Kou K 435

Creepage Discharge Behaviour and Numerical
Model Analysis at the Oil/Pressboard Interface

with the Effect of Nano Composite Coating
著者 Jang Kyunghoon
その他のタイトルナノコンポジットコーティングを効果を考慮した油
/プレスボード界面における沿面放電挙動及び数値
モデル解析
学位授与年度平成28年度
学位授与番号 17104甲工第435号
URL http://hdl.handle.net/10228/00006311
Doctoral Dissertation
Supervisor: Prof. Masayuki Hikita
Creepage Discharge Behaviour and

Numerical Model Analysis at the
Oil/Pressboard Interface with the Effect of
Nano Composite Coating
ナノコンポジットコーティングの
ナノコンポジットコーティングの効果を考慮した油/プレス
効果を考慮した油プレス
ボ―ド界面における沿面放電挙動及び数値モデル解析
February 2017
Department of Engineering
Graduate School of Engineering
Kyushu Institute of Technology
Jang Kyunghoon
Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering i

KYUSHU INSTITUTE OF TECHNOLOGY
ABSTRACT
DEPARTMENT OF ENGINEERING
Doctor of Philosophy
Creepage Discharge Behaviour and Numerical Model

Analysis at the Oil/Pressboard Interface with the Effect of
Nano Composite Coating
by Kyunghoon Jang
Electrical insulation has been described as the most important component of

electrical equipment such as transformer, circuit breakers, vacuum interrupters
and power cables. Above all, oil filled transformers are one of the basic elements
of a power system. They are connected to a large number of customers via power
transmission and distribution system. However, partial discharge and breakdown
phenomenon by lightning surge has proved to be one of the major factors and
weak point at the oil/pressboard (PB) interface. Besides, aged oil filled
transformers are increasing trend, there is need for replacement of aged
transformer. Especially, the life time of a transformer is mainly dependent on the
life of solid insulation. To enhance competitiveness and cost reduction of
transformer system, there is need for alternative conventional pressboard in terms
of miniaturization and simplification of insulation structure in oil filled
transformer. Besides, to improve of design electric field, dielectric strength as
well as suppression of creepage discharge occurrence, there is necessary to
improve the insulation performance to prevent discharge phenomena.
In this reason, a lot of work has been recently devoted to improve the dielectric
properties of insulation by adding specific functional additives in to the solid
polymer, GIS spacer as well as liquid oil. However, recently few studies have
been focused on modifications of the insulating paper using various specific
functional additives to improve dielectric and mechanical properties; it is because
streamer propagation in oil is greatly affected by the presence of pressboard.
Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering ii

Above viewpoints, this research work suggest the new coating method on
pressboard surface by thin solid layer coating such as epoxy resin, Teflon coating
and silica/epoxy nanocomposite to study discharge phenomena and behavior in
transformer composite insulation system. In experiment, five kinds of specimens
are used with different coating materials using rod-plane and needle-bar electrode
configuration. The creepage discharge properties of conventional pressboard in
oil/pressboard composite insulation are compared with that of the solid layer
coated pressboard newly in terms of discharge propagation length, discharge
occurrence probability and dielectric properties. Besides, to clarify the effect of
epoxy/silica nanocomposite, we investigated the creepage discharge behavior
using epoxy/silica nanocomposite plate with different silica filler loading and
different nano silica size.
Additionally, we investigated the creepage discharge behavior using COMSOL
Multiphysics 2D model for simulation. In simulation, to understand the creepage
discharge phenomena in composite insulation system depending on solid
insulation parameters such as work function and permittivity, creepage discharge
modelling and simulation have been investigated. The simulation is based on a
system of coupled general expression of governing equations that contain the
physics to model streamer development based on the hydrodynamic diffusion-
drift including ionization, dissociation, charge recombination and electron
attachment, combined with Poisson’s equation in dielectric liquids. The governing
equations for solid pressboard insulation are composed of Gauss’s Law. All the
results are reported and discussed.
Thesis Supervisor: Prof. Masayuki Hikita

Title: Professor, Department of Engineering
Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering iii

Contents
Abstract iv
Contents iiv
List of Figures iv
List of Tables iv
Acknowledgements iv
Symbols & Abbreviations iv
Chapter 1
1.1 OIL FILLED TRANSFORMER................................................................... 1
1.2 RESEARCH MOTIVATION AND PROBLEM ......................................... 3
1.3 RESEARCH OBJECTIVES AND CONTRIBUTIONS .............................. 4
1.4 THESIS STRUCTURE ................................................................................. 5
CHAPTER 2
2.1 MINERAL OIL ............................................................................................. 8
2.1.1 The effect of moisture on mineral oil ............................................................. 10
2.2 CELLULOSE-BASED PRESSBOARD .................................................... 11
2.2.1 The effect of moisture on cellulose-based pressboard ....................................... 14
2.3 CREEPAGE DISCHARGE PHENOMENA AT THE OIL/
PRESSBOARD INTERFACE .................................................................. 15
2.3.1 The influence of solid insulation in mineral oil ................................................ 15
2.3.2 Discharge propagation modes ....................................................................... 18
2.3.3 Effect of relative permittivity in composite insulation system ............................ 23
2.3.4 Permittivity and polarization ......................................................................... 25
CHAPTER 3
3.1 BASIC ELECTRODYNAMIC EQUATIONS ........................................... 28
3.2 CHARGE CARRIER GENERATION ....................................................... 29
Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering iv

3.2.1 Charge injection by electric field emission ...................................................... 29
3.2.2 Charge injection by thermal electron emission ................................................. 33
3.2.3 Molecular ionization by electric field ............................................................. 34
3.2.4 Dissociation of ion by electric field and temperature ........................................ 37
3.3 CARRIER RECOMBINATION ................................................................. 40
3.4 ELECTRON ATTACHMENT ................................................................... 41
3.5 ION AND ELECTRON MOBILITY AVALUES………………………..42
3.6 SUMMARY ................................................................................................ 42
CHAPTER 4
4.1 INTRODUCTION....................................................................................... 44
4.2 EXPERIMENT SETUP .............................................................................. 45
4.3 EXPERIMENTAL PROCEDURE………………………………………..47
4.3.1 Creepage discharge observation .................................................................... 47
4.3.2 Sample preparation&dielectric property measurement ...................................... 52
4.3.3 Surface roughness and nano silica distribution state ......................................... 55
4.3.4 Partial discharge inception............................................................................ 58
4.4 RESULTS (ROD-PALNE ELECTRODE)................................................. 61
4.4.1 Dielectric properties .................................................................................... 61
4.4.2 Discharge patterns and PDIV ................................................................................... 63
4.4.3 Discharge propagation length........................................................................ 68
4.5 RESULTS (NEEDLE-BAR ELECTRODE) .............................................. 69
4.5.1 Discharge patterns and PDIV ........................................................................ 69
4.5.2 Discahrge propagation length........................................................................ 74
4.6 DISCUSSION (EFFECT OF SOLID LAYER COATING) ...................... 76
4.6.1 Dry effect and permittivity effect ................................................................... 77
4.6.2 Surface modification effect ........................................................................... 79
4.6.3 Electron suppression effect ........................................................................... 81
4.7 SUMMARY ................................................................................................ 82
CHAPTER 5
5.1 EXPERIMENT METHOD ......................................................................... 85
5.1.1 Electrode setup ........................................................................................... 85
5.1.2 Sample preparation ..................................................................................... 87
5.1.3 Dielectric property measurement ................................................................... 91
5.2 RESULTS ................................................................................................... 92
Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering v

5.2.1 Dielectric properties .................................................................................... 92
5.2.2 Discharge patterns and discahrge occurrence probability ................................... 98
5.2.3 Discharge propagation length...................................................................... 102
5.2.4 DISCUSSION .......................................................................................... 105
5.3 SUMMARY .............................................................................................. 109
CHAPTER 6
6.1 INTRODUCTION..................................................................................... 111
6.2 GOVERNING EQUATIONS IN OIL ...................................................... 112
6.2.1 Hydrodynamic diffusion-drift model ............................................................ 112
6.2.2 Charge emission from electrode and solid insulation surface ........................... 113
6.2.3 Lightning impulse voltage .......................................................................... 117
6.3 NEEDLE-BAR GEOMETRY .................................................................. 118
6.4 SIMULATION RESULTS ....................................................................... 120
6.4.1 Creepage discharge under high electric field ................................................. 120
6.4.2 Poole-Frenkel model for solid insulation dependent on effective barrier
height ...................................................................................................... 125
6.4.3 creepage discharge velocity and bdv dependent on permittivity effect ............... 126
6.5 SUMMARY .............................................................................................. 127
CHAPTER 7
7.1 CONCLUSIONS ....................................................................................... 130
7.2 FURTHER WORK ................................................................................... 131
APPENDIX A
APPENDIX B
APPENDIX C
Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering vi

