittivity of conductive
semi-conducting paper layer in a frequency range up to 30 MHz ................................... 25
Figure 2.12: The relationship between ionization with different types of discharges ...... 31
Figure 2.13: Internal discharges ....................................................................................... 32
Figure 2.14: Surface discharges ....................................................................................... 32
Figure 2.15: Corona discharges. ....................................................................................... 33
Figure 2.16: A combination of internal discharge and surface discharge [19]................. 33
Figure 2.17: A combination of corona discharge and surface discharge [19]. ................. 33
Figure 2.18: Internal Discharge in a Solid Dielectric Bounded Cavity [25]. .................. 35
Figure 2.19: Corona Discharge. ....................................................................................... 35
Figure 2.20: Surface Discharge........................................................................................ 36
XVII
Table of Figures
________________________________________________________________________
Figure 2.21: Typical discharge from a cavity in insulation. ............................................. 37

Figure 2.22: Typical discharge across surface of insulation............................................. 37
Figure 2.23: Typical corona discharge in gas. .................................................................. 38
Figure 2.24: Irregular discharge pattern ........................................................................... 38
Figure 2.25: Ultrasonic PD detector. ................................................................................ 40
Figure 2.26: Basic Partial Discharge measurement circuits specified in the IEC-270
Standard ............................................................................................................................ 42
Figure 2.27: PD detector made from Robinson Electronic Instruments Company. ......... 43
Figure 2.28: Conventional Electrical Discharge Detection Equivalent Circuit............... 43
Figure 2.29: Application of HFCT and clamping sensor for PD detection in power cable.
........................................................................................................................................... 45
Figure 2.30: PD signal captured from the XLPE cable. ................................................... 48
Figure 2.31: An equivalent circuit of cable parameter that first introduced by Stone and
Boggs . .............................................................................................................................. 50
Figure 2.32: PD waveforms as a function of distance propagated through a cable with
frequency dependent attenuation. The distances propagated are 0, 100, 300 and 1000m.51
Figure 3.1: Equivalent circuit of a two conductor transmission line of length x............ 54
Figure 3.2: Decision tree for transmission line model selection....................................... 66
Figure 3.3: Nominal PI section. ........................................................................................ 67
Figure 3.4: Equivalent two port network for line with lumped losses.............................. 68
Figure 3.5: Equivalent two port network for half line section.......................................... 69
Figure 3.6: Bergeron transmission line model.................................................................. 69
Figure 3.7: Physical interpretation of the new weighting function.................................. 72
Figure 3.8: J-Marti frequency – dependent line models at nodes k and m. ...................... 73
Figure 3.9: Simulated results of four different frequency dependent cable models ......... 76
Figure 3.10: The cross section of coaxial cable................................................................ 78
Figure 3.11: Representation of a single core cable in ATP. ............................................. 79
XVIII
Table of Figures
________________________________________________________________________
Figure 3.12: the ATP simulated result for comparing the velocity of propagation without
semi-conducting layers. Red line is the sending signal and green line is the receiving
signal from the ATP simulation........................................................................................ 80
Figure 3.13: Comparison of two different methods of including the semi-conducting
layers in the cable model................................................................................................... 82
Blue line (a) is the result using first method and the green line (b) is that obtained using
the second method............................................................................................................. 82
Figure 3.14: Cable model in ATP. .................................................................................... 83
Figure 3.15: Geometrical configuration of single core cable in FEM program. The finite
element method can be used to include the actual geometry of the cable. ....................... 85
Figure 3.16: Mesh structure used in the 2D FEM model.................................................. 86
Figure 3.17: FEM result of the percentage increase of attenuation when a lead sheath is
used instead of a copper sheath......................................................................................... 87
Figure 3.18 a: FEM can model different structures and different number of wires in the
sheath layer. ...................................................................................................................... 88
Figure 3.18 b: the restricted form of sheath structure that ATP cable model assumes in its
model................................................................................................................................. 88
Figure 3.19: Attenuation factor versus frequency as a function of loss factor of insulation.
........................................................................................................................................... 89
Figure 4.1: 3-D model of a spherical void with its axis parallel to the electrical field
distribution. ....................................................................................................................... 95
Figure 4.2: 2-D model of a spherical void with varied radius a and b, which located in
insulation of thickness t and permittivity r. ..................................................................... 95
Figure 4.3: A void located in a uniform field. .................................................................. 95
Figure 4.4: Initialized mesh consists of 1618 elements. ................................................... 98
Figure 4.5: Refined mesh consists of 6472 elements........................................................ 98
Figure 4.6: Refine selection mesh consists of 3084 elements. ......................................... 99
Figure 4.7: Flow chart of the model................................................................................ 101
Figure 4.8: From left to right, the spherical (circle), elliptical and side-section of
cylindrical (rectangle or square) shape of gaseous cavity. ............................................. 102
XIX
Table of Figures
________________________________________________________________________
Figure 4.9: Field ratio (E / Eo) for a circular and elliptical shape of gaseous cavity
embedded in the insulation, r =2.5, effect of cavity size and insulation thickness. Top
figure shows the result in 2-D and bottom figure shows the result in 3-D. .................... 103
Figure 4.10: Field ratio (E / Eo) for a cylindrical section shape of cavity embedded in the
insulation, r =2.5, effect of cavity size and insulation thickness. Top figure shows the
result in 2-D and bottom figure shows the result in 3-D................................................. 104
Figure 4.11: Field ratio (E / Eo) for a circular and elliptical shape of cavity embedded in
the insulation with semi-conducting layers have been attached. .................................... 105
embedded in the insulation, effect of cavity position with its axis parallel to the applied
electrical field direction. ................................................................................................. 106
Figure 4.13: Cavity Field ratio (E / Eo) for a cylindrical section shape of cavity embedded
in insulation with various cavity orientations. ................................................................ 107
Figure 4.14: Points chosen to compute stress inside the cavity...................................... 108
Figure 4.15: The direction of the field distribution inside the void. ............................... 108
Figure 4.16: Field ratio (E / Eo) for two circle or elliptical shapes of gaseous cavities
embedded in the insulation. ............................................................................................ 110
Figure 4.17: Differential field value in p.u. for two circle or elliptical shapes of gaseous
cavities embedded in the insulation. ............................................................................... 110
Figure 4.18: Determination of signal attenuation versus frequency for the combinations
listed in Table 4.3............................................................................................................ 113
Figure 4.19: High frequency signal attenuation as a function of frequency for XLPE cable
for a range of tan values................................................................................................ 116
Figure 4.20: High frequency signal phase constant as a function of frequency for XLPE
cable (tan = 0.0001). ..................................................................................................... 117
Figure 4.21: High frequency signal attenuation as a function of frequency for paper-
insulated cable for a range of tan values. ...................................................................... 117
Figure 4.22: High frequency signal phase constant as a function of frequency for paper-
insulated cable (tan = 0.001). ........................................................................................ 118
Figure 4.23: High frequency signal attenuation as a function of frequency for Type I
cable and Type II cable with same permittivity and conductivity value of semi-
XX
Table of Figures
________________________________________________________________________
conducting and insulating layers (Both types of cables have dielectric constant of 3.5).
The only different between both types of cables is the structure of cable...................... 118
Figure 4.24: High frequency signal phase constant as a function of frequency for Type I
The only different between both types of cables is the structure of cable...................... 119
Figure 4.25: Skin depths for some metals....................................................................... 121
Figure 4.26: the skin depth of XLPE power cable as function of frequencies and
conductivity..................................................................................................................... 122
conductivity in the logarithmic scale. ............................................................................. 123
Figure 4.28: the graph of attenuation as function of different frequency in the
conductivity of outer semi-conducting layers in the range from 100000 to 0.00001. The
permittivity, of inner semi-conducting layer is equal to 1. .......................................... 127
Figure 4.29: the graph of velocity as function of different frequency in the conductivity of
outer semi-conducting layers in the range from 100000 to 0.00001. The permittivity, of
inner semi-conducting layer is equal to 1. ...................................................................... 128
conductivity of inner semi-conducting layers in the range from 100000 to 0.00001. The
permittivity and conductivity of outer semi-conducting layer is based on Steven Boggs’s
semi-con measurement ................................................................................................... 128
inner semi-conducting layers in the range from 100000 to 0.00001. The permittivity and
conductivity of outer semi-conducting layer is based on Steven Boggs’s semi-con
measurement . ................................................................................................................. 129
permittivity and conductivity of inner semi-conducting layer is based on Steven Boggs’s
outer semi-conducting layers in the range from 100000 to 0.00001. The permittivity and
XXI
Table of Figures
________________________________________________________________________
conductivity of inner semi-conducting layer is based on Steven Boggs’s semi-con

measurement . ................................................................................................................. 130
permittivity and conductivity of outer semi-conducting layer is carbon-polyethylene
compound........................................................................................................................ 131
Figure 4.35: the graph of phase constant as function of different frequency in the
compound........................................................................................................................ 131
conductivity of outer semi-conducting layer is carbon-polyethylene compound........... 132
permittivity and conductivity of inner semi-conducting layer is carbon-polyethylene
compound........................................................................................................................ 133
compound........................................................................................................................ 133
conductivity of inner semi-conducting layer is carbon-polyethylene compound........... 134
permittivity and conductivity of outer semi-conducting layer is conductive paper........ 135
XXII
Table of Figures
________________________________________________________________________
conductivity of outer semi-conducting layer is conductive paper. ................................. 136
permittivity and conductivity of inner semi-conducting layer is conductive paper........ 137
conductivity of inner semi-conducting layer is conductive paper. ................................. 138
Figure 4.46: the graph of attenuation as function of different frequency for each case in
Table 4.6. The permittivity and conductivity of inner and outer semi-conducting layer is
based on K.Steinbuich semi-con measurement . ............................................................ 140
Figure 4.47: the graph of phase constant as function of different frequency for each case
in Table 4.6. The permittivity and conductivity of inner and outer semi-conducting layer
is based on K.Steinbuich semi-con measurement .......................................................... 141
Figure 4.48: the graph of velocity as function of different frequency for each case in
based on K.Steinbuich semi-con measurement . ............................................................ 141
Table 4.6. The permittivity and conductivity of inner semi-conducting layer is based on
K.Steinbuich semi-con measurement and assume that outer semi-conducting layer
employs the same material as the one employed in the inner semi-conducting layer. ... 142
in Table 4.6. The permittivity and conductivity of inner semi-conducting layer is based
on K.Steinbuich semi-con measurement and assume that outer semi-conducting layer
XXIII
Table of Figures
________________________________________________________________________
K.Steinbuich semi-con measurement and assume that outer semi-conducting layer

Figure 5.1: Laboratory test measurement setup.............................................................. 148

Figure 5.2: Different type of sensors and their position on power cable........................ 149
Figure 5.3: The principle of capacitive sensor................................................................ 150
Figure 5.4: The sheath champing sensor......................................................................... 150
Figure 5.5: The principle of axial magnetic field sensor. ............................................... 151
Figure 5.6: lumped element model of axial magnetic field sensor. ................................ 151
Figure 5.7: the axial magnetic field sensor: (a) coaxial wire and (b) metal strap........... 152
Figure 5.8: The principle of inductive sensor. ................................................................ 153
Figure 5.9: The high frequency current transformers (HFCTs). HFCTs-A (Model: Ailtech
80434) has the lower sensitivity compared to others but HFCTs-A can give much more
stable performance; HFCT-B is made in the lab of the University of New South Wales: it
can overcome the commercial HFCT’s limitation of cost; HFCTs-C (Model: Altech
93511-1) gives the highest sensitivity and is suitable for the HV testing....................... 153
Figure 5.10: The principle of Rogowski coil. ................................................................. 154
Figure 5.11: Frequency response of four different sensors............................................. 155
Figure 5.12: the setup of semi-conducting samples measurement. ................................ 157
Figure 5.13: Flow chart for the verification procedure of the cable model. ................... 159
Figure 5.14: Comparison between measured and modeled signal attenuation, with and
without semicon, as a function of frequency. The full line shows measured results: the
green dashed line is numerical simulation with semi-conducting layers: the brown dashed
line is numerical simulation without semicon included.................................................. 161
Figure 5.15: Comparison of measured signal attenuation of the XLPE cable on each semi-
con layers with results of ATPDraw simulation ............................................................. 162
Figure 5.16: Attenuation versus frequency on the inner and outer semi-conducting layers
(136 m and 45 m XLPE cable). ...................................................................................... 163
Figure 5.17: Semi-conducting layer voltage at the receiving end versus frequency on
outer semi-conducting layer and comparison with HFCT measurements. From left: Cable
XXIV
Table of Figures
________________________________________________________________________
B (outer semi-con), Cable A (outer), Cable A (inner), Cable B (CT at rec end), Cable A
(CT at rec end). ............................................................................................................... 164
Figure 5.18: the ATP simulated result for comparing the velocity of propagation. ....... 166
Figure 5.19: The measured result at the sending end of 5 m PILC cable. ...................... 169
Figure 5.20: The measured result at the receiving end of 5 m PILC cable..................... 169
Figure 5.21: The ATPDraw simulation result of a 5 m length of paper insulated cable.
Red line is the sending side and Green line is the receiving side. .................................. 170
Figure 5.22: The measured result at the sending end of 136 m XLPE cable.................. 171
Figure 5.23: The measured result at the receiving end of 136 m XLPE cable. .............. 172
Figure 5.24: The simulated result at the sending end and receiving end of 136 m XLPE
cable. Red line is the sending side and Green line is the receiving side......................... 173
cable. Red line is the sending side and Green line is the receiving side......................... 173
Figure 5.26: The simulated result at the sending and receiving end of semi-conducting
layer voltage. Red line is the sending side and Green line is the receiving side. ........... 174
Figure 5.27: the graph of attenuation (dB/m unit) as function of different frequency in the
dielectric loss (tanG) range from 0.01 to 0.0005. ............................................................ 175
Figure 5.28: the graph of attenuation (dB) as function of different frequency over cable
length of 400 meters........................................................................................................ 177
Figure 5.29: the graph of attenuation (dB) at two specific frequencies of 1 MHz and 5
MHz over cable length of 400 meters............................................................................. 177
Figure 5.30: Explanatory diagram for the reconstruction of signal source..................... 179
Figure 5.31: Reconstruction of signal source by using Equation 5.6. ............................ 179
Figure 6.1: Laboratory test measurement setup.............................................................. 186

Figure 6.2: PD Detector for measuring the partial discharge (D) and voltage level (C)
from the input (B) and connecting to the ground (A). .................................................... 186
Figure 6.3: Cross section diagram of the artificial defect............................................... 187
Figure 6.4: Cross section of the simulated model........................................................... 187
Figure 6.5: Two different types of pin: sharp shape and spherical shape...................... 188
Figure 6.6: Ratio of electric field for the distance between insulation and head of pin. 188
XXV
Table of Figures
________________________________________________________________________
Figure 6.7: Comparison of PD signals captured by different types of HF-CTs ............. 189
Figure 6.8: The compared result of HFCT and External sensor. .................................... 190
Figure 6.9: Frequency spectrum of external sensor 3 and other type of HF-CTs........... 190
Figure 6.10: Comparison of three single PD signals captured at different location of
external sensors............................................................................................................... 192
Figure 6.11: PD pattern of measurement that is captured by Mtronix PD detector........ 193
Figure 6.12: Frequency spectrum of the PD measurement in figure 6.11. ..................... 193
Figure 6.13: PD pattern that is captured by external sensor 1. ....................................... 194
Figure 6.16: Example of the distorted PD pulse. q1, q2 and q3 are in positive parts of
pulse and q4, q5 and q6 are in negative parts of pulse. .................................................. 196
Figure 6.17: Evaluated of PD by different methods (time width of 0.05Ps). ................. 198
Figure 6.20: Results of measurement versus simulation as function of the cable length in
percentage. ...................................................................................................................... 200
Figure 6.21: 3D simulated plot of attenuation as a function of length of cable and of
frequency of PD pulse..................................................................................................... 200
Figure 7.1: Configuration of the 11kV three phase XLPE insulated cable. ................... 203
Figure 7.2: Equivalent circuit and general configuration of three phase cables............. 204
Figure 7.3: The cross section equivalent circuit of three phase cable. ........................... 205
Figure 7.4: Illustration of FEM calculation on three phase cable................................... 205
Figure 7.5: The result of self and mutual inductance in frequency range up to 1MHz. . 208
Figure 7.6: Laboratory test measurement setup for three phase XLPE cable................. 209
Figure 7.7: Construction of 3PARD signal analysis ...................................................... 211
Figure 7.8: Visualization of 3PARD .............................................................................. 212
Figure 7.9: Calibrating signal fed into the phase 1-Unit 1.1........................................... 213
Figure 7.10: Calibrating signal fed into the phase 2-Unit 1.2......................................... 213
XXVI
Table of Figures
________________________________________________________________________
Figure 7.11: Calibrating signal fed into the phase 3-Unit 1.3......................................... 214
Figure 7.12: High level of noise (phase 1) at 2.8 MHz centre frequency and 300 kHz
bandwidth........................................................................................................................ 215
bandwidth........................................................................................................................ 215
bandwidth........................................................................................................................ 216
Figure 7.15: 3PARD diagram at 2.8 MHz centre frequency and 300 kHz bandwidth
(under the noisy environment) ........................................................................................ 216
Figure 7.16: PD of phase 1 at 1.2 MHz centre frequency and 300 kHz bandwidth. ...... 217
Figure 7.19: 3PARD at 1.2 MHz centre frequency and 300 kHz bandwidth. ................ 219
Figure 7.20: 3PARD at 11.7 MHz centre frequency and 300 kHz bandwidth. .............. 219
Figure 8.1: The measurement setup of calibrating signals propagation. ........................ 225
Figure 8.2: The diagram of three phase XLPE cable in laboratory. ............................... 225
Figure 8.3: 22 kV three phase core belted cable............................................................. 226
Figure 8.4: 11kV three –core XLPE insulated three phase cable. .................................. 227
Figure 8.5: The circuit for Frequency Response Analysis over 300 MHz. .................... 227
Figure 8.6: A diagram of R&S & ZVL Vector Network Analyzer. ............................... 228
Figure 8.7: Frequency response magnitude (S21, Damping ratio, paper insulated cable) is
measured for ratio of Phase Asending (Input) to Areceiving (Output).................................... 228
Figure 8.8: Frequency response magnitude (S21, Damping ratio, XLPE cable) is measured
for Phase Asending (Input) to Areceiving (Output)................................................................. 229
Figure 8.9: 100pc slow calibrating voltage signal from the calibrator. .......................... 230
Figure 8.10: -100pc fast calibrating signal from the digital calibrator. .......................... 231
Figure 8.11: -100pc calibrating signal from the digital calibrator. The voltage responses
of inner semi-conducting layers (XLPE cable) on the sending ends are shown in graph.
......................................................................................................................................... 231
XXVII
Table of Figures
________________________________________________________________________
of inner semi-conducting layers (XLPE cable) on the receiving ends are shown in graph.
......................................................................................................................................... 232
Figure 8.13: 100pc calibrating signal from the calibrator. The voltage responses of inner
semi-conducting layers on the sending end are shown in graph..................................... 233
semi-conducting layers on the receiving end are shown in graph. ................................. 233
Figure 8.15: Three phase XLPE cable design in pipe type modeling............................. 238
Figure 8.16: The simulation circuit of three core belt paper cable. ................................ 239
Figure 8.17: The simulation circuit of three phase XLPE cable..................................... 240
Figure 8.18: The measured and simulated result of Phase Asending of three phase paper
belted type cable. ............................................................................................................ 242
Figure 8.19: The measured and simulated result of Phase Areceiving of three phase paper
Figure 8.20: The simulated result of Phase Bsending and Phase Breceiving of three phase paper
Figure 8.21: The measured result of Phase Bsending and Phase Breceiving of three phase paper
Figure 8.22: The simulated result of Phase Asending, Phase Areceiving and Phase Bsending,
Phase Breceiving of three phase paper belted type cable. ................................................... 244
Figure 8.23: The measured and simulated result of Phase Asending of three phase XLPE
cable. ............................................................................................................................... 245
Figure 8.24: The measured and simulated result of Phase Areceiving of three phase XLPE
cable. ............................................................................................................................... 245
Figure 8.25: The measured result of Phase Bsending and Phase Breceiving of three phase
XLPE cable. .................................................................................................................... 246
Figure 8.26: The simulated result of Phase Bsending and Phase Breceiving of three phase
XLPE cable. .................................................................................................................... 246
XXVIII
Table of Figures
________________________________________________________________________
Figure A1.1: the graph of attenuation as function of different frequency in the

conductivity of inner semi-conducting layers in the range from 0 to 100. The permittivity,
of outer semi-conducting layer is equal to 100 and its conductivity is equal to 17. .... 274
Figure A1.2: the graph of velocity as function of different frequency in the conductivity
of inner semi-conducting layers in the range from 0 to 100. The permittivity, of outer
semi-conducting layer is equal to 100 and its conductivity is equal to 17. .................... 274
permittivity, of outer semi-conducting layer is equal to 1 and its conductivity is equal to
0.00001............................................................................................................................ 275
of inner semi-conducting layers in the range from 100000 to 0.00001. The permittivity,
of outer semi-conducting layer is equal to 1 and its conductivity is equal to 0.00001... 275
100000............................................................................................................................. 276
of outer semi-conducting layer is equal to 1 and its conductivity is equal to 100000.... 276
permittivity, of outer semi-conducting layer is equal to 100 and its conductivity is equal
to 1. ................................................................................................................................. 277
of outer semi-conducting layer is equal to 100 and its conductivity is equal to 1.......... 277
to 1. ................................................................................................................................. 278
XXIX
Table of Figures
________________________________________________________________________
of outer semi-conducting layer is equal to 10 and its conductivity is equal to 1............ 278
of inner semi-conducting layers in the range from 100000 to 0.00001. The permittivity
and conductivity of outer semi-conducting layer is based on Steven Boggs’s semi-con
measurement . ................................................................................................................. 280
of outer semi-conducting layers in the range from 100000 to 0.00001. The permittivity
and conductivity of inner semi-conducting layer is based on Steven Boggs’s semi-con
measurement . ................................................................................................................. 281
compound........................................................................................................................ 282
Figure A1.16: the graph of phase constant as function of different frequency in the
compound........................................................................................................................ 283
and conductivity of outer semi-conducting layer is carbon-polyethylene compound. ... 283
XXX
Table of Figures
________________________________________________________________________

compound........................................................................................................................ 284
compound........................................................................................................................ 285
and conductivity of inner semi-conducting layer is carbon-polyethylene compound. ... 285
and conductivity of outer semi-conducting layer is conductive paper. .......................... 287
and conductivity of inner semi-conducting layer is conductive paper. .......................... 289
XXXI
Table of Figures
________________________________________________________________________
Figure A2.1: Frequency response magnitude (S21, Damping ratio) is measured for ratio of
Phase Asending (Input) to Areceiving (Output) ...................................................................... 290
......................................................................................................................................... 290
Figure A2.2: Frequency Spectrum of ratio of Phase Asending (Input signal) to Areceiving
(Output signal). ............................................................................................................... 291
Phase Bsending (Input) to Breceiving (Output)....................................................................... 292
Figure A2.4: Frequency Spectrum of ratio of Phase Bsending (Input) to Breceiving (Output).
......................................................................................................................................... 292
Figure A2.5: Frequency response magnitude (S21, Damping ratio) is measured for ratio
Phase Csending (Input) to Creceiving (Output)....................................................................... 293
Figure A2.6: Frequency Spectrum of Phase Csending (Input) to Creceiving (Output)........... 293
Figure A2.7: Frequency response magnitude (S21, Damping ratio) is measured for Phase
Asending (Input) to Breceiving (Output)................................................................................. 294
Figure A2.8: Frequency Spectrum of Phase Asending (Input) to Breceiving (Output). ......... 294
Asending (Input) to Creceiving (Output)................................................................................. 295
Figure A2.10: Frequency Spectrum of Phase Asending (Input) to Creceiving (Output). ....... 295
Bsending (Input) to Creceiving (Output). ................................................................................ 296
Figure A2.12: Frequency Spectrum of Phase Bsending (Input) to Creceiving (Output)......... 296
Asending (Input) to Areceiving (Output)................................................................................. 297
Figure A2.14: Frequency Spectrum of Phase Asending (Input) to Areceiving (Output). ....... 297
Bsending (Input) to Breceiving (Output). ................................................................................ 298
Figure A2.16: Frequency Spectrum of Phase Bsending (Input) to Breceiving (Output)......... 298
Csending (Input) to Creceiving (Output). ................................................................................ 299
Figure A2.18: Frequency Spectrum of Phase Csending (Input) to Creceiving (Output)......... 299
XXXII
Table of Figures
________________________________________________________________________
Figure A2.19: Frequency response magnitude (S21, Damping ratio) is measured for Inner
Semi-conducting layer of Phase Asending (Input) to Areceiving (Output)............................. 300
Figure A2.20: Frequency Spectrum for Inner Semi-conducting layer of Phase Asending
(Input) to Areceiving (Output)............................................................................................. 300
Figure A2.21: Frequency response magnitude (S21, Damping ratio) is measured for Outer
Semi-conducting layer of Phase Asending (Input) to Areceiving (Output)............................. 301
Figure A2.22: Frequency Spectrum for Outer Semi-conducting layer of Phase Asending
(Input) to Areceiving (Output)............................................................................................. 301
Figure A3.1: -100pc calibrating signal from the digital calibrator. The voltage responses
of inner semi-conducting layers on the sending ends are shown in graph...................... 302
of outer semi-conducting layers on the sending ends are shown in graph...................... 303
of inner semi-conducting layers on the receiving ends are shown in graph. .................. 303
of outer semi-conducting layers on the receiving ends are shown in graph. .................. 304
of inner semi-conducting layers on the sending and receiving ends are shown in graph.
......................................................................................................................................... 304
of outer semi-conducting layers on the sending and receiving ends are shown in graph.
......................................................................................................................................... 305
......................................................................................................................................... 305
......................................................................................................................................... 306
XXXIII
Table of Figures
________________________________________________________________________
Figure A4.1: 100pc calibrating signal from the calibrator. The voltage responses of inner
semi-conducting layers on the sending ends are shown in graph. .................................. 307
Figure A4.2: 100pc calibrating signal from the calibrator. The voltage responses of outer
semi-conducting layers on the sending ends are shown in graph. .................................. 308
semi-conducting layers on the receiving ends are shown in graph................................. 308
semi-conducting layers on the sending and receiving ends are shown in graph. ........... 309
XXXIV
Table of Tables
________________________________________________________________________
List of Tables
Table 2.1: Resistivity of conductor materials at 20oC [10]............................................... 14

Table 2.2: Physical parameters of cable conductor materials........................................... 14
Table 2.3: Dielectric constants of various materials used as electrical insulation............ 19
Table 2.4: Difference characteristics between Water trees and Electrical trees. .............. 22
Table 3.1: Simulation results of different permittivity of insulation. ............................... 80

Table 3.2: Attenuation in dB/m for specific frequencies determined by measurement,
ATP modeling and FEM modeling................................................................................... 90
Table 4.1: Field ratio (E / Eo) at point N inside the cavity versus b/a (the cavity
parameters), while a/t=0.1 .............................................................................................. 108
Table 4.2: Field value (E / Eo) at point N inside the cavity with different b/a and a/t,
increasing in x-axis direction. ......................................................................................... 109
Table 4.3: Combinations of thickness of insulation and semi-conducting layers used in
analysis............................................................................................................................ 113
Table 4.4: The skin depth for different materials at frequency of 50 Hz........................ 120
Table 4.5: The skin depth for copper materials at various frequencies. ......................... 121
Table 4.6: Thickness of and semi-conducting layers used in analysis. The permittivity and
conductivity of inner and outer semi-conducting layer is based on K.Steinbuich semi-con
measurement. .................................................................................................................. 140
Table 5.1: Detail of cable examined. .............................................................................. 156

Table 5.2: the result of semi-con samples measurement. ............................................... 158
Table 5.3: Brief comparison of measurement and simulation results for cal. PD signal of
1000 pC........................................................................................................................... 170
XXXV
Table of Tables
________________________________________________________________________
1000 pC........................................................................................................................... 175
Table 8.1: Characteristics of high voltage insulating materials [4]. ............................... 227
Table 8.2: Error in fitting with the experimental results in Section 8.5.......................... 248
XXXVI
Chapter 1: Introduction 1
________________________________________________________________________
Chapter 1
Introduction
1.0 Background
The history of power cables begins at the end of the 19th century [1]. Power cables form
the essential connecting link between one piece of electrical apparatus or machinery and
another. They have advantages in esthetic, environmental and safety aspects compared
with the overhead transmission lines. Therefore most distribution networks of medium
and low voltages are constructed with power cables rather than overhead transmission
lines. However, a majority of the distribution grid failures are attributed to the power
cables [1].
The main factor causing the power system failure is insulation breakdown. The insulation
lifetime of power cables is determined by several factors. One of the more important of
these is the occurrence of partial discharge (PD) in the dielectric or at the dielectric
surface. Partial discharge is a localized electrical discharge that does not bridge the main
insulation but can cause damage to the insulation material and degrade it. Various
defects, such as voids, contaminants and electrical trees can cause partial discharge (PD)
activity in the cable insulation [2]. PD activity can cause rapid progressive deterioration
and failure. To ensure the reliability of the whole cable system, PD testing and location
are becoming an essential part of the monitoring and testing of high voltage power
cables. The ability to detect and locate a PD source in a cable however is limited by
attenuation of the high frequency PD pulses as they propagate through the lossy dielectric
insulation of the cable [3]. Therefore it is necessary to understand the high frequency
response of power cables and its effect on PD propagation.
________________________________________________________________________
Further, to enable reconstruction of PD signals as emitted a viable high frequency model

for simulation is needed.
The work described in this thesis has been involved with development of such a model
for application to high frequency PD signal analysis, with the ultimate aim being to
enable reconstruction of emitted PD signals from the measured distorted signal
propagated to the sensor over some length of cable. In addition to the cable model
developments using the transient analysis software for simulation purposes a range of
experimental tests were performed with PDs on cables to test the validity of the model.
The simulation results are compared with the test measurement results and it was found
that the model developed and used was able to predict the measurement results
accurately. It was found that the semi-conducting screen of the cable insulation has
significant impact on the PD propagation characteristics and thus on the measured PDs.
1.1 Research Goals
The overall objective is to develop an accurate cable model for detecting and locating
degraded insulation regions on power cables, caused by partial discharge activities. To
achieve that goal, there are many areas that need to be covered. This project focused on
determining the high frequency characteristics of the partial discharge signal that is
emitted from a cavity in the insulation structure. The specific primary goals of this study
were:
• To develop high frequency single and three phase cable models and verify them with
test measurements on cable samples in the laboratory. The cable models utilize the
characteristics of materials used for the semi-conducting and insulating materials,
including XLPE and paper insulation types.
________________________________________________________________________
• To determine the high frequency characteristics of semi-conducting materials and

investigate the influence of such layers on the overall PD signal attenuation and velocity
in the cable.
• To determine the field distribution in the vicinity of a cavity with varied physical
conditions
• To develop a laboratory’s partial discharge monitoring systems and measurement

techniques for the simulation of partial discharge signal. The PD patterns are detected by
different sensor bandwidths.
• To develop a method to model cross-coupling of PD signals between phases in three

core cables and to locate the PD source in a three phase cable in order to compare with
the measurements.
1.2 Achievements
The main achievements of this thesis are:
• Development of measurement techniques for high frequency signal characterization on

power cable (Section 5.3 to 5.5).
• Field Distribution in a cavity with varied physical conditions (Section 4.3).
• High frequency characterization and modelling of semi-conducting layers (Section 4.4
and 4.5).
• Development of a high frequency single phase cable model and its verification with
measurements (Section 5.6).
• Development of a HF model for single phase cable allowing reconstruction of the
original PD source signal and location of the PD source (Section 5.7 and 5.8).
________________________________________________________________________
• Development of a method to detect cross-coupling of PD signals between phases of a

three core cable and location of the PD source on the three phase cable (Section 6.4, 7.2
to 7.5).
• Development of a high frequency three phase cable model and its verification with
laboratory measurements (Section 8.2 to 8.5).
1.3 Synopsis of this thesis
Chapter 1 - Introduction: This Chapter gives the background of the work, research goals
and summary of this thesis.
Chapter 2 - Cable and Partial Discharge: This chapter provides a brief historical
development of the power cable and the structure of the modern cable and reviews partial
discharge fundamentals, including their definition and classifications, for both low
frequency (Standard) and high frequency (Non-Standard) PDs following by a review of
the characteristics of partial discharge signal propagation in the lossy insulation of cables.
Chapter 3 - Partial Discharge Signals and Cable Modeling: This part of the thesis
presents the general fundamental theory of cable modeling methods. The discussions are
followed by an overview of the recent research on high frequency cable modelling and its
verification procedure.
Chapter 4 - Field Analysis: This chapter explores the application of finite element field
analysis methods to determine the effect of size, position, and permittivity on air cavities
in cable insulation that may be the source of PD activity. Field simulated results of the
effect of semi-conducting and insulating material of the typical power cable are also
presented in this chapter.
________________________________________________________________________
Chapter 5 - High Frequency Pulse Measurements and Analysis: This part of the thesis
addresses the laboratory experimental procedure and results. It gives a detailed outline of
test configurations used including the PD sensors and the semi-conducting layers. This
chapter also presents the experimental measurements on single core cables using time and
frequency domain methods. The measured results are compared with the simulated
results to determine the feasibility of the model developed. A comparison is given based
on PD signal tests using different sensor bandwidths.
Chapter 6 - Partial Discharge Detection: An investigation into the possibility of

detecting different PD patterns and signals when conducting PD tests using different
sensor bandwidths is presented in this chapter. The occurrence of discharge activity is
created by an artificial defect manufactured in the single core cable insulation.
Chapter 7 - PD Characteristics and Defect Patterns in Three Phase Power Cables: This
chapter presents PD patterns of three phase power cable by using 3-Phase Amplitude
Relation Diagram (3PARD) visualization. The 3PARD techniques are used to allow users
to locate the specific phase core in which the PD was generated.
Chapter 8 - High Frequency PD Signal Propagation in Three Phase Power Cables: An

investigation of partial discharge propagation on three phase power cables is presented
here. Detail of how the high frequency signal on one phase of the cable is induced on the
other two phases is explained at the beginning of the chapter. The measured results and
the comparison with simulation results from the new cable model are also presented in
this chapter.
Chapter 9 - Conclusion and Future Study: This chapter provides the conclusions of this
study and outlines the possibilities for future study.
Chapter 2: Cable and Partial Discharge 6
________________________________________________________________________
Chapter 2
Power Cables and Partial Discharge
2.0 Underground Transmission
Electricity is an essential requirement for modern society. There are basically two ways
by which electricity is transported from generating plants to load areas. These are
overhead transmission and distribution lines and underground power cables. Electricity
utilities in Australia have traditionally used bare-conductor overhead construction for
primary and secondary distribution circuits [4]. Cables have been used primarily for high
density urban areas and for other locations where overhead lines are inappropriate.
2.0.1 Cable versus Overhead transmission line
There are two factors that show why the overhead line is the preferred option. The first is
the cost of installation: a cable is typically 6 to 20 times more expensive than an
equivalent overhead transmission line [5]. The high cost of cable is due in part to the
cable cost itself and this therefore encourages efforts to reduce the manufacturing cost but
installation cost is taking 40% of the total cost of a cable and it requires considerable skill
and is therefore expensive. The second factor is the thermal capacity. When a cable is
buried in the ground, it is to all intents and purposes placed in a thermal insulation
blanket, which results in its transmission capability being thermally limited. However,
this is not the case for overhead lines which are much more easily cooled by winds and
thermal convection in the ambient air. Figure 2.1 shows the schematic design of
traditional overhead line.
________________________________________________________________________
Ground
Resistivity
Figure 2.1: Schematic design of overhead line.
The conditions just described are changing rapidly with modern technology and
environmental concerns. This is a consequence of number of factors, the first being the
awareness of the general public to the poor aesthetics of overhead transmission lines. The
second reason is the amount of real estate that high voltage overhead lines require. The
intrinsic value of this land is no longer inconsequential. The third reason is the influence
of electric and magnetic fields (EMF), which are generated by overhead lines,
particularly the magnetic field. Overhead lines can generate high magnetic fields, which
are a matter of some concern to most of the population. Most the Australian and New
Zealand transmission and distribution companies such as Transgrid, EnergyAustralia and
Transpower have already considered these issues and published a number of public
papers. The companies are using the exposure guidelines provided by the International
Commission for Non-Ionizing Radiation Protection (ICNIRP). The ICNIRP guidelines
set fundamental limits on electrical currents induced in the body by EMF. The limits on
induced body current are termed ‘basic restrictions’ and are measured in current per cross
sectional area of body tissue in units of milliamps per meter squared (mA/m2).
Since induced currents in the body are difficult to measure or calculate, ICNIRP also
provides reference levels as shown in Figure 2.2 [6]. Reference levels are given in terms
________________________________________________________________________
of the more easily measured ambient electric and magnetic fields that give rise to the
induced body currents.
Provided field strengths are below the reference levels, resulting induced currents will be
within the basic restriction. If exposures exceed the reference level, this does not
necessarily mean that the basic restriction is being exceeded, simply that a more careful
evaluation is required.
Figure 2.2: The ICNIRP guidelines set fundamental limits on electrical currents induced
in the body by EMF [6].
Underground cable systems, with the closer spacing of phases allowed by the cable
insulation do not generate as high a level of EMF as overhead lines for the same voltage
and current levels. The closer spacing of phases allows greater cancellation of fields due
to the temporal phase differences in the individual magnetic fields generated by the
current in each phase.
At the present time, we are seeing an increasing effort to overcome the more important
drawbacks of cable systems [7]. This effort is concentrated in five basic areas:
1. Reducing cost
2. Simplifying installation and jointing
________________________________________________________________________
3. Improving reliability
4. Increasing the efficiency of power transmission
5. Raising the transmission capacity of underground cable systems.
2.1 Power cables
Power cables are an ideal solution for electric power transmission and distribution. They
have advantages in aesthetic, environmental and safety aspects compared with the
overhead transmission lines. Therefore most distribution networks of medium and low
voltages are constructed with power cables. However, while cables are considered more
reliable than overhead lines, a failure of a cable is more difficult to detect and then to
repair than is the case for an overhead line. For this reason the development of techniques
for locating degraded insulation in cables is of considerable importance.
Cables have been in use for over one hundred years but in the last thirty years the number
of cables has increased substantially and more particularly there have been in that time a
dramatic change in the type of insulation materials used in cables. This has some
implications in terms of the development of diagnostic methods for cables.
The power cable is essentially an insulated conductor. Power cables are classified
according to their type of insulation such as paper, rubber, plastic or gas. Cables for
power transmission and distribution have used many different types of insulation,
conductors and sheathing materials. Over the years many different materials were used
as insulation: natural and synthetic rubber, vulcanized rubber, oil and wax, cotton, PVC,
polyethylene, impregnated paper, nitrogen, SF6, XLPE and others. However impregnated
paper was by far the most common until about 30 years ago when XLPE has taken over
as the dominant insulation material. In the current transmission and distribution system
both paper (PILC) and XLPE cables are the most common types in current use.
________________________________________________________________________
2.1.1 PILC cables
One of the most successful designs was paper insulated lead covered (PILC) cable. Use
of paper insulated power cables can be traced back to 1891 in London [1]. During the
years the paper impregnation was improved by changing vegetable substances, mineral
oil, and wax-filled compounds. The sheath protecting the cable from moisture ingress
progressed from lead to aluminum [8]. PILC cables are traditionally used for heavy
power works. The paper insulation is impregnated with oil or non-draining compound to
improve its long-term performance. Cables of this kind need special jointing methods to
ensure that the insulation remains sealed. Because of these problems, the weight of cable
usually is heavy and it has led to the widespread use of PVC (polyvinyl chloride) and
XLPE (cross-linked polyethylene) insulating cables replacing PILC cables. PVC can only
be used at low voltages because of its high dielectric loss and so XLPE is the preferred
option for modern high voltage cables.
Figure 2.3 and Figure 2.4 present the structure of single core PILC cable and three core
belted type cable. Single core cables are normally used on installations where armouring
is not required, e.g. power station cabling or where the conductor section is such that
three conductor cable would be inconveniently heavy. The un-screened cores of three
core belted type cables are laid up with only a single outer lead sheath, taped and
armoured and this gives the smallest overall diameter for conductor cross-section but is
not always so convenient to terminate and joint [1].
Figure 2.3: Structure of single core PILC cable with round conductor.
________________________________________________________________________
Figure 2.4: Structure of three core belted type cable with oval conductors.
2.1.2 XLPE cables
The growth of solid dielectric insulated medium voltage cables began in the early 1950s.
Introduction of cross-linked polyethylene (XLPE) as an insulation material in the mid-
1960s seemed to be very promising due to good electrical, thermal and mechanical
properties. XLPE has low permittivity, high dielectric strength and negligible dielectric
loss. Maximum continuous operating temperature of XLPE is 90°C, while during
emergency overload and short-circuit voltages the allowable temperature can reach
130°C and 250°C respectively without causing any insulation damage [9]. Good
mechanical properties eliminated the tendency to stress cracking. Therefore, introduction
of XLPE increased the capability of polymeric insulated cables because of their higher
temperature ratings, with the result that PILC cables have effectively been replaced by
XLPE.
The typical single core XLPE cable has a stranded aluminium core and the cable is
concentric with semi-conducting layers under and over the XLPE insulation. The semi-
conducting layers are designed to smooth out any local fluctuations of electrical field that
may be caused by surface irregularity of the insulation. Such surface irregularities may
enhance the electric field and cause partial discharge activity. The metallic shielding
comprises helical copper strands and the overall outer jacket is extruded polyvinyl
chloride (PVC). The structure of XLPE cable is shown in Figure 2.5. The semi-
conducting layers are polyethylene or ethylene copolymer mixed with conductive carbon
________________________________________________________________________
black [5]. The semi-con layers are often neglected in cable modelling but it has been
shown that they are very important in determining high frequency PD propagation
characteristics [3].
Figure 2.5: Structure of the 11kV single core XLPE cable.
2.1.3 Classification of cables
Cables can be classified based on several parameters, such as:

1. Voltage rating of the cable, e.g. Low voltage (rated up to 1,000 volts), medium
voltage (rated 1,001 through 36,000 volts), high voltage (rated 69,000 volts to
500,000 volts) and extra high voltage (rated over 500,000 volts).
2. Insulation of the cable, e.g. XLPE, PVC, EPR or oil-paper cable
3. The metallic Shield over the cable insulation, e.g. Shielded or non-shielded cable
4. Oil pressure level of oil filled cables, e.g. Self-contained or pipe-type cable
5. Number of cores of the cable, e.g. Single core, two core or three core cable
________________________________________________________________________
2.2 Cable Structural Materials

Cables are constructed using a variety of materials for conductor, insulation, screening
and armoring.
2.2.1 Conductors
The most common conductor materials are annealed copper and hard-drawn aluminum of
high purity (>99.5% pure). The resistivity of materials significantly increases with
impurity content. Their selection was based on the following considerations:
x Ready workability and ease of handling;
x High electrical conductivity = 1 / resistivity (U);
x Chemical stability;
x Mechanical strength and flexibility;
x Economic issue such as availability, initial and operating cost.
Table 2.1 compares resistivity values for few conductor materials used in cables. For a
given current rating, aluminium requires a larger conductor cross-sectional area than
copper. Considerations of availability and cost have led to increasing use of aluminium.
Hard-drawn aluminium is preferable to annealed aluminium because of its higher tensile
strength. Annealed aluminium has approximately 40% of the tensile strength of annealed
copper. Sodium is characterized by high electrical conductivity, low density and low cost,
which all favourable for use in cable conductors. Unfortunately, sodium is chemically
unstable and it reacts violently with water: it is no longer considered a useful cable
conductor material. More detail can be found in Table 2.2.
Lead and steel are included in the table because they are used as sheath and/or armour
materials in cables and because they are subject to the AC magnetic field of the cable
they can carry induced current and be a source of heat.
________________________________________________________________________
Table 2.1: Resistivity of conductor materials at 20oC [10].
Material U (P:-cm)
Copper 1.73
Aluminium 2.83
Sodium 4.68
Steel 10.20
Lead 21.50
Table 2.2: Physical parameters of cable conductor materials.
Annealed Hard-drawn Sodium Wire

Copper Wire Aluminium Wire
Chemical Symbol Cu Al Na
Atomic Weight 63.57 26.97 22.79
Density (g/cm3) 8.89 2.70 0.972
Melting Point (°C) 1083 652-657 97.8
Thermal Conductivity 0.934 0.55 0.32
(cal/cm.sec°C)
Tensile Strength (kg/ 24 16
mm2)
Specific resistivity (μ- 1.71 2.83 4.2
cm)
Specific Permittivity 0.9999905 1.0000208
Conductors offer greater resistance to the flow of alternating current than direct current.
The magnitude of the increase can be called “AC/DC ratio”. The reason of this increase
may due to several reasons such as skin effect, the tendency for alternating current to
crowd toward the surface of conductor. Other factors are proximity effect, which is the
________________________________________________________________________
distortion of current distribution due to the magnetic effects of other nearby currents such
as other phases, hysteresis and eddy current losses in nearby ferromagnetic materials and
induced losses in nearby non-ferromagnetic materials.
Conductors are generally stranded in order to give enough flexibility to the comparatively
rigid copper and aluminium conductors. For aluminium conductor, solid type conductor
has also been used due to reducing cost and outer diameter of cable. Most of the single
conductor cables are in concentric strand type and they can also be compact round,
annular-strand, segmental or hollow core as shown in Figure 2.6.
Compact conductors have been widely used. The main advantage of such conductors is
reduced overall diameter of cable; higher electrical breakdown due to elimination of
space between conductor and insulation; low AC resistance due to minimizing of
proximity effect. The segmental type is preferable for large-size conductor such as 600-
800mm2 or larger, for which the skin effect is not negligible. The hollow core type
conductor is less influenced by skin effect than stranded, round conductor.
Figure 2.6: conductor shape [9].