List of Figures
Figure 1.1 Structure of oil filled transformer ........................................................ 2

Figure 1.2 Flashover failure of insulation ............................................................. 4
Figure 2.1 Molecular structures of hydrocarbon groups (Paraffinic,
naphthenic and aromatic) .................................................................... 9
Figure 2.2 Measured dependency of the breakdown voltage on the relative
moisture saturation with an approximative model fitted to the data.
Each point corresponds to the average of six breakdown voltage
measurements performed according to the IEC 60156 standard ....... 10
Figure 2.3 Formed items from pressboard .......................................................... 12
Figure 2.4 Micro to sub-micro structure of wood fibers ..................................... 12
Figure 2.5 Cellulose polymers ............................................................................ 13
Figure 2.6 Microscopic view of pressboard insulation ....................................... 13
Figure 2.7 General orientations of oil-pressboard interface with respect to
dominant direction of electric field ................................................... 15
Figure 2.8 Vector plots of the electric field distribution for the streamer
propagating (a) in air, (b) along an insulating surface and (c)
electric field distribution when a layer of charge with density
σ = 50μC/m2 is applied to the dielectric surface ........................... 16
Figure 2.9 discharge propagation lengths in FR3 at 43 kV ................................ 16
Figure 2.10 Creepage discharge stages from discharge inception to breakdown . 17
Figure 2.11 Typical propagation modes in esters (Midel 7131, positive polarity,
d=50 mm) ......................................................................................... 19
Figure 2.12 Evolution of streamer shape along applied voltage (Midel 7131,
negative polarity, d=50 mm) ............................................................ 20
Figure 2.13 Average streamer velocity versus voltage in the liquid alone and
with a solid parallel to the field (d=10 cm) ...................................... 21
Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering vii

Figure 2.14 Positive streamer propagation along solid interface; Upper: the
positive streamer model, Lower: actual tracks observed on PB
surface ................................................................................................ 22
Figure 2.15 Streamer inception probability distribution function for point
cathode in transformer oil with and without solid interface .............. 24
Figure 2.16 Equivalent series model to represent solid-liquid interface ............... 24
Figure 2.17 The net charge enclosed within a differential sized volume of
dipoles has contributions only from dipoles that are cut by the
surfaces. All totally enclosed dipoles contribute no net charge. (b)
only those dipoles within a distance 𝑑⃗ ∙ 𝑛⃗⃗ of the surface are cut by
the volume ......................................................................................... 26
Figure 3.1 Free electron emission from ground electrode under high electric
field 1~10 × 108 V/m ...................................................................... 30
Figure 3.2 Mechanism of field emission from the metal surface ........................ 31
Figure 3.3 Schematic diagram of thermionic emission from the metal surface.. 34
Figure 3.4 Molecular ionization by electric field ................................................ 35
Figure 3.5 Ionic dissociation by electric field ..................................................... 39
Figure 3.6 Recombination process in mineral oil ............................................... 41
Figure 4.1 Rod-plane electrode configuration (type 1) ...................................... 45
Figure 4.2 Needle-bar electrode configuration (type 2) ..................................... 45
Figure 4.3 Needle electrode (a) before experiment, (b) after experiment and (c)
after breakdown ................................................................................. 46
Figure 4.4 Schematic diagram of the experimental setup and measuring
system ................................................................................................ 47
Figure 4.5 Photograph of the experimental system ............................................. 49
Figure 4.6 FINEPIX S100FS and image intensifier (I.I) .................................... 50
Figure 4.7 Photo multiplier (PMT) .................................................................... 51
Figure 4.8 Power supply ..................................................................................... 51
Figure 4.9 Dimensional outlines of PMT............................................................ 52
Figure 4.10 Schematic diagram to prepare for nano silica and epoxy/silica
coating on surface of pressboard ....................................................... 53
Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering

viii
Figure 4.11 Pressboard samples used in experiment ............................................ 54
Figure 4.12 Weight and thickness variation of each sample before and after
drying process .................................................................................... 55
Figure 4.13 Surface roughness before and after epoxy resin coating (a) DP and
(b) EP-F ............................................................................................. 56
Figure 4.14 Surface roughness of each type of sample measured using a surface
roughness tester (bottom: Ra, top: Rz) ............................................... 57
Figure 4.15 The dispersion state of silica nano size particles in the base epoxy
matrix (Cross section view) (a) 0.2 wt%, (b) 0.5 wt%, (c) 1 wt%,
(d) 3 wt%, 5 wt% and (f) 50 wt% ...................................................... 57
Figure 4.16 SEM image of SiO2 nanoparticles surrounded by red circle
dispersion in base epoxy matrix (Cross section view) (a) 0.2 wt%,
(b) 0.5 wt%, (c) 1 wt%, (d) 3 wt%, (e) 5 wt% and (f) 50 wt%.......... 58
Figure 4.17 Determination of creepage discharge inception voltage (a) without
creepage discharge inception at 25 kV and (b) with creepage
discharge inception at 55 kV ............................................................. 59
Figure 4.18 The reaction between the output of image intensified camera and
charge amount in pC as a function of lightning impulse voltage ...... 60
Figure 4.19 Frequency dependence of complex permittivity at room
temperature; (a) real part of permittivity of each specimen, (b)
dielectric loss ..................................................................................... 62
Figure 4.20 Relative permittivity 𝜀r′ at 100 kHz ................................................... 63
Figure 4.21 Discharge patterns of each sample at 100 kV using rod-plane
electrodes in mineral oil .................................................................... 64
Figure 4.22 Discharge binary images of each sample using image analysis
program at 80 kV ............................................................................... 65
Figure 4.23 The number of discharge branches of each sample at 80 kV and
100 kV ............................................................................................... 66
Figure 4.24 Discharge occurrence probability 𝑃d at each voltage level with
different coating material ................................................................... 67
Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering ix

Figure 4.25 PD occurrence probability ratio k1 against EP-F versus filler
loading of nano SiO2, where k1 is defined as the voltage of 10%
discharge occurrence probability of SP divided by that of EP-F ...... 67
Figure 4.26 Discharge propagation length ld at each lightning impulse voltage
(NP, DP, EP-A, EP-F and TP) .......................................................... 68
Figure 4.27 Discharge propagation length ld of SP specimens at each lightning
impulse voltage .................................................................................. 69
Figure 4.28 Discharge patterns of (a) NP, (b) DP, (c) EP-F, (d) SP-0.2 wt% and
(e) SP-3 wt% at Vi 45 kV ................................................................... 71
Figure 4.29 Discharge light intensity from PMTs waveforms as a function of Vi
for different kinds of specimen .......................................................... 72
Figure 4.30 Discharge occurrence probability at each voltage level for
different kinds of specimen ............................................................... 72
Figure 4.31 The ratio k2 of 30% discharge occurrence probability Pd of nano
silica filler loaded to that of EP and coated NP and relative
permittivity 𝜀r′ at 100 kHz as a function of nano silica filler loading
for epoxy/silica nanocomposite coated PB ........................................ 73
Figure 4.32 Discharge propagation length ld of NP, DP, EP-A and EP-F as a
function of Vi .................................................................................... 75
Figure 4.33 Discharge propagation length ld of SP specimens as a function of
Vi for the needle-bar electrode system ............................................... 75
Figure 4.34 The relationship between average discharge propagation length
ladve ratio k3 and permittivity 𝜀r′ of SP with different silica filler
loading at 45 kV ................................................................................ 76
Figure 4.35 Bonds of water molecules in a PB microcapillary ............................ 77
Figure 4.36 Relation between estimated discharge inception voltage and the
Specific capacitance before and after dry process of pressboard at
70 and 150℃ for 3 hours, respectively .............................................. 78
Figure 4.37 Surface protrusions on the surface of conventional pressboard (DP)
using microscope and scanning electron microscope image ............. 79
Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering x

Figure 4.38 Analysis model with presence of protrusions on the surface of PB
with surface roughness measured in DP ............................................ 80
Figure 4.39 The calculated of electric field distribution with presence of
protrusion of the surface of PB around high voltage electrode ........ 80
Figure 4.40 The electric field results of protrusion tip parts and pole parts ......... 81
Figure 4.41 Surface discharge concepture model in oil/PB composite
insulation system ............................................................................... 82
Figure 5.1 Needle-bar electrode configurations with nano silica/epoxy
composite plate .................................................................................. 86
Figure 5.2 Diagram of experimental setup to measure and observe PDIV and
creepage discharge properties in oil and SiO2/epoxy composite
insulation system ............................................................................... 87
Figure 5.3 Photos of samples with different nano silica filler loading and neat
epoxy (a) neat epoxy, (b) 0.1 wt%, (c) 0.5 wt%, (d) 1 wt%, (e) 3
wt%, (f) 5 wt%, (g) 10 wt%, (h) 20 wt% and (i) 45 wt% .................. 88
Figure 5.4 SEM images of surface and middle layer of nano silica/EP
composites with different nano silica ................................................ 89
composites with different nano silica ................................................ 90
Figure 5.6 The variation of weight and thickness of 5 wt% silica/epoxy nano
composite before and after drying process at 100℃ for 1hour.......... 91
Figure 5.7 Silica/epoxy nanocomposites for dielectric property measurement .. 92
temperature; (a) real part of permittivity of each specimen, (b)
dielectric loss .................................................................................... 93
Figure 5.9 Relative permittivity at 100 kHz with different nano silica filler
loading ............................................................................................... 94
temperature with different nano silica size with keeping the filler
loading 3 wt%; (a) real part of permittivity of each specimen, (b)
dielectric loss ..................................................................................... 96
Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering xi

temperature with different nano silica size with 10 wt%; (a) real
part of permittivity of each specimen, (b) dielectric loss .................. 97
Figure 5.12 Relative permittivity at 100 kHz as a function of nano silica size
for different filler loading ................................................................. 98
Figure 5.13 Discharge image of neat EP and SP 45-1 wt% at 75 kV ................... 99
Figure 5.14 Discharge occurrence probability at each voltage level with
different nano silica filler loading ..................................................... 99
Figure 5.15 The relationship between discharge occurrence probability Pd and
permittivity 𝜀r′ ................................................................................. 100
Figure 5.16 Discharge occurrence probability 3 wt% at each voltage level with
different nano silica size .................................................................. 101
permittivity 𝜀r′ of 3 wt% .................................................................. 101
Figure 5.18 Discharge propagation length ld of EP with different silica filler
loading as a function of Vi with keeping the 45 nm filler size......... 102
Figure 5.19 The relationship between average discahrge propagation length
ldave and permittivity 𝜀r′ of EP as a function of silica filler loading
at 75 kV ........................................................................................... 103
Figure 5.20 Discharge propagation length ld of EP with different silica filler
size as a function of Vi with keeping the 3 wt% filler loading ........ 103
Figure 5.21 The relationship between average discharge propagation length
ldave and permittivity 𝜀r′ of EP as a function of silica filler size at 75
kV with keeping the 3 wt% filler loading........................................ 104
Figure 5.22 Discharge propagation lengths of EP with different silica filler size
as a function of lightning impulse voltage with keeping the 10 wt%
filler loading .................................................................................... 104
Figure 5.23 Calculated value of relative permittivity for epoxy/silica
nanocomposite with different silica filler loading (relative
permittivity of matrix and filler=3.46 and 5.5) ................................ 107
Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering xii

Figure 5.24 E-d characteristics of each sample with different nano silica filler
loading ............................................................................................. 107
Figure 5.25 Surface flashover conceptual model in mineral oil ......................... 108
Figure 6.1 Relation between each current density (JFE, JPF) and electric field . 114
Figure 6.2 Molecule structure of cellulose paper [C6 H10 O5 ] .......................... 115
Figure 6.3 Quantum chemical calculation of molecule consists of PB ............ 116
Figure 6.4 Coonceptual model of quantum chemical calculation of PB .......... 117
Figure 6.5 Waveform of standard lightning impulse voltage for numerical
simulation. The rising time, 𝑇f is 1.2 𝜇𝑠 and the 50% of falling
time, 𝑇t is 50 𝜇𝑠. (amended from [119]).......................................... 118
Figure 6.6 Needle-bar electrode system with PB immersed in oil .................... 119
Figure 6.7 Computer-aided design representation of the needle-bar electrode
geometry used for streamer simulation and conceptual model of
electron emission from ground and PB bulk ................................... 119
Figure 6.8 Electric field wave propagation as a function of time at 25 kV ..... 121
Figure 6.9 Electric field distribution and electric potential as a function of
time teps .......................................................................................... 122
Figure 6.10 Electron and space charge density distribution as a function of time
steps ................................................................................................. 122
Figure 6.11 Temperature rise as a function of time as the needle tip ................. 123
Figure 6.12 Temperature distribution as a function of time step ........................ 123
Figure 6.13 Discharge current derived from the integration of current density
during the creepage discharge propagation on the surface of PB at
25 kV ............................................................................................... 124
Figure 6.14 Discharge velocity and breakdown time with different ΦD of solid
insulation under 100 kV lightning impulse voltage ......................... 126
Figure 6.15 Creepage discharge propagation velocity and BDV with different
𝜀r′ ...................................................................................................... 127

xiii
List of Table
Table 2.1 Creepage discharge on different pressboard condition...................... 14

Table 3.1 Characteristic values of pure transformer oil .................................... 36
Table 4.1 Specification of the C9016-2X .......................................................... 51
Table 4.2 Properties of each specimen .............................................................. 53
Table 4.3 Specification of PMT and image intensifier ...................................... 60
Table 5.1 Properties of prepared nano silica/EP composites ............................. 88

xiv
Acknowledgements
First and foremost, I would like to express my deepest gratitude to Prof.

Masayuki Hikita and Prof. Masahiro Kozako who were my thesis advisors for
unwavering support and guidance during my Ph.D. studies. They are an
outstanding professors and person and working with Prof. Hikita and Kozako has
been a great privilege, great honor and a truly positive experience in Japan.
I also wish to thank my thesis committee members, Prof. Mitani, Prof. Komori
and Prof. Toyoda for their valuable suggestions, guidance and encouragement that
have substantially helped in my thesis.
Secondly, I would like to convey my honest appreciation to Prof. June-Ho Lee

(Hoseo Univ.) and Se-Hee Lee (Kyungpook Univ.) for they guidance and
encouragement that have substantially helped in my research work. Above all it
has been a great honor to work with and learn on COMSOL MULTIPHYSICS
from Prof. Se-Hee Lee and his students Su-Hun Kim and Jin-Hyun Choi.
Next, I would like to thank the Nissan Chemical Industry., LTD for their
technical support of all the nano silica samples. I am thankful to all the members
of the Nissan chemical research team.
A big thank you to my colleagues and friends who are Rajab during Ph.D.
study, Impulse PD team Akahoshi and Kurachi for 3 years. And I would like to
thank to convey my friends in Korea.
I would like to thank my family, Jang doosoo, Lee Suil and Jang Kyungtak,
they have always been there to encourage and support me in what I have done,
and they have made many sacrifices to give me the opportunities. And welcome
back to Seoul, Korea for my cousin Minji Chang and her husband Jaisun Ihm :).
Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering xv

Symbols
Symbols used in thesis
∆ Molecular ionization energy [J]

𝜀0 Permittivity of vacuum or free space (8.854 × 10−12 F ∙ m−1 )
𝜀r Relative permittivity of a material
𝜇 Mobility of charge carrier [m2 ∙ s−1 ∙ V −1 ]
𝜌 Mass density [kg ∙ m−3 ]
σ Electric conductivity [S∙ m−1 ]
𝜏a Electron attachment time constant [s]
α Molecular separation distance [m]
𝐶p Specific heat capacity of a material [J ∙ kg −1 ∙ K −1 ]
𝐷 Diffusion coefficient [m2 ∙ s−1 ]
⃗⃗
𝐷 Electric displacement field vector [C ∙ m−2]
𝐸⃗⃗ Electric field vector [V∙ m−1 ]
EA Electron attachment rate [mol ∙ m−3 ∙ s−1 ]
f(|𝐸⃗⃗ |) Electric field dependent function
𝐹⃗ Total flux density vector [mol ∙ m−2 ∙ s −1 ]
G Generation rate term of particles [mol ∙ m−3 ∙ s −1]
G(|𝐸⃗⃗ |) Generation rate term of particles that electric field dependent
[mol ∙ m−3 ∙ s−1]
h Planck’s constant [6.626 × 10−34 J ∙ s]
I Electric Current
𝐼1 Modified Bessel function of the first kind
⃗⃗⃗
𝐽c Conduction current density vector in a volume [mol ∙ m−2 ∙ s −1 ]
or on surface [mol ∙ m−1 ∙ s−1]
⃗⃗⃗⃗
𝐽d Displacement current density vector [C ∙ m−1 ∙ s−1 ]
k Boltzmann’s constant [1.38 × 10−23 J ∙ K −1 ]
kT Thermal conductivity of a material [W ∙ m−1 ∙ K −1]
𝐾d0 Zero field dissociation rate constant [s −1 ]
Kd Dissociation rate constant [s−1 ]
Kr Recombination rate constant [𝑚3 ∙ mol−1 ∙ s−1 ]
𝑚∗ Effective electron mass [kg]
𝑛̂ Unit vector normal to the boundary