________________________________________________________________________
2.2.2 Conductor Resistance
The DC resistance of a conductor depends upon its resistivity, its length and its cross-
sectional area as well as its temperature. The resistance, in term of temperature is given
as:
R2 To T2
(Equation 2.1)
R1 To T1
Where R1 = Conductor resistance at temperature T1 in ºC, R2 = Conductor resistance at
temperature T2 in ºC and To = constant varying with conductor material and it is equal to
234.5, 241 and 228 for annealed copper, hard-drawn copper and hard-drawn aluminum,
respectively.
The AC resistance is higher than DC resistance due to skin effect. The difference
between the two values depends upon the frequency and the conductor cross-sectional
area. Typically it is less than about 2% at 50 Hz in typical sized cables.
2.2.3 Cable insulation

Different insulating materials are used in cable manufacture. These include: Paper, Cross-
linked polyethylene (XLPE), polyvinyl chloride (low voltage only), gases, natural and
synthetic rubber. Factors influencing choices are rated voltage and current and the type of
cable required for the particular installation.
General performance characteristics of cable insulation can be summarized as follows:

x It should be flexible and anti-abrasive;
x It should withstand high impulse voltage and should remain stable for a
reasonably long period of time;
x Its dielectric power loss should be as small as possible;
x It should be excellent in its treeing resistance and its corona (partial discharge)
resistance.
________________________________________________________________________
XLPE
Cross-linked polyethylene (XLPE) is a thermosetting compound, which has better

electrical properties than P.V.C. and is therefore used for medium and high voltage
cables. It is replacing PILC in most modern applications. XLPE has the lowest dielectric
loss of all cable insulation. It has a high operating temperature of 90 °C.
P.V.C.
Polyvinyl chloride (P.V.C.) is the most used low voltage cable insulation. It is clean to
handle and is reasonably resistant to oil and other chemicals. When P.V.C. burns, it emits
dense smoke and corrosive hydrogen chloride gas. The physical characteristics of the
material change with temperature. When the temperature is low, P.V.C. becomes hard
and difficult to strip. Therefore, it should not be worked at temperature below 5 °C.
However, some special P.V.C. is still available which remains flexible at temperature
below 20 °C. Most of general P.V.C. also must not exceed 70°C and there is some special
versions which will operate safely at temperature up to 85 °C.
Paper
Dry paper is hygroscopic and can absorb moisture from the air. Dry paper is an excellent
insulator but loses its insulating properties if it becomes wet. Therefore it must be sealed
to ensure that there is no contact with the air. Because of this, paper insulated cables are
usually sheathed with impervious materials such as lead.
Rubber
For many years, wiring cables were insulated with vulcanized natural rubber (VIR).
Many cables of this type are still in service although it is many years ago. Insulation is
________________________________________________________________________
organic material, which is subject to the normal ageing process, becoming hard and
brittle. In this condition, it will continue to give satisfactory service unless it is disturbed,
when the rubber cracks and loses its insulating properties. Synthetic rubber compounds
are used widely for insulation and sheathing of cables for flexible and heavy applications.
Mineral
Mineral insulation such as magnesium oxide is an excellent insulation when it is kept dry.
Since it is hygroscopic, this insulation needs to be kept sealed within a copper sheath.
Mineral insulation is totally fireproof and will operate at temperature of up to 250°C. It is
entirely inorganic and non-ageing.
Gas insulation
Gases for cable insulation have high breakdown strength, chemically stable and flame
retardant. Some examples of gas insulation are nitrogen (N2), sulphur hexafluoride (SF6)
and Freon-12 (CCI2F2). Nitrogen gas is usually selected for gas-filled cables. SF6 is used
for compressed gas insulated cable because of its low permittivity, virtually zero
dielectric dissipation factor and high breakdown strength. Cable capacitance is very
dependent on the dielectric constant of cable insulation. From this point of view, gas
insulated cables are an attractive alternative. Unfortunately, they are more expensive.
2.2.4 Electrical properties of insulating materials
Electrical properties of insulation materials can be classified into two categories: low
voltage operating stress and high voltage operating stress.
At low stress, the properties of interest relate to dielectric constant, dielectric dissipation
or power factor and conductivity. At high stress, the characteristic of importance is
dielectric strength and the insulation must be resistant to partial discharge.
________________________________________________________________________
Dielectric constant (permittivity)
Dielectric constant or relative permittivity of the insulation is dependent upon the

temperature and frequency of testing. At constant temperature, it is reduced slightly as
the frequency increases. At constant frequency, it increases slightly with temperature.
The dielectric constants of various common cable insulation materials are shown in Table
2.3.
Table 2.3: Dielectric constants of various materials used as electrical insulation.
Material Type Dielectric Constant

Polyethylene, low density 2.25-2.35
Polyethylene, high density 2.30-2.35
Polypropylene 2.20-2.60
Ethylene-Propylene copolymer 2.25-2.35
Epoxy 3.50-5.00
Nylon 3.90
P.V.C. 5.00-9.00
Dissipation Factor
The dissipation factor represents the energy lost as heat due to polarization effects in AC
electric field, and is a function of the efficiency of the insulation. It is commonly referred
to as TanG. The power factor is not identical to dissipation factor, but the values are close
________________________________________________________________________
to identical when TanG is smaller than 0.1. Loss factor is the product of dielectric
constant and the dissipation factor as shown in Equation 2.2.
Dissipation factor (D) = TanG = Loss factor / Dielectric constant (Equation 2.2)
Loss factor in Equation 2.2 is the relative loss factor, which is a measure of the loss of
energy in a dielectric material through polarization currents and other dissipative
phenomena when subject to an AC electric field.
Dielectric Strength
The dielectric strength can be defined as the limiting voltage stress beyond which the
dielectric can no longer maintain its integrity and it can also be defined as the voltage
gradient at which electrical failure occurs. However, the dielectric strength is not an
absolute number, the value of dielectric strength measured may depend upon many
factors. It is commonly expressed in stress per unit thickness, kV/mm.
Most likely failure may occur due to partial discharge activity and damage in small voids.
The gas (e.g. air) present in such small voids in the insulation has dielectric strength
significantly less than the surrounding polymeric insulation. Under these conditions, the
discharges that take place in small voids lead to erosion of the insulation surfaces that are
in contact with air in the void. This leads to gradual decomposition of the insulation and
eventual failure.
Electrical Trees
Electrical treeing is an electrical pre-breakdown phenomenon as shown in Figure 2.7.
Electrical trees are the fine erosion channels that propagate through the material from an
initiating point. These electrical trees will extend from defects or area of partial
breakdown in the insulation. When discharges occur in some practical systems it may not
lead to immediate failure, a blackened needle-shaped series of defects results, sometimes
resembling a tree limb and this is called an electrical tree. The tree contains carbonised
________________________________________________________________________
material in the channels. Discharge may occur repetitively and can detect by partial
discharges sensing (as opposed to water trees). When the trees become more visible, the
rapid growth of the discharges may lead to breakdown in an extremely very short period
[24]. The tree will bridge the entire insulation wall and cause failure.
Figure 2.7: Electrical Treeing and paper degradation below a dry cable termination [18].
Water Trees
Water trees are commonly encountered in XLPE insulation and can be initiated at sites
where there are small concentrations of moisture. The water trees grow under relatively
low operating electric stress and they do not require the presence of voids. That grows
along the electric field lines and may start from a site in the main bulk of insulation or at
the insulation surface. They have a long development phase and do not cause any major
problems with insulation for a long time after initiations (years). Eventually they turn into
electrical trees and then lead to a rapid reduction in dielectric strength. Water trees do not
generate partial discharges until they change into electrical trees. They are extremely
difficult to detect in cables.
Water is one of the necessary agents for water treeing. Therefore different cable designs
were introduced to protect against water ingress and propagation in the cable. Water
________________________________________________________________________
trees have a different shape compared to electrical trees as shown in Figure 2.8, and
different causes. The differences are summarized in Table 2.4.
Figure 2.8: Water trees in power cable insulation [12]. Note the growth along the E field
lines.
Table 2.4: Different characteristics of Water trees and Electrical trees.
Water Trees Electrical Trees

Water required Water not required
Fan or bush shaped Needle or spindle shaped
Grow for years Failure shortly after formation
Micro-voids connected by tracks Carbonized regions
Detecting methods: Detecting methods:
Pulsed electro-acoustic method [142] Identification from PRPD pattern [145]
Return voltage measurement [143] Numerical modelling [146]
Neutron activation analysis [144] Optical method [147]
Micro- Proton-induced X-ray emission
[144]
________________________________________________________________________
Thermal Breakdown
Another mechanism of insulation failure is known as thermal breakdown. This occurs

when the insulation temperature starts to increase as a result of aging under operating
stress, perhaps because of overloading or increased dielectric losses. Under voltage stress,
some insulation systems will start to generate heat due to these high dielectric (DDF)
losses. If the rate of heating exceeds the rate of cooling, then thermal runaway occurs and
the insulation will eventually fail.
2.2.5 Semi-conducting Screen Materials
Semi-conducting screens are provided on conductors and over insulation to increase the
stability of cable insulation. The semi-conducting layers are made of polyethylene or
ethylene copolymer mixed with conductive carbon black [11]. The semi-conducting
layers are used to smooth out the electric stress enhancements at the conductor strands
and thus prevent partial discharge activity at the interface between the conductor and the
XLPE insulation. Such semi-conducting material may cause some challenges in cable
modeling. High frequency losses are caused by the propagation of radial displacement
current through the resistance of these layers [43, 44]. The losses caused by the semi-
conducting layers depend on the dielectric constant and resistivity of the semi-conducting
layers, and these properties tend to be frequency dependent. The properties of the semi-
conducting materials employed in solid dielectric cable are restricted by standards to
assure their effectiveness as a conductor, i.e. only a minimum conductivity is specified.
Further discussion can be found in Chapter 5 and Chapter 6.
2.2.6 Carbon–polyethylene compound

The required conductivity of the material to provide a uniform field distribution in the
cable cross section is achieved by addition of black carbon to polyethylene. Figure 2.9
shows the surface of a carbon–polyethylene [3] compound taken with a microscope. In
________________________________________________________________________
the polyethylene, the conducting carbon particles of different forms and sizes can be
recognized. The carbon content of this mixture is about 38%. It is a furnace carbon of the
so-called p-type.
Figure 2.9: Surface of carbon- polyethylene compound [3].
This carbon forms small spherical particles, which agglomerate to chains. These chains
again form agglomerates. The conductive paths are formed by contact between the
individual particles. Figure 2.10a and Figure 2.10b give the permittivity and conductivity
of semi-conducting carbon-polyethylene compound in a frequency range up to 30 MHz.
The result shows that the complex conductivity has a constant value in the frequency
range up to 30 MHz.
(a) (b)
Figure 2.10 (a): Real (1) part and imaginary (2) part of the complex conductivity.
Figure 2.10 (b): permittivity of semi-conducting carbon-polyethylene compound in a
frequency range up to 30 MHz [3].
________________________________________________________________________
2.2.7 Conductive paper layer
The conductive paper layer was available as a rolled-up strip with a width of 60 mm and
a thickness of 0.2 mm. The real part of permittivity and the conductivity in a frequency
range of 30 MHz give the constant values as shown in Figure 2.11a and Figure 2.11b.
Real and imaginary parts of permittivity are usually constant above 10 MHz.
(a) (b)
Figure 2.11 (a): Real (1) part and imaginary (2) part of the complex conductivity:
Figure 2.11 (b): Real (r1) part and imaginary (r2) part of permittivity of conductive
semi-conducting paper layer in a frequency range up to 30 MHz [3].
________________________________________________________________________
2.2.8 Screens and Jackets

Shielded cables must have a metallic material over and in contact with the non-metallic
layer. The following options are available for the metallic material:
x Helically wrapped flat metal tape (usually copper)
x Longitudinally corrugated metal tape (usually copper)
x Wire shield (usually copper)
x Concentric neutral wires (usually copper)
x Flat straps (usually use copper wires with an overall jacket applied over the wires
for corrosion protection).
x Tape plus wires (usually apply on higher voltage cables such as 35kV to 138kV
where fault currents often may be greater than the capabilities of wires)
x Continuous welded corrugated metal sheath (copper, aluminium etc.)
Jackets are generally required over metallic shields for mechanical and corrosion
protection against effects such as physical abuse, sunlight, flame or chemical attack.
There are many jacketing materials such as:
x Polyethylene
x Semi-conducting polyethylene
x Polyvinyl chloride (PVC)
x Chlorosulphanated polyethylene
x Chlorinated polyethylene
2.3 Power cable diagnostics
Power cable diagnostics can be divided into two categories: offline and online
diagnostics.
________________________________________________________________________
Offline diagnostics are used when the operations are disconnected from the power grid
and online diagnostics are performed during cable operation.
2.3.1 Off-line diagnostics
Off-line diagnostics are performed on the cables disconnected from the power grid. A
wider range of tests than for online diagnostics can be used but they are not always the
most useful tests. Typical offline diagnostics include the following testing:
Loss factor The measurements can be performed using a classical Schering bridge
measurement of loss factor at a power frequency [13]. Modern methods use a transformer
ratio arm bridge.
Dielectric polarization measurements In dielectric polarization measurements tests are

performed at several frequencies enabling a frequency spectrum of permittivity to be
analyzed. The spectrum reflects the properties of the dielectric material in the measured
frequency range. Degrading of insulation can increase the loss and the capacitance of the
dielectric material sample. These two parameters are also voltage dependent.
Polarization / depolarization current measurements are performed by charging the

sample by DC voltage and measuring polarization current. After applying DC voltage for
a long period of time the sample is short-circuited and depolarization current is measured
[13]. This method has some potential for water tree detection.
Return voltage measurements are similar to depolarization current measurements. The

DC voltage charges the sample; after a relatively short period of time during which the
sample is short-circuited, the test object is left in open-circuit condition and the recovery
voltage is measured [13]. Not commonly used for cables, where the Isothermal relaxation
variation is more often used for water tree assessment.
________________________________________________________________________
Partial discharge diagnostics Partial Discharge (PD) diagnostics is a widely used

technique to detect discharges appearing in cavities or on surfaces of the insulation [13].
Off-line PD diagnostics on the power cables are usually performed by energizing the
cable with the High Voltage (HV) supply. The measuring equipment may be coupled to
the cable using a coupling capacitor [14] or the PD pulses may be detected with HFCTs
or externally coupled capacitors. The method enables the PDs to be detected and
localized in cables.
Time Domain Reflectometry Most techniques for PD analysis utilize the time domain
reflectometry (TDR) method for location of PD sites. The PD pulse in the cable travels
along the cable to both ends of the cable and when measurements of time delay of the
reflections along the cable are made the location can be determined. The location of the
PD is estimated by measuring the intervals between the pulses [15].
2.3.2 On-line diagnostics
On-line diagnostics are performed on the cables during operation and the test instruments
must comply with at three demands: 1) resisting-disturbance of electromagnetic wave
from the environment; 2) not affecting operation of power cable; 3) high voltage divider
ratio if directly coupled.
DC current measurement The method was possible to implement in Japan where the
power cables operate mostly at relatively low voltages such as 6.6kV and are non-
grounded. DC voltage is applied to the cable conductor through an inductance and is
superimposed on the grid voltage. A filter eliminates the AC component of the current
that passes thought the insulation of the cable and only the DC component is measured.
The reduction of the insulation resistance indicates the presence of water trees [16].
Partial discharge diagnostics On-line partial discharges on the cables are detected using
high frequency sensors [17]. The sensors can be divided into two basic types: capacitive
or inductive type. The capacitive sensors are usually made of conductive tape placed on
________________________________________________________________________
the insulation screen between the HV termination and the screen wires. Another option is
to place the capacitive sensor on the insulation screen in the cable joint, under the
metallic screen. The inductive sensors usually used for on-line PD diagnostics are high
frequency current transformers and Rogowski coils. They can be placed on the power
cable after the earth connection, before the high voltage termination, or on the power
cable's earth connection conductor.
2.4 Partial Discharge
A significant trend in the development of electrical power apparatus is the increase of the
power and the size of the units. This requires severe demands on increased reliability.
Therefore, high voltage insulation technology requires modern testing procedures. In this
respect increasing attention is being paid to the development of predictive diagnostic
tools. Against this background, there is no doubt that the recognition of partial discharge
(PD) is of great importance, because PD phenomena can be regarded as the forerunner
for ageing phenomena in electrical insulation.
2.4.1 Partial Discharge definition and terminology
According of IEC-60270 Standard, Partial Discharge (PD) is defined as localized

electrical discharge that only partially bridges in the insulation between conductors and
which can or cannot occur adjacent to a conductor [20]. Usually each partial discharge
appears as a fast electrical pulse with duration of a few to hundreds of nanoseconds. In
the IEC-60270 standard, the partial discharge pulse (PD pulse) is defined as a current or
voltage pulse that results from a partial discharge occurring within the object under test.
The pulse is measured using suitable detector circuits, which have been introduced into
the test circuit for the purpose of the test [20].
The PD pulse usually has relatively small magnitude and energy and appears as a
consequence of the enhancement of electrical field due to local stress concentrations or
low dielectric strength media. However, the low energy associated with the PD pulses
________________________________________________________________________
can over time deteriorate the dielectric material, which can then lead to a complete
breakdown of the insulation. In addition, the presence of PD produces localized heat and
generates a complex chemical reaction that may accelerate the material ageing.
Ultraviolet radiation from the PD may also cause unavoidable defects in the insulation.
Electrical trees filamentary carbonised tracks in solids or on surfaces may also be
developed as a result of high localized-stress within a defect site. Therefore, continuous
presence of the PDs will significantly increase the deterioration rate of the insulation
materials, and thus reduce their expected lifetime.
The PDs occur when the applied electric field stress exceeds a critical value. This
situation causes some localised discharge to occur. The corresponding voltage is the PD
inception voltage. The PD inception voltage is defined as: the applied voltage at which
the repetitive PDs are first observed in the test object, when the voltage applied to the
object is gradually increased from a lower value at which no PDs are observed [20]. The
associated field stress that occurs in the test object at inception voltage is designated as
the PD inception stress [21].
In a similar way, the PD extinction voltage is defined as: the applied voltage at which
repetitive PDs cease to occur in the test object, when the voltage applied to the object is
gradually decreased from a higher value at which PD pulse quantities are observed [20].
In practise, the PD inception and extinction voltages are greatly affected by the nature of
the test object. The environment factors such as temperature, humidity and pollution may
affect the PD inception and extinction voltages, particularly surface discharges on
outdoor insulators. Accordingly, the field stress that occurs in the test object at extinction
voltage is designated as the extinction stress [21].
________________________________________________________________________
2.5 Classification of Partial Discharge
The discharge may result from ionization (a process by which an atom becomes
electrically charged due to losing or gaining one or more of its extra nuclear electrons)
causing an electron avalanche (a cumulative ionisation in gases in liquid or solid
insulation caused by the collision of field-accelerated electrons and neutral atoms or
molecules). The movement of electrical charges through an insulating medium initiated
by electron avalanches and maintained by various secondary processes generate further
avalanches causing electrical discharge [22]. This is the so-called “Townsend Ionization”
breakdown mechanism. Figure 2.12 shows the correlation between the various terms.
Figure 2.12: The relationship between ionization with different types of discharges [19].
________________________________________________________________________
Partial Discharges can occur in voids in solid insulation (paper, polymer etc), gas bubbles
in liquid insulation or around an electrode in a gas (corona) [18]. Generally, there are
three types of partial discharges [19]:
x Internal discharges (Figure 2.13): Discharges in a cavity in a solid dielectric
(both electrodes are shielded from the discharges by solid).
x Surface discharges (Figure 2.14): Discharge on a surface (at least one
electrode is shielded by a solid dielectric).
x Corona discharges (Figure 2.15): Discharges around a sharp point at high
voltage (the discharge is shielded from one electrode by a column of non-ionized
gas).
Each of these types of discharges is briefly explained below.
Figure 2.13: Internal discharges [19].
Figure 2.14: Surface discharges [19].

________________________________________________________________________
Figure 2.15: Corona discharges [19].
It is not always possible to classify a discharge as simply as one of the above processes.
For example, the discharges shown in Figure 2.16-2.17 are combinations. In Figure 2.16,
an intermediate state between internal discharge and surface discharge is shown. If the
inter-space between smaller and eventually is closed at the sides, the discharge becomes
internal. In Figure 2.17, a combination of surface discharge and corona discharge is
shown.
Figure 2.16: A combination of internal discharge and surface discharge [19].
Figure 2.17: A combination of corona discharge and surface discharge [19].

________________________________________________________________________
2.5.1 Internal Discharge
Internal discharges normally occur in cavities and little cracks in a solid dielectric as
shown in Figure 2.18. The discharge generates high frequency pulses, which attenuate
quickly as they pass to ground because of HF losses. The discharges are effectively small
arcs occurring within the insulation system, conductor and insulation wall, therefore
deteriorating the insulation quality. This can result in eventual complete insulation
failure [23]. In addition, the dielectric constant of the gas in the cavities in insulation is
lower than that of the surrounding solid dielectric, so that the electric stress within the
cavity is higher than in the solid dielectric and thus breaks down even earlier because the
gas has a lower dielectric strength.
The cavity defects may be developed accidentally during fabrication because of

impurities of materials, imperfect dielectric mixture or insufficient vacuum impregnation.
Environment effects during service of the equipment such as temperature and vibration
may also contribute to the formation of the cavities.
PDs also occur in regions of high electric stress concentration in non-uniform electric
fields, particularly gaseous and liquid dielectrics. Within these areas of local weakness
PDs may occur at voltage levels below the operating voltage imposed upon the
insulation. The rate of decay of the insulation is a complex function of many variables
including: the level of electric stress acting upon the insulation; the type and rates of
breakdown mechanisms; i.e., heat, chemical decay etc; the internal and external
environment; the repetition rate of the discharge; the location of the PD and the precise
inception voltage.
________________________________________________________________________
Figure 2.18: Internal Discharge in a Solid Dielectric Bounded Cavity [25].
2.5.2 Corona Discharge
Corona discharge, which is simply partial discharge activity in a free gas, occurs when
there is ionization, or in the presence of strong inhomogeneous fields around the sharp
point of edges of the HV electrodes (see Figure 2.19). Corona discharges also occurs
when the potential gradient exceeds a certain value in a situation where sparking is not
likely to occur. It can also occur at sharp protrusions from a conductor at earth potential
or floating materials such as small-thin wire or loose bolts. Corona discharges generally
appear with ac voltage during the negative half cycle of the sine wave only. They are
usually found at the high voltage electrodes. They can sometimes occur at low voltage
(even earth) electrodes if the electric field is high enough at a sharp point at earth
potential [28-30].
Corona Discharge
Sharp Metallic Point
Figure 2.19: Corona Discharge.

________________________________________________________________________
Corona discharge occurs frequently in air in substations, for example, and is often form
of interference when carry out discharge tests on equipment insulation in the substation.
Hence, the test area must be free of sharp metallic objects, such as sharp edge metals,
pointed wire ends, then connection wires, and etc. This is to prevent corona discharge
when the test is in progress, not only at the high-voltage side, but also at the floor of the
test area [24].
2.5.3 Surface Discharge
Surface discharge is a discharge form that starts on the outside of the insulation and
gradually progresses over the surface of the insulation, often causing carbonization of the
surface (surface tracking). It occurs with existence of a stress component, which is
parallel to a dielectric surface. These discharges can affect the electric field and bridge
the potential gradient between the applied voltage and ground by cracks or contaminated
paths on the insulation surface [23]. This is illustrated in Figure 2.20.
Figure 2.20: Surface Discharge.

________________________________________________________________________
2.5.4 Partial Discharge Patterns
Generally, in its early traditional format, the pattern of partial discharge was usually
presented superimposed on an elliptical time-base display, which represents one complete
cycle of the test frequency applied to the test object. Elliptical representation helps to
define the type of fault causing the discharges. Figure 2.21-2.24 illustrate the typical
partial discharge patterns on elliptical display [19].
Figure 2.21: Typical discharge from a cavity in insulation.
Characteristics Origin
Fairly symmetric pattern, stationary or/and This is the usual pattern of internal
wandering impulses. discharges in voids and in impregnated
dielectrics.
Figure 2.22: Typical discharge across surface of insulation.
A few large discharges in the positive half Discharge adjacent to a conductor at earth
cycle, several smaller discharges in the potential.
negative half cycle.
Similar pattern, but with the large impulses Discharge adjacent to a conductor at high
in the negative half cycle voltage
Near the inception voltage these discharges may be intermittent.
________________________________________________________________________
Figure 2.23: Typical corona discharge in gas.
Equally spaced impulses of about equal Corona discharges around a sharp point at
height at the negative crest of the applied high voltage.
voltage.
Equally spaced impulses of about equal Corona discharges around a sharp point at
height at the positive crest of the applied low voltage.
voltage.
Figure 2.24: Irregular discharge pattern [19].
Irregular band of impulses, tending to Contact noise.
occur around the zero points.
It is essential to know the typical discharge patterns, which associate with various types
of faults. However, from a practical point of view, it is shown that two types of partial
discharge patterns are of interest by interpreting partial discharge measurement [26]:
x Regular discharge patterns, which are characteristic for a particular type of high-
voltage component with insulation in good condition;
________________________________________________________________________
x Irregular discharge patterns, which correspond to certain unacceptable discharge

sources as shown in Figure 2.24. These may be related to manufacturing defects,
ageing, or come from external factors such as environmental disturbance.
2.6 Partial Discharge Detection
Partial discharges have a significant effect on the life of insulation of high voltage (HV)
equipment. Every discharge event may deteriorate the insulation material by the energy
impact of high energy electrons or accelerated ions, causing many types of chemical
transformations. An eventual breakdown of the insulation of HV equipment while in
service may result in considerable damage to the equipment and to the system to which it
is connected. The technique of detecting and analysing PDs occurring in insulation
structures or assemblies can be used to detect weaknesses before they lead to catastrophic
failure [32]. The early detection methods such as the elliptical display shown above had
no recording facilities available, however modern computer-based techniques now allow
recording of each individual discharge event. Several detection and measurement
methods have been introduced in the market, and it has evolved into a significant tool for
use specifically in the HV electrical insulation industry.
Basically, partial discharge detection methods can be classified into two major types:
non-electrical methods, and electrical methods. The electrical and non-electrical methods
can be further divided into different subgroups, which are (a) intrusive and (b) non-
intrusive methods. An intrusive method requires the sensing element of the measurement
system to be placed inside the power equipment to detect the partial discharge pulses. A
non-intrusive method requires the sensor to be located outside of equipments.
The choice of detection method is dependent on the nature of the investigation. Another
important factor that may affect the choices of detection method is the characteristics of
the dielectric properties. The possible PD location is also a necessary consideration in
choosing the detection technique.
________________________________________________________________________
2.6.1 Non-Electrical Methods
Non-electrical discharge detection is detecting the discharge occurrence by means of

light, sound, heat that is produced as a result of the discharge activities. The non-
electrical methods cover acoustic, optical and chemical methods. Such detection methods
are able to detect the presence of discharges successfully, but are usually insensitive, and
the results are liable to get affected by environmental disturbance. Furthermore, non-
electrical detection does not permit quantitative measurements [24]. One of the reasons is
the difficulty of calibration of the equipment.
In figure 2.25 is shown one type of ultrasonic PD detector used for simple visual location
of corona activity. Other, more quantitative ultrasonic detectors, use piezoelectric sensors
attached to the tank of a transformer for example. Such piezoelectric sensors will pick up
ultrasonic waves from PD activity within the transformer. PD generates sound waves at
frequencies over the range from a few kHz to hundreds of kHz. Ultrasonic detection of
PDs in cables is not much used for the general run of the cable length because of the
damping of the solid material. However they are extensively used for detection of PDs in
the joints and terminations (accessories) of a cable, providing access is attainable to these
items.
Figure 2.25: Ultrasonic PD detector.

________________________________________________________________________
2.6.2 Electrical Methods
Electrical methods of detecting and measuring PDs are more convenient, sensitive and
simple to be applied compared to the non-electrical methods. There are a number of
different techniques that can be used to detect the presence of PDs electrically. The most
commonly used techniques are following:
x Direct coupling to the tested equipment through a HV connection [20, 34].
x By means of a high frequency current transformer or other external sensors [33].
x By application of VHF/UHF transducers [35].
2.6.3 Calibration
Before the high voltage PD testing, the measurement system is calibrated using an
external calibration source by injecting known quantities of electrical charge, in the form
of pulses into the measurement system. The purpose of calibration is to ensure that the
measured PD levels can be assessed in terms of their potential damage by comparison to
known PD damage standards. The calibration source must be removed before high
voltage can be applied and no calibration can be made during tests [31].
2.6.4 Direct Coupling to HV Connection
As the partial breakdown of the insulation causes a charge reduction (by electrical current
discharge across a void for example), this in turn produces a momentary voltage collapse
between two main terminals of the test object. A very short period and small amplitude
electrical current pulse is then produced [27]. An external circuit can be connected to
detect the high-frequency partial discharge signal, and it gives higher sensitivity. In
addition, the discharge magnitude, discharge patterns, and statistical distributions can be
measured from the detector.
________________________________________________________________________
A PD detector system that applies direct coupling to the HV connection of a tested

sample or apparatus is recognized as a conventional electrical discharge detector as
shown in Figure 2.26, which is the basic PD measurement circuits specified in the IEC-
270 Standard.
Conventional
HV PD detector
(a) Cb
LV CRT Meter
Cx Cq
Z Int. cal. Analogue

Amplifier output signal
External Input unit
calibrator
HV
(b) Cx
LV CRT Meter
Cb
Cq
Z Analogue
output signal
Input unit
HV
(c)
Cx Cb
LV CRT Meter
Z
Z Cq Analogue
output signal
Figure 2.26: Basic Partial Discharge measurement circuits specified in the IEC-270
Standard
A standard test circuit mainly consists of a coupling capacitor, measuring impedance and
a broadband amplifier that is enclosed in an observation unit. The observation unit is
equipped with a cathode ray tube (CRT) to exhibit the detected PD pulses within an
elliptical display of sinusoidal supply voltage [34]. In Figure 2.27 is shown the Robinson
Electronic Instruments Discharge Detector, which is an example of an old style elliptical
display PD detector.
________________________________________________________________________
Figure 2.27: PD detector made from Robinson Electronic Instruments Company.
Figure 2.28 illustrates the equivalent circuit diagram of the conventional electrical
discharge detection circuit. Conventional detection method is the most common
detection method to detect partial discharges since it is convenient, sensitive and easier to
calibrate.
Vy Cy CB Cy
Vx Cx Cx
Vz Cz Cz
Z
(a) (b)
Figure 2.28: Conventional Electrical Discharge Detection Equivalent Circuit.
From the circuit at Figure 2.28 (a), there are three capacitive elements, namely Cx, Cy and
Cz. Cx represents the capacitance of the body insulation, which is equal to the total
________________________________________________________________________
capacitance of the test object. Cz represents the small cavity present that generates PDs
while Cy is the insulation in series with the cavity.
By applying mesh analysis on the circuit, the voltages across the three elements can be
determined by the following equations:
Vx = Vy + Vz (Equation 2.3)
When the voltage across Vz is sufficient to cause a breakdown inside the cavity, Vz will
be short-circuited by breakdown, thus the voltage eventually will almost reach to zero,
and subsequently Vx and Vy are connected in parallel since both are not affected. In
order to correct this situation, small an amount of charge flows from Cx towards Cy to
neutralise the voltages, thereby causing small voltage dips across the test object.
Voltage dips cannot be detected directly in practical sense, thus by connecting an external
coupling capacitor of equal value to the test object, greater sensitivity is obtained (Figure
2.28 (b)). Small impedance Z is connected in parallel with Cx, and for discharges
occurring in Cz, some charge has to flow from Cx to Cy (current flow shown in blue). At
the same time, some discharges will flow in opposite direction from CB to Cy (current
flow shown in green). This flow of charge constitutes a small current to flow from
impedance Z back to CB, and it can be detected since it flows through Z.
For better accuracy in results, a large signal must be developed across Z for a given
discharge in the cavity. This can be done simply by increasing the value of CB as much
as is possible, so that most discharges can be neutralised by Cy.
2.6.5 PD detection using High Frequency Current Transformer or

other External Sensors
The high frequency current transformer (HFCT) has been broadly applied for on-line PD
detection systems in power cable, transformer and other accessories [36, 37]. Usually, the
HFCT is clamped around the low voltage connection of the equipment. The high
frequency electromagnetic pulse generated from the discharge location will propagate
________________________________________________________________________
through the earth conductor and be picked up by the HFCT. Sufficient signal
amplification may be required prior to the connection of the measuring devices such as
oscilloscopes, pulse counters or PD detectors.
In order to increase the signal to noise ratio (SNR), analogue filters are utilized as well as
various signal rejection methods. It has been shown that the technique is able to
successfully discriminate between PD pulses and disturbances [36]. Figure 2.29 (a)
shows an application of HFCT for PD detection in a power cable.
(a) (b)
Figure 2.29: Application of HFCT and clamping sensor for PD detection in power cable.
Instead of using HFCTs, another different type of sensor can be used PD detection in
power cables known as a clamping sensor, shown in Figure 2.29 (b) [38, 39]. This can be
used as either an inductive or capacitive sensor depending on how the two parts of the
clamp are connected. The signal attenuation characteristic of the inductive clamping
sensor is relatively higher than the HFCTs. Another disadvantage is that the technique
can only be applied to a cable with a helical metal screen. More details on HFCTs and
other type of sensors can be found in Chapter 4.
2.6.6 VHF/UHF PD Detection System

________________________________________________________________________
Partial discharge events can result in a radiated electromagnetic wave, in addition to the
simple galvanic voltage and current pulses. The electromagnetic wave may have a
frequency spectrum reaching up to the order of GHz. This fact leads to the development
of Very High Frequency (VHF)/ Ultra High Frequency (UHF) PD detection systems. The
sensors for VHF/UHF PD detection systems generally can be divided into two groups,
which are the intrusive VHF/UHF disk sensor [40, 41] and the non-intrusive VHF/UHF
antenna [41, 42]. The first type of sensor detects the electromagnetic signal within the
HV apparatus and the latter picks up the PD generated electromagnetic wave outside the
apparatus. Most of the VHF/UHF PD detection systems apply intrusive sensors due to
better sensitivity when compared to the non-intrusive one. However this is not possible
with cables and if such monitors are used with cables they must be of the non-intrusive
antenna type.
The detected electromagnetic signal is analysed by using a spectrum analyser (100Hz-

2.65GHz) [42]. It has been found that each of the different classes of PD sources
discussed earlier generates a unique electromagnetic spectrum. This fact may lead to
discrimination of different PD sources by analysing their high frequency spectra.
________________________________________________________________________
2.7 Partial Discharges in Cables
Power cables are the arteries of the transmission and distribution network. Ageing and
failure of insulating systems are initiated by electrical, mechanical, thermal and chemical
processes during operation. These create defects reducing locally the dielectric strength
of the insulation. Various defects, such as voids, contaminants and electrical trees can
cause PD activity in high voltage power cable [45]. Insulation is very sensitive to partial
discharges and PD activity will gradually degrade the insulation, eventually leading to
major cable breakdown. To ensure the reliability of the whole cable system, PD testing
and location are an essential part of the monitoring and testing of power cable. The
location of PD in solid dielectric cable through measurement of traveling wave
electromagnetic phenomena was first introduced in 1961 [46] and has become the
standard method for PD location in power cables. Over the years, many studies have been
aimed toward improved PD location accuracy and better sensitivity [46-49].
2.7.1 Partial Discharge Signal Characteristics
To correctly acquire partial discharge signals and interpret the measurements, the partial
discharge physics and the PD signal transmission need to be understood. When PD signal
occurs as shown in Figure 2.30, they travel from the source point on the cable in both
directions along the cable, which acts as a transmission line. Partial discharges in solid
dielectric can generate electrical signals with frequencies up to several hundred MHz,
usually of low magnitude. Depending on the local electromagnetic environment, they are
also often hidden in noise.
________________________________________________________________________
0.03
0.02
Voltage (V)
0.01
0
-0.2 -0.1 0.0 0.1 0.2
- 0.2 -0.1 0 0.1 0.2
-0.01
Time (s)
Figure 2.30: PD signal captured from the XLPE cable.
PD signal in an un-aged dielectric gives rise to an electromagnetic pulse with a risetime

in the ns range and a pulse width in the range of 1.5 ns. The optimum bandwidth for
detection is in the range of 300 MHz [50, 51]. Discharge from electrical trees in aged
solid dielectrics often takes the form of a cascade of such fast pulses. In aged cavities, the
PD signal can have a risetime up to several 10 of ns and pulse width up to some hundred
of ns [54]. For such a broad pulse, a bandwidth of detection in the range of 10 MHz is
sufficient. The properties of PD signal have been reviewed in the series of articles [50-
53].
2.7.2 Attenuation and Velocity Characteristics of Cables
There are many factors that affect the accuracy to locate the PD sources. These factors
are high frequency attenuation, which can distort the PD signal from inside and outside of
the cable, and the high interference levels often present. The location of the PD source by
a detector is limited by high frequency attenuation of the PD pulse as they propagate
through the cable. At such high frequencies, the cables are no longer adequately modeled
________________________________________________________________________
by the lumped parameter method. The cable parameters become frequency dependent.
The conductor, solid dielectric, screen and semi-conducting layers, all of these parts have
some minor or major effect on the propagation characteristics such as velocity and
attenuation of cable and thus modeling of the cable for high frequency pulse propagation
is an important task.
Work by Stone and Boggs [53] and Weeks [55] are examples in this area. In 1982, Stone
and Boggs [53] investigated the mechanism of high frequency propagation in plastic
insulated cables in order to evaluate the possibility of PD location.
Stone and Boggs showed the influence of the semi-conducting screens of extruded cables
on the attenuation by using an equivalent circuit approach as shown in Figure 2.31. They
used the cable model used by Breien and Johansen [56] in 1971 who studied the
attenuation of traveling waves in single phase high voltage oil and paper cables. Semi-
conducting layers in power cable are used to reduce stress enhancements at the
conductor-dielectric interface. The study by Stone and Boggs revealed that the pulse
attenuation in plastic insulated power cables was much greater than expected from Breien
and Johansen’s model [56]. The excess attenuation came from losses in the extruded
screens which separate the core conductor and metallic screen from the insulation. High
frequency losses are caused by the propagation of radial displacement current through the
resistance of these screens [43, 44]. The magnitude of this radial displacement current is
largely determined by the capacitance of the insulation. The model of Breien and
Johansen [56] was thus modified to include the semi-conducting layers in the shunt
admittance path.
________________________________________________________________________
Figure 2.31: An equivalent circuit of cable parameter that first introduced by Stone and
Boggs [8].
Stone and Boggs showed the influence of the semi-conducting layers on the attenuation
and velocity of extruded cables by using an equivalent circuit approach. Weeks [55] used
a more fundamental approach to arrive at almost the same results. Weeks used the
Maxwell’s equations to evaluate the effects of the semi-conducting layers, the conductors
and surrounding earth on the propagation characteristics of extruded cables. Figure 2.32
shows the change in the shape of a PD pulse as it propagated through cable, as a function
of the distance propagated [63]. As a result, the high frequency end of the PD spectrum is
lost as a function of distance as the PD pulse propagates down the cable, i.e., the
amplitude of PD pulse decreases and the width increases.
________________________________________________________________________
Figure 2.32: PD waveforms as a function of distance propagated through a cable with

frequency dependent attenuation. The distances propagated are 0, 100, 300 and 1000m.
2.7.3 Time and Frequency Domain PD Testing
In the earlier part of the 20th century, the PD detected technology only allowed the user to
detect the low frequency ranges such as few kHz to few MHz and this wasn’t able to
detect the full range of PD signals as partial discharge pulse can generate up to few
hundred MHz. Hence, the knowledge of PD in the full range of time domain is more
important. Time domain PD testing is normally carried out by triggering a PD detector
such as an oscilloscope by the first pulse. The pulse propagates from the PD source to the
cable termination at which the PD detector is located. Once a PD source is detected,
location is based on the relative time of arrival of the first and second pulses [85]. By
using phase resolved PD analysis, it can be undertaken to help identify the cause of the
PD. However, time domain testing is usually carried out under off line condition and time
domain testing requires triggering on the first PD pulse above the noise, which is difficult
to achieve under on-line condition.
However, Time domain PD testing cannot give the full picture of PD measurement. In the
late 20th century, large numbers of PD studies were based on the frequency domain
approach. The PD studies on frequency domain allows the user to determine where the
________________________________________________________________________
PD location is and to verify all different type of high frequency loss relating to cable
parameters such as semi-conducting layers. In the frequency domain PD testing, PDs can
be detected, characterized and located without having to trigger on the first pulse. The
frequency domain testing can carried out under in service condition and the PDs are
measured from various points.
In order to prevent any harm from the partial discharge and locate the PD source, it is
then necessary to reconstruct the partial discharge source pulse by using the PD pulse
captured at the termination of cable system. PD pulse reconstruction requires knowledge
of attenuation and velocity of PD type signals. Hence, an accurate cable model needs to
require for reconstructing the PD signal and predicting the any failure of cable system
and this is the primary purpose of this thesis.
Chapter 3: Cable Modelling 53
________________________________________________________________________
Chapter 3
Cable Modelling
3.0 Introduction
Due to increasing use of power cables in electrical utilities, the investigation of the power
cable and its insulation condition diagnostics becomes more important. Cable failures are
one of the primary reasons for interruption in the electricity supply and thus assessment
of condition is very important. For the purpose of PD signal analysis and discharge
location, the parameters of the partial discharge signal model must be measured prior to
PD detection.
Partial discharge analysis is one of the major diagnostic tools used for condition
monitoring of cables. Since the nature of partial discharge activity in cables is of the
streamer discharge type [57], PDs in cable have high frequency content and these signals
are attenuated significantly and influenced by the cable insulation material. Thus the
propagation characteristics of HF signals in cables are of importance. To determine PD
propagation characteristics, simulation of PD propagation phenomena is carried out. In
this regard, power cables are more complicated than overhead lines, with the cable
composed of a number of conducting, semi-conducting and insulating layers. All of these
parts have some effect on the propagation characteristics and thus modeling of the cable
for HF pulse propagation is an important requirement.
Techniques for assessing these influences can be divided into two major categories:
i) Analytical techniques where attempts are made to determine the solutions of the
appropriate field equations in term of functions such as the Bessel function [58, 59]
________________________________________________________________________
ii) Numerical techniques in which the solutions of the electric and magnetic field
equations are obtained numerically by finite element or finite difference methods [60,
61].
Analytical methods are not possible in some complex geometries. Therefore, numerical
electromagnetic solvers need to be implemented in those cases. The purpose of this
chapter is to contribute to the understanding of the application of the analytical and
numerical techniques for simulation of transient propagation in HV cables.
3.1 Theory of the Transmission line
In general, a transmission line refers to any pair of conductors that can transmit an
electrical signal from one end to the other. Some simple examples of transmission line
include parallel plate geometry, two-wire line and the coaxial cable. A transmission line
is a distributed parameter network and is described by circuit parameters that are
distributed through out its length. A cable of length 'x, shown in Figure 4.1, is described
by the following parameters:
R is the resistance per unit length; L is the inductance per unit length; G is the shunt
conductance per unit length and C is the shunt capacitance per unit length. For a length
'x, the series elements are made up of R'x and L'x, while C'x and G'x form the shunt
elements.
Figure 3.1: Equivalent circuit of a two conductor transmission line of length x.
________________________________________________________________________
By applying Kirchhoff’s current and voltage laws to the circuit shown in Figure 3.1, it
can be shown that:
wv ( x, t ) wi ( x, t )
Ri ( x, t ) L (Equation 3.1)
wx wt
wi ( x, t ) wv( x, t )
Gv ( x, t ) C (Equation 3.2)
wx wt
The Equation 3.1 and Equation 3.2 are a pair of first-order different equations referred to
as the general transmission line or telegrapher’s equations. Assuming harmonic
oscillations, the use of phasors simplifies the transmission line equations of ordinary
differential equations:
v(x,t) Re[V (x)e jZ t ] (Equation 3.3)
i(x,t) Re[I(x)e jZ t ] (Equation 3.4)
Where V(x) and I(x) are functions for the space coordinate x only and both may be
complex. Substitution of Equation 3.3 and Equation 3.4 into Equation 3.1 and Equation
3.2 yields the following ordinary differential equations:
dV (x)
(r jZ l)I (x) zI(x) (Equation 3.5)
dx
dI (x)
(g jZ c)V (x) yV (x) (Equation 3.6)
dx
where
z r jZ l (Equation 3.7)
________________________________________________________________________
is the series impedance of the transmission line and
y g jZ C (Equation 3.8)
is the shunt admittance of the transmission line. Equation 3.5 and 3.6 can be combined to
solve for V(x) and I(x) as follows:
d 2V ( x )
J 2V ( x ) (Equation 3.9)
dx 2
d 2 I (x)
J 2 I (x) (Equation 3.10)
dx 2
where
J D jE (r jZ l)(g jZ c) zy (Equation 3.11)
J is the propagation constant whose real part, D, is the attenuation constant in Nepers/m
and the imaginary part, E, is the phase constant in radians/m. it is common to express the
attenuation in decibels/m (dB/m). The quantities r, l, g and c may vary with frequency.
The solutions to Equation 3.9 and Equation 3.10 are:
V ( x ) V ( x ) V ( x ) V0 e J x V0 eJ x (Equation 3.12)
I (x) I ( x) I (x) I 0 e J x I 0 eJ x (Equation 3.13)