xvi
N Density of particles in a volume [mol ∙ 𝑚−3 ] or on surface
[mol ∙ 𝑚−2]
NA Avogadro’s number (6.023× 1023 mol−1 )
q Elementary charge (1.6022× 10−19 C)
Q Electrical power dissipation in a volume [W ∙ m−3 ] or on surface
[W ∙ m−2]
t time [s]
T Absolute temperature [K]
𝑢
⃗⃗ Velocity vector of a fluid [m ∙ s−1 ]
vi Velocity of each charge carrier [m ∙ s −1 ]
X Relative mass density
Wa Work function
Va Applied voltage
Abbreviations in thesis
AC Alternating current
DC Direct current
PD Partial discharge
BDV Breakdown voltage
PDIV Partial discharge inception voltage
HV High voltage
GND Ground
IEEE Institute of Electrical and Electronics Engineers
PMT Photomultiplier tube
wt% Weight percent
PB Pressboard
NP Neat pressboard or/and Neat epoxy
DP Dry pressboard
EP Epoxy coated pressboard
SP Silica/epoxy composite coated pressboard or/and Silica/epoxy
composite plate
TP Teflon coated pressboard
I.I Image Intensifier
𝑃d Discharge occurrence probability
𝑙d Discharge propagation length
𝑅z 10 point surface roughness
𝑅a Average surface roughness

xvii
𝑙dave Average discharge propagation length
Vi Impulse voltage
𝑆l Nano silica filler loading
k1 10% discharge occurrence probability ratio
k3 Discharge propagation length ratio of epoxy silica nanocomposite
coated PB
k5 Discharge propagation length ratio of epoxy/silica nanocomposite

xviii
xix
Chapter 1: Introduction
CHAPTER 1
1.1 OIL FILLED TRANSFORMER

Electrical insulation has been described as the most important component of
electrical equipment such as transformer, circuit breakers, vacuum interrupters
and cables. Above all, oil filled transformers are one of the basic elements of a
power system. They are connected to a large number of customers via power
transmission and distribution system. And also large transformers are one of the
basic elements of electric power system. When it comes to ageing and degradation
process, large power transformers are a major concern to any electricity utility [1].
A large number of transformers working in power systems in many countries of
the world have exceeded their technical lifetime. While designing, it is assumed
that a transformer’s technical lifetime is usually over 30 years. It is estimated that
in Central Europe over 35% of power transformers have exceeded the age of 30,
and about 20% have been working in the power system for over 35 years. The
technical lifetime of a transformer depends mainly on the technical lifetime of oil-
paper insulation. Especially, the life of a transformer is mainly dependent on the
life of solid pressboard insulation of the winding paper. The pressboard decreases
in tensile strength with the progress of aging and, at some point, can no longer
withstand the short circuit stresses and with passing years the insulation used in
the transformer undergoes ageing [3, 4].
Oil filled transformer as shown in Fig. 1.1 (a) and (b) basically adopts mineral
oil and pressboard (PB) composite insulation system as an insulating medium
which can be categorized into major insulation and minor insulation. It is because
cellulose materials have excellent insulating capability in insulating oil as well as
low cost and good performance. The major insulation consists of insulation at the
region between different windings, between windings and the core limb/yoke and
between leads and ground. In a large oil-immersed power transformer, the use of
cellulose-based pressboard for electrical insulation is required in the gap between
phase windings. Pressboard barriers that have higher dielectric strength compared
to the mineral oil are important in dealing with the pre-breakdown events by
increasing the partial discharge inception voltage (PDIV) [5, 6].
Many kinds of defects such as excessive moisture, high temperature and
extremely concentrated electrical stress often lead to partial discharge (PD) [7].
On the other hand solid-liquid interface of oil filled transformer is considered as a
Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering 1

weak point giving rise to failure modes in the oil/pressboard composite insulation
system, it is because creepage discharge tend to propagate along the interface
between oil and pressboard and then leading to flashover and breakdown.
(a) Oil filled power transformer
(b) Schematic of inner structure of oil filled transformer

Figure 1.1 Structure of oil filled transformer [1, 8]

1.2 RESEARCH MOTIVATION AND PROBLEM
Cellulose paper or pressboard and mineral oil are the main insulation materials
used in oil filled transformer as a composite insulation system to protect the
equipment from lightning surge. However, partial discharge and breakdown
phenomenon by lightning surge has proved to be one of the major factors and
weak point at the oil/pressboard interface. Besides, aged equipment is increasing
rapidly. To enhance competitiveness and cost reduction of oil filled transformer,
there is need for alternative conventional insulation, miniaturization, and
simplification of insulation structure and improvement of design electric field. For
above needs, the distance between insulation is narrowed, meaning that electric
field and probability of discharge at oil/PB interface increases than designing
condition. Thus, it is necessary to improve the dielectric property of conventional
pressboard [1-9].
Figures 1.2 (a) and (b) show flashover failure along the mineral oil and
pressboard interface. The creepage discharges tend to propagate along the
interface between mineral oil and pressboard due to different permittivity. Despite
advances in manufacturing and monitoring technologies, some unexpected faults
and failures still occur. These are usually dielectric in nature, and often without
any obvious cause [10].
Lightning impulse withstanding property is one of important characteristics in
power apparatus to design of oil filled transformer; it is because the lightning
impulse is a pure natural phenomenon. On the other hand, it is very difficult to
predict the actual wave shape of a lightning disturbance. In addition, engineers
want to predict the discharge initiation, development process in composite
insulation system for upgrading the insulation design of power apparatus under
lightning surge. Indeed, the conditions of the occurrence of discharges and the
parameters influencing discharge development up to flashover are very useful for
the design and dimensioning of high voltage components and systems such as
insulators, spacers, bushings and circuit breakers envelopes. For these reasons, a
lot of works have recently been devoted to improve the dielectric properties of
insulating oil by adding specific functional additives into the oil [11, 12].
However, few studies have been focused on modifications of the insulating paper
using various specific functional additives to improve dielectric and mechanical
properties [13-16], it is because creepage discharge phenomena has been reported
in many literatures that streamer propagation in oil is greatly affected by the
presence of pressboard and other solid insulation [1, 17, 18-22]. Thus there is a
growing interest in suppression of creepage discharge in oil filled transformer.

(a) (b)
(c)
Figure 1.2 Flashover failure of insulation (a) to (c) [1, 2]
1.3 RESEARCH OBJECTIVES AND

CONTRIBUTIONS
The main objective of this research is to study the behavior and characteristics
of creepage discharge at the oil/pressboard composite insulation system depending
on solid layer coating materials on PB surface such as epoxy resin and
epoxy/silica nanocomposite under standard lightning impulse voltage. Second
objective is to investigate the effect of nano size silica on suppression of creepage
discharge inception and discharge propagation process by silica/epoxy
nanocomposite coating method.
The research has involved two kinds of approach such as experimental results
and computer simulation studies. The experiments for creepage discharge
observation at the oil/pressboard interface have been conducted in an oil chamber
experiment setup. The creepage discharge characteristics of conventional

pressboard are compared with solid layer coated pressboard. A numerical analysis
method in composite insulation system has been developed in the COMSOL
Multiphysics 2D Model, a finite element analysis (FEA) software package to
understand the creepage discharge behavior and mechanism. With respect to all
these, contributions are followings:
1. The experimental work presented in this thesis contributes to a further

understanding on the surface discharge behavior at the oil-pressboard interface
dependent on surface condition, nano silica filer loading and different silica size.
2. This experiment results may assist in the prediction of discharge initiation

and development in mineral oil for insulation design as well as upgrading for large
power transformer and to allow study of their lifespan.
1.4 THESIS STRUCTURE
Chapter 1 discusses the problems and weak point in terms of partial discharge
within large transformers associated with the oil/pressboard interface that
motivates this research. The research objectives have been outlined.
Chapter 2 presents a review of the literature that is directly relevant to both the
experimental work and numerical simulation model development that has been
undertaken as part of this thesis.
Chapter 3 presents general electrohydrodynamics model in mineral oil in terms

of ionization, dissociation, carrier recombination and carrier attachment due to
extremely high electric field and temperature variation.
Chapter 4 explains about creepage discharge properties at the interface between

oil and epoxy/silica nanocomposite coated pressboard under the applied standard
lightning impulse voltage. In this chapter, the experimental set up (rod-plane
electrode and needle-bar electrode) and method are indicated and the creepage
discharge characteristics in terms of discharge occurrence probability, discharge
propagation length and dielectric properties are compared with those of
conventional pressboard and they have been discussed in this chapter.