________________________________________________________________________
Where the plus and minus superscripts denote waves traveling in the +x and –x
directions, respectively. For a wave in one direction, the ratio for the ratio of voltage and
current at any point x is called characteristic impedance of the line Z0.
z
Z0 (Equation 3.14)
y
When a traveling wave propagates along the transmission line, it will be attenuated due to
different sources of losses in the transmission line components. The attenuation depends
on the physical properties of cable material such as r(Z),c(Z), l(Z) and g(Z). In some
case, the r and g can be assumed to be very small and can be ignored (a lossless cable).
In order to measure the characteristic impedance of cable, the C and L parameters of a

cable can be measured by terminating it with an open circuit and then a short circuit
termination respectively. For example if a sample cable has the total inductance of about
2.8μH and the total capacitance is about 3.59nF. Ignoring the r and g in Equation 3.11,
the ratio of voltage and current waves’ amplitudes is the characteristic impedance, Z0 of
the (lossless) cable is defined as:
l
Z0 = Characteristic impedance = (Equation 3.15)
c
The Characteristic impedance can be determined as 28 s. For a typical overhead

transmission line, Z0 is about 400 s. For an underground cable it is in the range of 20-80
Ohms, because of the closer spacing of the conductors, increasing c and decreasing l.
________________________________________________________________________
3.2 Historical Development in Cable Modeling
In order to detect the PD in a cable, it is thus necessary to understand the PD propagation

characteristics and mechanism of high frequency propagation in extended lengths of
cable. In 1971, Breien and Johansen [56] introduced a cable model for measuring the
attenuation of traveling waves in a single core power cable. Their model consisted of the
series impedance z, which is determined from the theoretical considerations and the shunt
admittance y, which is determined by fitting empirical functions to experimentally
determined results. Later in 1982, Stone and Boggs [53] developed their equations for
characteristics of cable. Stone and Boggs [3] found that the pulse attenuation in power
cable is actually much greater than expected from Brieien and Johansen’s theory [56].
This is mainly because of the displacement current that flows radially through the semi-
conducting layers. This magnitude of the radial displacement current is largely
determined by the dielectric material’s capacitance. Thus, Stone and Boggs [53] modified
the theory of Breien and Johansen [56] in order to include the semi-conducting layers in
the shunt admittance. The function of semi-conducting layers is to act as an interface
between the conducting and insulating materials to enable smoothing of the electric field
that helps to limit localized stress on the insulation due to conductor stranding. This was
the first investigation to consider the effect of semi-conducting layers on loss.
3.2.1 Influences from the Cable Design
As the previous section mentions all parts of the cable have an influence on propagation
characteristics so that extra care is needed. Stone and Boggs [50-53] showed the
influence of the semi-conducting layers by using a new equivalent circuit able to take the
frequency range up to 100MHz. Stone and Boggs also mentioned that the permittivity of
the dielectric in XLPE cable is approximately constant up to at least 100 MHz. However,
Gustavsen [62] and Breien [56] showed that the permittivity of the dielectric does vary
with frequency and later Boggs [64] showed that the conductivity of the semi-conducting
________________________________________________________________________
layers is varied when the frequency changes. Papazyan [64] studied the effect of screen
wire spiralization and showed that the cable high frequency characteristics depend on the
surrounding medium. The cable wave propagation characteristics above approximately
70 MHz can be affected by the surrounding medium. This effect is mainly attributed to
the helical structure of metallic screen
3.2.2 Calculation of Propagation Characteristics
Since the nature of partial discharge activity in cables is of the streamer discharge type
[65], PD has high frequency content and these signals are attenuated significantly by the
cable dielectric itself. It is important to know how it will behave under different
conditions. To achieve this, simulation of the PD propagation phenomena in cables are
carried out.
It is important to know what the necessary high frequency cable parameters are. The
measurement of cable parameters can be carried out after the cable has been built, but this
is hard to do and is rarely done. The cable parameters: resistance, inductance, capacitance
and conductance are evenly distributed along the line. These parameters will vary with
frequency and this dependence has been extensively studied over the past 30 years.
Hedman [66] developed the relationship between resistance and inductance versus
frequency for typical transmission line, and showed that the resistance value increases by
a factor of 1000 as frequency varies from 60 Hz to 1 MHz.
In 1971, Breien and Johansen [56] gave the attenuation of traveling wave of high voltage
cable due to the combined effect of dielectric losses in the cable insulation and skin effect
in the core and sheath as function of frequency. The results are in good agreement with
the measurements when having a proper impedance and admittance per unit length.
________________________________________________________________________
Wedepohl and Wilcox [67] used mathematical models which included the effect of
frequency dependency of cable parameters to estimate the transient sheath overvoltage in
cables.
Ametani [68] developed a general formulation of impedance and admittance of single

core and pipe type cables. By using these formulations, it is possible to analyze wave
propagation characteristics and transients on any type of cable system. Some major
assumptions have made in Ametani [68] study, such as constant permeability,
displacement currents are negligible at low frequency 1 MHz and neglect of the dielectric
loss.
The Electromagnetic Transient Program (EMTP) [69] is currently the most widely-used
tool for the analysis of electromagnetic transient in power systems. It was created in
1986. Cable parameter calculations in the EMTP support programs are based on well-
understood formulae and assumptions outlined by workers such as J Marti [70], L Marti
[71], Noda [72] and Ametani [68].
All of the literatures above on cable models have not considered the semi-conducting
layers in their study. However in the 1990s, number of papers reported study of semi-
conducting layers and their effect and these issues will be discussed in the following
sections.
3.3 Methodology
Analytical techniques and Numerical techniques
________________________________________________________________________
3.3.1 Analytical method
Modeling the power cable for simulation of electromagnetic transient propagation

requires the calculation of cable parameters from geometrical data and material
properties, and then conversion of the existing cable parameter into a new set of
parameters for usage by the cable model. There are number of analytical methods
available to be used such as ATPDraw, PSCAD, EMTP-RV. A user must select the
model, which is most suitable to the particular simulation required. This software is based
on the theory of Dommel’s EMTP [73-75].
The EMTP offers a number of modeling choices, from simple lumped-parameter, -

circuit representation [76] to precise models that give some dependence with frequency
of cable parameters [77-79].
The semi-conducting layers can have a substantial effect on the propagation

characteristics of a cable in terms of velocity, characteristic impedance and the
attenuation [3, 55].
Weeks [55] used a fundamental approach based on Maxwell equations to evaluate the
effect of conductors, semi-conducting layers and surrounding medium on propagation
characteristics of cables. Weeks [55] show that the speed is reduced by the conductivity
of conductors and surrounding earth when frequency below about 104 Hz. The
propagation speed is reduced by the semi-conducting layer when the frequency is below
about 109 Hz. The speed is dependent upon the thickness of insulation and conductor
semi-conducting screen.
The EMTP cable model does not allow representation of semi-conducting layers in the
cable geometry. So, an approximate data conversion procedure must be applied. The first
method of such conversion for EMTP cable model was introduced by Gustavsen [80].
Gustavsen [80] gave the full detail on the first step of procedure, and his calculation of
________________________________________________________________________
cable parameters also showed how to handle the semi-conducting layers. Gustavsen [80]
proposed including the effects of semi-conducting layers in EMTP by replacing the
insulation with a changed thickness and with increased permittivity.
However, the simulated result using this method is not accurate when high frequency
(e.g. the partial discharge frequency range) signal propagation is involved. The
components of the cable cannot be simply modified when dealing with high frequency
attenuation. In order to overcome this problem, the actual structure of the cable is
modified as suggested in [81, 82]. The original sheath section is substituted for the semi-
conducting layer and there is a fictitious insulation of thickness 1 mm specified between
the core and the inner semi-conducting layer.
In 2007, Mugala [83] developed an approximate model which was adopted for separating
the effects from different parts of the conductors and dielectric system. Mugala [83]
showed that the insulation system dominates the cable loss at frequency above 10 MHz
and the signal losses in the core conductor are generally higher than the losses in the
metallic screen.
3.3.2 Background of Five EMTP Programs
In the late 1960s Dommel Hermann developed EMTP (Electromagnetic Transient

Program) and brought this program to BPA (Bonneville Power Administration). In the
1973, Dommel Hermann left BPA and joined UBC (The University of British Columbia).
Therefore, EMTP is divided into two versions, one is a commercial simulation tool that
has been developed by BPA and other one is a model-developed tool that has been
introduced in UBC.
The BPA version has been further developed by Scott Meyer, Tsu-Huei Liu and many
North America power engineers. In order to receive more funding for development, the
DCG (EMTP Development Coordination Group) was established at 1984 (Scott Meyer
________________________________________________________________________
didn’t join DCG). The earlier members included BPA㧘the US Bureau of

reclamation㧘WAPA㧔Western Area Power Administration), CEA (the Canadian
Electrical Association㧕, Ontario Hydro, and Hydro Quebec. Further detail can be
checked at http://www.emtp96.com. Several years ago, DCG licensed EMTP to IREQ
which led to the EMTP-RV version and this is the commercial EMTP software.
Scott Meyer’s philosophy in EMTP was developed under the funding of US government,
so that it should be available to anyone in anywhere at any time free of charge, provided
the purpose of application is not for any business in simulation tool
development/marketing. In 1986, Scott Meyer established an EMTP independent version,
ATP (Alternative Transients Program). ATP is supplied at no cost and is available to the
user to self-develop. The detail of this software can be found in
http://www.ee.mtu.edu/atp/.
In 1989, UBC began using the brand name, Microtran to sell the product. The product
details can be found at http://www.microtran.com. One of the most famous products in
Microtran is PSCAD software.
In the middle of 1980s, Manitoba HVDC Research Centre developed an EMT type
program called EMTDC, which is based on the trapezoidal format. However, there is no
tie at the code level between EMTP and EMTDC. The EMTDC was initiated by Dennis
Woodford in Manitota Hydro, and then developed by Manitoba HVDC centre. EMTDC
is used on DC high voltage transmission line system. The free trail version software can
be downloaded from http://www.hvdc.ca.
Overall, the principal developers/ leaders of those five EMTP versions are:
1. UBC (so called MicroTran), developed by Dommel Hermann and J Marti
2. ATP, developed by Scott Mayer and Tsu-Huei Liu.
3. DCG, developed by L Marti.
4. RV, developed by Jean Marsederijian
________________________________________________________________________
5. The EMTDC is an independent program, the leader in EMTDC development is Dennis

Woodford, with his associates/coworkers in HVDC centre, University of Manitoba,
RTDS and Electranix.
The following sections introduce features of EMTP.
3.3.3 Numerical Integrator Substitution
Numerical integration substitution constitutes the basis of Dommel’s EMTP [73-75] and
is the most generally accepted method for the solution of electromagnetic transients.
Most of the EMTP software use trapezoidal integrator as numerical integrator
substitution, due to its simplicity, stability and reasonable accuracy in most
circumstances. However, based on a truncated Taylor’s series, the trapezoidal rule can
cause numerical oscillations under certain conditions due to the neglected terms.
Therefore, for the three phase cable modeling, the backward Euler rule is used for the
numerical integrator substitution and this will be discussed further in Chapter 7.
Discretisation of system components are the other basic characteristic of Dommel’s

EMTP. Discretisations of components give a predetermined time step, which are then
combined in a solution for nodal voltages. Branch elements are represented by the
relationship that they maintain between branch current and nodal voltage.
3.4 Transmission lines model
There are many different ways to model transmission lines and cables. The limitations of
a simple model may produce useless simulations for some purposes. Although a complex
model may produce a better model representation, the accuracy of the result will depend
on the event that it is required to simulate. A simple lossless transmission line has been
described in the previous section in the form of a Norton equivalent. The model is equally
________________________________________________________________________
applicable to overhead lines and cables. The main differences arise from the procedures
used in the calculation of electrical parameters from their respective physical geometries.
In the following section a description of the available models is given and then several
simulation examples are presented to compare the models. The transmission line model
can be divided into two types: Frequency independent and frequency dependent.
Frequency independent model

x PI section model
x Bergeron’s model
x Constant distributed parameter models
Frequency-dependent model
x Semlyen model
x J Marti model
x L Marti model
x Taku Noda model
Figure 3.2 shows a decision tree for the selection of the appropriate transmission line
model. The minimum limit for travel time is (Length of cable)/ c, where c is the speed of
light 3 x 108 m/s, (but not if it is a solid dielectric cable) and this can be compared to the
time step to see if a PI section or traveling wave model is appropriate.
________________________________________________________________________
Figure 3.2: Decision tree for transmission line model selection.
3.4.1 Frequency independent model
PI Section Model
The PI circuit is a discrete approximation to the constant distributed parameter model.

Line models based on PI circuits were used as a first solution for transient studies either
using the EMTP program or Transient Network Analyzers.
________________________________________________________________________
The PI circuit is not generally the best model for transient studies. The solution of the
simulations using the distributed parameter model is faster and commonly gives more
accurate results.
The cascade connection of PI circuits can be useful for untransposed lines since it is not
necessary to consider approximations of the phase to mode transformation matrix. By
default, lines with frequency dependent parameter elements cannot be represented by PI
circuits. Moreover, spurious oscillations generated by lumped parameter elements must
be accepted. However, resistances connected in parallel with the R-L branches can
compensate the spurious oscillations. The correct number of PI circuits depends on each
particular system to be simulated.
Various PI section models exist, however the nominal/coupled PI as shown in Figure 3.3,
is the preferred option for transient solutions. The exact equivalent PI is only adequate for
steady-state solution where only one frequency is considered.
The main advantages of the PI circuit are:

a) The model does not condition the calculation time step.
b) The steady state solution is exact.
Figure 3.3: Nominal PI section.

________________________________________________________________________
Bergeron’s model
Bergeron’s model [84] is a simple, constant frequency method based on traveling wave
theory and it is basically the model described in previous section. The line is still treated
as lossless but its distributed series resistance is added in lump form. The lumped
resistance can be inserted throughout the line by dividing its total length into several
sections. The lumped resistance model is shown in Figure 3.4. It can give reasonable
results provide thatR / 4 Z o , where Z o is the characteristic impedance. However, for
the high frequency studies, the lumped resistance model may not be adequate.
Figure 3.4: Equivalent two port network for line with lumped losses.
By assigning half of the midpoint resistance to each line section as shown in Figure 3.5,
which is the model for half of the line:
1
ikm (t) vk (t) I k (t W / 2) (Equation 3.16)
Zo R / 4
1 § Z R / 4·
I k (t W / 2) vm (t W / 2) ¨ o im (t W / 2) (Equation 3.17)
Zo R / 4 © Z o R / 4 ¹̧
________________________________________________________________________
Figure 3.5: Equivalent two port network for half line section.
By cascading two half line sections and eliminating the mid point variables, the model is
shown in Figure 3.6. It has the similar expression as previous models but the current
source representing the History terms and it contains conditions from both ends on the
line at time (t W / 2) :
Z o
I k' (t W ) (vm (t W ) (Z o R / 4)imk (t W ))
(Z o R / 4)2
(Equation 3.18)
R / 4
(vk (t W ) (Z o R / 4)ikm (t W ))
(Z o R / 4)2
Figure 3.6: Bergeron transmission line model.

________________________________________________________________________
Constant distributed parameter models
The constant distributed parameter model calculates the different time propagation of the
decoupled mode components. In each extreme of the line, the values are converted from
mode domain to phase domain using a transformation matrix. For transposed lines, this
matrix is constant, however, for untransposed lines, the transformation matrix varies with
the frequency [111]. The variation with the frequency is more significant for cables than
for lines. This makes necessary to take cautions upon adopting the frequency value where
the parameters will be calculated. In addition, if the time step is not a sub-multiple of the
line propagation time, the results of the simulation will be incorrect. Moreover, since the
values are calculated by linear interpolation, the results may differ for different time step
calculation if the signal contains high peak values.
The main limitation of the constant distributed parameter models is the assumption of
considering the parameters constant with respect to the frequency variation. Larger error
is produced for the ground mode, i.e. for those transient signals in which the zero
sequence components of voltage and current are present.
For short lines or cables, the constant distributed parameter model requires that the time
step must be less than the propagation time. Therefore, it requires greater calculation
time.
Despite its limitations, this model improves substantially the results with respect to the
models based on PI circuits.
3.4.2 Frequency-dependent model
In the old EMTP program, the line model separates the propagation into low and high
frequency so that the line can have higher attenuation to higher frequencies. This was an
________________________________________________________________________
early attempt to provide frequency dependence. However, newer frequency dependent

models (in the phase domain) are now preferred and will be discussed in the following
sections.
Semlyen's Model
The Semlyen’s model approximates the characteristic admittance and the propagation
constant of each mode through two exponentials. Even though it was not the first model
that takes into account the variation of the parameters with the frequency, it is the oldest
model that is available in the ATP software [91].
The simplicity of the equations causes the approximation to become insufficient even for
line parameters without discontinuities. Therefore, the use of this model is becoming rare
and its availability in ATP software may be discontinued.
José Martí's Model
The José Martí's Model (J Marti) model approximates the characteristic admittance and
the propagation constant by rational functions [89]. Even though it has limitations, among
all the variable parameter models it is the most frequently used. One of the limitations is
that it uses a constant transformation matrix to convert from mode domain to phase
domain. For overhead lines this is not as important as it is for cables.
The development of this model can be best explained from the physical interpretation of
Meyer and Dommel’s weighting functions [150]. Figure 3.7 shows the new weighting
function, which developed by J. Marti [89] from Meyer and Dommel’s weighting
functions [150].
________________________________________________________________________
ik (t) im (t)
Vk (t) a1 (t) Vm (t) Zeq

Zeq
Figure 3.7: Physical interpretation of the new weighting function.
In order to define Zeq for the new weighting function, the forward and backward
traveling functions can be defined in the frequency domain as
Fk Z Vk Z Z eq Z I k Z (Equation 3.19)
Fm Z Vm Z Z eq Z I m Z (Equation 3.20)
and
Bk Z Vk Z Z eq Z I k Z (Equation 3.21)
Bm Z Vm Z Z eq Z I m Z (Equation 3.22)
Where Zeq()= impedance of linear network.

1
A1 Z e J Z l (Equation 3.23)
cosh ª¬J Z l º¼ sinh ª¬J Z l º¼
The time domain form of A1 () is the function of a1(t) show in Figure 3.7.
Comparing Equation 3.21 to 3.22 into the frequency domain, it follows that
Bk Z A1 Z Fm Z (Equation 3.24)
Bm Z A1 Z Fk Z (Equation 3.25)
The time domain form of Equation 3.14 and 3.15 is given by the convolution integrals:
f
bk t ³W f t u a (u)du
m 1 (Equation 3.26)
________________________________________________________________________
f
bm t ³W f t u a (u)du
k 1 (Equation 3.27)
From Equation 3.11 and 3.12,

vk ( t ) ek (t ) Ekh (Equation 3.28)
vm ( t ) em (t ) Emh ' (Equation 3.29)

Where ek(t) and em(t) are the voltage across the network Zeq. After converting to a modal
representation, Equation 3.28 and 3.29 give at each time step, the equivalent J-Marti line
models shown in Figure 3.8
ik (t) im (t)
k m
Vk (t) Zeq Ikh Imh Zeq Vm (t)
Figure 3.8: J-Marti frequency – dependent line models at nodes k and m.
The model presents an unstable behavior for very low frequencies; for example cases
including trapped charge. Furthermore, in some cases, the voltage can be increased
without limits.
One of the parameters required by this model is the mode conductance. The model is
sensitive respect to this parameter for trapped charge studies. Even though it is possible
to obtain accurate results, it requires data manipulation. For example, the fitting must
start from a very low frequency value to adjust the model (i.e. 0.0001 Hz).
________________________________________________________________________
L Martí's Model
This model is also known as the L Marti cable model [88]. It takes into account the
frequency dependence of the cable parameters, as well as the frequency dependence of
the modal transformation matrix Q [87]. It is the most accurate cable model available in
the EMTP, and unlike other distributed parameter models. It can accurately reproduce
high and low frequency phenomena in the same simulation. The L Marti model is
computationally demanding in terms of speed and storage, but not prohibitively so. In
fact it is only about 30% slower than a comparable J Marti line model.
Taku Noda's Model.
The Taku Noda’s model is different from the previous models because the calculation is
made directly in phase domain [90]. Therefore, it eliminates the approximation errors
produced by the use of the transformation matrix. The characteristic admittance and the
deformation coefficients are fitted through rational functions.
For a given line, to obtain an adequate model using Taku Noda's model is generally more
difficult, however it has the advantage that it allows one to define a time step independent
of the propagation time, but this demands one to also employ this time step for the
simulation. If another time step is necessary for the simulations, the model must be
recalculated. Another advantage is that it is able to calculate a three phase cable model
and to consider the coupling effect of each phase.
The creation of a Taku Noda’s model requires two steps; first from the line data an
auxiliary file is created using ATP software. Secondly this file is converted to the final
file that will model the line in the simulation. An adjustment program called ARMAFIT
that adjusts the values using rational functions is used to generate the final file [92].
________________________________________________________________________
To compare the performance of each one of the above frequency dependence line models,
a simulation with 11KV, 136 M XLPE cable (inner and outer semi-conducting layers are
included) was tested. The simulation was run by ATP. The energizing voltage in this
circuit is “1.5*sin(14000000*3.1415*timex)*exp(-7000000*timex)” with short circuit
termination. The simulated results are shown in Figure 3.9. The results produced by the
Semlyen’s model differ considerably respect to the results (40% more attenuation)
produced by the other models. In this frequency range (few MHz), Semlyen’s model is
not the good choice to do the simulation.
________________________________________________________________________
Figure 3.9: Simulated results of four different frequency dependent cable models
________________________________________________________________________
3.4.3 Underground Cable Parameters
The cross section of a coaxial cable can be simplified to that of Figure 3.10 and its series
per unit length impedance is calculated by the following set of loop equations:
§ dV1 ·
¨ dx ¸ § Z' Z12' 0 · § I1 ·
¨ ¸ ¨ '
11
¸
¨ dV2 ¸ ¨ Z 21
'
Z 22 '
Z 23 ¸ ¨ I2 ¸ (Equation 3.30)
¨ dx ¸ ¨ ¸¨I ¸
¨ dV ¸ ¨© 0 Z '
32
'
Z 33 ¸¹ © 3 ¹
¨ 3¸
© dx ¹
where
Z11' = the sum of the following three component impedances:
Zcore-outside = internal impedance of the core with the return path outside the core
Zcore-insulation = impedance of the insulation surrounding the core
Zsheath-inside = internal impedance of the sheath with the return path inside the sheath
'
Z 22 = Zsheath-outside + Zsheath/armour-insulation + Zarmour-inside
'
Z 33 = Zarmour-outside + Zarmour/earth-insulation + Zearth-inside
The coupling impedance, Z12' = Z 21

' '
and Z 23 '
= Z 32 are negative because of opposing
current directions
Z12' = Z 21
' '
= - Zsheath-mutual and Z 23 '
= Z 32 = - Zarmour-mutual
where Zsheath-mutual = mutual impedance of the tubular sheath between the inside
loop 1 and outside loop2.
Zarmour-mutual = mutual impedance of the tubular armour between the
inside loop 2 and outside loop3.
Z13' = Z 31
'
= 0 because loop 1 and loop 3 have no common branch. The impedances of
the insulation (:m-1) are given by
________________________________________________________________________
§r ·
'
Zinsulation jZ ln ¨ outside (Equation 3.31)
© rinside ¹̧
where routside is outside radius of insulation and rinside is inside radius of insulation.
Figure 3.10: The cross section of coaxial cable [86].
3.5 Issue of semi-conducting layers in ATP programs
ATP (Advanced Transient Program) is one of the EMTP (Electromagnetic Transient

Program) forms and is probably the most widely-used tool for the analysis of
electromagnetic transients in power systems. The basic method of cable modeling is to
calculate the cable parameters from the geometrical data and material properties and then
convert into a new set of parameters for use by available cable models.
The current EMTP cable model is limited for cable modeling in that it does not consider
certain cable features such as semi-conducting layers and wire screens. Instead, EMTP
(or ATP) assumes a cylindrically symmetrical current distribution in all conductors. The
assumed cylindrical distribution also means that the helical winding effect of the sheath
cannot be taken into account. Each conducting component of the cable has constant
permeability and the dielectric is considered as being non-dispersive and frequency
________________________________________________________________________
independent. In general, the dielectric loss is assumed to be negligible compared with

conductor and sheath loss.
However, it is quite apparent that these assumptions would give inaccurate results for
high frequency (several MHz) and high voltage applications.
Earth
Armor
Sheath
Core
Insulation
Figure 3.11: Representation of a single core cable in ATP.
Clearly, the real cable design is different from that used in the simple ATP configuration
shown in Figure 3.11. In particular, a realistic cable model needs to include the following
components: the core stranding, the inner & outer semi-conducting layers and the wire
screen (sheath).
3.5.1 Importance of including semi-conducting layers in ATP models
Neglecting the presence of semi-conducting layers, the results obtained using ATP are
satisfactory with the theoretical calculations in terms of propagation times, as shown in
Figure 3.12. The simulation is based on a (136 meter length) J Marti cable model with a
________________________________________________________________________
short circuit termination. The calculated theoretical velocity is obtained from the standard
lossless representation using an insulation permittivity of 2.2 (XLPE).
0.4
[mA]
0.2
Amplitude (mA) 0.0
-0.2
-0.4
Sending
-0.6 Receiving
-0.8
-1.0
-1.2
0.0 0.2 0.4 0.6 0.8 1.0 [us] 1.2
Time ( s)
Figure 3.12: the ATP simulated result for comparing the velocity of propagation without
semi-conducting layers. Red line is the sending signal and green line is the receiving
signal from the ATP simulation.
From the ATP simulated results, the velocity of propagation can be obtained by dividing
the length of the cable (136 M) by the ATP calculation of the time difference in sending
and received signals. Using the data from Figure 3.12 the velocity obtained by ATP is
2.08 x 108 m/s, which is in good agreement with the theoretical value of 2.021 x 108 m/s
(speed of light = 299792458 m/s).
Table 3.1 is shown the simulations results of this cable model (propagation velocity and
attenuation) when different permittivity of insulation is applied. The result shows the
higher permittivity of insulation, which give the lower propagation velocity.
Table 3.1: Simulation results of different permittivity of insulation.
Permittivity of Insulation Propagation velocity The ratio of I receiving end

/ I sending end
8
2.2 2.08 x 10 m/s 1.57
8
2.5 1.86 x 10 m/s 1.54
________________________________________________________________________
The semi-conducting layers can thus have a substantial effect on the propagation
characteristics of a cable in terms of velocity, characteristic impedance and the
attenuation. However ATP does not allow representation of semi-conducting layers in the
cable geometry. So, an approximate data conversion procedure must be applied. The first
method of such conversion for ATP was introduced by Gustavsen [93]. Gustavsen
proposed including the effects of semi-conducting layers in ATP by replacing the
insulation with a changed thickness and with increased permittivity, determined through
the conversion equation:
r= rins (ln(r2/r1)/ln(b/a)) (Equation 3.32)
Where rins is the relative permittivity of the main insulation (2.2 for XLPE), r1 is the
radius of the core, r2 is the sum of r1 plus the sum of the thickness of semi-conducting
layers and insulation and a and b are the inner and outer radius of the insulation.
Therefore, in the ATP cable geometry section (Figure 3.11), the insulation section is
modified by the new radius, r2 and permittivity, r:
r2 = 10.05 + (0.6 + 3.5 + 0.8) = 14.95 mm. Where 10.05mm is the radius of core, 0.6mm
is the thickness of inner semi-conducting layer, 3.5mm is the thickness of insulation and
0.8mm is the thickness of outer semi-conducting layer.
r = 3.2 (calculated by Equation 3.32)
However, the simulated result using this method is not accurate when high frequency
(e.g. the partial discharge frequency range) is involved. This method cannot include the
semi-conducting layers precisely and semi-conducting layers have the dominant role on
cable loss at high frequency. The components of the cable cannot be simply modified
when dealing with high frequency attenuation.
In order to overcome this problem, the actual structure of cable was modified as
suggested in [94]. Referring to Figure 3.11, the armor section is substituted for a sheath
________________________________________________________________________
layer. The original sheath section is substituted for the semi-conducting layer and there is
a fictitious insulation of thickness 1 mm specified between the core and the inner semi-
conducting layer.
Figure 3.13 shows the results of a comparison of results obtained using the two methods
outlined above. The simulation used is J Marti Cable model in ATP. The cable sample is
136 meter XLPE power cable with inner and outer semi-conducting layers. The set-up of
the models is as shown in Figure 3.14. The second method (including the fictitious layer)
gives good agreement with measurements. The pulse is approximated 5 MHz. The results
from the first method (modification of insulation thickness) give lower attenuation and
lower velocity. The first method is modified the thickness of insulation in order to
include the semi-conducting layers and second method (54%) give the higher attenuation
than first method (66%).
0.3
[mA]
0.1
Amplitude (mA)
-0.1
-0.3
b a
-0.5
-0.7
-0.9
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 [us] 1.6
Time ( s)
Figure 3.13: Comparison of two different methods of including the semi-conducting

layers in the cable model.
Blue line (a) is the result using first method and the green line (b) is that obtained using
the second method.
________________________________________________________________________
Figure 3.14: Cable model in ATP.
The problem that arises in PD propagation is that there are no distributed parameter
models available for such power cable insulation at such high frequencies (100 MHz)
Moreover, it is important to model the frequency dependence of power cables accurately
in order to minimize the cost of construction. In this thesis, the simulation work used the
ATP software as a basis to develop the existing J-Marti models for power cables to
simulate the PD propagation to 100 MHz.
In ATP, J-Marti model of the cable can only allow a three layer configuration to be used
in the model: the conductor, the sheath and the armor. Since the ATP cable model does
not have the capability of modeling the inner and outer semi-conducting layers, it had to
be modified to allow them to be incorporated. In order to develop a full XLPE cable
model, the original sheath layer was employed as the inner semi-conducting layer, the
armor layer is substituted for the outer semi-conducting layer and between them is the
insulation layer. The metal sheath was included by using an enclosing pipe in a pipe-type
cable model in the ATP and there is no change on the total radial distance.
3.6 Numerical Method
When the cable structure is complex, simulation through circuit theory (ATP)
approximations become difficult or impossible. The cable can only be modeled through
the use of software that computes electromagnetic phenomena, often known as Maxwell
________________________________________________________________________
solvers, which provide a solution to Maxwell’s equations. Numerical approaches to the

solution of Maxwell’s equations can be classified in several ways [95]:
i) Boundary approach, where only a boundary is discretized. These include Method of

Moments (MoM) and Boundary Element Method (BEM);
ii) Area approach, where all areas are discretized. These include Finite Element Method
(FEM), Finite Difference Time Domain method (FDTD) and Finite Integration
Technique (FIT). FDTD, FIT, FEM and MoM can evolve the solution through time and
FEM, BEM and MoM can evolve the solution through both frequency and time domains
Sakaguchi and Oyama [96] give second part of the series of Maxwell solvers [95].
Sakaguchi developed the application of field analysis, Method of Moments (MoM) in
GIS. For comparison, the current waveforms are compared with the EMTP results which
give reasonable agreement.
The Finite Element Method (FEM) is used to calculate the propagation characteristics.
The FEM technique originated in structural analysis [97] and it was applied to
electromagnetic problems in the late 1960s [98]. Lucas and Talukdar [99] estimated the
cable resistance and inductance by using the finite element techniques.
Cârstea [100] describe the applications of FEM to the computation of the transmission
line parameters such as resistance, self and mutual inductance and capacitance. The
results are compared with the analytical results.
The aim of the study by Papazyan [101, 102] on the wave propagation in power cables
was to analyze the effect on the attenuation constant by applying helical wire of screen
and different surrounding medium such as water, air and soil. Papazyan [101, 102] used
the method of field analysis, FEM and FDTD. The comparison results reveal marked
agreement between the analytical and FEM model results, both matching the
________________________________________________________________________
measurement result closely. The FDTD result contains relatively higher error, which is
attributable mainly to the volume discretizing (meshing) and the finite length of the used
cable model.
As mentioned previously, the full cable structure must be included in the modeling in
order to get accurate results. Numerical technique can overcome this problem easily
3.6.1 Finite Element Method
In this paper, the Finite Element Method (FEM) is used to calculate the propagation
characteristics. The FEM technique originated in structural analysis [97] and it was
applied to electromagnetic problems in the late 1960s [98]. As mentioned previously, the
full cable structure (conductor, insulation, semi-conducting layers and sheath, as shown
in Figure 3.15) must be included in the modeling in order to get accurate results and FEM
technique can overcome this problem easily, as the method can be applied to complex
conducting and dielectric configurations, as shown in Figure 3.15.
Sheath (copper/lead)
Outer semi-con layer
Insulation (=2.2)
Inner semi-con layer
Conductor (aluminium)
Figure 3.15: Geometrical configuration of single core cable in FEM program. The finite
element method can be used to include the actual geometry of the cable.
________________________________________________________________________
Conductors subject to skin and proximity effects have non-uniform current distributions.
Finite element approaches deal with this situation by dividing each conductor into a large
number of sub-conductors or filaments. It is then assumed that the current distribution
over each filament is constant (Figure 3.16). By making the number of filaments suitably
large the error in this approximation can be made arbitrarily small [103], but the
simulation time is longer.
Conductor
Figure 3.16: Mesh structure used in the 2D FEM model.
A general approach is to obtain second order differential equations for the electric and
magnetic field, known as the vector Helmholtz Equation or the wave equation. This is the
most general and complex equation for describing the propagation of electromagnetic
fields. It should be noted that the FEM numerical solver places stricter requirements on
the PC RAM usage of the computer.
The finite element simulations are solved in the two-dimensional perpendicular hybrid-
mode solver, due to the inherent nature of the inhomogeneity of the materials, the
solutions is a hybrid between Transverse Electric (TE) and Transverse Magnetic (TM)
field modes [104]. The analysis is made on a cross section in the x-y plane of the cable.
The wave propagation is in the z direction and has the form:
________________________________________________________________________
H(x, y, z, t) = H (x, y)ej(t-

z) (Equation 3.33)
E(x, y, z, t) = E (x, y)ej(t-

z) (Equation 3.34)
Where is angular frequency and

is the phase constant.
E(r, t) = Re(E(rT) ej(t-
z) )
= Re(E(rT) ej(t-z) ) (Equation 3.35)
= + j
= - (Equation 3.36)
The propagation constant includes the imaginary part

(rad/m) and the real part, ,
which represents the attenuation constant (dB/m) along the propagation path.
3.6.2 Advantage of FEM
ATP does not account for the spiralization of the metallic screen and the dielectric loss.
However, the FEM can overcome these problems. To illustrate the significance of the
greater accuracy with FEM, Figure 3.17 shows the percentage loss increase that occurs
after the copper sheath is replaced by a lead sheath.
0.6
% of attenuation increases
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50
Frequency (M Hz)
Figure 3.17: FEM result of the percentage increase of attenuation when a lead sheath is
used instead of a copper sheath.
________________________________________________________________________
The result shows that at low frequency, the percentage difference loss (after the copper
sheath is replaced by lead sheath) is larger than the loss difference at high frequency and
the percentage loss difference is thus decreasing as frequency increases. Moreover, FEM
can also obtain the results of different structure of sheath. For example, helical structure
and different number of copper wire included [see Figure 3.18 a & Figure 3.18 b].
(a) (b)
Figure 3.18 a: FEM can model different structures and different number of wires in the
sheath layer.
Figure 3.18 b: the restricted form of sheath structure that ATP cable model assumes in its
model.
The FEM can also include dielectric loss in the cable model. Figure 3.19 shows results
obtained for the attenuation factor of an XLPE cable when the dielectric loss factor (Tan
) of XLPE cable is varied from 0.01 to 0.0005. This result can give useful information
on the influence of dielectric loss in signal propagation over a wide frequency range.
More details on the modeling and simulated results can be found in Section 4.4.
________________________________________________________________________
0.25
0.2
Attenuation (dB/m)
Tan: 0.01
0.15 Tan: 0.001
Tan: 0.005
0.1 Tan: 0.0001
Tan: 0.0005
0.05
0
0 10 20 30 40 50
Frequency (M Hz)
Figure 3.19: Attenuation factor versus frequency as a function of loss factor of insulation.
The assumptions in this FEM model are infinite surroundings and infinite cable length.
So another technique is proposed that combines the versatility of the numerical
procedures with the efficiency of the analytical solutions. The idea of this technique is to
evaluate the self and mutual effects of both impedance and admittance using FEM field
calculation [105, 106] and using the values to understand the propagation of
electromagnetic waves.
3.7 Comparison between Both Methods
ATP is efficient only in analyzing simple geometrical configurations of cables.