Chapter 5 explains about the creepage discharge and dielectric properties of

epoxy/silica nanocomposite plate with different filler loading and particle size to
verify the effect of epoxy/silica nanocomposite on discharge suppression. In this
research, the discharge occurrence probability, discharge propagation length and
dielectric properties have been examined as a function of the voltage under
standard lightning impulse voltage. The experimental preparation, setup and
procedure are presented in detail.
Chapter 6 describes the numerical analysis method in composite insulation

system quantitatively to understand the creepage discharge behavior and
mechanisms depending on relative permittivity by epoxy/silica nanocomposite at
short gap. Analysis model fully is coupled with governing equations based on the
charge continuity equations for positive ions, negative ions and free electrons,
along with the energy balance equation for a temperature and Poisson’s equation
for electric fields. The validation of the model using experimental results is
discussed. In terms of simulation results, the role of electric field dependent
molecular ionization on the streamer development is explained. In addition, the
roles of electron attachment and other recombination process (ion-ion and ion-
electron). Furthermore, the possibility of creepage discharge model for long gap
distance in insulation oil is investigated.
Chapter 7 is the conclusions of this work, which is summarizing an important

outcome obtained by an accomplishment of the present research. The chapter
concludes with suggested future work to extend the theoretical work found in this
thesis.


Chapter 2: Introductions of Oil Filled Transformer & Creepage Discharge Phenomena
CHAPTER 2
Insulations of Oil Filled Transformer &

Creepage Discharge Phenomena
2.1 MINERAL OIL
A petroleum-based mineral oil has been most often employed in the electrical
insulation inside power transformers for more than a century for electrical
equipment including transformers, circuit breakers and power cables. Mineral oil
has an excellent performance as electrical insulation (electrical breakdown
strength and thermal conductivity than gaseous insulators) and cooling medium
[22, 23]. Unfortunately, as with all insulation, the failure of the liquid insulation
can cause catastrophic damage to not only the power equipment, but also the
surrounding environment. And also some problems such as scarcity and high price
of mineral oil, sulfide-induced corrosion of copper lead to be conductive in liquid
insulation, environmental pollution due to oil leakage, atmospheric pollution have
been pointed out for practical use [22].
Mineral oils are obtained from crude petroleum extracted from the earth.
Mineral insulating oil is a mix of hydrocarbons in the form of chain and ring
molecules with small amounts of sulphur, nitrogen, oxygen and trace metallic
elements [1, 23, 24]. Mineral oil is classified by the percentage content of the
following three hydrocarbon groups, which are paraffinic (C2n H2n+2 ), naphthenic
(C2n H2n ) and aromatic (Cn Hn ) as shown in Fig. 2.1 [23-26]:
 Paraffinic – straight and branched saturated hydrocarbons

 Naphthenic (cycloparaffins) – hexagonal ring of carbon-carbon single
covalent bonds
 Aromatic – hexagonal ring modelled as alternating single and double
carbon-carbon covalent bonds indicating delocalization of the carbon-
carbon bonding electrons.

Chapter 2: Insulations of Oil Filled Transformer & Creepage Discharge Phenomena
The paraffinic hydrocarbons are composed of straight and branched chains of

carbon and hydrogen atoms, whilst the naphthenes have a saturated ring-like
structure. The aromatics also have ring-like structure which is unsaturated that can
be identified by double bonds between some of the carbon atoms [1]. In general,
these mixtures determine the physical properties of transformer oil that make it an
excellent electrical insulator and cooling medium such as high resistivity, high
dielectric strength, high specific heat and thermal conductivity along with low
viscosity and pour point as well as low density. The other important parameter for
most transformer oils is the relative permittivity𝜀oil that typically has a value of
2.2 [1-4, 23].
Figure 2.1 Molecular structures of hydrocarbon groups (paraffinic, naphthenic and

aromatic) [1, 26]

2.1.1 THE EFFECT OF MOISTURE ON MINERAL OIL
Insulation oil is hydrophobic and oil and water do not readily mix but the real
problem with moisture in oil is that, even in small amounts, it has a marked effect
on the dielectric strength of the oil as shown in Fig. 2.2 [26]. The breakdown
voltages for each oil sample were normalized with the corresponding dry oil value
to account for the variation between different oil samples (60-88 kV). It is
observed that the breakdown voltage remains almost unchanged when the relative
moisture saturation is below 20%. Above that it decreases rapidly [27]. Guidelines
are given for acceptable levels of moisture according to transformer category in
the standard for insulating oils in electrical equipment which defines a minimum
breakdown of 55 kV/mm and a maximum water content of 10 ppm for
transformers rated at a minimum system voltage of 170 kV [1, 26].
Figure 2.2 Measured dependency of the breakdown voltage on the relative

moisture saturation with an approximative model fitted to the data. Each point
corresponds to the average of six breakdown voltage measurements performed
according to the IEC 60156 standard [26, 27]

2.2 CELLULOSE-BASED PRESSBOARD
Cellulose based insulation has been the preferred choice for the solid insulation
in power transformers, In 1939, G. T. Kohan of Bell Telephone Laboratories in
New York wrote an article in AIEE Transactions on cellulose as an insulating
material for various electrical application: paper capacitors, paper-insulated power
cables, oil filled cable [9]. Cellulosic insulation is used primarily in oil filled
transformers from distribution to large power units covering a wide range from 10
kVA to 1500 MVA and from line voltage to 1000 kV. In terms of physical size, it
ranges from the pole and pad mounted units on our streets to large substation units
that can have several tons of cellulosic insulation immersed in 40,000 to 100,000
L (~10,000 to 30,000 gallons) of oil [9]. The insulation structure consists of not
only the HV and LV insulation but also support structures, winding tubes, spacer
blocks and formed items for end closing. These are illustrated in Fig. 2.3.
Electrical grade paper and pressboard are mostly made from wood pulp
processed by the Kraft chemical process. The starting material is wood, both soft
wood and hard wood. Wood is a natural composite material that is made up of
flexible tubes of cellulose bound together by lignin, a brownish aromatic polymer
that is mostly removed during the pulping process. A schematic representation of
the fine structure of wood pulp is shown in Fig. 2.4 [9].
In general, fibres may consist of approximately 50-90% of carbohydrate group,
i.e. cellulose (40-80%) and hemicelluloses (10-40%). Fibres may also consist of
5-25% of lignin [28, 29]. Specially, Hemicellulose molecules are the second
major components that facilitate the hydrogen bonding process, but the
mechanical strength is reduced if their quantity exceeds about 10% [17]. Another
disadvantage of hemicelluloses is their ability to “hold on” to water which makes
the paper/pressboard more difficult to dry out and thus, this signifies its dominant
role in the hydrophilic behavior of the cellulose based solid insulation. Besides the
significant hygroscopic nature of cellulose paper is obvious from the moisture
sorption curves. Meanwhile, lignin is a component of wood that binds to cellulose
fibres to strengthen the cell walls of plants and add further to their matrix rigidity
[1, 17, 29].