Numerical modeling requires a good balance between accuracy and computation cost.
However, the numerical technique is, at least in theory, quite generally applicable.
However, it is subject to computational limitations that arise from the extremely large
numbers of finite elements that can be required to produce accurate results. ATP suffers
from a more fundamental limitation – it is not generally applicable, particularly to
complicated conductor arrangements.
________________________________________________________________________
Table 3.2: Attenuation in dB/m for specific frequencies determined by measurement,

ATP modeling and FEM modeling.
Frequency Measurement ATP FEM

(M Hz) Attenuation (dB/m) model model
5 M Hz 0.014545 0.001818 0.016076
10 M Hz 0.036364 0.027727 0.032266
15 M Hz 0.055455 0.048684 0.053909
20 M Hz 0.070909 0.059091 0.065444
30 M Hz 0.101818 0.090909 0.100455
40 M Hz 0.133636 0.123636 0.130237
50 M Hz 0.174545 0.172727 0.182754
Table 3.2 compares the measured and simulated results obtained using ATP and FEM
modeling. The results from ATP give attenuations that are slightly lower (approximate 7
to 12 %) than the measured results and the results that are obtained by FEM. One of the
main reasons for this is that the ATP program cannot take into account the semi-
conducting impact and the dielectric loss of the cable. Since the analytical model does not
account for the dielectric loss and irregular structure of the metallic screen, these should
be taken into account in order to accurately reproduce the experimental results. Because
of the complexity of the geometry, analytical field theory or circuit approximations
analysis become difficult and therefore numerical modeling is considered as a better and
more feasible approach.
3.8 Conclusion and Summary

Two major cable modelling techniques are introduced in this chapter. ATP Program is the
most popular tool for transient study of power system due to the wide range of theoretical
models employed. However still there are some capabilities which need to be developed
such as semiconducting layers in cable model, this is not directly available in cable model
of ATP. The propagation of pulses in cables can be simulated using equivalent circuit
models. However, as a result of the complex nature of cable, such an analysis cannot be
________________________________________________________________________
carried out with sufficient accuracy by equivalent circuits. Instead, a numerical

electromagnetic method (FEM) is applied. The simulated results of both methods are
given in Chapter 5, High Frequency Pulse Measurements and Analysis.
ATP and FEM techniques present the following main limitations:

i) ATP is very efficient only for analysis of simple geometrical configurations of the
cable;
ii) Numerical differential methods such as FEM are very efficient but they require a fine
discretization of the conductive regions to obtain accurate solutions.
3.8.1 ATP Program versus FEM Technique

The advantage and disadvantage of both techniques are outlined below.
ATP- Advantages:
- Free cost and low hardware requirements
- Allow the user to access the main part of software
- Simulated time is short
ATP- Disadvantages:
- Not straightforward to represent features: wire screens, semi-conducting layers and
lossy insulation
- Neglects the effect between parallel cables.
ATP- Suggested improvements:
- Cable model should be able to include more layers of cable
- Easier to access the ARMAFIT program for the Noda cable modeling
- Plotting function should include in the ATP program
- a friendly three phase cable model.
FEM- Advantages:
- Can handle the cable with irregular shapes.
________________________________________________________________________
- Included the loss factor of conductor, insulation and the skin effect.
- Allow 3-Dimensional model.
FEM- Disadvantages:
- Fine meshes need at high frequency to achieve accurate results
- Most FEM programs are commercial software so the cost is high.
- Good computer hardware is required.
FEM- Suggested improvements:
- Minimize the simulation time.
- FEM should be combined with the electric circuit program.
3.8.2 Factors that will affect the results of cable simulation:
1. Number of phases
2. In air or in ground or on the surface
3. The ground resistivity in ohm of the homogeneous earth.
4. Frequency at which the line parameters will be calculated (Bergeron and PI) or
the lower frequency point (J-Marti, Noda and Semlyen).
5. Length of the cable
6. Impedance and admittance matrix data (R, ZL and ZC).
7. Conductance between conductors
8. Total capacitance between conductors
9. Number of cables in the system. For example: three phase cable.
10. Relative permittivity of inner insulator
11. Relative permittivity of outer insulator
12. Relative permeability of conductor.
13. Total radius of the cable.
Chapter 4: Field Analysis 93
________________________________________________________________________
Chapter 4
Field Analysis
4.0 Introduction
In power cables using cross-linked polyethylene (XLPE), ethylene-propylene rubber

(EPR) and paper-oil insulations, voids may be formed during the fabrication process or
may form subsequent to energization and operation of the cables at significant load.
While paper cables are relatively resistant to PD activity, XLPE cables are not. The
electric field distribution in high voltage XLPE power cables with voids present is a
concern to electrical engineers. The peak stress value in an insulating system is an
important parameter to control because it influences discharge initiation and propagation.
In the study of the potential for PD generation, it is clear that the electric stress in a void
in XLPE should be calculated in terms of the general electric stress in the solid insulation.
Investigations into electrical stress within gaseous voids have been made over the past
forty years. Klein [115] investigated several types of voids in impregnated paper cables
and indicated that voids located in butt-gaps with a lens-shape are the most dangerous
one in terms of their PD potential. Nesseir [116] derived expressions for determining the
PD inception voltage in typical voids in cables. Cheng [117, 118] determined the
insulation thickness of a cable, which gave maximum inception voltage of the void. In
addition, Cheng [117, 119] shows that if the maximum size of voids decreases from 50 to
30 μm, the insulation thickness of a 15 kV cable in the current standard is adequate for
use at 30 kV. Therefore, decreasing the void size is a dominant consideration in
decreasing cable size, especially for higher voltage cables.
________________________________________________________________________
In all of these investigations, the effect of void parameters on the electric stress within the
void was computed using both analytical and numerical methods of solution.
In this Chapter, three simple geometrical shapes of voids are considered: spherical,
elliptical and cylindrical, using different orientations of the void axis. The electric field
distribution in the void is investigated as a function of different void shape, relative
position of void and its orientation. In addition the effects of insulation permittivity,
stress in the void and the number of voids on the stress distribution are also studied.
Higher electric stress in void can lead to the breakdown of cable insulation. Therefore, it
is important to find the electric stress in the void in relation to its dimension and to the
electrical properties of the layers of cable.
4.0.1 Partial Discharge in Cavities
The presence of voids can eventually lead to insulation breakdown since the breakdown
strength of the gaseous void is lower than that of solid dielectric and, in addition, the
electric field is stronger inside the void than in the insulation because of the lower
permittivity of the gas relative to the solid dielectric material. Cross-linked polyethylene
insulation of cable should be free from any void larger than 125 μm for cable rated 5
through 69kV [4].
4.1 Analytical method
A 3 Dimensional (3-D) model of a spherical void within the insulation is shown in Figure
4.1. The 3-D model can be reduced to a 2-D model as shown in Figure 4.2. Use of a 2-D
model can greatly reduce the calculation time. However the variety of void shapes mean
that a 2-D model is not always possible. With a 3-D model it can be possible to improve
the description of the void spatial extension and to model the influence between the
________________________________________________________________________
shapes of voids more realistically. The voids in this paper are assumed to be filled with
air (r = 1).
Figure 4.1: 3-D model of a spherical void with its axis parallel to the electrical field
distribution.
Figure 4.2: 2-D model of a spherical void with varied radius a and b, which located in
insulation of thickness t and permittivity r.
Figure 4.3: A void located in a uniform field.
The analytical formulae employed in this study have already been reported [120, 121].
Assuming there is no free surface change at the insulation surface and cavity boundary, a
________________________________________________________________________
cylindrical cavity placed in a uniform dielectric field Eo, with the cavity axis normal to
the direction of Eo, will have electric stress (E(r)) outside and inside the cavity
determined by Equation 4.1 and Equation 4.2 [120].
E(r) = Eo ( ir [1+ (r -0/r +0)a2/r2]cos - i [1-(r -0/r +0)a2/r2]sin ) for r > a
(Equation 4.1)
= Eo (2 0/r + 0)( ir cos - i sin ) for 0 r < a

(Equation 4.2)
E(r) = Eo ( ir [-1-(r -0/r +20)2a3/r3]cos + i [1-(r -0/r +20)2a3/r3]sin ) for r > a

(Equation 4.3)
= Eo ( - ( ir cos - i sin ) (3 0/r +2 0)) for 0 r < a

(Equation 4.4)
For a spherical cavity the stress outside and inside the cavity can be calculated by
Equation 4.3 and Equation 4.4 [120], where ir and i are unit vectors of the co-ordinate
system in the insulation. a is the radius of the cavity and r is the distance from the cavity
centre to the arbitrary point, P. r0 is the permittivity of the solid insulation and 0 is the
permittivity of the gaseous void. In the above formulae, the surface conductivity of the
cavity has been neglected. If finite conductivity is included, the electric stress in the void
may be significantly reduced. For material with high permittivity, the average field stress
inside the spherical cavity is 3/2 times the average applied field without the void as
shown in Equation 4.4.
In practice, cavities have irregular shapes and thus analytical solutions for such non-ideal
voids are impossible. But numerical solutions can be obtained by solving Maxwell’s
equations with appropriate boundary equations.
________________________________________________________________________
4.2 Numerical method
Due to the singularity point at the corner of a cylindrical cavity, the analytical method
cannot be used without simplifying assumptions [122]. In such cases, a Finite Element
Method (FEM) was chosen for computation use.
4.2.1 Finite Element Method
The finite element method (FEM) was first proposed in the 1940s and its first application
was in aircraft design in the 1950s. Its first application to electromagnetic problems was
not realized until the late 1960s and nowadays it is widely used in a large variety of
electromagnetic problems.
Since its early days, the FEM has evolved into well-developed numerical techniques for
obtaining approximate solution to boundary-value problems. The basic premise behind
this technique is the discretization of the solution domain into a finite number of
subdomains (elements), on which the solution is approximated by a piecewise
interpolation function [38].
The whole cross-section of the cable and the void cavity was included in the
computation. It consists of four main domains: the insulation material, the conductor
material, the cavity and the cavity surface. The cavity itself is subdivided into five small
channels which are used for determining the stress distribution inside the cavity. The
calculation is performed using a model with a surface layer of insulation 0.2 mm thick
around the cavity.
An initialized mesh with 1618 elements was made from the beginning as shown in Figure
4.4. In order to increase the accuracy, a higher order approximation is obtained by sub-
________________________________________________________________________
dividing each element into many smaller elements as shown in Figure 4.5 or refine
selection mesh which focuses on the cavity, is shown in Figure 4.6.
Figure 4.4: Initialized mesh consists of 1618 elements.
Figure 4.5: Refined mesh consists of 6472 elements.

________________________________________________________________________
Figure 4.6: Refine selection mesh consists of 3084 elements.
4.2.2 Model Overview
The model in this study is handled from a script and is described in a flow chart in Figure
4.7. The model contains two loops: geometry and parameter settings. Both setthings are
initialized once outside the loops. The first loop is used to estimate the accuracy. When
the simulation result is not accurate enough, the higher order mesh will be called as
shown in Figure 4.4 to Figure 4.6. The second loop is used for determining the results by
changing the position of cavity or shape of cavity and it is running separate from the
model. If the size of cavity is outside the thickness of insulation or the position of cavity
is located outside the insulation, then the loop will not continue. After two loops are
finished, the field distribution plot and related data can be obtained.
The permittivity of the dielectric materials is taken as 2.5 (XLPE insulation have the
permittivity as 2.2. Table 3.1 in Chapter 3 shows the comparison of permittivity of
dielectric materials, 2.2 and 2.5) and 10 (this value of permittivity, 10 is used to compare
the effect of higher permittivity) and the applied voltage is 12.7 kV. The insulation
radius, R of the cable chosen is 2.7 times to its conductor radius, rc, i.e. R/rc = 2.7. The
________________________________________________________________________
ratio of gaseous cavity diameter to the cable insulation thickness, a/t are selected as 0.05,
0.1, 0.2 till 1 respectively. These parameters correspond to the cable samples will be
indicated in the relevant subsections below.
In this study, three different shapes of cavities are considered. Under 2-D model,
spherical, ellipse and rectangle shapes of cavities are selected respectively as shown in
Figure 4.2, b and a represent the length of y-axis and length of x-axis of different cavity
shapes. In Figure 4.8, when b/a < 1, which is the case of elliptic cavity increasing its size
in a x-axis direction, when b/a = 1, which is the circle shape of cavity and when b/a >1,
which is the case of elliptic cavity increasing in y-axis direction. The cylindrical cavity
has singularities at the corners and is difficult to model analytically.
________________________________________________________________________
Figure 4.7: Flow chart of the model.

________________________________________________________________________
Figure 4.8: From left to right, the spherical (circle), elliptical and side-section of
cylindrical (rectangle or square) shape of gaseous cavity.
4.3 Results and Discussion
a) Influence of cavity size and thickness of insulation
For a permittivity of insulation, r = 2.5 with a single gaseous cavity at the center of the
insulation, the effect of cavity size is shown in Figures 4.9 and 4.10. Figure 4.9 shows
result for a circular or elliptical cavity and Figure 4.10 shows result for a cylindrical
cavity. The ratio of electrical field within the cavity, E to the mean field without the void
(Eo) is plotted as a function of the ratio a/t, with a being the void dimension along the
field direction a-axis and t being the insulation thickness. The results indicate that the
increase in void stress is stronger as a/t decreases, i.e. as the thickness of insulation
increases for a given set of cavity dimensions. Moreover, the stress ratio also increases as
b/a increases, i.e. the cavity diameter increases in b-axis.
Figure 4.11 shows the field ratio for a circular or elliptical cavity with a semi-conducting
screen used with the cable. E is in the void centre and Eo is the average field in void
volume without void, as before. The result indicates that the field ratio is reduced with
the semi-conducting layer, but the reduction is very small. The difference between the
two cases varies from 0.05% to 3% with the circular cavity having the highest reduction.
________________________________________________________________________
2.5
1.5
E / Eo
0.5
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
a/t
b/a = 0.5 b/a = 1 b/a = 2 b/a = 4 b/a = 6 b/a = 8 b/a = 10
embedded in the insulation, r =2.5, effect of cavity size and insulation thickness. Top
figure shows the result in 2-D and bottom figure shows the result in 3-D.
________________________________________________________________________
2.5
2
E / Eo
1.5
0.5
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
a/t
b/a = 0.25 b/a = 0.5 b/a = 1 b/a = 2 b/a = 4 b/a = 6 b/a = 8 b/a = 10
Figure 4.10: Field ratio (E / Eo) for a cylindrical section shape of cavity embedded in the
insulation, r =2.5, effect of cavity size and insulation thickness. Top figure shows the
result in 2-D and bottom figure shows the result in 3-D.
________________________________________________________________________
2.5
1.5
E / Eo
0.5
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
a/t
b/a = 0.5 b/a = 1 b/a = 2 b/a = 4 b/a = 6 b/a = 8 b/a = 10
Figure 4.11: Field ratio (E / Eo) for a circular and elliptical shape of cavity embedded in
the insulation with semi-conducting layers have been attached.
b) Influence of cavity position

Figure 4.12 shows the cavity field ratio variation along the radial (applied field) direction
within the insulation. The field ratio is decreasing as position of void move toward the
surface of insulation and b/a decreases. The stress decreases by about 50% of the peak
field ratio over the insulation width. When the values of b/a are over 6, the rate of
increase of stress becomes slower. The reason is that when b/a > 6, the cavity size is quite
large compared with the size of cable and covers more than ¼ of the area of the whole
cable insulation.
________________________________________________________________________
3.5
2.5
2
E / Eo
1.5
0.5
10
6
0
2 b/a
6.5 7 8 9 10 11 0.5
12 13
Position of Void 14
embedded in the insulation, effect of cavity position with its axis parallel to the applied
electrical field direction.
c) Cavity orientation
The cavity field ratio as a function of cavity orientation with respect to the radial
direction (x-axis) is shown in Figure 4.13. The stress in the cavity reaches a maximum
when the cavity axis is parallel to the applied field, (i.e. = 90o). When the value of b/a is
less than 1, the maximum stress can be reached at = 0o. Moreover, the maximum stress
change is seen to be restricted to a ratio of 2.
________________________________________________________________________
2.5
1.5
E / Eo
0.5
0
0 10 20 30 40 50 60 70 80 90
Degree
b/a = 0.3 b/a = 0.5 b/a = 1 b/a = 2 b/a = 4
b/a = 6 b/a = 8 b/a = 10
Figure 4.13: Cavity Field ratio (E / Eo) for a cylindrical section shape of cavity embedded
in insulation with various cavity orientations.
d) Field variation inside the cavity

Five points (C, D, E, F and G as in Figure 4.14) were chosen for computation of the
variation of the electric field within the void. The points are on the diameter of the cavity
with equal distance between each. The direction and magnitude of the electric field
distribution is shown vectorially in Figure 4.15. The cavity is located in middle position
in the insulation (midway between the conductor and outer sheath).
The results obtained are shown in Tables 4.1 and 4.2. The weakest electric stress
occurred when b/a = 0.25 with a/t = 0.1. Reducing the a/t and b/a value reduces stress
inside the void. In addition, the results indicate the influence of stress is more pronounced
for a/t variation compared to its dependence on b/a variation: i.e. at a/t = 1, the stress
ratio is decreased 61% from point C to point G, whereas the stress ratio is decreased from
9% to 5% over the range of values of b/a.
________________________________________________________________________
Figure 4.14: Points chosen to compute stress inside the cavity.
Figure 4.15: The direction of the field distribution inside the void.
Table 4.1: Field ratio (E / Eo) at point N inside the cavity versus b/a (the cavity
parameters), while a/t=0.1
Point: C D E F G
b / a = 0.5 1.31 1.28 1.25 1.22 1.19
b/a=1 1.49 1.46 1.42 1.39 1.36
b/a=2 1.72 1.69 1.65 1.62 1.59
b/a=4 1.95 1.92 1.89 1.86 1.83
b/a=6 2.05 2.02 2.00 1.97 1.94
b/a=8 2.11 2.09 2.06 2.02 2.01
b / a = 10 2.14 2.12 2.09 2.06 2.04

________________________________________________________________________
Table 4.2: Field value (E / Eo) at point N inside the cavity with different b/a and a/t,
increasing in x-axis direction.
Point: C D E F G
a / t = 0.05 1.57 1.55 1.52 1.50 1.48
b / a = 1.9
a / t = 0.1 1.49 1.46 1.42 1.39 1.36
b/a=1
a / t = 0.2 1.38 1.31 1.24 1.18 1.13
b / a = 0.5
a / t = 0.4 1.40 1.25 1.12 1.02 0.93
b / a = 0.25
a / t = 0.6 1.50 1.26 1.07 0.93 0.83
b / a = 0.17
a / t = 0.8 1.69 1.30 1.04 0.87 0.73
b / a = 0.125
a/t=1 1.68 1.34 1.02 0.81 0.65
b / a = 0.1
e) Influence of nearby cavities

Figure 4.16 shows the field (E / Eo) between two elliptical cavities as the gap between
the two cavities increases. The increase can be up to 88%. In addition, when b/a is larger
than 6 and the gap distance is less than 1 mm, the field ratio values are similar. When the
gap distance is smaller, i.e. distance gap < 1 mm, the stress inside the cavities is weaker,
but the strong stress occurs between the gap of the two cavities in series.
________________________________________________________________________
1.6
1.4
1.2
1
E / Eo
0.8
0.6
0.4
0.2
0
0 1 2 3 4 5 6
Distance between two voids
b/a = 0.5 b/a = 1 b/a = 2 b/a = 4 b/a = 6 b/a = 8 b/a = 10
Figure 4.16: Field ratio (E / Eo) for two circle or elliptical shapes of gaseous cavities
embedded in the insulation.
f) Comparison of cavity shapes

From Figures 4.10 and 4.11, the results indicate that under the 2-D model, the cylindrical
cavity has higher stress than the spherical and elliptical types. Moreover, when b/a < 1,
which is the case of an elliptic cavity, increasing the size a gives a higher stress in the
spherical shape than in the elliptical shape.
1.8
1.6
1.4
1.2
Er (p.u.)
0.8
0.6
0.4
0.2
0
0 1 2 3 4 5 6
Distance between two voids (mm)

b/a = 0.25 b/a = 0.5 b/a = 1 b/a = 2
b/a = 4 b/a = 6 b/a = 8 b/a = 10
Figure 4.17: Differential field value in p.u. for two circle or elliptical shapes of gaseous
cavities embedded in the insulation.
________________________________________________________________________
4.4 Influence of different parameters of conducting, semi-

conducting and insulating layers
4.4.1 Influence of the cable insulation
The insulation of a cable performs like the dielectric in a capacitor with the cable
conducting elements acting as electrodes [123]. The dielectric loss is thus a function of
the square of the system voltage. Consequently, the design of cables for high A.C.
voltage is greatly influenced by the total dielectric loss, in terms of both the current rating
and the high frequency (e.g. partial discharge) signal propagation characteristics.
In the following, the effects of insulation thickness and dielectric loss factor on
attenuation and propagation velocity are presented for both XLPE and impregnated paper
insulation.
A). Thickness of the Insulation

The capacitance of a power cable is an important consideration and is dependent on the
dielectric constant (relative permittivity) of the cable insulation, as shown in Figure 2.31
in Chapter 2. The capacitance of a cable per unit length is given by [148]:
C = / 18 ln (R/r) F/km (Equation 4.5)
Where is the specific dielectric constant of the cable insulation, r is the conductor radius
including the inner semi-conducting layer, and R is the cable radius up to the outer
surface of the insulation.
The insulation resistance of cable insulation is [148]:

Ri = / 2 (ln (R/r) x 10-11 ) M-km. (Equation 4.6)
________________________________________________________________________
Where is the volume resistivity of the insulation material (-cm). It is to be noted that
the capacitance and the resistance of a coaxial cable is a parameter independent of cable
dimensions.
Seven different combinations of insulation and semi-conducting layer thickness have

been investigated. These combinations are shown in Table 4.3. Figure 4.18 shows the
results of signal attenuation for each different case. The attenuation figure is obtained by
using finite element techniques, which are introduced in Chapter 3. For overvoltage
protection design knowledge of attenuation is very important. Travelling electromagnetic
waves in power cables are strongly influenced by frequency dependent attenuation.
Lightning and switching voltages can be reduced to an innocuous level when the
attenuation level is too high in some long cable systems. The results show that the
thickness of insulation is more significant than the thickness of the semi-conducting
layers. In general, when the thickness of insulation decreases the result is higher loss and
higher capacitance per unit length. As Figure 4.18 shows, Case 2 and Case 1 have
different thickness of insulation (by 18%) but the difference in attenuation is 60%. In
general, the thinner the insulation, the higher is the attenuation. Consider Case 6 and Case
7, which have the same thickness of insulation as Case 2 and Case 4, but with different
thicknesses of semi-conducting layers. For Case 2 and Case 6, the thickness of the outer
semi-conducting layers reduces which leads to a decrease in attenuation. For Case 4 and
Case 7, the thickness of inner semi-conducting layers increases which leads to the
attenuation increasing.
________________________________________________________________________
Table 4.3: Combinations of thickness of insulation and semi-conducting layers used in

analysis.
Inner Outer
semi-con Insulation semi-con
Case (cm) (cm) (cm)
1 0.09 0.29 0.11
2 0.06 0.35 0.08
3 0.04 0.39 0.05
4 0.0154 0.4191 0.0381
5 0.01 0.44 0.02
6 0.06 0.35 0.06
7 0.0381 0.4191 0.0381
0.8
Attenuation (dB/m)
0.6
0.4
0.2
0
0 20 40 60 80 10 0
F re q u e n c y ( M H z )
Ca s e 1 C a se 2 C as e 3
Ca s e 4 C a se 5 C as e 6
Ca s e 7
Figure 4.18: Determination of signal attenuation versus frequency for the combinations
listed in Table 4.3.
B) Dielectric loss tangent

The cable dielectric properties are the major determining factor in cable design insofar as
they dictate the voltage rating, the charging current and the dielectric loss. The overall
losses (copper and dielectric) determine the current rating of a cable [109]. As the
________________________________________________________________________
operating voltage increases, so too does the dielectric power loss, as indicated by the
following formula:
Dielectric loss = V2Ctan Watts (Equation 4.7)
Where V is the cable operating voltage, C is the capacitance, is the angular frequency
(2Sf) and tan is the dielectric loss factor. The dielectric loss is proportional to the square
of operating voltage, but this increase with voltage is partly compensated by the fact that
the insulation thickness increases as the operating voltage increases and this reduces the
cable capacitance. However, this can be negligible compared to the voltage squared.
In terms of impact on cable rating, dielectric loss is generally negligible up to about 33
kV. However in terms of the signal attenuation and distortion impact, the effect covers all
voltage ratings.
Considering the effect of the dielectric material, there are two main types of cable to
consider, the 11kV XLPE cable and the paper insulated cable. XLPE cable has a
dielectric constant of about 2.2 and for paper insulated cable, it is about 3.5. Paper
insulated cable has the thicker insulation than XLPE cable. The simulations in this
section use the FEM cable model, which is introduced in Section 3.61 and Section 4.2.
Figure 3.19 shows the geometrical configuration of cable. By changing the cable
geometry and dielectric constant, the simulated results show in Figure 4.20 to 4.25.
Figure 4.19 and Figure 4.20 present the results of signal attenuation and phase constant
(E) versus frequency for XLPE cable. XLPE cable has the thickness of insulation, 0.35cm
with inner semi-conducting layer, 0.06cm and outer semi-conducting layer, 0.08cm. In
addition, Figure 4.19 gives attenuation for a range of tanG values from 0.01 to 0.0005.
The signal propagation velocity, c is determinable from the phase constant and related to
the 2 frequency. XLPE cable usually has the low dielectric loss factor such as 0.0001.
Comparing the dielectric loss factor, 0.0001 and 0.001 at 50 MHz, the attenuation is
increased 20% at dielectric loss factor, 0.001.
________________________________________________________________________
Figure 4.21 and Figure 4.22 give similar plots for paper insulated cable, except that the
range of tanG values is different (0.1 to 0.001) to account for the higher base value of
paper tanG compared to XLPE tanG.
Comparison of Figure 4.19 and Figure 4.21 indicates that with increase of the
permittivity of the dielectric the attenuation of signal propagation increases for the same
loss factor. Comparison of Figure 4.20 and Figure 4.22 shows that the phase velocity is
slightly increased when the permittivity is increased with loss factor, 0.0001 for Figure
4.20 and loss factor, 0.001 for Figure 4.22.
The value of loss tangent, tan is greatly dependent on the type of insulating material and
on the condition of the dielectric material [8] and in particular whether it has deteriorated
due to service aging. Service ageing and/or contamination of electrical insulation will
increase the tanG of any insulation. In Figure 4.19, the dielectric loss tangent (tan ) is
varied from 0.01 to 0.0005. Usually a tan value of 0.01 would not apply even for aged
XLPE as the usual (new) value is about 0.0001. Figure 4.19 indicates that the results for
tan values of 0.0001 and 0.0005 are close each other and having an attenuation of 0.17
dB/m at 50 MHz.
For the paper insulated type cable, the loss tangent is higher than that of XLPE cable so
that the range of tan values used is 0.1 to 0.001. As with XLPE, service ageing will
increase tanG. A tan value of 0.1 would be very unlikely to occur in paper insulated
cables. Figure 4.20 indicates that the attenuation values for tan values of 0.1 and 0.04
are very high. The results indicate that other loss tangents in the range 0.01 to 0.001 have
around attenuation values of about 0.29 at 50 MHz.
Figure 4.23 and Figure 4.24 show high frequency signal attenuation and phase constant
as a function of frequency for Type I cable and Type II cable. Assume both types of
cables have the same permittivity and conductivity value of semi-conducting and
________________________________________________________________________
insulating layers (Both types of cables have dielectric constant of 3.5). The only different
between both types of cables is the structure of cable. The main purpose of Figure 4.24 is
to investigate the influence of structure of cable. The results indicate the Type I cable
with the structure of XLPE have higher attenuation than Type II cable with the structure
of paper insulated cable.
The results show that dielectric loss has a great influence on the signal propagation
characteristics of power cables. In particular the attenuation of signals such as PD signals
would be much higher in paper insulated cables than in XLPE cable.
The results also show that the phase constant (and hence velocity) is not much affected
by the tanG value. The main effect on velocity comes from the semi-conducting layers.
0.25
0.2
Attenuation (dB/m)
Tan: 0.01
0.15 Tan: 0.001
Tan: 0.005
0.1 Tan: 0.0001
Tan: 0.0005
0.05
0
0 10 20 30 40 50
Frequency (M Hz)
Figure 4.19: High frequency signal attenuation as a function of frequency for XLPE cable
for a range of tan values.
________________________________________________________________________
2.5
2
Phase constant (rad/m)
1.5
0.5
0
0 10 20 30 40 50
Frequency (M Hz)
Figure 4.20: High frequency signal phase constant as a function of frequency for XLPE
cable (tan = 0.0001).
0.6
0.5
Tan: 0.1
Attenuation (dB/m)
0.4 Tan: 0.04

Tan: 0.01
0.3
Tan: 0.005
0.2 Tan: 0.0014
Tan: 0.001
0.1
0
0 10 20 30 40 50
Frequency (M Hz)
Figure 4.21: High frequency signal attenuation as a function of frequency for paper-
insulated cable for a range of tan values.
________________________________________________________________________
2.5
Phase constant (rad/m) 2
1.5
0.5
0
0 10 20 30 40 50
Frequency (M Hz)
Figure 4.22: High frequency signal phase constant as a function of frequency for paper-
insulated cable (tan = 0.001).
0.4
0.35
0.3
Attenuation (dB/m)
0.25
Type I cable
Type I
0.2 Type II cable
Type II
0.15
0.1
0.05
0
0 10 20 30 40 50
Frequency (M Hz)
Figure 4.23: High frequency signal attenuation as a function of frequency for Type I
The only different between both types of cables is the structure of cable.
________________________________________________________________________
2.5
Phase constant ß (rad/m)
2
Type I cable
Type
Type II Icable
1.5
Type II
1
0.5
0
0 10 20 30 40 50
Frequency (M Hz)
Figure 4.24: High frequency signal phase constant as a function of frequency for Type I
The only different between both types of cables is the structure of cable.
4.4.2 Influence of skin effect

The skin effect has practical consequences in the design of radio-frequency and
microwave circuits and to extent in AC electrical power transmission and distribution
systems. Also, it is of considerable importance when designing power cable. Skin effect
is a phenomenon which accounts for the increase in resistance of a conductor due to self-
inductance. This effect causes the current density in the conductor to be higher towards
the outer surface.
The current density J in an infinitely thick plane conductor decreases exponentially with
depth d from the surface, as follows [120]:
J = Jse-d/ (Equation 4.8)
________________________________________________________________________
Where is a constant called the skin depth. This is defined as the depth below the surface
of the conductor at which the current density decays to 1/e (about 0.37) of the current
density at the surface (JS). It can be calculated as follows [120]:
2U
G (Equation 4.9)
ZP
Where
= resistivity of conductor
= angular frequency of current = 2 × frequency
= absolute magnetic permeability of conductor = 0 r, where 0 is the
permeability of free space (4×107 N/A2) and r is the relative permeability of
the conductor.
It is easy to estimate the skin depth from the conductivity, permeability, and frequency.
Table 4.4 lists the skin depth for the different materials at a frequency of 50 Hz.
Table 4.4: The skin depth for different materials at frequency of 50 Hz.
Material Rel. Permeability Conductivity (S/m) Skin Depth (mm)

Copper 1 5.998 x 107 9
Aluminum 1 3.775 x 107 12
Stainless Steel 1 2.247 x 106 67
Iron 4000 1.120 x 107 0.34
In copper, the skin depth at various frequencies is shown below (Table 4.5) and Figure
4.25 shows the skin depth of various materials (e.g. copper, Aluminum, iron and steel) as
function of frequencies from 0.001 to 10M Hz. The basic impact of skin effect when the
skin depth is less that the conductor size is that it will increase effective resistance and
losses and thus cause significant effect on HF signal attenuation. Proximity effect due to
other phases in a three core cable can also have a similar effect.
________________________________________________________________________
Table 4.5: The skin depth for copper materials at various frequencies.
Frequency Skin Depth (mm)

50 Hz 9 mm
60 Hz 8.47 mm
10 kHz 0.66 mm
100 kHz 0.21 mm
1 MHz 66 μm
10 MHz 21 μm
Figure 4.25: Skin depths for some metals.
Figure 4.26 and Figure 4.27 show the skin depth of XLPE power cable as function of
frequency and conductivity. The skin effect becomes increasingly important at the higher
frequencies and higher conductivity. Skin effects can usually be disregarded for cable
smaller than 150mm2 cross section. However, the effect becomes significant for the
larger conductors: the larger size of conductor has smaller resistivity, which can lead to
high skin effect. Skin effect can be significantly reduced by the ‘Milliken’ type conductor
________________________________________________________________________
design. In this design, the cable conductors are divided into several segments. Each
segment consists of layers of wires stranded around a centre and hence over one lay
length of these wires the average distance of each stranded from the centre of cable is
nearly equal. The use of enameled stranded wires prevents the passage of current from
one strand to another and equalizes the self-inductance of each individual strand, thus
minimizing the skin effect.
100
90
80
70
Skin Depth (m)
60
1M Hz
10M Hz
50
30M Hz
40
50M Hz
30
20
10
0
0.00001 0.001 0.1 10 1000 100000
Conductivity
conductivity.
________________________________________________________________________
100
10
1
Skin Depth (m)
0.00001 0.001 0.1 10 1000 100000 1M Hz

10M Hz
0.1
30M Hz
50M Hz
0.01
0.001
0.0001
Conductivity
conductivity in the logarithmic scale.
________________________________________________________________________
4.5 Influence of Semi-conducting Layers
A PD simulation of a cable is easily carried out by a transient program such as EMTP,

provided that all the required data for the PD simulation are evaluated by the subroutine
cable parameters of EMTP program. One of the major factors that affect the cable PD
propagation characteristics significantly is a semi-conducting layer [3, 52]. Thus it is
necessary to detail carefully the semi-conducting layer electrical properties. The semi-
conducting layer can have various permittivity and conductivity. According to IEEE
standard, the inner semi-conducting layer has higher conductivity than outer semi-
conducting layer. For example, the measured results of conductivity give the inner semi-
conducting layer, 0.2-0.333 S/m and outer semi-conducting layer, 0.0349-0.045 S/m.
XLPE insulated cables consist of a copper or aluminum conductor, an inner semi-

conducting layer extruded over the inner conductor, an extruded XLPE-insulation layer,
an outer semi-conducting layer (usually co-extruded with the XLPE), a metallic wire
screen or aluminum sheath and an outer polyethylene sheath. The extruded semi-
conducting layers provide a well-bonded and smooth interface between the conductors
and the dielectric, avoiding electric field strength enhancements, which would lead to
partial discharge (PD) and a premature breakdown.
4.5.1 The Parameter of Semi-Conducting Layers
Traveling electromagnetic waves such as PD pulses in power cables are strongly

influenced by frequency- dependent attenuation. An in-depth analysis of the attenuation
of the transmission line is very important. In particular, the XLPE cable, composed of
insulating and semi-conducting materials, is very important for the attenuation of high
frequencies. Due to the significance of semi-conducting layers on propagation
characteristics found in the investigation, knowledge of the conductivity and permittivity
and their dependence on frequency is important to optimize cable design and for other
________________________________________________________________________
applications. This section presents the results of the relationship between six different
factors that influence PD signal propagation: attenuation, phase constant and velocity as
functions of frequency, conductivity and permittivity of semi-conducting layers.
The results can be divided into three sections: Section A (Figure A1.1 to Figure A1.10 in
Appendix I) used the measured results of semi-conducting layers; further detail on
measurement method can be found in Section 5.5.3.; Section B (Figure A1.11 to Figure
A1.14 in Appendix I) used the Boggs’ (permittivity and resistivity values of semi-
conducting layers) measurement of semi-conducting layers in [8] and Section C (Figure
A1.15 to Figure A1.26 in Appendix I) used Steinbrich’s measurement of semi-conducting
layers in [124] (permittivity and resistivity values of semi-conducting layers can be found
in Section 2.2.6).
a) Figure 4.28 and Figure 4.29 show the results of determinations of attenuation and
velocity as functions of frequency (from 100 kHz to 100 MHz) with the conductivity of
inner/outer semi-conducting layer in the range from 100000 to 0.00001. In addition the
permittivity, and conductivity of other semi-conducting layer (outer/inner semi-
conducting layer) are set to be constant for each calculation. For example, the
conductivity of inner semi-conducting layer is 0.3 S/m and conductivity of outer semi-
conducting layer is 0.037 S/m. The relative permittivity value of semi-conducting layers
in Figure 4.28 and Figure 4.29 is 1.0, the minimum possible value. For other permittivity
values of semi-conducting layer (between 1 and 100) the results can be found in
Appendix I.
b) Figure 4.30 to Figure 4.33 show the result of attenuation, velocity and phase constant
as function of frequency (from 5 MHz to 50 MHz) with the conductivity of inner/outer
semi-conducting layer in the range from 100000 to 0.00001. In addition the permittivity,
and conductivity of other semi-conducting layer (outer/inner semi-conducting layer) are
________________________________________________________________________
based on the Boggs’s frequency dependent measurement of semi-conducting layers in

[8]. Boggs gives the frequency dependent measurement of dielectric constant, loss angle
and conductivity of semi-conducting materials. The measurements show the measured
values are varied at different frequencies.
c) Figure 4.34 to Figure 4.45 show the result of attenuation, velocity and phase constant
as function of frequency (from 5 MHz to 30 MHz) with the conductivity of inner/outer
semi-conducting layer in the range from 100000 to 0.00001. In addition the permittivity,
and conductivity of other semi-conducting layer (outer/inner semi-conducting layer) are
based on the K. Steinbrich’s frequency dependent measurement of semi-conducting
layers in [124]. K. Steinbrich included two types of semi-conducting layers in [124],
which are carbon-polyethylene compound (Figure 4.34 to Figure 4.39) and conductive
paper layer (Figure 4.40 to Figure 4.45). The measured results can be found in Section
2.2.6 and Section 2.2.7 in Chapter 2.
The results show the attenuation is highest when the conductivity of the semi-conductor
material is 0.01 and at high frequency such as 100 MHz. The highest velocity occurs at
the highest conductivity of semi-conducting layers and the lowest velocity occurs when
the conductivity of semi-conducting layers is between 0.1 and 10 Siemens.
When the inner semi-conducting layer conductivity is changed and outer semi-conducting
layer conductivity kept constant, the graph of attenuation appears to have two peaks at
conductivities equal to 100 and 0.01 (the highest attenuation). In addition the velocity is
decreased at conductivity from 0.1 to 10000 S/m. Contrarily, when the outer semi-
conducting layer conductivity is changing and the inner semi-conducting layer
conductivity is kept constant, the graph of attenuation displays only one peak point at the
conductivity equal to 0.01. In addition the velocity is increased at when conductivity
from 10 to 100000 S/m with the lowest velocity at minimum conductivity, which is
shown as the minimum on the “U” curve in the diagram.
________________________________________________________________________
When using the semi-conducting sample of Boggs [8] and K. Steinbrich [124], the
attenuation calculated is lower than in the other cases, such as Figure 4.34 to Figure 4.45.
However, the trend line on the graph of attenuation and velocity are still the same.
0.7
0.6
0.5
Attenuation (dB/m)
0.4
0.3
100M Hz
0.2 80M Hz
6 0M Hz
0.1 50M Hz
40M Hz
0 20M Hz
1E- 0 5 5 M Hz
1
100000
Conductivity

permittivity, of inner semi-conducting layer is equal to 1.
________________________________________________________________________
2 00
150
Velocity (m/s)
100
50
0
0 .0 0001 0.00 1 0.1 10 1000 100 000
Conduc t iv it y
5M Hz 20 M Hz 40M Hz 50M Hz 60M Hz 80M Hz 100M Hz
outer semi-conducting layers in the range from 100000 to 0.00001. The permittivity, of
inner semi-conducting layer is equal to 1.
0.25
0.2
Attenuation (dB/m)
0.15
0.1
0.05 50M Hz
30M Hz
0 10 M Hz
1E- 05
1 5M Hz
100000
C o n d u c ti vit y

semi-con measurement [63].
________________________________________________________________________
200
150
Velocity (m/s)
100
50
0
0 .00001 0.001 0.1 10 100 0 100000
Conduct ivit y
5M Hz 10M Hz 30M Hz 50M Hz
conductivity of outer semi-conducting layer is based on Steven Boggs’s semi-con
measurement [63].
0.25
0.2
Attenuation (dB/m)
0.15
0.1
0.05 50M Hz
30M Hz
0 10M Hz
0.00001 5M Hz
1
100000
C o nd u c t ivity

________________________________________________________________________
200
150
Velocity (m/s)
100
50
0
0 .00001 0.001 0.1 10 100 0 100000
Conduct ivit y
conductivity of inner semi-conducting layer is based on Steven Boggs’s semi-con
measurement [63].
________________________________________________________________________
The following results are using the semi-conducting layer measurement results from K.
Steinbrich [124]
1
0.9
0.8
Attenuation dB/m
0.7
0.6
0.5
0.4
0.3
0.2
30
0.1
20
0
10 Frequency (MHz)
100000
1000
1 5
0.01
Conductivity 0.00001

compound.
1.4
1.2
1
Phase constant
0.8
0.6
0.4
0.2
0
0.00001 0.001 0.1 10 1000 100000
Conductivity
compound.
________________________________________________________________________
190
185
180
175
Velocity (m/μs)
170
165
160
155
150
145
140
0.00001 0.001 0.1 10 1000 100000
Conductivity
conductivity of outer semi-conducting layer is carbon-polyethylene compound.
________________________________________________________________________
0.45
0.4
0.35
Attenuation dB/m
0.3
0.25
0.2
0.15
0.1
30
0.05
20
0
10 Frequency (MHz)
100000
1000
1
0.01 5

compound.
1.4
1.2
1
Phase constant
0.8
0.6
0.4
0.2
0
0.00001 0.001 0.1 10 1000 100000
Conductivity
compound.
________________________________________________________________________
190
180
170
Velocity (m /μs)
160
150
140
130
120
0.00001 0.001 0.1 10 1000 100000
Conductivity
conductivity of inner semi-conducting layer is carbon-polyethylene compound.
________________________________________________________________________
0.5
0.45
0.4
0.35
Attenuation dB/m
0.3
0.25
0.2
0.15
0.1
30
0.05
20
0
10 Frequency (MHz)
100000
1000
1
0.01 5

permittivity and conductivity of outer semi-conducting layer is conductive paper.
1.4
1.2
1
Phase constant
0.8
0.6
0.4
0.2
0
0.00001 0.001 0.1 10 1000 100000
Conductivity
________________________________________________________________________
190
185
180
Velocity (m/μs)
175
170
165
160
155
150
0.00001 0.001 0.1 10 1000 100000
Conductivity
conductivity of outer semi-conducting layer is conductive paper.
________________________________________________________________________
0.5
0.45
0.4
0.35
Attenuation dB/m
0.3
0.25
0.2
0.15
0.1
30
0.05
20
0
10 Frequency (MHz)
100000
1000
1
0.01 5

permittivity and conductivity of inner semi-conducting layer is conductive paper.
1.4
1.2
1
Phase constant
0.8
0.6
0.4
0.2
0
0.00001 0.001 0.1 10 1000 100000
Conductivity
________________________________________________________________________
190
185
180
Velocity (m /μs)
175
170
165
160
155
150
0.00001 0.001 0.1 10 1000 100000
Conductivity
conductivity of inner semi-conducting layer is conductive paper.
________________________________________________________________________
4.5.2 The Thickness of Semi-Conducting Layers
In the previous Section 4.5.1, the influence of semi-conducting layer electrical parameters
have been discussed. This section presents results concerning the significance of
changing the thickness of the semi-conducting layers. Table 4.6 shows the different
thickness of semi-conducting layers used in the calculations. Seven different
combinations of insulation and semi-conducting layer thickness have been investigated.
The permittivity and conductivity of inner and outer semi-conducting layer is based on
K.Steinbuich semi-con measurement [124]. The conductivity of inner semi-conducting
layer is greater than outer semi-conducting layer. Figure 4.46 to Figure 4.48 show the
result of attenuation, velocity and phase constant as function of frequency for each case
is based on K.Steinbuich semi-conducting measurement [124]. Further details on
measured results can be found in Chapter 2. The outer semi-conductivity layer is
assumed to be same material as the inner semi-conductivity layer.
Figure 4.46 shows that Case 2 and Case 1 have different thickness of semi-conducting
layers (by 30%) but the difference in attenuation is 60%. Comparing Case 2 and Case 6,
both cases have the same thickness of insulation but the outer semi-conducting layer of
Case 2 is thicker, which gives the higher attenuation and slight higher phase constant.
Figure 4.49 to Figure 4.51 show the result of attenuation, velocity and phase constant as
function of frequency for each case in Table 4.6. Consider Case 6 and Case 7, which have
the same thickness of insulation as Case 2 and Case 4, but with different thicknesses of
semi-conducting layers. For Case 2 and Case 6, the thickness of the outer semi-
conducting layers is reduced, which leads to the attenuation decreasing. For Case 4 and
Case 7, the thickness of inner semi-conducting layers increases which leads to the
attenuation increasing.
________________________________________________________________________
Table 4.6: Thickness of and semi-conducting layers used in analysis. The permittivity and
conductivity of inner and outer semi-conducting layer is based on K.Steinbuich semi-con
measurement.
Case Inner semi-con Insulation Outer semi-con