Figure 2.3 Formed items from pressboard [12]
Fig 2.4 Micro to sub-micro structure of wood fibers [12]

Cellulose, the essential component of paper and pressboard is a polymer of

glucose units linked to one another in a special manner as shown in Fig. 2.5. It
may be represented simply as [C5 H10 O5 ]𝑛 , ignoring the extra atoms on the end
groups. The number of glucose units (n) which form individual cellulose polymer
chains for new electrical paper and pressboard is typically 1200. The length of the
polymer chain is characterized by the number of links and this is the degree of
polymerization [9, 30].
Figures 2.6 show fibrous, porous structure with non-homogeneous pressboard
surface and numerous protrusions of cellulose fibers [1]. The pores diameter
within the fibre structures of pressboard varies from 10 nm to 7μm. These pores
provide paths for mineral oil and water to penetrate by diffusion and swelling
mechanisms. Besides, the pores within the pressboard allow oil absorption during
impregnation process [1, 31-33]. The ratio of fiber and oil changes as the material
structure changes from a medium of bulk oil-pressboard composite toward the
bulk oil medium. The porosity of pressboard can also result in impurities and
small air bubbles within the oil being drawn into the pressboard. The material
properties and geometry of pressboard thus lead to a complex oil-pressboard
interface [33].
Figure 2.5 Cellulose polymers [12]
(a) Loose structure of cellulose fibres in (b) Numerous protrusions of cellulose

pressboard [1] fibers
Figure 2.6 Microscopic view of pressboard insulation [1]

2.2.1 THE EFFECT OF MOISTURE ON CELLULOSE-

BASED PRESSBOARD
Cellulose based pressboard generally absorb moisture rapidly until reaches

equilibrium according to the ambient conditions. At room temperature (20-25℃)
cellulose can hold from 4 to 10% moisture in the relative humidity range of 30 to
70% typical on factory floors in winter and summer conditions [29]. The moisture
level in insulation in a newly built transformer should be about 0.5%. This makes
it mandatory to dry out scrupulously the insulation in electrical equipment. Wet
insulation is a dielectric hazard in several ways:
 PD inception becomes significant above 3% moisture level and may

result in gas bubbles and release of hydrogen,
 Progressed of paper degradation and aging process.
For regular Kraft paper the life would be lowered by half for every doubling of
moisture content, though for upgraded paper the loss of life is not so drastic. From
a temperature perspective, the temperature increases above 75 ℃ for the
mechanical strength to decrease to half depending on the moisture content of the
paper [29, 34].
Table 2.1 shows the creepage discharge on different pressboard condition in 30
mm gap distance [35]. The results can be deduced that the moisture content in
pressboard plays a more critical role in creepage discharge at the oil-pressboard
interface. On the other hand, moisture in pressboard bulk has been said to play
important role in surface discharges [1]. Therefore, transformer insulation paper
requires rigorous dry out before oil impregnation. Drying of the cellulose is
required to reduce the moisture content, which if not completed can decrease the
dielectric strength and accelerate aging [22].
Table 2.1 Creepage discharge on different pressboard condition [35]

2.3 CREEPAGE DISCHARGE PHENOMENA AT

THE OIL/PRESSBOARD INTERFACE
2.3.1 THE INFLUENCE OF SOLID INSULATION IN

MINERAL OIL
Creepage Discharge phenomena has been reported in the literature that streamer
propagation in mineral oil is greatly affected by the presence of pressboard and
other kinds of solid insulation [1, 22, 36-37, 41, 42]. Especially, different
insulation components or characteristics of oil-solid systems have been shown to
have serious impact on discharge inception and propagation. They are the
orientation of the oil-solid interface with respect to the dominant electric field
direction and the permittivity difference between oil and pressboard as shown in
Fig. 2.7 [22, 42]. And some researchers consider the electric filed distribution
with/without presence of the solid insulation [42]. Figure 2.8 (a) shows the
electric field distribution in the system without the presence of the insulating
surface. The electric field magnitude has a maximum in the direction of the axis of
the streamer. However, as seen in Figure 2.8 (b), when an insulating surface
(relative permittivity 𝜀𝑟 = 3.5) is placed in the gap parallel and close to the axis of
the streamer, the maximum field is no longer along the streamer axis, but points
towards the insulating surface. The maximum field is also larger with the presence
of the insulating surface. The streamer velocity in the presence of the insulator is
also increased moderately compared to no presence of solid insulator [43].
(a) Transformer oil-pressboard system where the (b) Transformer oil-pressboard system where
Interface is oriented parallel to the dominant the interface is oriented perpendicular to the
direction of the electric field. dominant direction of the electric field.
Figure 2.7 General orientations of oil-pressboard interface with respect to
dominant direction of electric field [1, 22]

(a) (b) (c)

Figure 2.8 Vector plots of the electric field distribution for the streamer
propagating (a) in air, (b) along an insulating surface and (c) electric field
distribution when a layer of charge with density σ = 50μC/m2 is applied to the
dielectric surface [43].
Z. D. Wang et al., [42] have reported the influence of discharge propagation

length and PDIV with/without three kinds of solid insulation under AC voltage
using needle-bar electrode configuration in Figure 2.9. As can be seen, in an open
gap, the longest negative streamer is merely 6 mm, when the Perspex surface is
present, the negative streamer channel can be as long as 26 mm, more than four
times that in open gap. These results verify the promotion effect by presence of
solid insulation on discharge.
(a) In open gap
(b) On Perspex
Figure 2.9 Discharge propagation lengths in FR3 at 43 kV [42].

Figure 2.10 shows the breakdown process at the oil-pressboard interface in four
distinct stages. The experiment was conducted by increasing the AC voltage
gradually [1, 28, 44]. As seen in Fig. 2.10, the first event occurs at PDIV level.
During this first stage, no visual activity has been observed but the occurrence of
PD activity is known based on the monitoring data from the PD measurement
system. As the voltage increases, the second stage occurs when a corona like
event is visible at the needle tip. This stage is observed to occur at a voltage level
that is above the PDIV level and lower than the flashover voltage, which is also
called the sustained PD voltage range. Surface discharges can be sustained for
long periods without surface flashover or breakdown as long as the voltage is
within this range. Next, as the voltage is increased closer to the flashover voltage,
the third stage occurs when streamers are observed from the ground bar electrode
and distinct crackling is heard. Finally, as the voltage is increased further, a
surface flashover occurs and causes electrical breakdown [1].
Figure. 2.10 Creepage discharge stages from discharge inception to breakdown [1,
26]

2.3.2 DISCHARGE PROPAGATION MODES
Discharge stopping length, average velocity and discharge shape characteristics

including channel width and number of branches can offer additional information
to help distinguish and understand various modes of streamer. In terms of the
shape of streamer propagation, streamers are generally classified as slow and
“bushy” for streamers emanating from the negative electrode or fast and
“filamentary” for streamers emanating from the positive electrode [45]. The
streamer velocity can vary from tens m/s to hundreds km/s. In terms of the
velocity of streamer propagation, streamers are classified into different modes (1st,
2nd, 3d and 4th). However can be clearly divided, slow mode, fast mode and
combination of fast and slow modes, similar to the classification of 2nd, 3rd and
3th+2nd mode in depending on the electrodes geometry, applied voltage, polarity,
molecular structure and hydrostatic pressure [46].
Z. D. Wang [47] have reported similar observation in Midel 7131 insulation oil
without solid insulation as shown in Fig. 2.11 in which, as the propagation time
increases, the streamer propagates at a constant speed of less than 3km/s for slow
mode (a) ant that has relatively more branches. For fast mode, streamer
propagates much faster at the speed from 5 km/s to 50 km/s and that has bright
filamentary branches. On the other hand, for combination mode, streamer initiates
at fast mode and then quickly changes into slow mode. At the beginning of this
combination mode, only the filamentary branch with few ramified branches
propagates at a high speed and then many small ramified branches start to grow
around the main streamer branch, and the propagation speed falls into the slow
mode propagation range. Figure 2.12 shows the shape evolution of slow streamer
and fast streamer. In this case, the length of captured streamer is scattering due to
the increasing velocity of fast streamers. Therefore unified streamer area, ratio of
area to length (width) is deduced to compare the slow streamer and fast streamer
with fairness. This pseudo width can indicate the branching or expanding of the
streamer shape. It is found that the pseudo width of fast streamer and slow
streamer follows the same relationship with applied voltage: the higher voltage,
the more branches [47].