1 0.09 0.29 0.11
2 0.06 0.35 0.08
3 0.04 0.39 0.05
4 0.0154 0.4191 0.0381
5 0.01 0.44 0.02
6 0.06 0.35 0.06
7 0.0381 0.4191 0.0381
0.9
0.8
0.7 Case1
Attenuation (dB/m)
Case2
0.6
Case3
0.5
Case4
0.4 Case5
0.3 Case6
0.2 Case7
0.1
0
0 50 100
Frequency (M Hz)
based on K.Steinbuich semi-con measurement [124].
________________________________________________________________________
5.5
4.5 Case1
Case2
Phase Constant
3.5 Case3
Case4
2.5 Case5
Case6
1.5 Case7
0.5
0 50 100
Frequency (M Hz)
is based on K.Steinbuich semi-con measurement [124].
175
Case1
165
Case2
Case3
Velocity
155 Case4
Case5
Case6
145 Case7
135
0 50 100
Frequency (M Hz)
based on K.Steinbuich semi-con measurement [124].
________________________________________________________________________
0.25
0.2 Case1
Attenuation (dB/m)
Case2
0.15 Case3
Case4
0.1 Case5
Case6
0.05 Case7
0
5 10 15 20 25 30
Frequency (M Hz)
K.Steinbuich semi-con measurement [124] and assume that outer semi-conducting layer
employs the same material as the one employed in the inner semi-conducting layer.
1.5
Case1
Phase Constant
Case2
1
Case3
Case4
Case5
0.5 Case6
Case7
0
5 10 15 20 25 30
Frequency (M Hz)
in Table 4.6. The permittivity and conductivity of inner semi-conducting layer is based
on K.Steinbuich semi-con measurement [124] and assume that outer semi-conducting
layer employs the same material as the one employed in the inner semi-conducting layer.
________________________________________________________________________
175
Case1
165
Case2
Case3
Velocity
155 Case4
Case5
Case6
145 Case7
135
5 10 15 20 25 30
Frequency (M Hz)
K.Steinbuich semi-con measurement [124] and assume that outer semi-conducting layer
employs the same material as the one employed in the inner semi-conducting layer.
4.6 Conclusion
In general, the analytical method can calculate the electric field strength in the cavity and
dielectric. However, the field distribution in the vicinity of a cavity of non-standard shape
is difficult to calculate by analytical methods. In this Chapter a numerical method is
discussed for the field calculation. The results show the important influence of the shape
of cavity on the electric stress within it. As is well-known, the stress in the void cavity is
enhanced and this is an important consideration for PD activity. The cavity stress
enhancement increases as the permittivity of the dielectric increases. The increase is
greater for cavities with large diameter to thickness ratios. A cavity with its axis parallel
to the applied field direction has a higher stress enhancement. The stress in a cavity is
influenced by the presence of any nearby cavities in the radial direction and the stress
________________________________________________________________________
increases with increase in the gap between the two cavities. The increase in stress is more
pronounced with change of a/t than with variation of b/a. and the stress distribution in the
cavity is smaller for spherical voids than for cylindrical types
The above analysis provides a reasonable indication of the skin depth for different
materials at various frequencies. Copper at 10 MHz has a skin depth of 21 μm. At a
conductivity of 0.00003 the skin depth is higher and is independent of frequency as
shown in Figure 4.26. This chapter also investigated the factors such as PDs in cables and
the impact of skin effect, dielectric loss of insulation, thickness of semi-conducting layers
and the conductivity and permittivity for semi-conducting materials. The results show
that dielectric loss of the insulated material has great influence on the signal propagation
characteristics of power cable. XLPE cable with tan = 0.0001 will have attenuation of
0.18 dB/m at 50 MHz, compare with the paper insulated cable, which give the attenuation
of 0.28 dB/m. The value of tan of cable insulation is dependent on the type of insulating
material and its condition and this will thus have an effect on PD propagation. This is of
particular interest in the use of PD diagnostics of service-aged cable, where the loss
factor may be increased due to ageing.
The higher loss factor of impregnated paper as compared to XLPE means that PD signal
attenuation along paper insulated cables will be more severe than in XLPE insulated
cable. Thickness, permittivity and conductivity of semi-conducting layer have dominant
effect on attenuation and velocity. Conductivity of semi-conducting layer at 0.01 can give
the higher attenuation, which is up to 1.2 dB/m. The lowest velocity (150-160 m/ s) can
be obtained with the conductivity of semi-conducting layer closed to 0-0.1. In general,
when the thickness of semi-conducting layers increases the result is higher loss. The
phase constant (
) and attenuation () have great reduction when the materials of inner
and outer semi-conducting layers are the same. The effect of semicon electrical
characteristics on velocity is also an important consideration in doing site location
calculations by time of flight measurements.
Chapter 5: High Frequency Pulse Measurements and Analysis 145
________________________________________________________________________
Chapter 5
High Frequency Pulse Measurements and

Analysis
5.0 Introduction
As previously mentioned in Chapter 2, partial discharge (PD) measurement and location

has become an essential part of insulation condition monitoring of high voltage cables.
However, such PD detection and location techniques require the monitoring of very high
frequency pulses as they propagate through the cable. Such pulses will suffer attenuation
due to lossy cable insulation and this will then complicate measurement and location of
the PD source.
5.1 Characteristics of Power Cable
Modern power cables are suffering increased loads and this presents an increased
problem to their insulation viability. In particular, to the generation of partial discharge
activity that can degrade the insulation. Partial discharge (PD) activity in power cables is
caused by a number of possible defects, such as voids and water trees that have
progressed to the electrical tree stage in cable insulation [2]. The PDs will gradually
degrade the insulation material, eventually leading to full and final insulation breakdown.
While the older style paper-insulated cables (e.g. PILC cable) tolerate PDs of moderate
level for long periods, modern cross linked polyethylene (XLPE) power cables are very
________________________________________________________________________
sensitive to any level of partial discharge activity. Because XLPE cables are increasingly
being used due to a number of technical and economical advantages, it is imperative that
they not be subject to any PD activity for any length of time. This requires monitoring of
PD activity and this in turn requires knowledge of the PD propagation characteristics in
such cables.
5.2 Purpose of the test
Partial discharges in solid insulation can generate electrical signals with frequencies up to
several hundred MHz. The ability to detect and locate a PD source is limited by
attenuation of the high frequency PD pulses as they propagate through the cable.
Therefore it is necessary to understand the high frequency response of such cables.
Further, to enable reconstruction of PD signals as emitted a viable high frequency model
for simulation is needed. The aim of the work outlined in this Chapter is to develop an
Cable model and understanding of the propagation characteristics so that the original site
of the PD and the original emitted signal can be determined. This work involved
experimental tests and development of a simulation model able to represent faithfully the
cable characteristics in electrical PD signal analysis.
5.3 Test Setup for Cable Measurements
Three cables of different length were employed in the tests. Cable A and cable B were
11kV single core XLPE insulated cables with semi-conducting layers. Cable A had
length of 136 meters and cable B 45 meters. Cable B has a much thinner inner semi-
conducting layer (0.7mm thickness) than Cable A (1.2 mm thickness). Further detail is
shown in Table 5.1. Cable C was 22kV single core PILC cable with the cable length of
3.5m. Those three cables have the copper sheath out of insulation. Three types of
measurements were performed for the comparison with the cable model predictions.
________________________________________________________________________
The first test used a high frequency sinusoidal voltage source to provide variable
frequency signals applied to the cable. The test setup is shown in Figure 5.1. The signal
was applied between the conductor at the left side (the sending end) and the (earthed)
cable sheath. The other end of the cable (the receiving end) has the conductor short-
circuited to the sheath (earth). The measurement of the sending and receiving end current
signals was done using two commercial high frequency current transformers (HF-CTs).
PD monitoring normally uses such HF-CTs to monitor current pulses associated with the
PDs in the sheath-earth connection. The Lecroy 9362C oscilloscope is used for
measurement and 500 MS/s sample rate is used. The first test was simply a measure of
attenuation degree and propagation velocity in the cables.
The second test involved a measurement of the voltage response on the outer and inner
semi-conducting layers of the cable. As the layers represent a capacitively-coupled
connection, their voltage will be related to the signal voltage level. The purpose of this
test was to determine the viability of using the semi-conducting layers as capacitive
probes to detect PD signals. This test used the sinusoidal voltage source at the sending
end with a short circuit at the receiving end of the test cable.
In the final test, a standard PD calibrating signal source was used to inject a simulated PD
pulse into the cable to simulate the practical situation. The sending and receiving end
current signals were monitored with the HF-CTs to determine the real attenuation level of
the signals. The receiving end was short-circuited again.
________________________________________________________________________
Figure 5.1: Laboratory test measurement setup
1. Signal generator, 2. Inner semi-conducting layer, 3. Outer semi-conducting layer,

4. Sheath, 5. (HF-CTs), 6. Oscilloscope (Lecroy 9362C, maximum sample rate: 10 GS/s),
7. Computer.
5.4 Sensor Selection
PD sensor is defined as a system, which is able to detect the PD signals, and is comprised
of a coupling device. The PD sensor selection is important because PD and noise are
greatly affected by the sensor characteristic. The optimal sensor has to be high frequency
and broadband as PD pulse is in the high frequency band such as few hundred MHz: also
the sensor needs to have good sensitivity to capture the small PD signal against the
background noise. In addition, the installation of sensor is expected to be easily done
without interruption to normal service.
5.4.1 Sensor Types
The type of PD sensors used for cables can be divided into three groups according their
coupling mechanism to the power cable. (i) Capacitive sensor which couples through
electric field, (ii) inductive sensors which couple through the magnetic field, and (iii)
directional couplers which couple through both electric and magnetic field. Usually
directional couplers are installed between the insulation and metallic screen of cable. The
________________________________________________________________________
directional technique can be seen as invasive and so the directional couplers were not
considered as part of the scope of thesis. For the purpose of PD detection, the capacitive
sensor and inductive sensor were used to measure voltage or current, respectively. The
positions of different type of sensors on the power cable are shown in Figure 5.2.
Capacitive sensor couples to the cable conductor Sheath champing sensors capture the
through electric field between the sensor and signal directly from the metical screen.
cable conductor.
Rogowski coil and Current

transformer couples through
the magnetic field induced
from currents in the ground
wire.
Figure 5.2: Different type of sensors and their position on power cable.
Capacitive sensor
The Capacitive sensor is used for PD offline and online diagnostics on the power cables.
The capacitive sensor is formed by a conductive layer with a main capacitance to the
conductor and a stray capacitance to the screen. The capacitive sensor picks up PD
signals through the electric field caused by the PD and transfers it to the data acquisition
unit. The principle is shown in Figure 5.3.
The main advantage of using capacitive sensors is that its production and use is very
simple and cheap. In addition, the capacitance can be selected to be relatively high,
providing good sensitivity. The sensor can be applied to any cable independent of the
screen wires or HV termination design.
________________________________________________________________________
Figure 5.3: The principle of capacitive sensor.
Sheath champing sensor

This type of sensor is attached around the cable sheath and PD signals can be picked up
at the sheath of cable and access can be easily provided at the terminals and also at some
specific distances from the terminals, i.e. this type of sensor can be applied at a desired
number of points along the length of the cable. This sensor can provide a practical and
sensitive monitoring system for life management and reliable operation of cables at a
reasonable cost. The sensor picks up the PD signal though the series impedance. This
type of sensor is either made by aluminum or copper and the practical of sensor is shown
in Figure 5.4. This type of sensor can give high sensitivity due to it attached on the sheath
directly.
Figure 5.4: The sheath champing sensor.

________________________________________________________________________
Axial magnetic field sensors

The axial magnetic field sensor is designed to be used only on cables with helical wire
shielded construction. The axial magnetic field is generated by current flowing in the
helical structure of the metal screen. Return current Is flowing in the metal screen can be
decomposed into axial Iz and radial I components. Axial magnetic field Hz resulting
from the current I induces a voltage in the axial magnetic field sensor. The principle of
the sensor is shown in Figure 5.5. Figure 5.6 shows the lumped element model of axial
magnetic field sensor. Where Z0 is the characteristic impedance of power cable, M is the
mutual inductance between the helical metal screen and the sensor, L is the self
inductance of the sensor, C is the capacitance between metal screen and the sensor and
Rm is measuring equipment impedance. The sensor is a coil, which is applied over the
jacket of cable.
Figure 5.5: The principle of axial magnetic field sensor.
Figure 5.6: lumped element model of axial magnetic field sensor.

________________________________________________________________________
This type of sensor can be either made by several turns of coaxial wire or by a simple
copper strap as shown in Figure 5.7 (a) and Figure 5.7 (b) respectively. The main
advantage of the axial magnetic field sensor is its wide bandwidth and low production
cost. The possibility to move the sensor along the cable during the diagnostics can
sometimes be useful. However, this type of sensor has low sensitivity which is caused by
the small mutual inductance, M. It was not used in the work for this thesis.
(a) (b)
Figure 5.7: the axial magnetic field sensor: (a) coaxial wire and (b) metal strap.
High frequency current transformer
A CT is an inductive sensor can be designed to measure the current in the conductor, in

the screen or in both. The CT inductive sensor is sensitive to the magnetic field
component of the current and the magnitude of the output for a given PD is dependent on
the mutual inductance and hence of the permeability of the magnetic core. This type of
sensor can pick up the PD current that propagates through the earth conductor to earth.
The magnitude can be adjusted by varying the number of turns of the winding. The
configuration of the CT inductive sensor is shown in Figure 5.8. The current transformer
has an iron core that usually improves the coupling efficiency. In the PD measurement,
the high frequency current transformer (HFCT) is normally adopted and this has ferrite
core to reduce losses and increase the HF bandwidth. The practical current transformer is
________________________________________________________________________
shown in Figure 5.9. The disadvantage of this type of sensor is the high cost of
commercial units, but it is possible to make such CTs with good characteristics and this
has been done in the HV laboratory.
Figure 5.8: The principle of inductive sensor.
B C
Figure 5.9: The high frequency current transformers (HFCTs). HFCTs-A (Model: Ailtech
80434) has the lower sensitivity compared to others but HFCTs-A can give much more
stable performance; HFCT-B is made in the lab of the University of New South Wales: it
can overcome the commercial HFCT’s limitation of cost; HFCTs-C (Model: Altech
93511-1) gives the highest sensitivity and is suitable for the HV testing.
________________________________________________________________________
Rogowski coil
The Rogowski coil basically consists of a winding wound on a toroid shape core, usually
of non-magnetic material. The current carrying conductor goes though the center of
toroid and the magnetic field created by the current circulates around the conductor and
in the toroid core. The magnitude field in toroid core induces the voltage in the Rogowski
coil windings. Rogowski coil is usually shielded by metallic enclosures in order to
provide the shielding from noise interference and to form the constant capacitance. The
Rogowski coil can be optimized for the required bandwidth and the lack of losses in the
core make its bandwidth very high, but at the expense of signal sensitivity.. The coils can
be clamped on the operating cable but it gives rise to safety issue. Figure 5.10 shows the
principle of Rogowski coil. Figure 5.11 shows the frequency response of Sheath
champing sensor, Green Sensor "A" (Model: Aitech 80434), Silver Sensor "B" (Made in
the lab) and Black Sensor "C" (Model: Aitech 93511-1) in Figure 5.9. The result shows
sheath champing sensor as having the greatest sensitivity and green sensor "A" having
lowest sensitivity, which is more suitable for the low voltage offline experiment test.
Figure 5.10: The principle of Rogowski coil.

________________________________________________________________________
0
-10 0 20 40 60 80 100
-20
-30
Ratio in dB
-40
-50
-60
-70
-80
Frequency (M Hz)
Black sensor “C”

Black (Model:inAiltech
Sensor"C" 93511-1) Silver Sensor"B"
Figure 5.13 Silver sensor “B” 5.13
in Figure (Made in Lab)
Sheath champing sensor Green Sensor"A" in Figure 5.13
Sheath champing sensor Green sensor “A” (Model Ailtech 80434)
Figure 5.11: Frequency response of four different sensors.
5.5 Measurement of Semi-Conducting Material
As mentioned in the previous sections, semi-conducting layers have the dominant role on
propagation. In order to have accurate cable model, precise parameter value need to
include such as permittivity of semi-conducting layers. The parameter value of semi-
conducting layers is varied from various manufactures so the measurement of semi-
conducting material is need.
5.5.1 Cable Sample
An 11 kV XLPE cable is investigated and all the semi-conducting samples measured are
obtained from it. The geometric arrangement of this cable is shown in Figure 5.1. The
dimensions of the cable are shown in Table 5.1:
________________________________________________________________________
Table 5.1: Detail of cable examined.
Parameter XLPE cable

Conductor Ø 20.1 mm, Aluminum
Inner semi-conducting layer Ø 21.3 mm
XLPE insulation Ø 28.3 mm, = 2.3
Outer semi-conducting layer Ø 29.9 mm
Wire screen Ø 32.6 mm
Jacket Ø 37.9 mm
Nature of the semi-con layers Inner and outer semi-con layers both can be
stripped off
Type of cable 11 kV XLPE
5.5.2 Preparation of the semi-conducting sample
The materials to be measured are the inner semi-con and outer semi-con layers of XLPE
cable. The outer semi-conducting layer is easily taken out since it could be stripped off.
The inner semi-con layer is thinner than outer semi-con layer so it had to be sliced off
manually with a scalpel. The non-uniformity in the thickness of samples may affect the
measurement so need to make sure both samples have uniform thickness. It is critical that
the sample should not be contaminated by any other material.
5.5.3 Measurement Setup and Results
XLPE insulation is very sensitive to PD. One common method to detect PD is the
measurement of the fields of the PD with sensitive field sensors such as sheath champing
sensor, which was introduced in previous section. This sensors couples PD signals
through the outer semi-conducting layer and therefore the properties of the semi-
conducting layer have a significant influence on the coupling mechanisms. Therefore,
knowing the semi-conducting conductivity is important to optimize PD field sensors. The
aim of this measurement is to obtain the electrical conductivity of both inner and outer
semi-conducting layers.
________________________________________________________________________
To measure the semi-con samples, a good electrical contact between the sample and the
measuring probe is critical. To ensure uniformity in the dimension of the semi-con
samples, all the samples need to be cut to be the required diameter and the edges are
trimmed to ensure there are no short circuits caused by the electrical contact.
The basic setup is shown in Figure 5.12. There are two stranded copper plates that
connect both end of testing sample (semi-conducting layer) with the measuring probe.
One measuring probe is connected to 5 DC voltage source and other measuring probe act
as reference and it is connected to the detector (voltage/current meter) directly.
Comparing both sources with both measuring probes, solve the Equation 5.1 with the
measurement and the resistivity of the semi-con sample can be obtained, which is shown
in Table 5.2.
A
U R (Equation 5.1)
A
Where:
is the static resistivity (measured in ohm-metres, m);
R is the electrical resistance of the testing material (measured in ohms, );
is the length of the piece of material (measured in metres, m);
A is the cross-sectional area of the specimen (measured in square metres, m²).
1 cm 1cm
2.5cm long
Thickness: 2.5 mm
8cm
Figure 5.12: the setup of semi-conducting samples measurement.

________________________________________________________________________
Table 5.2: the result of semi-con samples measurement.
Resistivity *m Conductivity

Inner Semi-con layer 3~5 0.2 ~ 0.333
Outer Semi-con layer 22.05 ~ 28.689 0.03486 ~ 0.04535
5.6 Verification of the High Frequency Cable Model
Since cable PDs have high frequency content and these signals are attenuated
significantly by the cable itself. Thus the propagation characteristics of cables are of
some importance. In this regard, power cables are more complicated than overhead lines,
with the cable composed of a number of conducting, semi-conducting and insulating
layers. All of these parts have some effect on the propagation characteristics and thus
modeling of the cable for HF pulse propagation is an important requirement.
i) Analytical techniques (Circuit Analysis) where attempts are made to determine the
solutions of the appropriate field equations in term of functions such as the Bessel
function [58, 59] and
ii) Numerical techniques (Field Analysis) in which the solutions of the electric and
magnetic field equations are obtained numerically by finite element or finite difference
methods [60, 61].
In this chapter, the measured results are verified with the simulated result by the proposed
cable models, which are introduced in Chapter 3. In the Chapter 3, Section 3.5 introduces
the proposed cable model by using analytical techniques and Section 3.6 introduces the
cable model by using numerical method. It is important to know whether the proposed
cable model is good or not. This can be done by comparing the results from the high
frequency cable model with the measurements made on the cable. The verification
procedure of the cable model is illustrated by the diagram shown in Figure 5.13.
________________________________________________________________________
Figure 5.13: Flow chart for the verification procedure of the cable model.
________________________________________________________________________
5.6.1 Verification of the results in frequency domain

a) High frequency sinusoidal source measurement
Test measurements were conducted at several frequencies in the range 0.1 to 50 M Hz
with a sinusoidal waveform using a function generator and captured signal by using high
frequency measurement current transformers. The variation of source voltage was
effectively negligible. Figure 5.14 shows the measured attenuation compared with
theoretical calculations using the cable model (J-Marti cable model, 136 meter XLPE
cable) described in Section 3.5.1 and Section 3.6. The modeling is done both with and
without the semi-conducting layers included and their effect on signal attenuation is very
significant as can be seen. As expected the result shows that the receiving end CT signal
is more attenuated as frequency increases. The signal propagation through the cables
depend more on the frequency than on the length of cable. The result shows the major
influence of the effects of dissipation of semi-conducting layers on signal propagation.
The degree of attenuation may be affected also by temperature and by the pressure
applied on the semi-conducting layers. It will also be affected by the loss factor and
permittivity of the insulation material. The measured and simulated results with semi-
conducting layers included give good agreement, an indication that the cable model
developed is a valid and accurate simulation.
________________________________________________________________________
0.2
0.15
Attenuation (dB/m)
Attenuation dB/m
0.1
0.10
0.05
0.05
0
0 5 10 15 20 25 30 35 40 45 50
F (M H ) Frequency (MHz)
Measured result (136 M XLPE cable) Simulated result by ATPDraw Simulated result by ATPDraw (without
semi-conducting layers)
Figure 5.14: Comparison between measured and modeled signal attenuation, with and
without semicon, as a function of frequency. The full line shows measured results: the
green dashed line is numerical simulation with semi-conducting layers: the brown dashed
line is numerical simulation without semicon included.
b) High frequency sinusoidal signals on the semi-conducting layers

The second test is a determination of the voltage response of the semi-conducting layers
to the signals. The semi-conducting layers in the cable can smooth and decrease the
electric field and protect against any PD generation between the conductor and insulation
or between the insulation and the metallic screen [140]. The semi-conducting layers also
have an effect on the propagation characteristics of the cable in terms of the attenuation
and velocity [141]. Using the measurement setup as shown in Figure 5.1, measurements
of signal attenuation on the inner and outer semi-conducting layers are presented in
Figure 5.15, together with the simulated result obtained using the analytical cable model,
which is described in the previous Section 3.5.1. The signals on the semi-conducting
layers were measured with voltage probes: Sheath champing sensor in Chapter 5.4. The
________________________________________________________________________
sensors are used and are connected to the semi-conducting layers at the sending and
receiving end.
0.3
0.25
Attenuation dB/m
Attenuation (dB/m)
0.2
0.15
0.1
0.05
0
0 10 20 30 40 50
Frequency (M Hz)
Outer semi-con Inner semi-con Simulated result on semi-con
Figure 5.15: Comparison of measured signal attenuation of the XLPE cable on each semi-
con layers with results of ATPDraw simulation
As shown in Figure 5.17, it can be seen that the voltage response of the semi-conducting
layers has greater sensitivity for high frequency signals and the outer semi-conducting
layer has a higher sensitivity with respect to the inner semi-conducting layer. The results
of the analytical simulation give agreement of the measurement of semi-conducting
layers and the difference between the simulated and measured results is small.
Figure 5.16 shows the attenuation versus frequency as measured on the inner semi-
conducting layer and outer semi-conducting layer of the 136 m XLPE cable (Cable A)
and also the outer semi-conducting layer of the 45 m XLPE cable (Cable B). The results
show that the receiving end signal becomes more attenuated as the frequency increases.
During the high frequency tests, the cable set-up can be considered as a resonant circuit.
The loss increases dramatically with increase of frequency.
Figure 5.17 shows the inner and outer semi-conducting layer voltage for the cable A and
cable B compared with the output level of the current transformer at the receiving end
________________________________________________________________________
(T2) versus frequency. The voltage level output of the semi-conducting layers has greater
sensitivity than the current transformer monitors and the outer semi-conducting layer has
higher sensitivity with respect to the inner semi-conducting layer. In addition the results
show that the semi-conducting voltage values are increasing as frequency increasing to
the test limit of 40 M Hz.
1. Outer semi-con layer (136 m)

2. Inner semi-con layer (136 m)
3. Outer semi-con layer (45 m)
Figure 5.16: Attenuation versus frequency on the inner and outer semi-conducting layers
(136 m and 45 m XLPE cable).
________________________________________________________________________
Figure 5.17: Semi-conducting layer voltage at the receiving end versus frequency on
outer semi-conducting layer and comparison with HFCT measurements. From left: Cable
B (outer semi-con), Cable A (outer), Cable A (inner), Cable B (HFCT at receiving end),
Cable A (HFCT at receiving end).
5.62 Verification of the results in frequency domain
The problem that arises in PD propagation is that there are no distributed parameter
models available for such power cable insulation at such high frequencies. Moreover, it is
important to model the frequency dependence of power cables accurately in order to
minimize the cost of construction. In this Chapter, the simulation work used the ATP
program as a basis to develop the existing J-Marti models for power cables to simulate
the PD propagation.
Analytical technique: ATP Cable Modelling

ATP (Advanced Transient Program) is one of the EMTP (Electromagnetic Transient
Program) forms and is probably the most widely-used tool for the analysis of
________________________________________________________________________
electromagnetic transients in power systems. The basic method of cable modeling is to

calculate the cable parameters from the geometrical data and material properties and then
convert into a new set of parameters for use by available cable models. More detail on
this cable model can be found in Section 3.5.
The general parameters used by cable models are the series impedance and admittance:
Z() = R() + jL () (Equation 5.2)
Y() = G() + jC () (Equation 5.3)
and the frequency dependent characteristic impedance of the cable, defined as follows:
Zc() = Z().Y() (Equation 5.4)
Where R, L, C, G are respectively the series resistance, series inductance, shunt

capacitance and shunt conductance per unit length of the cable. The quantities are
frequency dependent as shown.
The Z and Y are easily evaluated by the ATP supporting routine. However, the current
ATP cable model does not consider certain cable features such as semi-conducting layers
and wire screens. Instead, ATP assumes a cylindrically symmetrical current distribution
in all conductors. The assumed cylindrical distribution also means that the helical
winding effect of the sheath can not be taken into account. Each conducting component
of the cable has constant permeability and the dielectric is considered as being non-
dispersive and frequency independent. In general, the dielectric loss is assumed to be
negligible comparing with conductor and sheath loss.
________________________________________________________________________
Importance of including semi-conducting layers in ATP models

Neglecting the presence of semi-conducting layers, the results obtained using ATP are
satisfactory with the theoretical calculations in terms of propagation times, as shown in
Figure 5.18 where the calculated theoretical velocity is obtained from the standard
lossless representation using an insulation relative permittivity of 2.2 (XLPE).
0.4
[mA]
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
-1.2
0.0 0.2 0.4 0.6 0.8 1.0 [us] 1.2
Figure 5.18: the ATP simulated result for comparing the velocity of propagation.
The ATP simulation is done with J-Marti Cable model with short circuit termination,
which is described in Section 3.5.1. From the ATP simulated results, the velocity of
propagation can be obtained by dividing the length of the cable (136 m) by the ATP
calculation of the time difference in sending and received signals. Using the data from
Figure 5.22 the velocity obtained by ATP is 2.08 x 108 m/s, which is in good agreement
with the theoretical value of 2.021 x 108 m/s (3% difference).
In ATP program, the J-Marti model of the cable can only allow a three-layer
configuration to be used in the model: the conductor, the sheath and the armor. Since the
ATP cable model does not have the capability of modeling the inner and outer semi-
conducting layers, it had to be modified to allow them to be incorporated. In order to
develop a full XLPE cable model, the original sheath layer was employed as the inner
semi-conducting layer, the armor layer is substituted for the outer semi-conducting layer
and between them is the insulation layer. The metal sheath was included by using an
________________________________________________________________________
enclosing pipe in a pipe-type cable model in the ATP and there is no change on the total
radial distance.
Another cable model was tried with only the included conductor, insulation and sheath
(without Inner and outer semi-conducting layers). Table 3.1 shows higher permittivity of
insulation, the lower propagation velocity and the ratio of I receiving end / I sending end.
a) Verification of paper insulated cable model
Paper insulated cable has different characteristics than the XLPE cable. This is due to the
different material of insulation and layers such as oil are included in paper insulated
cable. The oil has a higher electric strength than other normal materials. It is interesting
to notice the higher attenuation (increasing frequency) in the case of paper insulated cable
comparing with other short length XLPE cable, the reason for this is that the loss factor is
higher than XLPE. The results are contradicts with the Figure 4.23 in Section 4.4. The
simulation of Figure 4.23 is assumed both types of cables to have the same dielectric
constant of 3.5 and the main purpose of Figure 4.23 is to investigate the influence of
structure of cable.
The paper insulated cable testing set-up is shown in Figure 5.1. The PD calibrating signal
(1000 pC) is generated by an electronic PD calibrator. The sending end and receiving end
currents are detected by high frequency current transformers (HF-CTs). The output signal
from the HF-CTs is fed back to the oscilloscope via a 50 coaxial cable for recording.
________________________________________________________________________
Due to the short length of this cable with short circuit termination, interference effects are
dominating the measurement and the receiving side shows few reflected pulses. The
sending signal of paper insulated cable shows that it has 16.95mV voltage as shown in
Figure 5.19 and the receiving signal is about 7.50mV voltage as shown in Figure 5.20.
The ratio between receiving side current and sending side current is 0.443 and the
traveling time between sending and receiving ends is 58 ns. The traveling speed from
measurement:
l
v (Equation 5.5)
t
Where l is length of cable, t is the measurement travel time. Therefore the traveling speed
on paper insulated cable is v = 3.5 / (58 x 10-9) = 0.603 x 108 m/sec.
The measured velocity is a little different to the theoretical velocity. The higher loss
factor of paper at high frequencies may be a cause of this difference and the short length
of cable also increases the measurement uncertainty.
________________________________________________________________________
12
10
6
Amplitude (mA)
0
0.5 1 1.5 2 2.5 3 3.5 4
-2
-4
-6
-8
T ime (us)
Figure 5.19: The measured result at the sending end of 5 m PILC cable.
4
3
2
Amplitude (mA)
1
0
0 1 2 3 4 5 6 7 8 9 10
-1
-2
-3
-4
-5
T ime (us)
Figure 5.20: The measured result at the receiving end of 5 m PILC cable.
b) Simulation of PD propagation by ATP
It is thus necessary to develop a model for paper insulated cable to simulate the PD
propagation correctly. ATP is the best choice used for simulation of the PD propagation
on single phase cable. For simulation of the PD responses of the paper insulated cable,
the frequency dependent (J Marti’s) cable model is employed. The ATP program can
________________________________________________________________________
model the calibrated PD signals by applying the equation in TACS function. In order to
use the original PD signal that collected from the Lecroy 9362C oscilloscope with sample
rate: 500 MS/s, the source type 1 of ATP need to apply and through modification of atp.
output file of ATP program. The result of ATP ratio of I Receiving end and I sending end (Figure
5.21) and the travel time has a similar result as the measured values (about 2%
difference).
1000 pC.
ATPDraw Measured
results results
Ratio between IReceiving and 0.416 0.443
ISending
Traveling time 57ns 58ns
The compared resultant signal.

80
[mA]
50
Amplitude (mA)
20
-10
-40
-70
0.0 0.4 0.8 1.2 1.6 [us] 2.0
Time (us)
Figure 5.21: The ATPDraw simulation result of a 5 m length of paper insulated cable.
Red line is the sending side and Green line is the receiving side.
C) Verification of XLPE cable model
Using the setup shown in Figure 5.4, a PD calibrator signal, simulating a 1000pC partial
discharge is applied to the sending end of the cables with the receiving end short
________________________________________________________________________
circuited. The current pulse of the calibrator is shown in Figure 5.22. The calibrator
signal is attenuated, as expected during propagation along the cable. From the
measurements, the ratio of I receiving end / I sending end is calculated to be 1.028 which is more
than 1, which is as expected because of the termination short circuit. For a loss-less cable
the ratio would be 2 giving a current doubling at the short circuit. The travel time of
signal over the 136 meters is 923 nano-seconds. Therefore the propagation velocity on
this cable is v = 136 / (923 x 10-9) = 1.47 x 108 m/sec which is less than the speed of light,
3 x 108 m/s, as expected for such insulation. However it is only 49% of light speed,
whereas using the relative permittivity of XLPE (Hr =2.2) would give the velocity as 67%
of light speed as mentioned in previous section. This is an indication of the importance of
the semi-conducting layer in modifying the propagation characteristics.
Figure 5.22 shows the sending end pulse while Figure 5.23 shows the receiving end pulse
(the left hand negative pulse). The wave shape at the receiving end is deformed due to the
attenuation and dispersion of the cable. (The positive pulse seen at about 1.7
microseconds is the pulse reflected back from the (high impedance) source after two
further transits of the cable length).
3
C u rre n t (m A )
-1
-0.3 -0.1 0.1 0.3 0.5 0.7
Current (mA)
-5
-9
-13
Second (us)
Figure 5.22: The measured result at the sending end of 136 m XLPE cable.
________________________________________________________________________
C urren t (m A ) 0
Current (mA)
-0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0

-4
-8
-12
Second (us)
Figure 5.23: The measured result at the receiving end of 136 m XLPE cable.
d) Simulation of PD propagation in XLPE cables by ATP
In this section the new modified XLPE J-Marti cable models (detail can be found in
Section 3.5.1 and Section 5.6.2) were employed for simulation purposes and their
calculated responses are compared with the measurement results on the real cables.
The cable model results demonstrated that XLPE cable can in fact be fully modeled,
including the conductor, two semi-conducting layers, the XLPE-insulation and the sheath
as mentioned in previous Section 3.5. From the simulation results of the 136 m XLPE
cable (cable A) shown in Figure 5.24, the travel time is 866 nano-seconds and the ratio of
I receiving end /I sending end is 1.04. The pulse at left is the sending end pulse and the second
pulse is the receiving end pulse.
________________________________________________________________________
0.3
[mA]
0.1
-0.1
Current (mA)
-0.3
Sending
-0.5
-0.7
Receiving
-0.9
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 [us] 1.6
Time (us)
cable. Red line is the sending side and Green line is the receiving side.
Comparing the simulated result and measured result, there is 1% error in the ratio and
6.5% error in the travel time. The high travel time error may be due to the distortion of
the pulse affecting the rise time somewhat. Further, the ATPDraw assumes that all the
common earths are perfectly grounded. However, in the practical situation, a perfect
ground is not realizable. This may affect the timing error also. Figure 5.25 shows the
simulation of the other cable, the 45 meter Cable B.
Current (mA)
Sending
Receiving
Time (us)
cable. Red line is the sending side and Green line is the receiving side.
________________________________________________________________________
Figure 5.26 shows the simulation of the PD signal propagation as monitored by the semi-
conducting layers. The ratio of I receiving end / I sending end is about 0.15 which compared with
the measured value, 0.147. This gives useful information for understanding how the PD
signal passes through the semi-conducting layer.
6
[mV]
3
0
1
-3
Current (mA)
2
-6
-9
-12
0.0 0.5 1.0 1.5 [us] 2.0
1: Sending end current 2: Receiving end current
Time (us)
Figure 5.26: The simulated result at the sending and receiving end of semi-conducting
layer voltage. Red line is the sending side and Green line is the receiving side.
A summary of the results of the various lengths of cable are shown in Table 5.5. The
errors in the 136 and 45 meter lengths are acceptable in the circumstances. The 45 meter
cable had a very thin inner semi-conducting layer and this caused difficultly in modeling
the cable in ATPDraw since the thickness was not precisely known and this may generate
some error. Comparing the results of Table 3.1 and Table 5.4, the semi-conducting layers
have a very significant effect on the propagation velocity and attenuation at high
frequency.
Clearly, the simulation model gives good agreement of the test results. This is despite
neglect of the dielectric loss which has less effect on attenuation than the semi-
conducting layers. This is in agreement with the discussion in Chapter 4. Figure 5.27
________________________________________________________________________
shows the attenuation (dB/m unit) as function of different frequency in the dielectric loss
(tanG) range from 0.01 to 0.0005. The dielectric loss of XLPE insulation is usually in the
range of 0.001 to 0.005, which the different of attenuation is quite small.
1000 pC.
Ratio of I Traveling
receiving end / Travel speed (m
Length of XLPE cable: I sending end time (s) /sec)
136 m Simulation: 1.04 866ns 1.53 x108
Measurement: 1.028 923ns 1.47 x108
Error: 1% 6.5% 4%
45m Simulation: 1.4 265ns 1.69 x108
Measurement: 1.33 277ns 1.62 x108
Error: 5% 4.5% 4.1%
0.25
0.2
0.15
Attenuation
(dB/m)
0.1
0.05 40
0 20 frequency (M
Hz)
0.01
0.001
10
0.005
0.0001
0.0005
tan
Figure 5.27: the graph of attenuation (dB/m unit) as function of different frequency in the
dielectric loss (tanG) range from 0.01 to 0.0005.
________________________________________________________________________
5.7 Propagation Constant vs. Cable Length
It is important to investigate the relationship of the propagation constant as function of

the cable length. The longer length of cable has the higher attenuation. In Figure 5.28 is
shown the attenuation (in dB) over the XLPE cable length of 400 meters. The attenuation
is quite high when the signal is high frequency and with long cable distance. Higher
frequency and longer cable distance can thus increase the loss greatly as would be
expected. However, the calibrating signals and PD signals that captured in the
measurement have a low frequency range, about 5 MHz. For example, high frequency PD
signals may have 50 MHz bandwidth. Figure 5.29 shows a more detailed look at the
frequency below 5 MHz. the results are in reasonable agreement with the measurements
that were discussed in the previous sections.
Figure 5.28 shows that the relationship of attenuation and distance is reasonably linear
(below 50 MHz). From the Figure 5.28, the slope and intercept can be obtained. In order
to find the maximum cable length (Lmax) as function of sensor sensitivity, the following
equation can be used:
Lmax = (Q max/ Q min- Y ) / y (Equation 5.6)
Where Q max is the maximum apparent charge/voltage that can be tolerated by the cable
and Q min is the minimum charge/voltage that can be observed. Y is the value of
intercept from Figure 5.28 and y is the slope (in dB/m) from Figure 5.28. The typical
cable length limit can up to 1000 meters for the ATP analysis and hundred km cable
length for the FEM analysis.
________________________________________________________________________
Figure 5.28: the graph of attenuation (dB) as function of different frequency over cable
length of 400 meters.
Figure 5.29: the graph of attenuation (dB) at two specific frequencies of 1 MHz and 5
MHz over cable length of 400 meters.
________________________________________________________________________
5.8 Reconstruction of Signal Source
Online partial discharge testing is usually allowed to capture the PD signals from both
terminations of cable. One of the project aims is to reconstruct the PD source signal by
using the received signal, which is captured from the sensor at the received end. The
reconstruction of PD source signal can also give the location of PD and this information
is helpful for prediction of the power system failure. Figure 5.30 shows the detail on how
to reconstruct the signal source. Firstly, the propagation characteristics need to obtain
from the simulated results (by using developed cable model). The important propagation
characteristics are attenuation and velocity. The characteristics can be assumed to be X(f)
or X(t), which are defined as characteristics in frequency domain and time domain.
Secondly, by using the high frequency (HF) sensors, the received signals can be captured.
Finally, apply Equation 5.7 and recalculate the original signal.
I original signal = X (f)(or X (t)) * I receiving signal (Equation 5.7)
Figure 5.31 shows the reconstruction of the source signal pulse shape by using the
Equation 5.7. The ratio of I receiving end and I sending end of the previous simulation is
0.596 and travel time is 0.896 s. The propagation characteristic with frequency
attenuation is 2.25 dB and the velocity is 151 m/ s. The green line in Figure 5.31 is the
signal after reconstruction. The result shows the original signal source shape agrees with
signal reconstructed with the error of fitting ±1~3%. To rebuild the source signal is
important. From the source signal, the PD location and type can be estimated. As long as
the propagation characteristics are known, the engineer can locate the fault in the short
time and low cost.
________________________________________________________________________
Figure 5.30: Explanatory diagram for the reconstruction of signal source.
Figure 5.31: Reconstruction of signal source by using Equation 5.6.