(a) Propagation length versus propagation time
(b) Photographs of streamer propagation

Figure 2.11 Typical propagation modes in esters (Midel 7131, positive polarity,
d=50 mm) [47]

(a) Ratio of streamer area to length versus applied impulse voltage
(b) Correlated streamer shape at different stages in (a)

Figure 2.12 Evolution of streamer shape along applied voltage (Midel 7131,
negative polarity, d=50 mm) [47]

O. Kesaint and G. Massala [48] have reported the streamer average velocity
with respect to the applied voltage in the transformer oil with and without
pressboard as shown in Fig. 2.13. However, Lundgaard et al., [49] has also
reported that slow velocity streamer propagation could change to a fast mode. The
slow velocity streamer was found to cause weak darkening path on the pressboard
surface, whilst the fast velocity has led to a burnt track on the pressboard surface,
as shown in Fig. 2.14 [1].
Figure 2.13 Average streamer velocity versus voltage in the liquid alone and with
a solid parallel to the field (d=10 cm) [48]

Figure 2.14 Positive streamer propagation along solid interface; Upper: the
positive streamer model, Lower: actual tracks observed on pressboard surface [1]
The polarity effect in insulating oil in inhomogeneous electric fields manifests
differences in discharge initiation, structure and propagation velocity of streamers.
The physicochemical processes implicated in the development of streamers in
each polarity can be succinctly explained as follows [47, 49]:
(1) In the vicinity of the positive electrode, either free electrons or electrons
detached by field emission are accelerated towards the anode. Whereas the
electrons rapidly drift to the anode, the positive ions essentially stand still. The
liquid increases its temperature due to the interaction between the electrons and
the fluid molecules. This local heating causes a low-density region, which
becomes the origin of positive streamer branches. Inside this region, ionization of
fluid molecules generates more free electrons which arrive at the anode. The
remaining positive ions build a space charge representing the new, virtual tip. This
pulsed process repeats periodically and the positive pre discharge propagates in
the direction of the electric field until the inception conditions for subsequent
stages are attained.
(2) The initiation process for negative streamers differs from the positive one.
At sufficiently high fields, electrons are injected into the liquid and due to the
interaction between the electrons and fluid molecules and the associated energy
release, a low density region is formed. Whereas the positive streamer appears as
a three stage process, only two propagation stages of the negative pre discharge
could be observed whereby secondary modes can only be detected at large

electrodes gap. Depending on the growth rate of electric field, different structures
can occur (leafless tree-top, frequently ramified or filamentary structure). Each
mode has a typical minimum propagation velocity.
2.3.3 EFFECT OF RELATIVE PERMITTIVITY IN

COMPOSITE INSULATION SYSTEM
At present, insulation in oil filled transformers consists of mineral insulating oil

and solid insulation that is Kraft paper and pressboard. Cellulose materials have
excellent insulating capability in mineral insulating oil and are low in cost.
However the permittivity of pressboard is over twice that of oil. If this
permittivity were reduced, the insulating distance in transformers can be
decreased because uniform electric field distributions can be made in oil-paper-
pressboard composite insulation systems [50]. Hence, a lot of researchers on
creepage discharge for different material properties and modification of
pressboard immersed in transformer oil [13, 14, 42, 50-52] as well as SF6 gas
condition for reduction permittivity by nanoparticles [42, 53-58]. Dielectric solid
materials such as glasses, plastics and epoxy resin are characterized as having a
compact structure and not being porous. These factors make them different when
compared to the pressboard that is fibrous and porous form of solid insulation [1].
These factors suggest that the type of materials and surface roughness may offer
different results in creepage discharge phenomenon [1].
David Ariza and Claire Pitois et al., [51, 52] have reported the discharge
inception probability, discharge propagation length and charge injected by
conduction currents for negative polarity in mineral oil with different solid
interfaces. The inception voltage for mineral oil without interface is the lowest in
all the considered cases at about 13 kV as shown in Fig. 2.15. The inception
voltage for the impregnated paper (𝜀𝑟 = 3.0), PET (𝜀𝑟 = 3.0) and PVDF (𝜀𝑟 =
8.0) interfaces is nearly the same about 13.3kV despite of the differences in their
relative permittivity and surface properties, whereas a significantly larger
inception voltage is found for the PTFE (𝜀𝑟 = 2.1) interface at about 15.5 kV. The
discharge propagation length of low permittivity solid materials is also shorter
than that of impregnated paper [52]. These results indicate that creepage discharge
at the oil/solid interface depends on solid insulation properties and surface
geometry. In addition, different discharge currents have also been observed.
Hence, when considering the pressboard with a porous, fibrous and non-

homogenous surface structure, one can also hypothesise different results

compared to other solid materials [1].
Figure 2.15 Streamer inception probability distribution function for point cathode
in transformer oil with and without solid interface [51]
Figure 2.16 shows the liquid/solid interface with the protrusions and troughs on
the solid surface that can be modelled in the figure as liquid and solid parts in
series. The interface breakdown should be induced by a partial breakdown of the
liquid part due to its lower dielectric strength, and factor K was introduced to
describe the reduction of the creepage breakdown strength ∆𝑈𝑠 to the breakdown
strength of open gap ∆𝑈𝑂 [42, 59].
(a) Idealized series model (b) Actual condition

Figure 2.16 Equivalent series model to represent solid-liquid interface [59].

∆𝑈𝑠 ∆𝑥𝑠 𝜀𝑜
K= = 1− ∙ (1 − ) (2.1)
∆𝑈𝑂 ∆𝑥0 𝜀𝑠
The limit of K can be deduced as:
𝜀𝑜
lim∆𝑥𝑠 →0 𝐾 = 1 and lim∆𝑥𝑠 →∆𝑥 𝐾 = 𝜀𝑠
𝜀𝑜
Therefore K should be larger than and less than 1 assuming that the liquid
𝜀𝑠
permittivity is smaller than that of solid materials, and it should vary with the
combination of liquid and solid. The larger the permittivity mismatch between
liquid and solid insulation, the more reduction in creepage discharge and
breakdown strength [42, 59]. The supporting experimental evidences can be found
in references [50- 52, 60, 61].
2.3.4 PERMITTIVITY AND POLARIZATION
The relative permittivity of a linear dielectric material describes the ability of

tightly bound electrons in the material to be displaced from the positive nucleus
under an applied electric field to form electric dipoles which results in a net
electronic polarization. The greater the relative permittivity of a material, the
greater the polarization charges that develops in a region where there is a local
imbalance of dipoles, as shown in Fig. 2.17. This polarization charge is a source
of electric field that affects electrodynamics in a system such as a composite
liquid-solid insulation structure. The net polarization charge produces the
polarization field:
𝑃⃗⃗ = (𝜀𝑟 − 1)𝜀0 𝐸⃗⃗ (2.2)
This is dependent on the electric field. A linear combination of the polarization

field and the electric field in the form.
⃗⃗ = 𝜀0 𝐸⃗⃗ + 𝑃⃗⃗
𝐷 (2.3)
= 𝜀𝑟 𝜀0 𝐸⃗⃗ ,

Results in the displacement field, which is a true measure of the total field from
both free and polarization charges that can act upon free charge. Thus, the greater
a material’s relative permittivity 𝜀𝑟 , the greater the total force that acts upon free
charge in a system [17].
(a) (b)
Figure 2.17 (a) The net charge enclosed within a differential sized volume of
dipoles has contributions only from dipoles that are cut by the surfaces. All totally
enclosed dipoles contribute no net charge. (b) Only those dipoles within a distance
𝑑⃗ ∙ 𝑛⃗⃗ of the surface are cut by the volume. [17]


Chapter 3: Electrohydrodynamics
CHAPTER 3
Electrohydrodynamics
3.1 BASIC ELECTRODYNAMIC EQUATIONS
The fundamental equations of the electrodynamic model include Poisson’s

equation and the charge continuity equations. Eqn. 3.1 shows a form of Poisson’s
equation assuming that both a positive and a negative charge carrier exist in the
system. Eqn. 3.2 and 3.3 show the general form of the charge continuity equations
for both positive and negative charge carriers, respectively, assuming drift
dominated charge transport [62].
−∇ ∙ (ε∇V) = 𝜌+ + 𝜌− , where 𝐸⃗⃗ = −∇𝑉 (3.1)
∂𝜌+
+ ∇ ∙ 𝐽⃗+ = 𝐺𝐶+ + 𝑅𝐶+ (3.2)
∂t
∂𝜌−
+ ∇ ∙ 𝐽⃗− = 𝐺𝐶− + 𝑅𝐶− (3.3)
∂t
In Poisson’s equation, V represents the electric potential, 𝜀 represents the

permittivity of the liquid, 𝐸⃗⃗ is the electric field and 𝜌+ and 𝜌− represent the
positive and negative charge densities, respectively. In the charge continuity
equations, 𝐽⃗+ = 𝜌+ 𝜇+ 𝐸⃗⃗ and 𝐽⃗− = −𝜌− 𝜇− 𝐸⃗⃗ represent the positive and negative
drift current densities, where 𝜇+ and 𝜇− are the positive and negative charge
mobility values in the liquid respectively, and 𝐺𝐶+ , 𝐺𝐶− , 𝑅𝐶+ and 𝑅𝐶− represent
the generation and recombination terms for the positive and negative charge
carriers. Mathematically, Poisson’s equation describes the electric field
distribution in an electrode system in the presence of free charge carriers, while
the charge continuity equations describe how the distributions of free charge
carriers are modified by the electric field. Therefore, these electrodynamic
equations are coupled and must be self consistently solved. The electrodynamics
model taking place in transformer oil or any other dielectric liquid during periods

of significant electrical stress, and then the next step in the process is the
identification of sources and sinks of free charge carriers. The sources and sinks
of free charge carriers can be broadly categorized as either being bulk effects or
boundary effects. The injection of electrons from an electrode into the liquid via
Fowler-Nordheim tunneling is an example of a boundary effect and such a source
of charge would be accounted for by the setting of boundary and initial conditions
for the charge continuity equations. The electric field dependent dissociation of
weakly bonded neutral ion pairs in transformer oil under high electric field
stresses is an example of a bulk charge generation effect and is accounted for by
the generation terms of the charge continuity equations. The recombination of
positive and negative ions to form neutral species is accounted for by the
recombination terms of Eqn. 3.2 and 3.3.
3.2 CHARGE CARRIER GENERATION
3.2.1 CHARGE INJECTION BY ELECTRIC FIELD