________________________________________________________________________
5.9 Conclusion
This chapter presents the characteristics of partial discharge and high frequency signal
response in different lengths of XLPE cables and their semi-conducting layers. It
demonstrates the ATP cable model can model the single core cable. The results are
shown that the XLPE cable can be fully and accurately modeled in ATPDraw and that the
inclusion of the semi-conducting layer influence is very important in the modeling for
proper simulation. The semi-conducting layers affect signal propagation very
significantly and must be taken into account in any PD analysis work. The match of the
frequency dependent cable model response with measured data demonstrates the validity
of the ATP frequency dependent cable models for PD propagation study. However, it
must be noted that practical realities such as earth resistances, proximity of other cables,
temperature and semi-conducting layer pressure may also affect the results.
In addition this chapter presents results of high frequency signal response, using PD
simulations, in two different lengths of XLPE cables and their semi-conducting layers.
The high frequency response results show that attenuation of the HF-CTs based
measurements is increasing as frequency increases. However when monitoring the semi-
conducting layer voltage, it was found that this increased as frequency increased, as
would be expected from a capacitive coupled sensor. This indicates that use of the semi-
conducting layer can have a high sensitivity to partial discharge monitoring and needs to
be considered.
The model used for reconstruction which includes the effect of semicon material in the
losses provides accurate reproduction of the propagation characteristics of high frequency
PD pulses and the thesis work had used such a model to reconstruct PD waveforms of site
PDs from the measured signal for the first time. The use of the original waveform is
important for PD identification and location in the practical situation.
________________________________________________________________________
Quantitative Summary
Types of cables:
XLPE cable
Paper insulated cable
Length of cables:
136 meter of XLPE cable
45 meter of XLPE cable
3.5 meter of paper insulated cable
Diameter of semi-conducting layers of XLPE cable:

Inner semi-conducting layer: 21.3mm
Outer semi-conducting layer: 29.9mm
Resistivity of Semi-conducting layers

Inner semi-conducting layer: 3~5
Outer semi-conducting layer: 22.05~28.69
Dielectric constant of cables

XLPE cable: 2.2
Paper insulated cable: 3.5
Temperature of testing area:

20~23 degree
Effect of frequency range:

0.1 MHz ~100 MHz
Error between simulation and measurement:

±1~6.5%
Chapter 6: Partial Discharge Detection 182
________________________________________________________________________
Chapter 6
Partial Discharge Detection
6.0 Introduction
The lifetime of XLPE power cables is determined by several factors. One of the more
important of these is the breakdown in a solid dielectric. There are number of possible
mechanisms for electrical breakdown in a solid dielectric [112]:
(a) Conductive - Ionic charge carriers move in an electric field, producing a leakage
current. The large ion and electron flow represents a conducting path and
breakdown occurs.
(b) Disruptive – The high energy fields cause molecular disintegration of the material
associated with high energy release.
(c) Thermal – The electric resistance and dielectric loss created result in an increase
in temperature and there is a further increase in loss which produces higher
temperatures.
(d) Partial charges - Most often electric breakdown is the occurrence of Partial
Discharges (PDs) at the dielectric. Electric discharges which do not bridge
electrodes are called partial charges and this will be discussed further in the
following sections.
Various defects, such as voids, contaminants and electrical trees can cause Partial
Discharge (PD) activity in high voltage XLPE power cable. Although the magnitudes of
partial discharges are usually small however XLPE insulation is very sensitive to partial
discharges and PD activity can cause progressive deterioration and ultimate failure. To
________________________________________________________________________
ensure the reliability of the whole cable system, PD testing and location are an essential
part of the monitoring and testing of XLPE cable.
Partial discharges in solid dielectric can generate electrical signals with frequencies up to
several hundred MHz [3]. The ability to detect and locate a PD source is limited by
various factors such as external interferences of noisy background, loss from the different
parameters of cable and the limited bandwidth of detection sensors. Therefore it is
necessary to understand the high frequency response of power cables and there are
number of publications on cable modelling already available [3] [63]. In order to have
more understanding on realistic PD signal propagation, the experiments used the artificial
defect to generating PD signal instead of generating by PD calibrator. Moreover, the
external sensors are used for PD detection instead of the conventional sensors, High
Frequency Current Transformers (HF-CTs).
Our aim of the work described in this thesis is to test the sensitivity and bandwidth of the
external sensors, investigate the best method to estimate PD pulse and the propagation of
PD pulses for different length of the cable.
6.0.1 Characteristics and structures of XLPE cable
As previously mentioned in Chapter 2 and Chapter 5, XLPE cables have excellent

electrical insulation performance. Their thermal resistivity and the dielectric dissipation
factor are low, giving good thermal rating. They are economical and easy to handle and
install. However they have the problem of susceptibility to water trees.
In the 22 kV single core XLPE cable used for the work described in this Chapter, the
conductor is stranded aluminium and the cable is concentric with inner and outer semi-
conducting layers and the main (XLPE) insulation between the semi-conducting layers.
The metallic shielding comprises helical copper strands and the overall jacket is extruded
polyvinyl chloride (PVC). The semi-conducting layers are made of polyethylene or
________________________________________________________________________
ethylene copolymer mixed with conductive carbon black [11]. The semi-conducting
layers are used to smooth out the electric stress enhancements at the conductor strands
and thus prevent partial discharge activity at the interface between the conductor and the
XLPE insulation.
6.0.2 High frequency attenuation
XLPE cable systems are now almost universally used in the electrical supply industry and
this trend is likely to continue. In order to understand the PD signal propagation
characteristics of XLPE cable, there is a need of having clear understanding on the loss
parameters of cable. Those different loss parameters can be the skin effect, loss of the
dielectric, and the semi-conducting layers.
XLPE cables have very low dielectric loss factor at power frequency so that one might
expect very low high frequency signal attenuation. However, in common with all HV
cables, XLPE cables incorporate semi-conductive layers and significant high frequency
signal losses are caused by the propagation of a radial electric displacement current
(which increases with frequency) through the resistance of these layers [63].
6.1 Measurement
The measured and simulated results in Chapter 5 are based on calibrating signal and
sinusoidal signals. In this Chapter, the measurement and simulation are used the PD
signals that are generated by an artificial PD defect with capacitance-coupled sensors
used for the detection of PD, rather than a high frequency current transformer (HFCT). In
addition, PD pulse waveforms were evaluated by three different integral methods in order
to estimate the best way to characterize the PD pulses.
________________________________________________________________________
6.1.1 Measurement Set up
A 22 kV single core XLPE cable is employed in the measurements. The detail of the
cable investigated is given in Figure 6.1. The total length of cable is about 4 meters. An
artificial defect is employed at a location 1/3 of the way along the total length of cable.
The reason to chosen defect location at 1/3 of total length of cable is to determine the
signal propagation at different length of the cable. Four external sensors are attached on
the cable at various locations. External sensors used were capacitive coupling sensors.
Sensor 1 and sensor 2 are attached at the both terminated ends of cable and Sensor 3 is
attached at the side of artificial defect. From the position of sensor 2 to sensor 3 the
separation is 1/3 of the total cable length. From the position of sensor 3 to sensor 1 the
separation is 2/3 of the total cable length. The major benefit of using those external
sensors is that they can be put at any desired number of points along the length of cable
on/off site. This is the major advantage for studying PD signal propagation along the
cables. Figure 6.2 shows the PD detector for test measurements.
The artificial defect is created by a pin which is located close to the insulation surface.
The closer to the insulation surface, the higher electric stress will be. The cross section of
the experiment set up is shown in Figure 6.3.
The test voltage 10 kV is applied to the cable, and an internal discharge is generated
within the artificial defect, where the discharge inception voltage is 4 kV. PD pulses are
generated at the artificial defect and they travel to both ends of the cable. The external
sensors are used to capturing the PD pulses along cable.
________________________________________________________________________
Figure 6.1: Laboratory test measurement setup.
A B C D
A: Connecting to ground
B: Input
C: Voltage level
D: Measuring PD
Figure 6.2: PD Detector for measuring the partial discharge (D) and voltage level (C)
from the input (B) and connecting to the ground (A).
________________________________________________________________________
Dimension
Controller
Brass
Sheath
Air
Semi-con
Insulation
Figure 6.3: Cross section diagram of the artificial defect
A number of electric field models have been used to run the simulation for the artificial
defect experiment. Figure 6.4 shows the cross section of the simulated model. In this
figure, the sharp pin has been used. In fact, two different types of pin are used: sharp head
and spherical shape head (in Figure 6.5) for the simulation. In order to increase the
accuracy, refined selection mesh is used in the air gap area and the refined mesh consists
of 56944 elements. Figure 6.4 indicates the maximum electrical field is on the interface
of the air, semi-conducting layer and insulation. Furthermore, the conductivity and
permittivity of semi-conducting layers and insulation need to be chosen carefully.
Different parameters of the semi-conducting layers and insulation can give quite different
results.
1.6 mm
Sharp 2.02 mm
shape
2.6 mm
Sheath
4.2 mm
Air Semi-con
Insulation
Figure 6.4: Cross section of the simulated model.

________________________________________________________________________
Figure 6.5: Two different types of pin: sharp shape and spherical shape.
Figure 6.6 is the result of the simulation. The vertical axis of Figure 6.6 is the ratio of
maximum electric field with the defect to the maximum of electric field without the
artificial defect. The horizontal axis of Figure 6.6 is the percentage of the distance
between the insulation surface and the head of the pin. The result indicates that the sharp
pin can give the maximum electric field when the pin is close to the insulation surface.
However, after passing 50% of the distance between the insulation surface and the head
of the pin (horizontal axis of Figure 6.6), the two different shapes: spherical and sharp pin
have the same field ratio. It means that this distance is the optimal distance to produce an
artificial defect due to spherical and sharp pin will give the same result.
Figure 6.6: Ratio of electric field for the distance between insulation and head of pin.
________________________________________________________________________
6.2 Selection of detection sensors

6.2.1 Conventional detection sensor
In order to find out the sensitivity level of the capacitance coupling external sensors,
High Frequency Current Transformers (HF-CTs) are used for the comparison with
external sensors. In this case, three HF-CTs with different bandwidths are used. An 11kV
single core XLPE cable with length of 4 meter long is used to test the sensitivity of the
three HF-CTs. This XLPE cable has surface discharges at the cable terminations as the
signal source. Figure 6.7 illustrates the result of the captured PD signal by using the three
different types of HF-CTs.
Figure 6.7: Comparison of PD signals captured by different types of HF-CTs
6.2.2 Comparison result
In order to find out which sensor is more suitable for the ‘artificial defect’ experiment the
external sensor and HF-CT 1 are applied on the measurement. The measurement set up is
shown in Figure 6.1 and the compared result of HF-CT 1 and the external sensor is
shown in Figure 6.8. The result indicates clearly the external sensor has a better
sensitivity than HF-CTs under the ‘artificial defect’ experiment, where the internal
discharge is generated. Due to the attenuation effects of semi-conducting layers on high
________________________________________________________________________
frequency PD signals, the detection process of the HF-CT is easily distorted by the
attenuation factors. Figure 6.9 shows the frequency spectrum of all different kinds of HF-
CTs and external sensor. The Network Analyzer (9 kHz to 3000 MHz) was used to make
the two port measurements of the scattering parameters on all the detection sensors. HF-
CT 1 and HF-CT 2 are used for high voltage testing due to good sensitivity. The result
illustrated that external sensors have better sensitivity than the HF-CTs. Both HF-CTs
and the external sensor have good bandwidth.
Figure 6.8: The compared result of HFCT and External sensor.
Figure 6.9: Frequency spectrum of external sensor 3 and other type of HF-CTs.
________________________________________________________________________
From the experience on using the external sensor, the following advantages can be listed:
1. Better sensitivity than HF-CTs.
2. Allow to capture PD pulse at any position of cable on and off site.
3. Suitable for most of the applications due to its easy installation.
6.3 Measurement Results

6.3.1. PD signal propagation
To establish a reliable monitoring system for the cable, a comprehensive knowledge of

PD signal propagation characteristics is required. Three individual PD signal were
detected from the output of external sensor 1 to external sensor 3. As each sensor is
located in the different position of cable so this gives great information on the PD signal
propagation. A comparison of three different PD signals is shown in Figure 6.10. Sensor
3 is located near the artificial defect so the PD signal has the greatest amplitude. Sensor 2
is located about 1/3 of the total cable length and sensor 1 is located about 2/3 of the total
cable lengths. Therefore, the PD signals of sensors 1 and 2 have 1/3 and 2/3 of travelling
distance compared with signal of sensor 3. There is a clear attenuation of the PD signals
and higher attenuation at PD signal of sensor 1 and 2 when comparing them with PD
signal of sensor 3. Moreover, Figure 6.10 shows that there are few reasonable reflection
pulses from the terminations. The oscillation in the wave tail is not significant and is
attenuated quickly.
________________________________________________________________________
Figure 6.10: Comparison of three single PD signals captured at different location of

external sensors.
6.3.2. PDs pattern of the measurement
A typical PD pattern was recorded from an Mtronix PD monitoring system MPD540

[149] and the result is shown in Figure 6.11 for the defect. The MPD 540 measurement
system consists of one or more acquisition units, an optical interface and a PC including
measuring software. The PD signals are filtered, amplified and digitized. The time
accuracy of detection of a PD signal is about 2 ns with the amplitude quantization of 14
Bit and sampling rate of 64 MS/s. A digital band-pass filter realizes the quasi-integration.
The centre frequency for the digital filter can be chosen in frequency range up to 20 MHz
with the bandwidth between 9 kHz and 3 MHz. The selection of the bandwidth of the PD
measurement system is very important for noise suppression. The bandwidth of PD
measurement used is 700 kHz to 1 MHz. The calibration setting is to connect a charge
calibrator in parallel to cable. Then enter the calibrator charge in the QIEC (target) field
and compute a divider factor, which is multiplied with the charge values. After
calibration, the measured field show the same values as the calibrator charge. In Figure
6.12 is indicated the frequency spectrum of the PD measurement and the bandwidth of
the PD measurement (700KHz-1MHz), taken at the first peak of frequency response. The
frequency spectrum is corresponding to the FFT of the single PD pulse which be
captured. Figure 6.11 illustrates the discharge level is over 100pC. The characteristics
________________________________________________________________________
are a fairly symmetric pattern with stationary wandering impulses and this is the usual
pattern of internal discharge in voids in impregnated dielectrics.
Figure 6.11: PD pattern of measurement that is captured by Mtronix PD detector.
Figure 6.12: Frequency spectrum of the PD measurement in figure 6.11.
6.3.3. PDs pattern from the external sensors
The same PD activity was also recorded using the external sensors at three different
positions of cable, as shown in Figure 6.1. Figure 6.13 is the record of the PD pattern at
sensor 1, Figure 6.14 is the record of PD pattern at sensor 2 and Figure 6.15 is the record
of PD pattern at sensor 3. The voltage of 8 kV is required to be applied on the cable in
order to display the PD activity on the PD measuring system. The voltage level can
indicate the sensitivities of sensors are high and the shape of a pin (in Section 6.1.1)
________________________________________________________________________
affects the sensitivity as well. The selections of the bandwidth of those three PD
measurements are between 8-8.5MHz, which is higher than the PD measurement in
Section 6.3.2. Figure 6.10 in Section 6.3.1 indicates the individual PD pulses in time
domain that is captured by sensor 1, sensor 2 and sensor 3. Figure 6.10 shows the clear
attenuation and velocity difference between sensor 3 (closest to the defect, V3peak to peak =
0.0315 V) and sensor 2 (V2peak to peak = 0.0221 V) / sensor 1 (V1peak to peak = 0.0195 V).
Moreover, Figure 6.13- 6.15 have the similar PDs pattern. The sensor 3 has the highest
PD level but the differences of PD level between those three sensors are relatively low.
The high noise levels cover some of the PDs patterns and some of the important
information is lost due to the integrated function in Mtronix monitoring system. The
maximum PD magnitude is 50pC which is captured by external sensor 3. The HF-CTs is
also used to detect any PDs pattern however the HF-CTs cannot captured any PDs pattern
by using Mtronix monitoring system under the ‘artificial defect’ measurement.
Figure 6.13: PD pattern that is captured by external sensor 1.

________________________________________________________________________
6.4 Evaluation of Partial Discharge

6.4.1. Introduction of three different methods
The shape of a partial discharge pulse is influenced significantly when passing through
the cable. It can be distorted and its apparent charge becomes difficult to evaluate and the
use of frequency-domain methods for fault type identification become questionable. The
aim of this section is to find out which parts of the distorted PD pulse are the most
________________________________________________________________________
suitable for evaluation so that the apparent charge of the PD pulse can show the smallest
change.
According the Australian Standard AS60270 [113], the charge q generated by the
calibrator is determined from the relation:
1
q ³ | i(t) | dt R³
| u(t) | dt (Equation 6.1)
Where i(t) is the current pulse generated by the calibrator, u(t) is the voltage pulse
measured by the oscilloscope and R is the load resistance (R =50 : in this case). The
charge q is equal to the integration of current.
This charge determination formula is generally used in low voltage calibrated signal
testing. When the Equation 6.1 was applied on the high voltage testing, the charge value
dropped to a relatively small value (in unit, pC). In order to apply the Equation 6.1 for
determination of the charge under high voltage testing, the percentage of charge was used
instead of the actual charge unit (pC).
Voltage (V)
q1
q2 q3
q4 q5 q6
Figure 6.16: Example of the distorted PD pulse. q1, q2 and q3 are in positive parts of
pulse and q4, q5 and q6 are in negative parts of pulse.
________________________________________________________________________
An example of a distorted PD pulse is shown in Fig 6.16. In this case, there are three
methods available to evaluate the PD characteristics of the distorted PD pulse (in
Figure6.16):
a) The first peak of the waveform is evaluated, total charge: q=q1.
b) The waveform sections with the same polarity are evaluated and summed (e.g. the
entire positive parts of pulse), total charge: q=q1+q2+q3.
c) All sections of the waveform are evaluated regardless of the polarity and summed,
total charge: q=q1+q2+q3+q4+q5+q6.
6.4.2. Measurement results of different location of sensors
Some measurements of pulse shape were taken during the experiment of artificial detects.
The results are shown in Figure 6.17 to Figure 6.19. On the horizontal axis, L is indicated
as the percentage of total length of power cable. 0% indicates the position of the artificial
defect point and the PD pulse was generated at that position. 100 % indicates the
termination of the power cable. The first peak waveform has the pulse width of
approximate value, 0.02Ps which indicates a frequency about 50 MHz. The results of
each method (method a, b and c) are shown in Figure 6.17 to 6.19. Figure 6.17 included
all the waveform within the time width of 0.05Ps. Figure 6.18 included all the waveforms
within the time width of 0.04Ps and Figure 6.19 included all the waveforms within the
time width of 0.03Ps. The results show that method b and c have the highest percentage
(3-6 %) difference compared with method a. This is due to the fact that the result in
Figure 6.17 and Figure 6.19 is taking into account the reflected pulse from the
termination and some distortions from external noise environment. In most of the cases,
method a is the suitable method for characterizing the PD pulse. However, method b is
also acceptable when dealing with the high frequency PD pulse.
Furthermore, the results confirmed that the apparent charge of the PD, as a diagnostic
parameter is not very sensitive when comparing to other diagnostic parameters such as
the shape of PD pulse and its size. The results of Figure 6.17 to Figure 6.19 can also be
________________________________________________________________________
used to evaluate the attenuation of the PD pulse along the cable. The percentage of charge
measured is decreased from 100 to 80% at half of the 4 meter cable length. The result
indicates that the attenuation is very significant after the PD pulse generated from the
position of artificial defect point. This is due to the high electric field between insulation
and semi-conducting layer at the artificial defect point. From the 20-100% positions of
the cable length, the percentage drop of apparent charge is only 20%. At the termination
of the cable, the percentage charge can drop to approximately 60%.
Figure 6.17: Evaluated of PD by different methods (time width of 0.05Ps).

________________________________________________________________________
6.4.3. Simulated results
The attenuation from the measurements on the cable is compared to the attenuation from
the cable model and is shown in Figure 6.20. The simulated result is close the
measurement with error of only r 3%. Figure 6.21 indicates the 3-Dimension plot of
attenuation of PD pulse with function of length of cable from 0 to 100 meter and function
of frequencies of PD pulses. Maximum length of 100 meter is chosen to give better
display in the graph. The PD pulses in the measurement have a frequency of 50 MHz,
which give about 7dB attenuation at 100 meter length of cable. The calibrated signal has
a frequency about 10 MHz which give the lower attenuation with the PD pulse in this
measurement (1.1dB). Figure 6.20 and Figure 6.21 indicates the model can give
measured results correctly in terms of charge and length of cable. The longer distance
between position of defect and detecting point, the lower percentage of charge is
displayed and a higher percentage of error occurs (approximate error is 3~5%). The
validity of frequency range can be up to 100 MHz. The frequency has more influence on
the attenuation than the length of cable and the higher attenuation level gives the lower
________________________________________________________________________
level of charge. The results indicate that the PD pulse attenuation can be modeled as
functions of increasing cable length and frequency of the pulse.
Figure 6.20: Results of measurement versus simulation as function of the cable length in
percentage.
Figure 6.21: 3D simulated plot of attenuation as a function of length of cable and of

frequency of PD pulse.
________________________________________________________________________
6.5 Conclusion
Partial Discharge can generate electrical signals with frequencies up to few hundred
MHz. In order to have a realistic measurement test in laboratory, an artificial defect is
applied on the short length of XLPE cable, which can generate an internal discharge and
that was used to investigate the PD signal propagation on XLPE cable. The capacitance
coupling external sensors have been applied for the PD detection measurements and the
results are shown that the external sensors have a number of advantages over the HF-CTs
under the detection of PD pulse such as 50% higher of sensitivity. Three methods were
applied for the evaluation of PD waveforms. Most researchers use the method (a) to
capture the PD pulse [15, 44]. However, the results show that method (b) is also
acceptable to evaluate the PD waveform. The percentages of charges decrease from
100% to 85% when the percentages of total length of cable increase from 0% (defect
position) to 33% (sensor 2). The attenuation of PD pulse propagation is increasing (from
1.1dB to 7 dB) with the frequency of pulse (from 10 MHz to 50 MHz) and the distance of
cable (from 10 meter to 100 meter). In particular, increasing the frequency of the PD
pulse will give much higher attenuation. In addition, the simulation results are compared
with the test measurement results and it was found that the model used was able to
predict the measurement results accurately.
202
Chapter 7: PD Characteristics and Defect Patterns in Three Phase Power Cables
________________________________________________________________________
Chapter 7
PD Characteristics and Defect Patterns in Three

Phase Power Cables
7.0 Introduction
Power cables are a major and increasingly important asset for the electrical utility. Cables
are being loaded to higher levels and this has increased the potential for cable failure due
to dielectric breakdown. In particular, the generation of partial discharge (PD) activity in
the cable can degrade the insulation over time and cause major insulation breakdown. As
a result, PD measurement and location diagnostics have become an essential part of
insulation condition monitoring for MV and HV cables.
In single phase (single core) cables PD detection is relatively simple and location is
possible if the propagation veleocity is known. However in three phase (3-core) cables
cross-coupling of PD signals between phases causes problems with PD detection and
location. The cross-coupled signals can give false indications of PD detection. This study
describes a method to solve the problem of PD location in the three phase cable, using
techniques that have been applied to motors and transformers. An artificial internal
partial discharge is created in one phase of the three phase cable and all three phases are
monitored for PDs. The result from the three-channel PD measurement is an amplitude
relation between adjacent phase signals that can be displayed in a 3-phase amplitude
relation diagram (3PARD) [149]. Different internal PDs can appear in definite
distinguishable frequency clusters with each separate cluster representing a different
apparent PD source. Therefore separate PD patterns for each PD source can be obtained
by re-calculation of the related cluster to determine PD characteristics.
203
________________________________________________________________________
7.1 Structures and Performance of XLPE Cables
The cable configuration is shown in Figure 7.1. The conductors are aluminium with
concentric stranding. Each core has inner and outer semi-conducting layers (0.3 mm
minimum wall and 1x103-106 :-cm resistivity). Insulation is 6.35/11 kV XLPE with
minimum 3.4 mm average wall, individually helical copper wire screened. PVC (5V-90)
sheathed (1.3 mm minimum average wall), HD PE sheathed (1.4 mm minimum average
wall). The thickness and conductivity of the semi-conducting layers has a dominant effect
on PD signal attenuation and velocity [82, 8]. The inductance and capacitance parameters
of each phase to screen conductor of the cable are: C=0.311μF/km and L=0.185mH/km.
Therefore using these parameters, the approximated velocity, assuming the cable to be
lossless and ignoring the effect of other phases can be determined as:
1 1
v 1.32 u 10 8 m / s (Equation 7.1)
15
LC 0.311 u 0.185 u 10
Figure 7.1: Configuration of the 11kV three phase XLPE insulated cable.
204
________________________________________________________________________
7.2 Three phase cable parameters
Simulation of cable model necessitates the accurate values of the parameters of the lines.
In this section, as an attempt in estimating such parameters, the parameters of high
voltage three phase power cable are calculated. Numerical simulations were used to
calculate the influence of the various parameters of interest on the induced signals in the
cores of a three phase cable. These parameters, self and mutual, of both inductance and
capacitances are traditionally determined either analytically or experimentally. Analytical
techniques are quick but they are only applicable for structures with a very simple shape
and with linear materials. In contrast, the experimental techniques are used for any type
of structure but they are costly and very time consuming. Furthermore, they suffer from a
fundamental constraint: bad precision. Numerical modeling techniques are a good
alternative for both analytical and experimental methods. The first to use such an
approach to cable characterization was Lucas [133]. Later workers have shown that it is
possible to use techniques such as finite element and boundary element methods [134].
Transmission line models are normally based on the transmission line theory very often
based on the lumped equivalent circuit as shown in Figure 7.2. The mutual inductance
(not shown) and mutual capacitances can have a dominant influence on the three phase
cable modeling for high frequency PD models and they have to be taken into account
when modeling PD propagation characteristics. The inductances of such lines are
dependent on frequency and the discrete capacitance values can be calculated
independent of frequency.
L1 R1
L2 R2 C12 G12 C13 G13
L3 R3 C23 G23
L4 R4 C14 G14 C24 G24 G44
Figure 7.2: Equivalent circuit and general configuration of three phase cables.
205
________________________________________________________________________
The finite element method (FEM) is one of the most powerful of the numerical
techniques and is particularly well-adapted to perform this type of calculation. The three
phase cable has self capacitance (S) and mutual capacitance (Q) which is shown in Figure
7.3. In this case, the port voltage is used for FEM calculation of capacitance values as
shown in Figure 7.4.
Figure 7.3: The cross section equivalent circuit of three phase cable.
Port 1 (voltage source)
Port 3
Port 2
Port 4
Figure 7.4: Illustration of FEM calculation on three phase cable.

206
________________________________________________________________________
7.2.1 Method I: Calculating parameters with Ohm’s law
To calculate the parameters, there must be at least two electrodes in the system, where
one is grounded and the other one can be either voltage or current on the electrodes.
When solving the time-harmonic situation, the capacitance value can be obtained from:
Im(Y )
C (Equation 7.2)
Z
Y11 = -2.125684e-4 + 3.339564e-8 i [S]

-18 -9
Y21 = 5.854788e - 9.296035e i [S]
-17 -9
Y31 = 3.218531e - 9.26667e i [S]
-15 -8
Y41 = -2.298837e - 3.247986e i [S]
Z 2S f 314.159265 [rad/s] where f = 50 Hz (Equation 7.3)
Im(Y11 )
= 1.063016e-10 C
Z
Im(Y21 )
= -2.95902e-11 C
Z
Im(Y31 )
= -2.949673e-11 C
Z
Im(Y41 )
= -1.033866e-10 C
Z
7.2.2 Method II: Calculating parameters using the energy method
Using this method, the voltage or current is one on one or two ports at a time and the
energy density is integrated over the whole geometry so the geometry can be in any
configuration. The following equations show how to calculate the capacitance matrix
from the integral of electric energy density:
207
________________________________________________________________________
Cii 2 ³ We d: (Equation 7.4)

:
1
Cij ³ W d: 2 (C
:
e ii Cij ) (Equation 7.5)
where We is the electric energy density. The self and mutual capacitance can be obtained:
C11 = 1.019611e-10 C
C21 = -2.95902e-11 C
C31 = -2.949673e-11 C
C41 = -1.033866e-10 C
By using the same methods as the capacitance calculation, the self and mutual
inductances can be obtained by integration of magnetic energy density:
Lii 2 ³ Wm d: (Equation 7.6)

:
1
Lij ³W
:
m d: (Lii Lij )
2
(Equation 7.7)
As the cable consists of large diameter conductors and carries currents with high
harmonics. The current density is hence not homogeneous. The cable parameters are
depended on the frequency. The result of self and mutual inductance variation with
frequency is shown in Figure 7.5. The mutual inductance is reduced with frequency due
to loop currents. In addition, the results in Figure 7.5 show that inductance may not have
a strong influence on the transient or partial discharge analysis of the cable because the
inductances decrease to small values at high frequencies.
208
________________________________________________________________________
Self inductance
Mutual inductance
Figure 7.5: The result of self and mutual inductance in frequency range up to 1MHz.
7.3 Experiment Setup
A 11kV three core XLPE cable was employed in the measurements. The test details are
given in Figure 7.6. The total length of cable was 5 metres with individual semi-
conducting layers for each phase of cable. An artificial PD source was used on one core
at a site one third of the way along the total length. One third of the total cable length can
give more information on PD signal propagation versus cable length. Seven external PD
sensors of the capacitive coupling type were devised by attaching to the sheath at various
locations of the cable. Sensors “a” to “f” are attached at each terminated end of the three
cable cores and sensor “g” is attached at the PD defect site. From the position of the PD
site, sensor “a”, sensor “b” and sensor “c” are located 1/3 of the total cable length away.
From the position of the PD site sensor “d”, sensor “e” and sensor “f” are located 2/3 of
the total cable length away. The benefit of using external sensors is that it allows them to
be put at any desired points along the length of the cable on site. Due to the strong
209
________________________________________________________________________
attenuation of the high frequency PD pulses along the cable, the location of external
sensors along the cable allows good sensitivity of the on-site PD measurement. External
sensors have a major advantage for studying PD signal propagation along the cables such
as higher sensitivity and ease of attachment, further detail in Chapter 5.
Figure 7.6: Laboratory test measurement setup for three phase XLPE cable.
Figure 6.9 (in Section 6.2.2) shows a comparison of the frequency response of available
high frequency current transformers (HF-CTs) and the external sensors used as obtained
with a Network Analyzer (9 kHz to 3000 MHz). As can be seen, the external capacitive
sensors have better sensitivity than the HF-CTs. Both types of sensor have good
bandwidth for PD detection.
When measuring partial discharges online, the three phases are energized simultaneously
with three phase voltage, the resulting electric field becomes elliptical and varies spatially
in time [132]. In order to minimize the effect of electric field variation caused by three
phase excitation only one phase is energized at a time so that the electric field distribution
inside the cable insulation is constant.
The artificial PD defect can generate two PD pulse signals propagating along both cable
ends. The artificial PD defect is created by a sharp pin which can be extended and
210
________________________________________________________________________
retracted relative to the insulation surface. The cross section of the PD source set up is
shown in Figure 6.3 (Section 6.1.1).
With maximum test voltage of 7.5 kV per phase applied to the core with the PD source,
PDs were generated within the defect with an inception voltage of 5.8 kV. The PD pulses
then travel to each end of the cable core. The external sensors are used to capture the PD
pulses as they propagate along the cable.
7.4 PD Evaluation Technique
The PD pattern obtained with the sensors is recorded with an Mtronix MPD540 three
channel PD monitoring system. The Mtronix 540 PD system enables the free choice of
centre frequency and bandwidth of detected signals. Signals from all three channels can
be used to clearly distinguish between noise and PD. The selection of the bandwidth of
the PD measurement system is very important for noise suppression. Due to the complex
propagation and coupling characteristics of the three phase cable, every source of PD
signal has a unique impulse coupling magnitude ratio and consequently a characteristic
signature. To visualize the recorded data gained by the PD measurement of the three
phase cable test, an evaluation technique, called 3 Phase Amplitude Relation Diagram
(3PARD), has been used. This calculates the magnitude ratios of an original test pulse
signal and its nearly simultaneous measured couplings at two other phases and displays it
in a diagram as shown in Figure 7.7.
211
________________________________________________________________________
Figure 7.7: Construction of 3PARD signal analysis [135].
In order to calculate the PD cross-correlation, the PD signal with the highest PD level
(e.g. Unit 1.2) is divided by the PD level of other phases (e.g. Unit 1.2/Unit 1.3 and Unit
1.2/Unit 1.1 respectively) and is transferred into 120o 3PARD. The subtraction of one
(e.g. Unit 1.2/Unit 1.3 -1 and Unit 1.2/Unit 1.1 -1 respectively) makes sure that common
mode signals (same PD impulse levels on all three phases) are visualized at the origin of
the 3PARD as shown in Figure 7.8.
212
________________________________________________________________________
Figure 7.8: Visualization of 3PARD [136].
7.5 Experimental Results
7.5.1 100pC calibrating signal

The measurements were made on 11kV three phase XLPE cable by applying the 3PARD
evaluation method. At the first measurement, a 100pC, calibrating signal was fed to
different phases and used to verify the accuracy of the visualization scheme. Figure 7.9
illustrates the calibrating signal fed into the phase 1 (Unit 1.1), Figure 7.10 shows the
calibrating signal fed into the phase 2 (Unit 1.2) and Figure 7.11 shows the signal fed into
phase 3 (Unit 1.3), respectively. The results show that 3PARD system can be used to
clearly display which phase is polluted by induced coupling of the calibrating signal and
which phase has the direct signal. As the calibrating signal is strong and is easy to
213
________________________________________________________________________
distinguish from noise, the clusters are displayed far from the origin of the 3PARD
diagram.
Figure 7.9: Calibrating signal fed into the phase 1-Unit 1.1.
214
________________________________________________________________________
7.5.2 Measurement under the noisy environment

Another measurement was made with the signal superimposed on a high level of noise
and the PD signal was able to be detected from the polluted signal environment. The PD
signal is created by the artificial defect and the experiment setup is shown in Section 7.3.
The results of this are shown in Figures 7.12 to Figure 7.14. The test voltage used was 6.6
kV and centre frequency was 2.8 MHz with a 300 kHz bandwidth.
215
________________________________________________________________________
bandwidth.
bandwidth.
216
________________________________________________________________________
bandwidth.
Figure 7.15 shows the 3PARD diagram of the PD signals of Figures 7.12, Figure 7.13
and Figure 7.14. The results show that the phase 3 has the highest value of discharge and
this is verified from Figures 7.12 to Figure 7.14. In addition, it can clearly be seen that
the results of Figure 7.15 are extracted from a high background noise due to the lack of a
strong red cluster (a fast and high level pulse such as a PD signal) in the diagram.
Figure 7.15: 3PARD diagram at 2.8 MHz centre frequency and 300 kHz bandwidth
(under the noisy environment)
217
________________________________________________________________________
7.5.3 Measurement under the laboratory
The purpose of the experiment is to distinguish which phase of the 3-core cable has the
artificial PD defect. Each data set (pulse) is assigned to a single dot in the 3PARD
diagram. A colour coded statistical map of the electrical origin of the PD fault is then
generated. Multiple faults can be separated as different clusters of points. Figures 7.16 to
Figure 7.18 show the PDs of each phase of the cable at 1.2 MHz centre frequency and
300 kHz bandwidth. 1.2 MHz centre frequency is chosen in this Section 7.5.3 rather than
2.8 MHz centre frequency in Section 7.5.2 due to better noise rejection and better
sensitivity. Figure 7.16 to Figure 7.18 cannot clearly indicate which phase of cable
contains the artificial defect. Figure 7.19 is the 3PARD diagram and it can indicate PD
source in phase 1. However, the centre frequency at 11.7 MHz can give better indication
of the location of PD source, which is in phase 1 (Unit 1.1). The 3PARD diagram of 1.2
MHz centre frequency and 11.7 MHz centre frequency are shown in Figure 7.19 and
Figure 7.20, respectively. Choosing the correct centre frequency and bandwidth can give
the better results in 3PARD diagram.
Figure 7.16: PD of phase 1 at 1.2 MHz centre frequency and 300 kHz bandwidth.
218
________________________________________________________________________
219
________________________________________________________________________
Figure 7.19: 3PARD at 1.2 MHz centre frequency and 300 kHz bandwidth.
Figure 7.20: 3PARD at 11.7 MHz centre frequency and 300 kHz bandwidth.
220
________________________________________________________________________
7.6 Conclusion
For tests on power cables requiring PD measurements with high sensitivity, capacitance-
coupled sensors attached externally to the cable surface provide the best solution.
Laboratory tests with an artificial defect in one phase generating PD applied on a short
length of 3-core XLPE cable was used to investigate cross-coupling of PD signals
between the phases in the three phase XLPE cable. Understanding of the cross-coupling
of PD signals between the phases helps to develop an accurate cable model for PD
propagation characteristics. The accuracy of defect location depends on how accurate the
propagation time and attenuation of the entire cable is known. Efforts involving
simulation of cable model necessitate the accurate values of the parameters of cable. The
cable parameters of three core XLPE cable such as capacitance and inductances are
shown in Section 7.2. The capacitance coupling external sensors were applied for the PD
detection measurements and the results show that such external sensors have a number of
advantages over HF-CTs when used for detection of PD pulses. In order to visualize the
recorded data obtained from the PD measurements of the three phase cable under test, the
3PARD diagram obtained with an Mtronix PD detector was used. Each data pulse from a
3-channel monitor is assigned to a single dot one the one diagram. The measured results
show that the 3PARD diagram allowed the user to verify the true location of the fault
between phases with PD location, even in the presence of high levels of signal noise.
221
Chapter 8: High Frequency PD Signal Propagation in Three Phase Power Cables
________________________________________________________________________
Chapter 8
High Frequency PD Signal Propagation in Three

Phase Power Cables
8.0 Introduction
This chapter presents the investigation of modeling of partial discharge propagation in

three phase power cables and shows how the high frequency signal on one phase of the
cable is induced on the other two phases. This chapter is a continuation of the modeling
of partial discharge propagation in single phase cables in Chapter 5.
Location and detection of PDs in three-core type three phase cables is rather more
complex than in single phase cables because of the inter-phase coupling of the high
frequency signals. Therefore it is necessary to understand the high frequency signal
propagation and coupling effects in the three phase cable.
One of the problems with three phase power cables is that there is no real viable high
frequency model available for characterizing PD propagation in such cables. This study
presents results of measurements of PD type pulses and high frequency sinusoidal wave
propagation in three phase cables. It also provides details of the development of a
wideband model applicable to the three phase cables for PD propagation studies. For
transient pulse studies, the most important and also most difficult part of the simulation is
inclusion of the frequency dependence of the cable transmission line, because the
simulation requires high accuracy in a wide frequency range, from power frequency to
about a few hundred MHz.
222
________________________________________________________________________
8.1 Historical Development in Three Phase Power Cable

Modelling
In 1970, Hedman [126] evaluated the effect of distributed, constant resistance on single
and three phase transmission lines. The results demonstrate how the resistance produces
distortion in the travelling wave and shows that the ground component resistance has a
significant influence on the coupled voltage.
Brown, Dugan and Rocamora [127, 128] give details of a series of three phase pipe type
cables investigation in 1976-77. In part I [127], the general equations for the impedance
of conductors inside a pipe are presented and used to calculate the step response of an
idealized pipe type cable. In part II [128], the theory is applied to develop a surge
propagation model for an actual pipe type cable and proximity effects in the cable
elements are represented in the model.
Darcherif [129] showed the application of finite element methods to the computation of
three phase cable characteristics in the frequency domain by means of the solution of the
diffusion and Laplace equations. In order to improve the solution accuracy and the
computational times, Darcherif and Raizer [129, 138] introduced two techniques,
discretization and Infinite space technique. Both techniques are used to improve the
solution accuracy and the simulation time. The first one is the so-called P and H type
refinement procedure, which is discretizing the mesh into smaller elements and by
increasing the order of the trail function while using the same mesh. In adaptively
increasing the order of the approximations, the advantage is that high order trial functions
have better convergence properties compared to simple first order elements. The second
technique estimated the result by dividing the space into two subdomains: one of them in
finite and contains the whole studied geometry; the other one, which extends to infinity
and contain air only is mapped onto a finite subdomain by the mean of a spatial
transformation. Further detail can be found in J. F. Imhoff’s article [139].
223
________________________________________________________________________
Felizani [130] demonstrated an FEM-Impedance network boundary conditions (INBC)

procedure to calculate efficiently the impedance matrix of multi-transmission line which
can overcome the limitation of the traditional FEM in modelling lossy conductors at high
frequency.
P.C.J.M. Van Der Wielen [131, 132] measured the PD activity on-line in a three phase
belted medium voltage cable system results in number of differences as compared to the
off-line measurement where only one phase is energized. The results show that the effects
of induced currents and charges in the phase conductors and earth screen upon a PD not
only depends on the discharge site but also become phase angle dependent.
Mulitconductor lines have been used in the EMTP by a transformation to natural modes
to diagonalise the matrices involved. The problems in EMTP are casued by inaccuracies
in the modal domain. Noda [72] in the 1990s have introduced the multi-conductors
possibility on phase domain and Noda’s cable model have beem applied on the well-
known EMTP progam: ATPDraw [72]. However, the Noda’s cable model in ATPDraw
still has not been tested carefully and give results that are unstable during simulation.
Gustavsen[125] has developed the model based on the phase domain which gives the
stable results.
This Chapter described the novel investigations on three phase cable modelling, which is
based on the wideband cable model [137] in EMTP-RV software. The compatibility of
the cable model includes the necessary means for PD propagation study such as the semi-
conducting layers which need to be modeled and the implemented on the paper insulated
cable and XLPE cable.
224
________________________________________________________________________
8.2 Propagation Measurements
8.2.1 Meaurement Setup
Two types of three phase cables were employed in the laboratory tests: one was a 22 kV
three core paper insulated lead covered (PILC) cable of 10 metres length and the second
was an 11 kV three phase cross linked polyethylene (XLPE) cable of 30 metres length.
Three types of measurements were performed for the comparison of experimental results
with the cable model predictions.
The first test used a network analyzer with frequency response analysis (FRA) circuit
measurement to determine the frequency response magnitude (S21) and the frequency
spectrum on each phase of the three phase cables. The measurements were carried out
from several kHz to 300 MHz. The circuit was matched by using resistors of 82 : in
parallel with the input of Network Analyzer to simulate an impedance of about 33 : to
match that of the cable.
The second test involved a measurement of the voltage response on the outer and inner
semi-conducting layers of the cable, as for the single phase cables. As the layers represent
a capacitively coupled connection, their voltage will be related to the signal voltage level.
The purpose of this test was to determine the viability of using the semi-conducting
layers as capacitive probes to detect PD signals. This test used the 100 pC calibrating
signal as the voltage source with a short circuit at the receiving end of the cable.
In the final test, a standard PD calibrating signal source was used to inject signals into the
cable to simulate the practical situation. The sending and receiving end signals were
monitored with the capacitance-coupled sensor to determine the attenuation level and
propagation velocity of the signals. The test setup is shown in Figure 8. 1. The signal was
applied between the conductor at left (the sending end) and the (earthed) cable sheath.
The other end of the cable (the receiving end) had the conductor short-circuited to the
225
________________________________________________________________________
sheath. Each phase is monitored with a capacitance coupled sensor (the sheath) to capture
the PD signals.
Figure 8.1: The measurement setup of calibrating signals propagation.
Capacitance coupled
sensor
Figure 8.2: The diagram of three phase XLPE cable in laboratory.
8.2.2 Structure of Three Phase Cables
In this experiment, two different types of three phases cable were used:
226
________________________________________________________________________
A. Three core belted paper cable (Figure 8.3)

Three single conductors in oval shape with paper insulation over the conductors. There is
no individual core sheath. A single lead sheath is extruded over all three cores and fillers
and the cable is aluminium-wire armored over the lead sheath. There is no semi-
conducting layers over the conductors and insulation.
Figure 8.3: 22 kV three phase core belted cable
B. Three phase XLPE cable

The three conductors are stranded compacted aluminum with circular cross-section as
shown in Figure 8.4. Each concentric phase insulation layer has inner and outer semi-
conducting layers on the XLPE. (resistivity: 1x103 to 106 :-cm). The thickness and
conductivity of the semi-conducting layers has a dominant effect on PD signal
attenuation and velocity [8] and this is an important consideration in the modeling. There
is an outer metallic (copper tape) screen on each phase. There is a PVC sheath over the
filler material and a high density polyethylene outer jacket.
Table 8.1 shows a comparison of the insulating materials such as Oil paper and XLPE
which are the components of XLPE and PILC cables.
227
________________________________________________________________________
Figure 8.4: 11kV three –core XLPE insulated three phase cable.
Table 8.1: Characteristics of high voltage insulating materials [4].
Performance
Material Dielectric Loss Factor Thermal Operating Softening
Constant tanG (%) Resistivity Temperature Temperature
o
(H) U (therm-ohm) T0 ( C) Tmax ( oC)
XLPE unfilled 2.2 0.10 350 90 135
EPR 3.3 2.25 610 90 135
Oil Paper 3.5 1.00 500 80 -----
8.2.3 Frequency Response Measurements
Figure 8.5: The circuit for Frequency Response Analysis over 300 MHz.
228
________________________________________________________________________
The frequency response measurement set up is shown in Figure 8.5. The measurement is
carried by using a wide band (R&S–Rohde & Schwarz ZVL) Vector Network Analyzer
as shown in Figure 8.6. For 30 meter (6.35/11 kV) three phase XLPE cable, the
frequency range is from several kHz to 300 MHz, for S 21 which simulated irec / isend when
the circuit is matched by using the resistors of 80 in parallel with the inputs of Network
Analyzer to simulate an Impedance of about 33 Ohms of the cable.
Figure 8.6: A diagram of R&S & ZVL Vector Network Analyzer.
0.80
0.60
Ratio
0.40
0.20
0.00
0 100 200 300
Frequency (MHz)
Figure 8.7: Frequency response magnitude (S21, Damping ratio, paper insulated cable) is
measured for ratio of Phase Asending (Input) to Areceiving (Output).
229
________________________________________________________________________
0.80
0.60
Ratio
0.40
0.20
0.00
0 100 200 300
Frequency (MHz)
Figure 8.8: Frequency response magnitude (S21, Damping ratio, XLPE cable) is measured
for Phase Asending (Input) to Areceiving (Output).
Figure 8.7 and Figure 8.8 show the frequency response magnitiude of three phase belted
cable and three phase XLPE cable. More measured results can be found in Appendix II.
In the Appendix, Figures A2.1 to Figure A2.12 show the measurements of frequency
response magnitude (S21, Damping ratio) and frequency spectrum for the three core paper
belted cable when sinusoidal signals are injected. Figure A2.13 to Figure A2.18 show the
same quantities for the three phase XLPE cable. Both cables demonstrate a high
damping ratio when a high frequency signal (greater than about one MHz) is propagatng
through the cable. This damping ratio is higher in the three phase belted paper cable
compared to the XLPE cable due to the higher dissipation loss factor of paper. For
example for 25 Mhz, the paper belted cable has damping ratio of 0.2 and XLPE cable has
damping ratio of about 0.4 generated in the semi-conducting layers. Both Figure 8.7 and
Figure 8.8 indicate that the attenuation of high frequency signals in paper insulated cable
is higher than in XLPE cable. The results match with the conclusion in Section 5.6.2.
Figures A2.7 to Figure A2.12 show the strong coupling effect between each phase of
cable and the figures indicate low damping ratio between the lowest frequency and 20 M
Hz. Figure A2.19 to Figure 2.22 show the damping ratio of semi-conducting layers and
the results indicate the highest damping ratio at 50 MHz. As the dissipation loss factor of
230
________________________________________________________________________
paper is higher than that of XLPE, the reverse would be the case without the semi-
conducting layers where cable damping would be higher in that case.
8.3 Measurement of Voltage Response to PD calibration signals

using the Semi-conducting Layers
The high frequency partial discharge pulse that propagates through the cable is an
electromagnetic wave. Its electric field charges the equivalent capacitances between
different layers of cable such as semi-conducting layers. Chapter 5 showed that semi-
conducting layers damp the high frequency signal significantly and attenuation rises
nonlinearly with increase of frequency. A calibrator voltage signal, simulating 100 pC
partial discharge, as shown in Figure 8.9 and a -100pC digitial of calibrating signal, as
shown in Figure 8.10. 100pC calibrator signal has wider signal duration, 0.1μs and lower
amplitude. -100pc calibrator signal duration is very short, 0.01μs but it has higher
amplitude, which is 0.78 V.
0.03
0.02
Voltage (V)
0.01
0.00
-0.40 -0.20 0.00 0.20 0.40
-0.01
Time (μs)
Figure 8.9: 100pc slow calibrating voltage signal from the calibrator.
231
________________________________________________________________________
0.2
0
-0.10 -0.05 0.00 0.05 0.10
Voltage (V)
-0.2
-0.4
-0.6
-0.8
Time (μs)
Figure 8.10: -100pc fast calibrating signal from the digital calibrator.
0.15
0.10
0.05
Voltage (V)
0.00
-1.00E-07 1.00E-07 3.00E-07 5.00E-07 7.00E-07 9.00E-07
-0.05
-0.10
-0.15
Time (s)
SendA_innersemi SendB_innersemi SendC_innersemi
of inner semi-conducting layers (XLPE cable) on the sending ends are shown in graph.
232
________________________________________________________________________
0.08
0.06
0.04
Voltage (V)
0.02
0.00
-1.00E-07
-0.02 1.00E-07 3.00E-07 5.00E-07 7.00E-07 9.00E-07
-0.04
-0.06
-0.08
Time (s)
RecA_innersemi RecB_innersemi RecC_innersemi
of inner semi-conducting layers (XLPE cable) on the receiving ends are shown in graph.
The measured voltage response on inner semi-conducting layers of phase A, phase B and
phase C of XLPE cable are shown in Figure 8.11 and Figure 8.12, respectively. -100pc
calibrating signal (Figure 8.10) from the digital calibrator is energized on phase A. The
ratio of receiving end and sending end voltage on phase A is 0.07/0.11 =0.63.
Considering that the induction of voltage on inner semi-conducting layer of phase A with
respect to the applied voltage is 0.11/0.75 = 0.147 and the voltage of inner semi-
conducting layer of phase A at receiving end has a ration of 0.147 x 0.63 = 0.095 0.1.
This result can demonstrate the advantage on using semi-conducting layer as a sensor for
partial discharge detection, which is discussed in Chapter 5. In addition, the results show
the semi-conducting layers have dominant role on attenuation of power cables. More
results on voltage response by applied the -100pc calibrating signal can be found in
Appendix III. Comparing the voltages captured from the inner and outer semi-conducting
layers (in Appendix III), it can be seen that the inner semi-conducting layers give higher
sensitivity than outer semi-conducting layer.
233
________________________________________________________________________
0.06
0.04
0.02
Voltage (V)
0
-1.00E- 1.00E-07 3.00E-07 5.00E-07 7.00E-07 9.00E-07 1.10E-06
-0.02
07
-0.04
-0.06
Time (s)
semi-conducting layers on the sending end are shown in graph.
0.06
0.04
0.02
Voltage (V)
0
-1.00E- 1.00E-07 3.00E-07 5.00E-07 7.00E-07 9.00E-07 1.10E-06
-0.02
07
-0.04
-0.06
Time (s)
SendA_innersemi RecA_innersemi
RecB_innersemi RecC_innersemi
semi-conducting layers on the receiving end are shown in graph.
234
________________________________________________________________________
The voltage of phase B and C can be compared with the voltage at phase A on inner
semi-conducting layer of cable at sending end and receiving end are shown in Figure 8.11
and 8.12, respectively. Phase A is energized by -100pc, fast calibrating signal from the
calibrator, which is shown in Figure 8.10.
Vsending B/ Vsending A =0.52 (Equation 8.1)
Vsending C/ Vsending A =0.40 (Equation 8.2)
Equation 8.1 and 8.2 show the voltage is inducted on semi-conducting layers of other
phase approximately with a ratio of about 0.5~0.4.
Vreceiving B/ Vreceiving A =0.158 (Equation 8.3)
Vreceiving C/ Vreceiving A =0.33 (Equation 8.4)
The coupling ratio at receiving end reduced from about 0.5~0.4 to about 0.3~0.16 in
average for 30 meter length of cable.
The voltage of phase B and C can be compared with the voltage at phase A on inner
semi-conducting layer of cable at sending end and receiving end and are shown in Figure
8.13 and Figure 8.14, respectively. Phase A is energized by 100pc, slow calibrating signal
from the calibrator, which is shown in Figure 8.9.
Vsending B/ Vsending A =0.52 (Equation 8.5)
Vsending C/ Vsending A =0.57 (Equation 8.6)

235
________________________________________________________________________
Equation 8.5 and 8.6 show the voltage is inducted on semi-conducting layers of other
phase approximately with a ratio of about 0.5.
Vreceiving B/ Vreceiving A =0.34 (Equation 8.7)
Vreceiving C/ Vreceiving A =0.47 (Equation 8.8)
The coupling ratio at receiving end is reduced from about 0.5 to about 0.4 in average for
30 meter length of cable. This shows an intensive coupling effect of any PD signal
existing on one phase to the other two phases. There is an 8% difference between phase B
and C at the receiving end due to the different characteristic, length of the coaxial cables.
The low percentage of error can be accepted. Further measured results can be found in
Appendix IV. Moreover, the ratio of receiving end and sending end can be obtained by
Equation 8.9:
Vreceiving A / Vsending A (Equation 8.9)
For the 100pC slow calibrating signal in time domain (lower frequency in frequency
domain), the Vreceiving A / Vsending A is 0.72. For the -100pC fast calibrating signal in time
domain (higher frequency in frequency domain) , the Vreceiving A / Vsending A is 0.504.
Comparing the Vreceiving A / Vsending A with the damping ratio in Figure 8.8, both results
show the similar trend: Vreceiving A / Vsending A decrease as frequency increases and
damping ratio decreases as frequency increases.
Comparing the results of fast (-100 pC) and slow (100 pC) calibrating signal, both results
show the strong coupling effect of each phase at the sending end: Phase B and phase C
have approximate half (value =0.5) of the Phase A (phase A is energized by the
calibrating signal). For the -100 pC fast calibrating signal at receiving end, phase B and
Phase C have approximate 0.3~0.2 of phase A, which show the smaller coupling effect
than sending end. For the 100 pC slow calibrating signal at receiving end, phase B and C
236
________________________________________________________________________
have approximate 0.4 of phase A, which is higher than -100 pC fast calibrating signal at
receiving end. The fast calibrating signal has a higher attenuation than the slow
calibrating signal. Both calibrating signals have the same charges but one is positive and
other one is negative. Moreover, the 100 pC slow calibrating signal has the wider pulse in
(time) X-axis, shorter voltage level in (voltage) Y-axis, stronger coupling effect and
higher attenuation.
Figures A3.1 to Figure A3.4 show voltage responses of semi-conducting layers by

applying -100 pC fast calibrating signal from the calibrator, the outer semi-conducting
layers have lower sensitivity than inner semi-conducting layers, which is 60% less
sensitivity. Figures A4.4 to Figure A4.6 show voltage responses of semi-conducting
layers by applying 100 pC slow calibrating signal from the calibrator, the outer semi-
conducting layers have approximate 67% less sensitivity than inner semi-conducting
layers. The longer length of cable and higher dielectric constant of insulation give the
lower damping ratio. Further detail can be found in the Section 4.4.1 and Section 5.7. All
the information is helpful for the design of measurement testing and cable modelling.
In summary, the measurement of voltage at semi-conducting layers, when applying

calibrating signal of 100 and -100 pC to phase A demonstrates a high mutual induction of
signal on phase B and phase C, which appears along the length of cable. Those other
cable parameters such as the length of cable, the cable geometry and the frequency
content of PD signal have a dominant effect on damping ratio and attenuation. Therefore
a valid three phase cable model can help to estimate the induced voltage properly (in
Section 8.4) and be used in three phase cable PD location.
237
________________________________________________________________________
8.4 Cable Model
Partial Discharge contains a wide range of high frequency components and the cable
models should be frequency dependent and have the capability of modeling the semi-
conducting layers. The cable model in this study of a three phase cable used the wideband
cable model of EMTP RV. It is the most sophisticated three phase cable modeling
technique currently available in EMTP. It represents the complete frequency dependence
of parameters and constitutes a wideband model. The backward Euler integration methed
is used in this cable model and the simualted result can be achieved through a phase
domain formulation where the model characteristics have been utilized in the
apporximation for the propagation matrix. This cable model is based on fitting the
matrics for propagation H and characteristic admittance Yc in the phase domain by
solving a linear matrix equation as a least squares problem. The matrices for propagation
H and characteristic admittance Y c can be expresed as [125]
H exp( YZl (Equation 8.10)

1
Yc Z ZY (Equation 8.11)
Where Z and Y are the series impedance and shunt admittance per unit length, and l is the
length of transmission line. For an n-conductor system, Z and Y are n n matrices. Using
modal decomposition, H can be expressed as:
n n
¦ * (Z )e O ¦H
/ (Z ) l
(Z ) l
H (Z ) Te T 1 k
k '
k (Z )e jZW k (Equation 8.12)
k 1 k 1
Where is diagonal, and T os the eigenvector matric. k is the square matrix, which
results from multiplying the kth column of T with the kth row of T-1. Each modal
contribution to H can be further expressed as:
* k (Z )e Ok (Z ) l * k (Z )e Ok (Z ) l (Z )e jZW k
'
H k (Z ) (Equation 8.13)
238
________________________________________________________________________
The fitting of H is based on poles and time delays obtained from the modes. At each
frequency point 1, H can be written as:
n
ª Nk cmkij º jZ1W k
H ij ( jZ1 ) | ¦ « ¦ »e (Equation 8.14)
k 1 ¬ m 1 jZ1 pmk ¼
Where Nk is the number of poles for mode k. In addition, very accurate simluated result
can be achieved with a low number of poles per mode, as compared to currently used
models in EMTP. The inconvenience is the computational speed. The cable model can be
accessed with the semi-conducting layers and it is a big advantage for three phase XLPE
cable modelling. More theoretical details can be found in [137].
Figure 8.15 shows a pipe type cable configuration consisting of three single core cables
enclosed within a pipe. This design can be directly modeled using the “pipe type”
representation which is available in EMTP RV software. Each single core cable can be
either individually sheathed (three phase XLPE cable as shown in Figure 8.30) or without
sheathed (e.g. three phase paper belted cable). The pipe type system requires the
geometrical and electrical property of cable in order to simulate an accurate simulation.
Figure 8.15: Three phase XLPE cable design in pipe type modeling.
239
________________________________________________________________________
Figure 8.16 and Figure 8.17 show the simulation circuit of three core belt paper cable and
three phase XLPE cable. The wideband cable models are used to model both cables. Due
to three core belt paper cable doesn’t include the semi-conducting layers, the cable model
has relatively simple circuit as shown in Figure 8.16. The cable core conductors for phase
A, B and C are shown from top to bottom alternatively.
Figure 8.17 shows that there are nine blue tubes in the middle of diagram and they are
represented the three conductors of cable (first, forth and seventh), three inner semi-
conducting layers (second, fifth and eighth) and three outer semi-conducting layers (third,
sixth and ninth). Only core conductor of phase A is energized by calibrator voltage signal
and the simulated results are shown in both sending and receiving end. The simulated
results can be found in the following section.
Figure 8.16: The simulation circuit of three core belt paper cable.
240
________________________________________________________________________
Figure 8.17: The simulation circuit of three phase XLPE cable.
8.5 Use of PD calibrator signal for measurements
A standard PD calibrator voltage pulse signal, as described above in section 8.2,

simulating a single 1000pC partial discharge was used in the cable tests as shown in
Figure 8.1. The calibrator was well regulated in its output and it was not loaded to any
significant degree by the various cable sections tested. In this section, the simulation of
wideband cable model is presented. In order to establish the validity of cable model, error
computation between measurement and simulation is implemented: Compare the ratio of
receiving and sending end of V peak to peak for both measurement and simulation.
241
________________________________________________________________________
8.5.1 Three core paper belted type cable

The calibrator voltage signal attenuation during propagation was measured. In order to
determine the coupling effect of PD signal exciting on one phase to the other two phases,
only phase A of cable is energized by PD calibrator pulse signal. The voltage of the phase
B and phase C can be compared with the voltage at phase A at sending and receiving end.
The measured voltage responses at the sending end and receiving end of three phase
paper belted type cable (10 meter long with the dielectric constant, 3.3) are shown in
Figures 8.18 to Figure 8.22. From the measurements, the ratio of V receiving end / V sending end
is 0.78 and the travel time of the signal is 0.09 microseconds. The simulated results with
the cable model give a ratio of V receiving end / V sending end as 0.76 and the travel time of the
signal as 0.11 microseconds. The experimental propagation velocity was determined from
the cable length and the travel time of the measurements [8]. The velocity of propagation
on paper belted cable is 1.11x108 m/sec.
Comparing the experimental and simulated results, both results are verified and match
almost perfectly. In Figure 8.19, the simulated result of phase A receiving side does not
give exactly same outcome as measurement and the % error of peak to peak amplitude
and time are 4% and 9%. In Figure 8.21, the measurement of phase B signal is in noisy
background due to its low voltage value and the capacity of oscilloscope. However, the
result of measurement and simulation are approximated.
As described in Chapter 7, three phase (3-core) cables cross-coupling of PD signals

between phases can give false indications of PD detection. The peak signal in phase A is
about 0.01 as shown in Figure 8.18. About 15 % are induced in phase B and C at the
sending end. The conductors of paper belted type cable is surrouded by insulation
(without the semi-conducting layers) and there are capacitance between each core of
cable. This is important to know the capacitance value in order to take the coupling effect
into account when doing the cable modelling. The calculation of those capacitance values
are shown in Section 7.2 in Chapter 7.
242
________________________________________________________________________
Figure 8.18: The measured and simulated result of Phase Asending of three phase paper
belted type cable.
Figure 8.19: The measured and simulated result of Phase Areceiving of three phase paper
belted type cable.
243
________________________________________________________________________
Figure 8.20: The simulated result of Phase Bsending and Phase Breceiving of three phase paper
belted type cable.
Figure 8.21: The measured result of Phase Bsending and Phase Breceiving of three phase paper
belted type cable.
244
________________________________________________________________________
Figure 8.22: The simulated result of Phase Asending, Phase Areceiving and Phase Bsending,
Phase Breceiving of three phase paper belted type cable.
8.5.2. Three phase three phase XLPE cable

The measured voltage responses at the sending end and receiving end of three phase
XLPE cable are shown in Figures 8.23 to Figure 8.26. Three phase XLPE cable
(6.35/11kV) has the length of 30 meter and dielectric constant of 2.2. From the
measurements, the ratio of Phase A receiving end / Phase A sending end is 0.499 and the travel
time of the signal is 0.18 microseconds. The simulated results show the ratio of Phase A
receiving end / Phase A sending end is 0.5 and the travel time of the signal is 0.18 microseconds.
The measurement shows the ratio of Phase B receiving end / Phase B sending end is 0.43 and the
simulated result is 0.38. The velocity of propagation on the XLPE cable was thus v = 30/
(0.18 x 10-6) = 1.66 x 108 m/sec.
Comparing the measurement and simulated results of Phase A and B, Phase A results are
matched perfectly and Phase B results are approximately matched. Measured results of
Phase B have more damping than the simulated results and there is less than 4% error of
peak to peak amplitude and 9.7% error time. The voltage of the phase B can be compared
with the voltage at phase A. The peak to peak amplitude of phase A is about 0.3 V as
shown in Figure 8.23. About 27% are induced in phase B at the sending end. XLPE
245
________________________________________________________________________
cable has higher coupling effect to the other phases than Paper belted type cable, which is
15%.
Figure 8.23: The measured and simulated result of Phase Asending of three phase XLPE
cable.
Figure 8.24: The measured and simulated result of Phase Areceiving of three phase XLPE
cable.
246
________________________________________________________________________
Figure 8.25: The measured result of Phase Bsending and Phase Breceiving of three phase
XLPE cable.
Figure 8.26: The simulated result of Phase Bsending and Phase Breceiving of three phase
XLPE cable.
For paper belted cable, paper insulation is between each phase and the mutual
capacitance between each phase can have a dominant influence on the three phase cable
modeling for high frequency PD models. Therefore, mutual parameters such as mutual
inductance and capacitance have to be taking into account when modeling PD
247
________________________________________________________________________
propagation characteristics and this is the main difference between Single core cable
modeling and three core power cable modeling. The XLPE cable has the individual semi-
conducting layers and sheath for each conductor. As previous discussion in Chapter 5 and
Section 8.3, the signals of semi-conducting layers are high enough to treat as a new
sensitive sensor of partial discharge. Therefore, each phase of XLPE cable can have the
strong inter-phase coupling.
The attenuation of the paper belted type cable is smaller than XLPE cable due to the
longer cable length of XLPE cable and the reflection from the cable termination of paper
belted type cable. The longer length of cable can give higher attenuation. In addition, the
loss from semi-conducting layers from XLPE cable can be one of the reasons. This is an
indication of the importance of the semi-conducting layer in modifying the propagation
characteristics. In addition, the traveling speed of XLPE cable is higher than paper belted
speed.
Compare Figure 8.18 to Figure 8.22 with Figure 8.23 to Figure 8.26, the XLPE modeling
give a more accurate cable model than paper belted type cable. The results can show the
viability of this cable model and the measured results are verified with simulated results.
In addition, the results show this cable model can handle inter-phase coupling adequately.
Table 8.2 shows the % error between the measurement and simulation. The results show
this cable model having accurate modeling on amplitude of signals, which is less than 5%
error in average. However, for the time-axis, the %error is more than 10% error on
average. Due to the low % error of amplitude, this cable modeling can give accurate
analysis on attenuation. The phase B has the higher %error of time than Phase A, which
is mainly because of the coupling effect and this problem can be considered as the future
improvement for this cable model.
248
________________________________________________________________________
Table 8.2: Error in fitting with the experimental results in Section 8.5.
% Error between measurement and

simulation
Amplitude peak-peak Time
Paper belted Phase A Sending end 0.2% 0.01%
type cable
Receiving end 4% 9%
Phase B Sending end 1.8% 10%
Receiving end 4.7% 19.5%

XLPE cable Phase A Sending end 3% 1%
Receiving end 6% 3%
Phase B Sending end 4% 9.7%
Receiving end 7% 4.2%
8.6 Conclusion
This Chapter presents the characteristics of partial discharge in two different types of
three phase power cables: XLPE cable and paper belted type cable. The high frequency
response results show that the attenuation increases dramatically in XLPE as frequency
increases. This indicates that the semi-conducting layer has a high sensitivity to partial
discharge effects and will both attenuate and slow the propagation. Thus, the semi-
conducting layers must be considered when developing a cable model for high frequency
signal propagation. The results demonstrate that the developed cable model can model
three phase power cables in terms of their PD propagation characteristics and it can
handle the inter-phase coupling perfectly. The percentage of error between measurement
and simulation are less than 15%. The three phase cable modeling needs to consider the
effect of semi-conducting layers and the inter-phase coupling effect such as mutual
parameters (capacitance) between each phase. In addition, the results also show that the
attenuation of high frequency signals in an XLPE cable is much higher than in paper type
cable due to the influence of the cable length, reflection from the cable termination and
249
________________________________________________________________________
semi-conducting layers. This result contradicts the results shown in Figure 8.7 and Figure
8.8 (Section 8.2.3), which indicate the attenuation of three core paper type cable are
higher than for XLPE cable. Section 8.2.3 shows the results in frequency domain and
Section 8.5 shows the results in time domain. In Section 8.2.3., the average attenuation
level of paper type cable is higher than XLPE cable. However, in Figure 8.7 and Figure
8.8 at low frequency range: 0 to 10 MHz indicates the XLPE cable has higher attenuation
than paper type cable and the attenuation may be affected by many factors as the previous
section mentioned. In Section 5.6.2, it is shown that the attenuation level is affected by
the length of cable and the termination of cable such as measurement impedance may
also affect the result by the reflection of signals, In Section 4.4, Figure 4.19 to Figure
4.22 show how the dielectric constant of insulation and material of semi-conducting layer
affect the velocity and attenuation level in wide frequency wide. The higher dielectric
constant gives higher loss. In addition, Figure 4.23 and Figure 4.24 assume the paper type
cable and XLPE cable have the same material but difference geometry and the results
show the cross section geometry of cable can have some influence on attenuation and
velocity, which is attenuation of XLPE cable higher than paper type cable.
Chapter 9: Conclusion and Future Study 250
________________________________________________________________________
Chapter 9
Conclusion and Future Study
The ultimate aim of this thesis work was to develop an accurate frequency dependent
cable model for detecting, categorizing and locating degraded insulation regions on
power cables, caused by partial discharge activities. The ultimate aim of this thesis is
achieved and it is presented in Section 5.6, Section 5.7, Section 5.8, Section 6.3, Section
7.5, Section 8.2, and Section 8.5. A number of challenges arise from any attempt to
achieve such partial discharge detection and location methods. These include optimal
sensor selection and application, material parameter investigation and propagation
characteristics of the signals. This required investigations of sensor sensitivities,
measurement of cable material properties and a large number of laboratory tests. In
addition, the thesis work involved the development of theoretical models, using the
measured characteristics, to establish a valid high frequency cable model that describes
the propagation characteristics of the cables. The model validity is tested by comparison
with experimental results.
The research results presented in this thesis are primarily concerned with cable modeling
of and experimental tests on single and three phase power cables. The results show that
the high frequency characteristics of semi-conducting materials plays an important role in
the propagation of high frequency PD signals and investigation of such semi-conducting
layers provide detail of the quantitative effect on the overall PD signal attenuation and
velocity of PD signals in the cable.
________________________________________________________________________
Finally, the research results demonstrate the successful development of a method to

model cross-coupling of PD signals between phases in three core cables and how to
locate the PD source in a three phase cable using a three channel PD monitoring system
and comparing the three signals obtained.
The novel contributions in this thesis are Section 4.3: the field analysis of cavity within
dielectric; Section 4.4 and Section 4.5: Influence of different parameters of semi-
conducting and insulating layers; Section 5.6: High frequency pulse measurement and
analysis; Section 5.7: Propagation constant with cable length; Section 5.8: Reconstruction
of signal source; Section 6.3 and Section 6.4: Detection and evaluation of artificial defect
and partial discharge pulses; Section 7.2, Section 7.4 and Section 7.5: Investigation on
PD characteristics and defect detection in three phase power cables and Section 8.2,
Section 8.3 and Section 8.4: PD signal propagation in three phase power cable and cable
modeling.
The five research goals in Section 1.1 are achieved. The first goal is to develop both high
frequency single and three phase cable models and verify them with test measurements
on cable samples in the laboratory. The cable models utilize the characteristics of
materials used for the semi-conducting and insulating materials, including XLPE and
paper insulation types. The simulated results give good match with the measurements,
which show the validity of cable model. In the future, the cable models may apply on
different type of cables to verify the validity on other types of cables.
The second goal is to determine the high frequency characteristics of semi-conducting

materials and investigate the influence of such layers on the overall PD signal attenuation
and velocity in the cable. The results are presented in the Chapter 4. The simulated results
also presented in different type materials and geometry of cable, which covered most of
the different condition of both paper type and XLPE cables. For the improvement, some
of the simulations can compare with the measurements.
________________________________________________________________________
The third goal is to determine the field distribution in the vicinity of a cavity with varied
physical conditions and the results are present in Section 4.3. For the improvement, the
works can involve more than two cavities and investigate the field distribution with
different physical conditions.
The fourth goal is to develop a laboratory’s partial discharge monitoring systems and
measurement techniques for the simulation of partial discharge signal. The PD patterns
are detected by different sensor bandwidths. In the future, the measurement set up can be
improved by better equipments such as oscilloscope and better grounding of
measurement.
The final goal is to develop a method to model cross-coupling of PD signals between

phases in three core cables and to locate the PD source in a three phase cable in order to
compare with the measurements. The results in Chapter 7 show the method to locate the
PD source in three phase cable by analysis the coupling effect between phase and their
characteristics. In the future, this method can try to apply on site and verify the validity of
the method under different environment.
9.0 Partial Discharges in Cavity
Partial discharges in a cavity often appear in extruded solid dielectric insulated cables.
Such PD activity is significant because extruded dielectrics are more prone to damage
from partial discharge than oil-paper insulation. In the case of partial discharges in a
cavity, the discharges develop from one part of the cavity surface, through the air in the
cavity, to another part of the cavity surface. Thus the discharges only bridge the cavity
and they do not develop through the solid dielectric to reach the electrodes. However they
do cause continual damage that eventually can result in total insulation failure. One
necessary condition for an electric discharge to start in air is the electric field strength
must be high enough. Thus the electric field in the cable must be determined and this can
only be done by calculation.
________________________________________________________________________
In general, the analytical method can calculate the electric field strength in the cavity and
dielectric. However, the field distribution in the vicinity of a cavity of non-standard shape
is difficult to calculate by analytical methods. A numerical method is discussed and
applied for the cavity field calculation in Chapter 4. The simulated results show the
important influence of the shape of the cavity on the electric stress within it. As is well-
known, the stress in the cavity is enhanced because of the lower permittivity and this is
an important consideration for PD activity. The cavity stress enhancement increases as
the permittivity of the dielectric increases. The increase is greater for cavities with large
diameter to thickness ratios. A cavity with its axis parallel to the applied field direction
has a higher stress enhancement. The stress in a cavity is influenced by the presence of
any nearby cavities in the radial direction and the stress increases with increase in the gap
between the two cavities. The increase in stress is more pronounced with change of
horizontal axis of cavity vs. thickness of insulation than with variation of vertical axis of
cavity vs. horizontal axis of cavity. In addition the stress distribution in the cavity is
smaller for spherical cavities than for cylindrical types.
9.1 Influence from Semi-Conducting layers
The research results show the development of a model for high frequency
characterization and with inclusion of semi-conducting layers and the dielectric loss of
the insulation material in the model. The high frequency response results using the
voltage response of the semi-conducting layers show that the semi-conducting layers
response voltage increases as frequency increases. This indicates that the semi-
conducting layers can have high sensitivity for detection of partial discharge signals and
this may be a useful feature to incorporate in the design of cables and in the application
of cable models. While the impact of the semicon layer on propagation has been known
to previous workers, it has not been included in detailed models previously.
________________________________________________________________________
The results also show that dielectric loss of the insulation material also has great
influence on the signal propagation characteristics of power cable, as expected. The value
of tan of cable insulation is greatly dependent on the type of insulating material and its
condition. This is of particular interest in the use of PD diagnostics of service-aged cable,
where the loss factor may be increased due to ageing.
The higher loss factor of impregnated paper as compared to XLPE means that PD signal
attenuation along paper insulated cables will be more severe than in XLPE insulated
cable if consider the loss factor alone. The thicker of semi-conducting layers can have
better stress control also higher PD signal loss.
9.2 Sensor Selection
The most widely applicable method for PD detection in cables utilizes a current
transformer installed around the cable sheath or clipped on the earth strap. In general,
underground cables are applied for power distribution and electromagnetic waves
generated by partial discharges can only be detected easily at the cable ends. However,
the sheath champing sensors can be applied any point for detection along the cable and
sheath champing sensors can have better sensitivity than current transformers. By using
the sheath champing sensors, the voltage response of semi-conducting layers can be
obtained.
________________________________________________________________________
9.3 Cable Modeling
Chapter 3 of this thesis presents the general fundamental theory of cable modeling
methods. The discussions are followed by an overview of the recent research on high
frequency cable modelling and its verification procedure. An ATP cable model and a
Finite Element Method analysis were used to model the cable characteristics. The results
demonstrate that the ATP cable model can model the single core cable adequately. By
using the EMTP-RV software, the three phase cable model is developed and its simulated
results are good agreement with the laboratory measurements.
9.4 HF Measurement and Simulation
Chapter 5 of the thesis addresses the laboratory experimental procedure and results. It
gives a detailed outline of test configurations used including the PD sensors and the semi-
conducting layers. This chapter also presents the experimental measurements on single
core cables using time and frequency domain methods. The measured results are
compared with the simulated results to determine the feasibility of the model developed.
A comparison is given based on PD signal tests using different sensor bandwidths.
The results show that the XLPE cable can be fully and accurately modeled in ATP and
that the inclusion of the semi-conducting layer influence is very important in the
modeling. The semi-conducting layers affect signal propagation and must be taken into
account in any PD analysis work. The match of the frequency dependent cable model
response with measured data demonstrates the validity of the ATP frequency dependent
cable models for PD propagation study. However, it must be noted that practical realities
such as earth resistances, proximity of other cables, temperature and semi-conducting
layer pressure may also affect the results.
________________________________________________________________________
Chapter 8 presents the characteristics of partial discharge and high frequency signal
response in paper cable and XLPE cables with their semi-conducting layers. It
demonstrates the developed EMTP-RV cable model can model the three phase power
cables and simulated results give a good agreement with measured results on both paper
and XLPE power cables. One of the major aims of the thesis project was to use modeling
to reconstruct the PD source signal by using the received signal, which is captured from
the sensor at the received end, and using the model to de-convolute the received signal.
By using the developed cable model in this thesis, the reconstruction of signal source can
be achieved with reasonable accuracy and the detail is shown in the thesis. This is an
important result because the waveform detail of partial discharges in both time and
frequency domains is an important property in identifying PD types and the source of the
PD activity.
The simulations in ATP are carried out using circuit theory approximations. However,
when the device structure is complex, such as in lossy dielectric cables with semi-con
inclusion, simulation through circuit theory approximation becomes difficult or
impossible because of the distributed parameters and losses. Finite Element Method
techniques can overcome all the above limitations from ATP cable modelling.
ATP and FEM techniques present the following main limitations:

i) ATP is very efficient only for analysis of simple geometrical configurations of the
cable;
ii) Numerical differential methods such as FEM are very efficient but they require a fine
discretization of the conductive regions to obtain accurate solutions.
The advantages and disadvantages of both techniques are:
ATP- Advantages:
- Free cost and low hardware requirements
- Allows the user to access the main part of software
________________________________________________________________________
- Simulation time is short

ATP- Disadvantages:
- Not straightforward to represent cable features: for example wire screens, semi-
conducting layers and lossy insulation
- Neglects the effect between parallel cables in multi-core types.
FEM- Advantages:
- Can handle the cable with irregular shapes.
- Includes the loss factor of conductor, insulation and the skin effect.
- Allow 3-Dimensional model development.
FEM- Disadvantages:
- Fine meshes need at high frequency to achieve accurate results
- Most of the FEM program is commercial software so the cost is high.
- Good quality computer hardware is required.
9.5 PD Detection on Cables
An investigation into the possibility of detecting different PD patterns and signals when
conducting PD tests using different sensor bandwidths is presented in Chapter 6. The
occurrence of discharge activity is created by an artificial defect manufactured in the
single core cable insulation. The artificial defect can be generated by internal discharge
and used to investigate the PD signal propagation on XLPE cable.
The capacitance coupling external sensors have been applied for the PD detection
measurements and the results show that the external sensors have a number of advantages
over HF-CTs when used for the detection of PD pulses. The attenuation of PD pulse
propagation is increasing with the increasing frequency of pulse and travel distance along
________________________________________________________________________
the cable. In particular, this is the case when increasing the frequency of the pulse, to
levels, which are within the domain of typical PDs, propagation can give much higher
attenuation. In addition the simulation results are compared with the test measurement
results and it was found that the model used was able to predict the measurement results
accurately.
Chapter 7 shows the investigation of the cross coupling of PD signals between phases in
three phase XLPE cable. The capacitance coupling external sensors have been applied for
better PD detection measurements than HF-CTs. In order to visualize recorded data
gained by PD measurement of three phase cable under test, the 3PARD diagram was
used. Each data pulse is assigned to a single dot in diagram. The measured results show
that the 3PARD diagram allowed the user to verify the fault between phases with PD
location.
9.6 Suggestions for Further Work
High frequency cable models presented in this thesis are based on the EMTP techniques
and FEM techniques. Instead of using these two techniques, there are a number of other
methods, which can be used for cable modelling such as JMAG and PSCAD. Time did
not allow these to be investigated. These model bases should be investigated for their
accuracy and suitability.
In addition, whereas the tests reported here were performed in the laboratory and with a
simulated PD sources, the verification of cable models can be used in on-line test
measurement results to check the agreement between simulated and measured results in
the practical situation with real PD signals.
________________________________________________________________________
The available cable model developed in this thesis was unable to accept the source signal
at any frequency over a GHz. This limit on frequency can be improved in the future by
using a combination of the FEM techniques with analytical methods.
The measurement of the semi-conducting layers’ parameters can also include the
measurement of permittivity with the conductivity. This can give more accurate
simulated results.
For future work on partial discharges in cavities there are many things of interest that can
be investigated:
Partial discharges in a cavity can be modeled in three-dimensions. In the two-dimensional
model presented in this work the discharges have a constant shape of cylindrical shells.
With a three-dimensional model it would be possible to improve the description of the
discharge spatial extension and to model the influence between different discharges more
realistically.
Partial discharge in cavities can be investigated in their frequency dependence. Such

frequency dependent behavior was observed in some measurements presented in this
thesis work.
References 260
________________________________________________________________________
References
[1] C. C. Barnes, “Power cables: their design and installation”, Chapman & Hall,
London, 1953.
[2] T. R. Blackburn, R.E. James, B.T. Phung and Z. Liu, “Partial discharge characteristics
in polymeric cable accessories”, 2001 International Symposium on Electrical Insulating
Materials (ISEIM01), Japan: pp532-535, 2001.
[3] G. C. Stone and S. A. Boggs, ‘‘Propagation of Partial Discharge Pulses in Shielded

Power Cable’’, IEEE Conf. Electr. Insul. Dielectr. Phenomena, pp. 275-280, 1982.
[4] T. Tanaka and A. Greenwood, “Advanced power cable technology”, Volume I, Basic
concepts and testing, 1983, CRC Press, Inc. Page(s): 2-10.
[5] R. Bartnikas and K. D. Srivastava, Power and Communication Cables. IEEE

press series on power engineering, 2002.
[6] EMF-Fact sheet, Transpower, New Zealand Ltd 2008
[7] W. A. Thue , “Electrical Power Cable Engineering”, New York : Marcel

Dekker, c1999
[8] S. Boggs, , A. Pathak, and P. Walker, “Partial discharge. XXII. High frequency
attenuation in shielded solid dielectric power cable and implications thereof for PD
location”, Electrical Insulation Magazine, IEEE, Volume: 12 , Issue: 1 , Jan.-Feb. 1996,
pp9 – 16.
[9] T. Tanaka and A. Greenwood, “Advanced power cable technology volume II present
and future”, CRC press, 1983, pp 4-7.
[10] N. H. Malik, A. A. Al-Arainy, and M. I. Qureshi , “Electrical Insulation in Power

Systems”, New York : Marcel Dekker, c1998
[11] T. C. Yu and J. R. Marti, "A Robust Phase-Coordinate Frequency-Dependent

Underground Cable Model(zCable) for the EMTP", IEEE Transactions on Power
Delivery, Vol. PD-18, No. 1, January 2003, pp189-194.
[12] E. Steennis and F. Kreuger, \Water treeing in polyethylene cables," IEEE

Transactions on Electrical Insulation, vol. 25, no. 5, pp. 989{1028, 1990.
References 261
________________________________________________________________________
[13] E. Kuffel, W. S. Zaengl, and J. Kuffel, High Voltage Engineering: Fundamentals.

Butterworth-Heineman, 2000.
[14] R. Bartnikas, “Partial discharges. their mechanism, detection and measurement,”

IEEE Transactions on Dielectrics and Electrical Insulation, vol. 9, no. 5, pp. 763{808,
2002.
[15] R. Papazyan, Techniques for localization of insulation degradation along medium-

voltage power cables, Ph.D. dissertation, Trita-ETS-1650-674X , Royal Institute of
Technology, 2005.
[16] S. I. Yamaguchi, S. Soda, and N. Takada, \Development of a new type insulation

diagnostic method for hot-line xlpe cables," IEEE Transactions on Power Delivery, vol.
4, no. 3, pp. 1513{1520, 1989.
[17] N. Ahmed and N. Srinivas, “On site partial discharge diagnostics in power cables,”
in Reliability of distribution & transmission cable systems, 2002.
[18] IPEC High Voltage Limited, Empress Business Centre, Chester Road, Manchester.
M16 9EA. UK. http://www.ipechv.co.uk/technical/partial-discharge.html
[19] F. H. Kreuger, Discharge detection in high voltage equipment, A Heywood Book,

Temple Press Books Ltd, London, 1964
[20] IEC Standard – Publication 60270, High Voltage Test Techniques – Partial
Discharge Measurements, 2000.
[21] F. H. Kreuger, Industrial High Voltage, Delft University Press, The Netherlands,
1991.
[22] Robinson Electronic Instruments, “E. R. A. Discharge Detector Model 3: System

Manual and Operation Manual”, Cheshire, United Kingdom.
[23] G. Paoletti, and A. Golubev. Partial Discharge Theory and Applications to

Electrical Equipment [Online]. Available: http://www.wmea.net/partial_
discharge_theory.htm. Western Mining Electrical Association. [September 2004]
[24] F. H. Kreuger (1989). Partial Discharge Detection in High Voltage Equipment.

Butterworth & Co. (Publishers) Ltd, United Kingdom.
[25] D. A. Nattrass (May/June 1998). Partial Discharge and Measurement and

Interpretation, IEEE Electrical Insulation Magazine, vol. 4, no. 3, pg10-23.
References 262
________________________________________________________________________
[26] M. G. Turner, and E. Gulski (Nov 1998). Pattern Recognition for Partial Discharge
Measurement. Tettex Instruments Division, Haefely Test AG, Zurich, Switzerland.
[27] B. T. Phung (2002). Use of Computer-based Systems for Partial Discharge

Monitoring and Analysis, 2002 ESAA Short Course on Insulation Condition Monitoring
and Partial Discharge Technique.
[28] C. Chang and Q. Su, "Analysis of Partial Discharge Patterns from a Rod to Plane
Arrangement", Conference Record of the IEEE International Symposium on Electrical
Insulation, Anaheim, USA, pp. 439-443, 2-5 April 2000.
[29] R. Bartnikas and J. E. McMahon, “ Engineering Dielectrics Volume: 1, Corona

Measurement and Interpretation”, American Society for Testing and Materials (ASTM),
Philadelphia, USA, 1979.
[30] M. Randoux and R. Wilputte, “Lesson Drawn from the Routine Testing of Insulation
Oils Used in Power Transformer on the Belgian Network”, CIGRE 12-07, 1986.
[31] IEC Standard, Publication 60270, "High Voltage Test Techniques-Partial Discharge
Measurements", 2000.
[32] R. Bartnikas, “Partial Discharges: Their Mechanism, Detection and Measurement”,

IEEE Transactions on Dielectrics and Electrical Insulation, Volume: 9 Issue: 5, pp. 763-
808, October 2002.
[33] B. T. Phung, Computer-Based Partial Discharge Detection and Characterisation,

Ph.D. Thesis, the University of New South Wales, Sydney, 1997
[34] Robinson Electronic Instruments, E. R. A. Discharge Detector Model3: System

Manual and Operation Manual, Chesire, UK.
[35] T. Ozaki, K. Abe and T. Umemura, “Partial Discharge Detection using Ferrite
Antenna”, IEEE International Symposium on Electrical Insulation, Baltimore, pp 31-374,
June 1992
[36] B. T. Phung, “On-line Partial Discharge Measurement on High Voltage Power

Cables”, 11th International Symposium on High Voltage Engineering (ISH), London, Vol
4, pp. 328-332, August 1999.
[37] T. R. Blackburn, “On-line Partial Discharge Measurement on Instrument

Transformers”, International Symposium on Electrical Insulating Materials, Toyohashi,
pp. 497-500, September 1998.
[38] B. T. Phung, “Recent Development for On-line Partial Discharge Detection in

Cables”, Australasian Universities Power Engineering Conference, Perth, pp 413-418,
September 2001.
References 263
________________________________________________________________________
[39] E. Pultrum, “On Site Testing of Cable System After Laying, Monitoring with HF
Partial Discharge Detection”, IEE Two day colloquium on supertension, London, pp21/1-
21/4, November 1995.
[40] L. De Maria, E. Colombo and W. Koltunowicz, “Comparison among PD Detection

Methods for GIS On site Testing”, 11th International Symposium on High Voltage
Engineering (ISH), London, Vol 4, pp. 590-593, August 1999.
[41] B. Baumgartner, “Partial Discharge X: PD in Gas-Insulated Substations-

Measurement and Practical Considerations”, IEEE Electrical Insulation Magazine, Vol.
8, No. 1, pp 16-27, Jan/Feb, 1992.
[42] T. Hoshino, “Frequency Characteristics of Electromagnetic Wave Radiated From

GIS Apertures”, IEEE Transactions on Power Delivery, Vol 16, No. 4 pp. 552-557,
October, 2001.
[43] G. C. Stone and S. A. Boggs, “Propagation of Partial Discharge Pulses in Shielded

Power Cable”, 1982 Annual Report of the Conference on Electrical Insulation and
Dielectric Phenomena, National Academy of Science, Washington, DC, p. 275
[44] S. A. Boggs, J. M. Braun and G. C. Stone, “Attenuating Voltage Surges in Power

Cable by Modifying the Semiconductive Shields”, Proceedings of the 1992 IEEE
International Symposium on Electrical Insulation. IEEE Publication 92CH3150-0. P.491.
[45] T. R. Blackburn, R.E. James, B.T. Phung and Z. Liu, “Partial discharge
characteristics in polymeric cable accessories”, 2001 International Symposium on
Electrical Insulating Materials (ISEIM01), Japan: pp532-535, 2001.
[46] F. H. Kreuger, Detection and Location of Discharges. Ph.D. Thesis, Technical

University of Delft, 1961. Chapter 6.
[47] M. S. Mashikian, R. Bansal and R. B. Northrop, “Location of Discharge Sites in

Power Cables”, IEEE Trans. PD-5, No. 2, April 1990. P.833.
[48] L. Wieringa, “Location of Small Discharges in Plastic Insulated High Voltage

Cables”, IEEE Trans. PAS-101, No. 7. Jan. 1985. P.2.
[49] M. Beyer, “A New Method for Detection and Location of Distributed Partial
Discharges in High Voltage Cables”, IEEE Trans PAS-101, No.9. Sept, 1982, p. 3431.
[50] S. A. Boggs, “Partial Discharge- Part II: Detection Sensitivity”, IEEE Electrical
Insulation Magazine, 6. No. 5, September/ October 1990. P.35
[51] S. A. Boggs and G. C. Stone, “ Fundamental Limitations to the Measurement of

Corona and Partial Discharge”, IEEE Trans EI-17 April, 1982, p. 143
References 264
________________________________________________________________________
[52] S. A. Boggs, “Partial Discharge- Part I: Overview and Signal Generation”, IEEE
Electrical Insulation Magazine, 6, No. 4, July/August 1990, P.33
[53] G. C. Stone and S. A. Boggs, ‘‘Propagation of Partial Discharge Pulses in Shielded

Power Cable’’, IEEE Conf. Electrical. Insulation. Dielectrics. Phenomena, pp. 275-280,
1982.
[54] P. Morshvis, Partial Discharge Mechanism Leading to Breakdown in Polymers,

Ph.D. Thesis, Delft University , 1993.
[55] W. L. Weeks and Y. M. Diao, ‘‘Wave Propagation Characteristics in Underground

Power Cable’’, IEEE Trans. Power App. Syst., Vol. PAS-103, No. 10, pp. 2816_2826,
October 1984.
[56] O. Breienn and I. Johansen, “Attenuation of travelling waves in single-phase high

voltage cables.” IEE Proceedings, volume 118, no. 6, June 1971.
[57] J. P. Steiner, Digital Measurement of Partial Discharge, PhD Thesis, Purdue

University, School of Electrical Engineering, West Lafayette, Ind., USA, 1988
[58] G. Mugala; R. Eriksson and P. Pettersson, “Dependence of XLPE insulated power

cable wave propagation characteristics on design parameters”, IEEE Transactions on
Electrical Insulation, Volume 14, Issue 2, Page(s):393 – 399, 2007
[59] A. Ametani, “A General Formulation of Impedance and Admittance of Cables”,

IEEE Transactions on Power Apparatus and Systems, Volume PAS-99, Issue 3,
Page(s):902 – 910, 1980
[60] D. Pommerenke and S. Sakaguchi, “Application of Maxwell solvers to PD

propagation. I. Concepts and codes”, Electrical Insulation Magazine, IEEE, Vol 18, issue
5 pp. 15- 21, 2002.
[61] R. Papazyan and D. Pommerenke,; R. Eriksson, “Modeling the Wave Propagation

Properties of Power Cables using Numerical Simulations”, Precision Electromagnetic
Measurements Digest, 2004 Conference, Page(s):412 – 413, 2004
[62] B. Gustavsen, J. A. Martinez and D. Durbak, “Parameter determination for modeling

system transients-Part II: Insulated cables”, Power Delivery, IEEE Transactions on
Volume 20, Issue 3, July 2005 Page(s):2045 - 2050
[63] S. Boggs, A. Pathak and P. Walker, “Partial discharge. XXII. High frequency
attenuation in shielded solid dielectric power cable and implications thereof for PD
location”, Electrical Insulation Magazine, IEEE, Volume: 12, Issue: 1 , Jan.-Feb. 1996
Pages:9 – 16
References 265
________________________________________________________________________
[64] R. Papazyan, “Wave Propagation on Power Cables with Special Regard to Metallic
Screen Design”, Precision Electromagnetic Measurements Digest, 2004 Conference,
Page(s):412 – 413, 2004
[65] J. P. Steiner, Digital Measurement of Partial Discharge, PhD Thesis, Purdue

University, School of Electrical Engineering, West Lafayette, Ind., USA, 1988
[66] D. E. Hedman, “Propagation on overhead transmission line, Part II: Earth

Conduction effects and practical results”, IBID, Vol. PAS 84, May 1965, pp. 205-211.
[67] L. M. Wedephol and D. J. Wilcox, “Estimation of transient sheath overvoltage in

power cable transmission systems”, Proc. IEE, Volume 120, No. 8, August 1973.
[68] A. Ametani, “A General Formulation of Impedance and Admittance of Cables”,

IEEE Transactions on Power Apparatus and Systems, Volume PAS-99, Issue 3, May
1980 Page(s):902 - 910
[69] Electro-Magnetic Transients Program theory book, Chapter 5-underground cables

pages 150-192, 1986, BPA.
[70].J. R. Marti, “Accuarte Modelling of Frequency-Dependent Transmission Lines in

Electromagnetic Transient Simulations”, IEEE Transactions on Power Apparatus and
Systems, Volume PAS-101, Issue 1, Jan. 1982 Page(s):147 - 157
[71] L. Marti, “Simulation of transients in underground cables with frequency-dependent

modal transformation matrices”, Power Delivery, IEEE Transactions on, Volume 3,
Issue 3, July 1988 Page(s):1099 -1110
[72] T. Noda, N. Nagaoka and A. Ametani, “Phase domain modeling of frequency-

dependent transmission lines by means of an ARMA model” Power Delivery, IEEE
Transactions on , Volume: 11 , Issue: 1 , Jan. 1996 , Pages:401 - 411
[73] H. W. Dommel, “Digital computer solution of electromagnetic transients in single

and multi-phase networks”, IEEE Transactions on Power Apparatus and Systems, 1969,
88 (2), pp. 734-41.
[74] H. W. Dommel, “Nonlinear and time-varying elements in digital simulation of

electromagnetic transients”, IEEE Transactions on Power Apparatus and Systems, 1971,
90 (6), pp. 2561-7.
[75] H. W. Dommel, “Techniques for analyzing electromagnetic Transients”, IEEE

Computer Applications in Power, 1997, 10(3), pp. 18-21.
[76] N. Nagaoka, A. Ametani, “Transient calculations on crossboned cables”, IEEE

Transactions on Power Apparatus and Systems, pp. 779-787, April 1983.
References 266
________________________________________________________________________
[77].J. R. Marti, “Accuarte Modelling of Frequency-Dependent Transmission Lines in

[78] L. Marti, “Simulation of transients in underground cables with frequency-dependent

modal transformation matrices”, Power Delivery, IEEE Transactions on, Volume 3,
Issue 3, July 1988 Page(s):1099 -1110

[80] B. Gustavsen, “Panel session on data for modeling system transients insulated
cables”, Power Engineering Society Winter Meeting, 2001. IEEE, Volume: 2, 28 Jan.-1
Feb. 2001, Pages:718 – 723.
[81] J. Elwardt, “Over Voltage Study on a 400 kV Cable Screen”, EEUG meeting, P. 13,
2003.
[82] M. Vakilian, T.R. Blackburn, B.T.Phung and H. Zhang, “Investigation of PD signal

propagation characteristics in XLPE Cables”. PowerCon, Singapore, 2004
[83] G. Mugala; R. Eriksson and P. Pettersson, “Dependence of XLPE insulated power

cable wave propagation characteristics on design parameters”, IEEE Transactions on
Electrical Insulation, Volume 14, Issue 2, Page(s):393 – 399, 2007
[84] L. Bergeron, “Water hammer in hydraulics and wave surges in electricity”, ASME
ommittee, Wiley, New York, 1961.
[85] S. Boggs, “The case for frequency domain PD testing in the context of distribution
cable”, IEEE Electrical Insulation Magazine, Vol: 19, Issue: 4, Year: July-Aug. 2003
[86] N. Watson and J. Arrillaga, Power Systems Electromagnetic Transient Simulation,

IEE, UK, 2003.
[87] L. Marti, "Simulation of Transients in Underground Cables with Frequency-

Dependent Modal Transformation Matrices". IEEE Transactions on Power Deliver), vol.
3, No. 3, pp.1099-1110, July 1988.
[88] L. Marti, “Simulation of electromagnetic transients in underground cables using the

EMTP”, 2nd International Conference on Advances in Power System Control, Operation
and Management, 1993. APSCOM-93, 1993
References 267
________________________________________________________________________
[89] J. R. Marti, “Accuarte Modelling of Frequency-Dependent Transmission Lines in


[91] A. Semlyen and A. Dabuleanu, "Fast and accurate switching transient calculations
on transmission lines with ground return using recursive convolutions", IEEE
Transactions on Power Apparatus and Systems, Vol. PAS- 94(2), p. 561-571, 1975.
[92] T. Noda, "User Instructions of Noda Setup in ATP", ATP monthly report, march 28,
1997
[93] B. Gustavsen, “Panel session on data for modeling system transients insulated
cables”, Power Engineering Society Winter Meeting, Volume: 2, On page(s): 718-723
vol.2, 2001
[94] J. Elwardt, “Over Voltage Study on a 400 kV Cable Screen”, EEUG meeting, P. 13,
2003.
[95] D. Pommerenke,; S. Sakaguchi, “Application of Maxwell solvers to PD propagation.

I. Concepts and codes”, Electrical Insulation Magazine, IEEE, Vol 18, issue 5 pp. 15- 21,
2002.
[96] S. Sakaguchi and M. Oyama, “Application of Maxwell solvers to PD Propagation.

III. PD propagation in GIS” Electrical Insulation Magazine, IEEE , Volume: 19 , Issue: 1
, Jan.-Feb. 2003 , Pages:6 - 12
[97] A. Hoorfar and Y. Jamnerjad, “Electromagnetic Modeling and Analysis of Wire-less

Communication Antennas,” IEEE Trans. Microwave, vol. 12, pp. 51-67, 2003.
[98] J. Jin, The Finite Element Method in Electromagnetic, 2nd ed., New York, Willey,
2002.
[99] R. Lucas and S. Talukdar, “Advances in Finite Element Techniques for Calculating
Cable Resistances and Inductances” Page(s): 875-883
[100] D. Cârstea, I. Cârstea, “Finite element techniques for coaxial cable parameters
Telecommunications in Modern Satellite, Cable and Broadcasting Service” 5th
International Conference on TELSIKS 2001, Volume 2, 19-21 Sept. 2001 Page(s):644 -
646 vol.2
References 268
________________________________________________________________________
[101] R. Papazyan, D. Pommerenke andR. Eriksson, “Modeling the Wave Propagation

Properties of Power Cables using Numerical Simulations”, Precision Electromagnetic
Measurements Digest, 2004 Conference, Page(s):412 – 413, 2004
[102] R. Papazyan, Wave Propagation on Power Cables with Special Regard to Metallic
Screen Design
[103] A. Darcherif, A. Raizer, G. Meunier, J. F. Imhoff and J. C. Sabonnadiere, “New

techniques in FEM field calculation applied to power cable characteristics computations”,
Magnetics Conference 1990 Digests of INTERMAG '90. International, pp. 17-20, 1990.
[104] Comsol User guide, RF module- Lossy Eigenvalue Calculations section, 2006,
pp36-40.
[105] M. Feliziani and F. Maradei, “A FEM approach to calculate the impedances of

shielded multiconductor cables” Electromagnetic Compatibility, 2002 IEEE International
Symposium, Vol 2, 19-23, pp.1020 – 1025,2002
techniques in FEM field calculation applied to power cable characteristics computations”,
Magnetics Conference 1990 Digests of INTERMAG '90. International, pp. 17-20, 1990.
[107] T. Tanaka and A. Greenwood, Advanced power cable technology, Volume II,
Present and Future, 1983, CRC Press, Inc.
[108] G. Mugala, R. Eriksson, U. Gafvert and P. Petterson, “Measurement technique for

high frequency characterization of semiconducting materials in extruded cables”
Dielectrics and Electrical Insulation, IEEE Transactions, Volume: 11, Issue: 3, Jun 2004,
pp 471 – 480
[109] G. Mugala and R. Eriksson, “High frequency characteristics of a shielded medium

voltage XLPE cable” Electrical Insulation and Dielectric Phenomena, 2001 Annual
Report, 14-17 Oct. 2001, pp 132 – 136
[110] Kerite Company, Power cables single core [Online]. Available (Picture):
http://www.kerite.com/catalog/catalogfiles/medium_voltage_pilcreplacement.htm
[111] O. P. Hevia, “Alternative Transients Program - Comparison of Transmission Line

Models”, Comité Argentino de Usuarios del EMTP/ATP (CAUE), Argentina
[112] C. C. Barnes, “Power cables: their design and installation,” Champam & Hall,
London, 1953 pp. 45-46.
References 269
________________________________________________________________________
[113] Australian Standard 60270: High-voltage test techniques- Partial discharge

measurements, pp33-34.
[114] H. N. O, T. R. Blackburn, B. T. Phung, M. Vakilian, H. Zhang and M. S. Naderi

“The propagation characteristics of high frequency partial discharge pulses in XLPE
cables”. To be presented at Australasian Universities Power Engineering Conference
(AUPEC’05), Tasmania, Australia, 2005.
[115] N. Klein, “Gas-filled cavities in solid type cables”. Tr. AIEE, Vol. 79, Pt. 3, 1960,
pp. 77-87.
[116] A. Nosseir, “Calculation of discharge inception voltage due to presence of void in

power cables”, IEEE Trans., Vol. EI-14, No. 2, April 1979, pp 117-120.
[117] F. C. Cheng, “Insulation thickness determination of polymeric power cables”, IEEE

Trans. Dielectrics EI, Vol. 1, No. 4, 1994, pp 624-629.
[118] F. C. Cheng (in reply), “Discussion on [4]” (by Danikas, M.G.), ibid., Vol. 2, No.
6, 1995, pp. 1161-1165.
[119] F. C. Cheng, “Electric field distribution distorted by voids inside power cables”, ,
2000 International Conference on Advances in Power System Control, Operation and
Management, 2000. APSCOM-00.Volume: 2, 30 Oct.-1 Nov. 2000 Pages:311 - 316 vol.2
[120] R. M. Fano, L. J. Chu and R. B. Adler, Electromagnetic fields, energy and forces,
Wily, New York, 1960, Chapter 5.
[121] L. M. Magid, “Electromagnetic field, energy and waves”, Wiley, New York, 1972,
Chapter 7.
[122] G. Mitra and B. Salvage, “Electric stress in a circular cylindrical gaseous cavity in
a solid dielectric, the axis of the cylinder being parallel to the field”, Proc. IEE, Vol. 113,
1966, pp. 931-955.
[123] S. Y. King and N. A. Halfter, Underground power cables, Longman House, New
York, U.S.A.,1982, pp 170-177.
[124] K. Steinbrich, “Influence of semiconducting layers on the attenuation behaviour of

single-core power cables Generation”, Transmission and Distribution, IEE Proceedings-
Volume 152, Issue 2, 4 March 2005 Page(s):271 – 276
[125] A. Morched, B. Gustavsen and M. Tartibi, “A universal model for accurate

calculation of electromagnetic transients on overhead lines and underground cables”,
IEEE Transactions on Power Delivery, Volume 14, Issue 3, July 1999 Page(s):1032 -
1038
References 270
________________________________________________________________________
[126] D. E. Hedman, Attenuation of Traveling Waves on Three-Phase Lines, Power

Technologies, Inc., 1970
[127] G. W. Brown and R. G. Rocamora, “Surge propagation in three-phase pipe-type

cables, part I : Unsaturated pipe”, IEEE Transactions on Power Apparatus and Systems,
Volume: 95, Issue: 1, Part 1, pp. 89- 95, Jan. 1976
[128] R. C. Dugan, R. G. Rocamora and G. W. Brown, “Surge propagation in three-phase

pipe-type cables: Part II: Duplication of field tests including the effects of neutral wires
and pipe saturation”, Power Apparatus and Systems, IEEE Transactions on, Volume 96,
Issue 3, Part 1, May 1977 Page(s):826 - 833

techniques in FEM field calculation applied to power cable characteristics computation”,
IEEE Transactions on Magnetics, Volume 26, Issue 5, Sep 1990 Page(s):2388 - 2390
[130] M. Feliziani, “A FEM approach to shielding effectiveness in braided-shield cables”

IEEE Transactions on Magnetics, Volume 26, Issue 2, Mar 1990 Page(s):929 - 932
[131] P. C. J. M. van der Wielen, E. F. Steennis and P. A. A. F. Wouters, “Fundamental

aspects of excitation and propagation of on-line partial discharge signals in three-phase
medium voltage cable systems”, IEEE Transactions on Dielectrics and Electrical
Insulation, Volume 10, Issue 4, Aug. 2003 Page(s):678 - 688
[132] P. C. J. M. van der Wielen, P. A. F. Wouters and E. F. Steennis, “Signal

interpretation of partial discharges in three-phase medium voltage cable systems
measured on-line”, Electrical Insulation, 2002. Conference Record of the 2002 IEEE
International Symposium on , 7-10 April 2002, Pages:542 - 545
[133] R. Lucas and S. Talukdar, “Advance in finite element techniques for calculating
cable resistance and inductance”, IEEE Trans on PAS vol. 97, no3. Pp875-883,
May/June, 1978.
[134] Y. Yin and H. Dommel, “Calculation of frequency dependent impedance of

underground power cables with FEM”, IEEE Trans on MAG, vol.25, No. 4, July, 1989.
[135] W. Kalkner, A. Obralic, M. Kaufhold and R. Plath, “ Investigation of PD Pulse

propagation and coupling between three phases of a stator of a rotating machine”, In 14th
ISH, Tsinghua University, Beijing, China, G-042, 2005.
[136] S. Schaper, W. Kalkner and R. Plath, “ Synchronous multi terminal on-site PD

measurements on Power Transformers”, In 14th ISH, Tsinghua University, Beijing,
China, G-121, 2005.
References 271
________________________________________________________________________
[137] A. Morched, B. Gustavsen and M. Tartibi: “A universal model for accurate

calculation of electromagnetic transients on overhead lines and underground cables”.
Power Delivery, IEEE Transactions, Volume: 14, Issue: 3 , July 1999, pp. 1032 -1038.
[138] A. Raizer, “P and H type adaptive mesh generation”, 34th MMM conference, May
1990.
[139] J. F. Imhoff, “Original solution of unbounded problems in Electromagnetism”,

IEEE Trans on Mag., Sep 1990.
[140] H. N. O, T. R. Blackburn, B. T. Phung, H. Zhang and R. H. Khawaja,

“Investigation of Electric Field Distribution in Power Cables with Voids”, 8th
International Conference on Properties and applications of Dielectric Materials, 2006.
June 2006 Page(s):637 – 640
[141] H. N. O, T. R. Blackburn, B. T. Phung, M. Vakilian, M. S. Naderi and H. Zhang,

“Investigation of high frequency signal propagation characteristics on HV XLPE cables”,
The 7th International Power Engineering Conference, 2005, Nov. 29 2005-Dec. 2 2005
Page(s):776 - 781 Vol. 2
[142] S. Kentaro, T. Yasuhiro, T. Tatsuo, O. Yoshimichi and T. Chikashi, “Detection of

Water Tree Position in Cable Insulation Wall by using PEA method” Technical Meeting
on Dielectrics and Electrical Insulation, IEE Japan, Vol.DEI-99;No.41-52;Page.13-18,
1999
[143] B. Jonuz, P. H. F. Morshuis, H. J. Van Breen, J. Pellis and J. J. Smit, “Detection of

water trees in medium voltage XLPE cables by return voltage measurements” Electrical
Insulation and Dielectric Phenomena, 2000 Annual Report Conference on Volume 1, 15-
18 Oct. 2000 Page(s):355 - 358
[144] P. F. Hinrichsen, A. Houdayer, G. Kajrys, A. Belhadfa, G. Kennedy, J. St.-Pierre

and J. P.Crine, “Comparison of NAA and micro-PIXE for impurity detection in water
trees grown in XLPE cables” IEE Proceedings of Physical Science, Measurement and
Instrumentation, Management and Education, Reviews, Volume 137, Issue 5, Sep 1990
Page(s):255 - 260
[145] A. Cavallini, M. Conti, G. C. Montanari, C. Arlotti and A. Contin, “PD inference

for the early detection of electrical treeing in insulation systems” IEEE Transactions on
Dielectrics and Electrical Insulation, Volume 11, Issue 4, Aug. 2004 Page(s):724 - 735
[146] A. El-Zein, M. Talaat and M. M. El Bahy, “A numerical model of electrical tree

growth in solid insulation” IEEE Transactions on Dielectrics and Electrical Insulation,
Volume 16, Issue 6, December 2009 Page(s):1724 – 1734
References 272
________________________________________________________________________
[147] C. Laurent and C. Mayoux, “Analysis of the Propagation of Electrical Treeing

Using Optical and Electrical Methods” IEEE Transactions on Electrical Insulation,
Volume EI-15, Issue 1, Feb. 1980 Page(s):33 - 42
[148] W. Schubert, Communication cables and transmission systems, Siemens

Aktiengesellschaft, 3rd Edition, 1985
[149] Mtronix Precision Measuring Instruments, Advanced Partial Discharge Measuring

System, Mtronix Precision Measuring Instruments, Berlin, August 17, 2006
[150] W. S. Meyer and H. W. Dommel, “Numerical modelling of frequency-dependent

transmission line parameters in an electromagnetic transients program” IEEE Trans.
Power Apparatus and Systems, Vol. PAS-93, Sept/Oct 1974, pp. 1401-1409
Appendix I 273
________________________________________________________________________
Appendix I
The Parameter of Semi-Conducting Layers.

Further discussion can be found in Chapter 4.
The results of semi-conducting layers’ propagation effect can be divided into three
sections:
Section A use the measured results of semi-conducting layers, further detail on
measurement method can be found in Chapter 5;
Section B use the Steven Boggs’ measurement of semi-conducting layers in [8] and
Section C use K. Steinbrich’s measurement of semi-conducting layers in [124].
Appendix I 274
________________________________________________________________________
2.5
2
Attenuation (dB/m)
1.5
70M Hz
0.5
50M Hz
30M Hz
0 10M Hz
0
1 1M Hz
100
Conduc t ivit y

of outer semi-conducting layer is equal to 100 and its conductivity is equal to 17.
300
200
Velocity (m/s)
Velocity
100
0
0.001 0.01 0.1 1 10 100
Conductivity
1M Hz 10M Hz 30M Hz 50M Hz 70M Hz
Figure A1.2: the graph of velocity as function of different frequency in the

Appendix I 275
________________________________________________________________________
0.18
0.16
0.14
Attenuation (dB/m)
Attenuation (dB/m)
0.12
0.1
0.08
0.06 100M Hz
80M Hz
0.04 60M Hz
50M Hz
0.02
40M Hz
0 20M Hz
0.00001 5M Hz
1
100000
Conductivity

0.00001.
250
200
Velocity Velocity
(m/s)
150
100
0.00001 0.001 0.1 10 1000 100000
Conductivity
5M Hz 20M Hz 40M Hz 50M Hz 60M Hz 80M Hz 100M Hz
of outer semi-conducting layer is equal to 1 and its conductivity is equal to 0.00001.
Appendix I 276
________________________________________________________________________
0.07
0.06
Attenuation (dB/m)
0.05
0 .04
0. 03
100M Hz
0. 02
80M Hz
60M Hz
0.01 50M Hz
40M Hz
0 20M Hz
1E- 05 5M Hz
1
100000
Conductivity

100000.
2 50
2 00
Velocity (m/s)
150
100
0 .00001 0.001 0. 1 10 1000 1000 00
Cond uctivit y
5M Hz 20M Hz 40M Hz 50M Hz 60M Hz 80M Hz 100M Hz
Appendix I 277
________________________________________________________________________
0.8
0.7
0.6
Attenuation (dB/m)
0.5
0.4
0.3
100M Hz
0. 2 80M Hz
6 0M Hz
0.1 50M Hz
40M Hz
0 20M Hz
1E- 05 5M Hz
1
100000
Co n d u c ti vit y

to 1.
200
180
Velocity (m/s)
160
140
0.00001 0. 001 0 .1 10 1000 100000
Conductivity
5M Hz 2 0M Hz 40M Hz 50M Hz 60M Hz 80M Hz 100M Hz
Appendix I 278
________________________________________________________________________
0 .7
0.6
0.5
Attenuation (dB/m)
0.4
0.3
100M Hz
0.2 80M Hz
60M Hz
0.1 50M Hz
40M Hz
0 20M Hz
1E- 05 5M Hz
1
10000 0
C o n d u ct ivity

to 1.
200
150
Velocity (m/s)
100
50
0. 00001 0.001 0.1 10 1000 100 000
Conducti vity
5M Hz 20M Hz 40M Hz 50M Hz 6 0M Hz 80 M Hz 100M Hz
Appendix I 279
________________________________________________________________________
The following results are using the semi-conducting layer measurement results from
Steven Boggs [8]
0.25
0.2
Attenuation (dB/m)
0.15
0. 1
0. 05 50M Hz
30M Hz
0 10 M Hz
1E- 05
1 5M Hz
100000
C o n d u c ti vit y

Appendix I 280
________________________________________________________________________
200
150
Velocity (m/s)
100
50
0
0 .00001 0. 001 0.1 10 100 0 100000
Conduct ivit y
and conductivity of outer semi-conducting layer is based on Steven Boggs’s semi-con
measurement [63].
0.25
0. 2
Attenuation (dB/m)
0. 15
0.1
0.05 50M Hz
30M Hz
0 10M Hz
0. 00001 5M Hz
1
100000
C o nd u c t iv i ty

Appendix I 281
________________________________________________________________________
200
150
Velocity (m/s)
100
50
0
0 .00001 0. 001 0.1 10 100 0 100000
Conduct ivit y
and conductivity of inner semi-conducting layer is based on Steven Boggs’s semi-con
measurement [63].
Appendix I 282
________________________________________________________________________
The following results are using the semi-conducting layer measurement results from K.
Steinbrich [124]
1
0.9
0.8
Attenuation dB/m
0.7
0.6
0.5
0.4
0.3
0.2
30
0.1
20
0
10 Frequency (MHz)
100000
1000
1 5
0.01

compound.
Appendix I 283
________________________________________________________________________
1.4
1.2
1
constant
0.8
Constant
PhasePhase
0.6
0.4
0.2
0
0.00001 0.001 0.1 10 1000 100000
Co n d uctivity
5M Hz 10M H z 20M Hz 30M H z
compound.
190
185
180
175
(m/μs)
170
(m/s)
165
Velocity
Velocity
160
155
150
145
140
0.00001 0.001 0.1 10 1000 100000
Co ndu cti vity
and conductivity of outer semi-conducting layer is carbon-polyethylene compound.
Appendix I 284
________________________________________________________________________
0.45
0.4
0.35
Attenuation dB/m
0.3
0.25
0.2
0.15
0.1
30
0.05
20
0
10 Freque ncy (MHz)
100000
1000
1
0.01 5
Conductivi ty 0.00001

compound.
Appendix I 285
________________________________________________________________________
1. 4
1. 2
1
Phase constant
Phase Constant
0. 8
0. 6
0. 4
0. 2
0
0. 00001 0.001 0.1 10 1000 100000
Conductivi ty
compound.
190
180
170
Velocity (m /μs)
160
150
140
130
120
0.00001 0.001 0.1 10 1000 100000
C o n du cti vity
and conductivity of inner semi-conducting layer is carbon-polyethylene compound.
Appendix I 286
________________________________________________________________________
0.5
0.45
0.4
0.35
Attenuation dB/m
0.3
0.25
0.2
0.15
0.1
30
0.05
20
0
10 Frequency (MHz)
100000
1000
1
0.01 5

Appendix I 287
________________________________________________________________________
1.4
1.2
1
constant
Constant
0.8
Phase
0.6
Phase
0.4
0.2
0
0. 00001 0.001 0.1 10 1000 100000
Conductivity
190
185
180
Velocity (m/μs)
175
170
165
160
155
150
0. 00001 0.001 0. 1 10 1000 100000
Conductivi ty
and conductivity of outer semi-conducting layer is conductive paper.
Appendix I 288
________________________________________________________________________
0.5
0.45
0.4
0.35
Attenuation dB/m
0.3
0.25
0.2
0.15
0.1
30
0.05
20
0
10 Frequency (MHz)
100000
1000
1
0.01 5

Appendix I 289
________________________________________________________________________
1. 4
1. 2
1
Phase constant
0. 8
0. 6
0. 4
0. 2
0
0. 00001 0.001 0.1 10 1000 100000
Con d uctivi ty
190
185
180
Velocity (m /μs)
175
170
165
160
155
150
0.00001 0.001 0.1 10 1000 100000
Conducti vity
and conductivity of inner semi-conducting layer is conductive paper.
Appendix II 290
________________________________________________________________________
Appendix II
The Result of Frequency Response Measurements in Chapter

8.2.3
8.2.3 A. Three core paper belted cable
Phase Asending (Input) to Areceiving (Output)
0.80
0.60
Ratio
0.40
0.20
0.00
0 100 200 300
Frequency (MHz)
.
Appendix II 291
________________________________________________________________________
0
0 100 200 300
-20
dB
-40
-60
-80
Fr eque ncy (MHz)
Figure A2.2: Frequency Spectrum of ratio of Phase Asending (Input signal) to Areceiving
(Output signal).
Appendix II 292
________________________________________________________________________
Phase Bsending (Input) to Breceiving (Output)
0.8
0.6
Ratio
0.4
0.2
0.0
0 100 200 300
Frequency (MHz)
Phase Bsending (Input) to Breceiving (Output).
0
0 100 200 300
-20
dB
-40
-60
-80
Frequency (MHz)
Figure A2.4: Frequency Spectrum of ratio of Phase Bsending (Input) to Breceiving (Output).
Appendix II 293
________________________________________________________________________
Phase Csending (Input) to Creceiving (Output)
0.8
0.6
Ratio
0.4
0.2
0.0
0 100 200 300
Frequency (MHz)
Figure A2.5: Frequency response magnitude (S21, Damping ratio) is measured for ratio
Phase Csending (Input) to Creceiving (Output).
0
0 100 200 300
-20
dB
-40
-60
-80
Frequency (MHz)
Figure A2.6: Frequency Spectrum of Phase Csending (Input) to Creceiving (Output).

Appendix II 294
________________________________________________________________________
Phase Asending (Input) to Breceiving (Output)
0.3
0.2
Ratio
0.1
0.0
0 100 200 300
Frequency (MHz)
Asending (Input) to Breceiving (Output).
0
0 100 200 300
-20
dB
-40
-60
-80
Frequency (MHz)
Figure A2.8: Frequency Spectrum of Phase Asending (Input) to Breceiving (Output).

Appendix II 295
________________________________________________________________________
Phase Asending (Input) to Creceiving (Output)
Ratio 0.2
0.1
0.0
0 100 200 300
Frequency (MHz)
Asending (Input) to Creceiving (Output).
0
0 100 200 300
-20
dB
-40
-60
-80
Frequency (MHz)
Figure A2.10: Frequency Spectrum of Phase Asending (Input) to Creceiving (Output).

Appendix II 296
________________________________________________________________________
Phase Bsending (Input) to Creceiving (Output)
0.3
0.2
Ratio
0.1
0.0
0 100 200 300
Frequency (MHz)
Bsending (Input) to Creceiving (Output).
0
0 100 200 300
-20
dB
-40
-60
-80
Frequency (MHz)
Figure A2.12: Frequency Spectrum of Phase Bsending (Input) to Creceiving (Output).

Appendix II 297
________________________________________________________________________
8.2.3 B. Three phase XLPE cable
0.80
0.60
Ratio
0.40
0.20
0.00
0 100 200 300
Frequency (MHz)
Asending (Input) to Areceiving (Output).
0
0 100 200 300
-20
dB
-40
-60
-80
Fr eque ncy (MHz)
Figure A2.14: Frequency Spectrum of Phase Asending (Input) to Areceiving (Output).

Appendix II 298
________________________________________________________________________
Phase Bsending (Input) to Breceiving (Output)
0.80
0.60
Ratio
0.40
0.20
0.00
0 100 200 300
Frequency (MHz)
Bsending (Input) to Breceiving (Output).
0
0 100 200 300
-20
dB
-40
-60
-80
Freque ncy (MHz)
Figure A2.16: Frequency Spectrum of Phase Bsending (Input) to Breceiving (Output).

Appendix II 299
________________________________________________________________________
Phase Csending (Input) to Creceiving (Output)
0.80
0.60
Ratio
0.40
0.20
0.00
0 100 200 300
Frequency (MHz)
Csending (Input) to Creceiving (Output).
0
0 100 200 300
-20
dB
-40
-60
-80
Frequency (MHz)
Figure A2.18: Frequency Spectrum of Phase Csending (Input) to Creceiving (Output).

Appendix II 300
________________________________________________________________________
8.2.3 C. Frequency response measurement on Inner semi-conducting layers (Three

phase XLPE cable)

Lin Mag vs. freq
3.50E-01
3.00E-01
2.50E-01
2.00E-01
1.50E-01
1.00E-01
5.00E-02
0.00E+00
0.00E+00 1.00E+08 2.00E+08 3.00E+08 4.00E+08
-5.00E-02
Figure A2.19: Frequency response magnitude (S21, Damping ratio) is measured for Inner
Semi-conducting layer of Phase Asending (Input) to Areceiving (Output).
dB vs. freq
0.00E+00
-1.00E+01
0.00E+00 1.00E+08 2.00E+08 3.00E+08 4.00E+08
-2.00E+01
-3.00E+01
-4.00E+01
-5.00E+01
-6.00E+01
-7.00E+01
-8.00E+01
-9.00E+01
-1.00E+02
Figure A2.20: Frequency Spectrum for Inner Semi-conducting layer of Phase Asending
(Input) to Areceiving (Output).
Appendix II 301
________________________________________________________________________
8.2.3 D. Frequency response measurement on Outer semi-conducting layers (Three

phase XLPE cable)
Lin Mag vs. freq
1.20E-01
1.00E-01
8.00E-02
6.00E-02
4.00E-02
2.00E-02
0.00E+00
0.00E+00 1.00E+08 2.00E+08 3.00E+08 4.00E+08
Figure A2.21: Frequency response magnitude (S21, Damping ratio) is measured for Outer
Semi-conducting layer of Phase Asending (Input) to Areceiving (Output).
dB vs. freq
0.00E+00
0.00E+00 1.00E+08 2.00E+08 3.00E+08 4.00E+08
-1.00E+01
-2.00E+01
-3.00E+01
-4.00E+01
-5.00E+01
-6.00E+01
-7.00E+01
-8.00E+01
-9.00E+01
Figure A2.22: Frequency Spectrum for Outer Semi-conducting layer of Phase Asending
(Input) to Areceiving (Output).
Appendix III 302
________________________________________________________________________
Appendix III
Measured Voltage on Semi-conducting Layer of Phase A, B

and C (-100pC)
Further discussion can be found in Chapter 8.3
0.15
0.10
0.05
Voltage (V)
0.00
-1.00E-07 1.00E-07 3.00E-07 5.00E-07 7.00E-07 9.00E-07
-0.05
-0.10
-0.15
Time (s)
of inner semi-conducting layers on the sending ends are shown in graph.
Appendix III 303
________________________________________________________________________
0.10
0.05
Voltage (V)
0.00
-1.00E-07 1.00E-07 3.00E-07 5.00E-07 7.00E-07 9.00E-07
-0.05
-0.10
-0.15
Time (s)
SendA_outersemi SendB_outersemi SendC_outersemi
of outer semi-conducting layers on the sending ends are shown in graph.
0.08
0.06
0.04
Voltage (V)
0.02
0.00
-1.00E-07
-0.02 1.00E-07 3.00E-07 5.00E-07 7.00E-07 9.00E-07
-0.04
-0.06
-0.08
Time (s)
RecA_innersemi RecB_innersemi RecC_innersemi
of inner semi-conducting layers on the receiving ends are shown in graph.
Appendix III 304
________________________________________________________________________
0.10
0.05
Voltage (V)
0.00
-1.00E-07 1.00E-07 3.00E-07 5.00E-07 7.00E-07 9.00E-07
-0.05
-0.10
Time (s)
RecA_outersemi RecB_outersemi RecC_outersemi
of outer semi-conducting layers on the receiving ends are shown in graph.
0.15
0.10
0.05
Voltage (V)
0.00
-1.00E-07 1.00E-07 3.00E-07 5.00E-07 7.00E-07 9.00E-07
-0.05
-0.10
-0.15
Time (s)
Appendix III 305
________________________________________________________________________
0.10
0.05
Voltage (V)
0.00
-1.00E-07 1.00E-07 3.00E-07 5.00E-07 7.00E-07 9.00E-07
-0.05
-0.10
-0.15
Time (s)
SendA_outersemi RecA_outersemi
RecB_outersemi RecC_outersemi
\
0.15
0.10
0.05
Voltage (V)
0.00
-1.00E-07 1.00E-07 3.00E-07 5.00E-07 7.00E-07 9.00E-07
-0.05
-0.10
-0.15
Time (s)
SendA_innersemi SendB_innersemi
Appendix III 306
________________________________________________________________________
0.10
0.05
Voltage (V)
0.00
-1.00E-07 1.00E-07 3.00E-07 5.00E-07 7.00E-07 9.00E-07
-0.05
-0.10
-0.15
Time (s)
SendA_outersemi SendB_outersemi
Appendix IV 307
________________________________________________________________________
Appendix IV
Measured Voltage on Semi-conducting Layer of Phase A, B

and C (100pC)
Further discussion can be found in Chapter 8.3
0.06
0.04
0.02
Voltage (V)
0
-1.00E- 1.00E-07 3.00E-07 5.00E-07 7.00E-07 9.00E-07 1.10E-06
-0.02
07
-0.04
-0.06
Time (s)
semi-conducting layers on the sending ends are shown in graph.
Appendix IV 308
________________________________________________________________________
0.06
0.04
0.02
Voltage (V)
0
-1.00E-07 1.00E-07 3.00E-07 5.00E-07 7.00E-07 9.00E-07 1.10E-06
-0.02
-0.04
-0.06
Time (s)
SendA_outsemi SendB_outsemi SendC_outsemi
semi-conducting layers on the sending ends are shown in graph.
0.04
0.03
0.02
Voltage (V)
0.01
0
-1.00E-07
-0.01 1.00E-07 3.00E-07 5.00E-07 7.00E-07 9.00E-07 1.10E-06
-0.02
-0.03
-0.04
Time (s)
RecA_outsemi RecB_outsemi RecC_outsemi
semi-conducting layers on the receiving ends are shown in graph.
Appendix IV 309
________________________________________________________________________
0.06
0.04
0.02
Voltage (V)
0
-1.00E- 1.00E-07 3.00E-07 5.00E-07 7.00E-07 9.00E-07 1.10E-06
-0.02
07
-0.04
-0.06
Time (s)
semi-conducting layers on the sending and receiving ends are shown in graph.
0.06
0.04
0.02
Voltage (V)
0
-1.00E- 1.00E-07 3.00E-07 5.00E-07 7.00E-07 9.00E-07 1.10E-06
-0.02
07
-0.04
-0.06
Time (s)
SendA_outersemi RecA_outersemi
Appendix IV 310
________________________________________________________________________
0.06
0.04
0.02
Voltage (V)
0
-1.00E- 1.00E-07 3.00E-07 5.00E-07 7.00E-07 9.00E-07 1.10E-06
-0.02
07
-0.04
-0.06
Time (s)
SendA_outersemi SendB_outersemi
0.06
0.04
0.02
Voltage (V)
0
-1.00E- 1.00E-07 3.00E-07 5.00E-07 7.00E-07 9.00E-07 1.10E-06
-0.02
07
-0.04
-0.06
Time (s)
SendA_innersemi SendB_innersemi
Appendix IV 311
________________________________________________________________________

Partial Discharge Testing

Uploaded by

Copyright:

Available Formats

Partial Discharge Testing

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

Partial Discharge Testing

Uploaded by

Copyright:

Available Formats

PROPAGATION OF HIGH FREQUENCY PARTIAL

DISCHARGE SIGNAL IN POWER CABLES

HIO NAM JOHNSON, O

Supervisor: Assoc. Prof. Trevor R. Blackburn

School of Electrical Engineering and Telecommunications

Surname or Family name: O

Abstract 350 words maximum:

Declaration relating to disposition of project thesis/dissertation

…………………………………………………………… ……………………………………..……………… ……….……………………...…….…

FOR OFFICE USE ONLY Date of completion of requirements for Award:

THIS SHEET IS TO BE GLUED TO THE INSIDE FRONT COVER OF THE THESIS

In addition, development of a method to detect cross-coupling of PD signals between

I have either used no substantial portions of copyright material in my thesis or I have

I gratefully acknowledge the support and suggestions of Professor Mehdi Vakilian

Additionally, I wish to express my appreciation to the School of Electrical Engineering

Hio Nam Johnson, O

CERTIFICATE OF ORIGINALITY................................................................................ III

AUTHENTICITY STATEMENT ..................................................................................... V

LIST OF FIGURES ...................................................................................................... XVII

Dielectric constant (permittivity)...................................................................... 19

3.4.3 Underground Cable Parameters ................................................................. 77

4.4.2 Influence of skin effect .............................................................................. 119

d) Simulation of PD propagation in XLPE cables by ATP ............................ 172

9.2 Sensor Selection.......................................................................................... 254

References ............................................................................................... 260

Figure 2.1: Schematic design of overhead line................................................................... 7

Figure 2.21: Typical discharge from a cavity in insulation. ............................................. 37

conductivity of inner semi-conducting layer is based on Steven Boggs’s semi-con

K.Steinbuich semi-con measurement and assume that outer semi-conducting layer

Figure 5.1: Laboratory test measurement setup.............................................................. 148

Figure 6.1: Laboratory test measurement setup.............................................................. 186

Figure A1.1: the graph of attenuation as function of different frequency in the

Figure A1.18: the graph of attenuation as function of different frequency in the

Table 2.1: Resistivity of conductor materials at 20oC [10]............................................... 14

Table 3.1: Simulation results of different permittivity of insulation. ............................... 80

Table 5.1: Detail of cable examined. .............................................................................. 156

Further, to enable reconstruction of PD signals as emitted a viable high frequency model

1.1 Research Goals

• To determine the high frequency characteristics of semi-conducting materials and

• To develop a laboratory’s partial discharge monitoring systems and measurement

• To develop a method to model cross-coupling of PD signals between phases in three

The main achievements of this thesis are:

• Development of measurement techniques for high frequency signal characterization on

• Development of a method to detect cross-coupling of PD signals between phases of a

1.3 Synopsis of this thesis

Chapter 6 - Partial Discharge Detection: An investigation into the possibility of

Chapter 8 - High Frequency PD Signal Propagation in Three Phase Power Cables: An

Power Cables and Partial Discharge

2.0 Underground Transmission

2.0.1 Cable versus Overhead transmission line

2.1 Power cables

2.1.1 PILC cables

2.1.2 XLPE cables

Figure 2.5: Structure of the 11kV single core XLPE cable.

2.1.3 Classification of cables

Cables can be classified based on several parameters, such as:

2.2 Cable Structural Materials

Table 2.1: Resistivity of conductor materials at 20oC [10].

Table 2.2: Physical parameters of cable conductor materials.