EMISSION
In generally, electrons are released from the metal electrode surface due to
high electrostatic fields over 1~10× 108 V/m as shown in Fig. 3.1. A theory to
describe field emission was first developed by Fowler and Nordheim in 1928,
hence the process of field emission is often referred to as Fowler-Nordheim field
emission, of Fowler-Nordheim charge injection [62, 63]. Fowler and Nordheim’s
work involved the development of an equation to describe the electric field
dependent current density in a vacuum, due to the quantum-mechanical tunneling
of electrons from the metal through the potential barrier at the metal/vacuum
interface. The level of electron tunneling, which takes place increases
exponentially with the electric field level. This is due to the fact that as the field
level increases the potential barrier at the metal/vacuum interface narrows, leading
to an exponential increase in the probability of electrons tunneling through the
barrier. The electron current was measured by Fowler and Nordheim. Electrons in
the proximity of the Fermi level may have the probability of passing through the
barrier – a phenomenon known as the “tunnel effect”. The basic equation for the
emission current by the influence of electric field can be expressed as follows:

3
⃗⃗⃗⃗⃗⃗2
𝑒 3 |𝐸| 8𝜋√2𝑚∅2
𝐽FE = exp (− ) (3.4)
8𝜋ℎ∅ ⃗⃗⃗⃗⃗⃗
3ℎ𝑒|𝐸|
Where, 𝐽FE is the emitted current density, e is the magnitude of electron charge,
⃗⃗⃗⃗⃗⃗ is the electric field, h is Planck’s constant and ∅ is 4.5 eV. This equation is
|𝐸|
based on the potential energy distribution at the metal/vacuum interface. Figure
3.2 shows the mechanism of field emission from the metal surface, W(x) is
potential barrier with applied field: 𝑊(𝑥) = −𝑒|𝐸⃗⃗ |𝑥, 𝑥 > 0. This potential
energy distribution is known as the “Triangular Barrier Approximation”, which
overestimates the potential barrier height at the metal/vacuum interface. This
overestimation arises from the fact that the “Triangular Barrier Approximation”
does not take into account the effects of image force barrier lowering caused by
the presence of emitted charge in the vacuum. The one dimensional potential
energy distribution, W(x), external to the metal/vacuum interface taking into
account the image force barrier lowering effects has the following form: 𝑊(𝑥) =
𝑒2
−𝑒|𝐸⃗⃗ |𝑥 − (16𝜋𝜀𝑥) , x > 0. The shape of the potential energy distribution near the
metal/vacuum interface, when image force effects are taken into account is
illustrated by the lower trance in Fig. 3.2.
Figure 3.1 Free electron emission from ground electrode under high electric field
1~10× 108 V/m

Figure 3.2 Mechanism of field emission from the metal surface [64].
To derive an expression for the emission current let us consider an electron as

it leaves the surface in the direction x as shown in Fig. 3.2. Electric field can be
approximated as that between a point charge and the equipotential planar surface.
The field lines here are identical to those existing when an image charge of +e is
thought to exist at a normal distance of –x on the other side of the equipotential
metal surface. Applying Coulomb’s law, the force on the electron in the x-
direction is given by follows:
−𝑒 2 −𝑒 2
𝐹(𝑥) = = (3.5)
4𝜋𝜀0 (2𝑥)2 16𝜋𝜀0 𝑥 2
The potential energy at any distance x is obtained by integrating the above

equation from ∞ to x.
−𝑒 2
𝑊𝑒1 = (3.6)
16𝜋𝜀0 𝑥
which gives a parabola shown by curve 1 of Fig. 3.2. The effect of the
accelerating external field when applied at right angles to the cathode surface
gives the electron a potential energy

𝑊E = −𝑒𝐸𝑥 (3.7)
which is a straight line shown by Fig. 3.1 (red line). The total energy is then
−𝑒 2
W = 𝑊a + 𝑊E = − ( ) − 𝑒𝐸𝑥 (3.8)
16𝜋𝜀0 𝑥
which shown the resultant blue line (Fig. 3.2). Thus a marked reduction ∆𝑊 in the
potential barrier is obtained. The maximum reduction at 𝑥𝑚 is obtained by
differentiating equation (3.8) or
𝑑𝑊 𝑒2
= 2
− 𝑒𝐸 = 0
𝑑𝑥 16𝜋𝜀0 𝑥m
𝑒
𝑥𝑚 = √
16𝜋𝜀0 𝐸
Inserting this value into Eqn. (3.8) the lowering in the work function becomes
𝑒𝐸
∆W = −e√ (3.9)
4𝜋𝜀0
Hence, the effective value of the work function is
𝑒𝐸
𝑊𝑒𝑓𝑓 = 𝑊a − √ (3.10)
4𝜋𝜀0
and the saturation current due to electron emission using Eqn. (3.14) in the
presence of field E becomes
𝑒 𝑒𝐸
𝐽s = 𝐴𝑇 2 𝑒𝑥𝑝 [− (𝑊a − √ )] (3.11)
𝑘𝑇 4𝜋𝜀0

which is known as the Schottky’s equation. If the current density in the absence of
external field is 𝐽0 (Eqn. (3.14)) then rearranging (3.11)
𝑒 𝑒𝐸
𝐽𝑠 = 𝐽0 𝑒𝑥𝑝 [ (−√ )] (3.12)
𝑘𝑇 4𝜋𝜀0
To obtain emission current J significantly higher than 𝐽0 , E must be of the

magnitude of 10 MV/cm or higher. In practice a significant field emission current
may be observed at lower fields. A few electrons in a metal will have energy
slightly above the Fermi level and thus will have a greater probability to penetrate
the potential barrier “tunnel effect”.
3.2.2 CHARGE INJECTION BY THERMAL ELECTRON

EMISSION
In metals at room temperature the conduction electrons will not have sufficient
thermal energy to leave the surface. The emission current is related to the
temperature of the emitter by the Richardson relation for thermionically emitted
saturation current density:
4𝜋𝑚𝑒𝑘 2 2 𝑊𝑎
𝐽TH = 3
𝑇 𝑒𝑥𝑝 [− ] (3.13)
ℎ 𝑘𝑇
where e and m are the electronic charge and mass respectively, h is Planck’s
constant, k Boltzmann’s constant, T the absolute temperature and Wa the surface
4𝜋𝑚𝑒𝑘 2
work function. A is Richardson constant (𝐴 = ).
ℎ3
the above expression becomes
𝑊𝑎
𝐽TH = A𝑇 2 𝑒𝑥𝑝 [− ] (3.14)
𝑘𝑇

Figure 3.3 Schematic diagram of thermionic emission from the metal surface [64]
3.2.3 MOLECULAR IONIZATION BY ELECTRIC FIELD
Electric field dependent molecular ionization, also known as field ionization,

is a direct ionization mechanism, where an extremely high electric field level
results in the elevation of a valence band electron in a neutral molecule to the
conduction band, thus generating both a free electron and positive ion. This differs
from electric field dependent ionic dissociation where the dissociation of a neutral
molecule produces a free negative ion and positive ion, which are both relatively
large and immobile compared to electron. For field ionization, the asymmetry in
the generated carriers mobility values leads to the formation of significant space
charge densities in electrically stressed dielectric liquids within the timescales
associated with streamer initiation and growth [16]. This process is illustrated
graphically in figures 3.4 (a) and (b)

(a) Case of low electric field (b) Case of high electric field
Figure 3.4 Molecular ionization by electric field [64].
In the case of low external electric field (Fig. 3.4 (a)), when the combined
force between electron and positive ion is stronger than that of external applied
force, the charges are chemically bonded stably in the insulating oil. Therefore,
there are a lot of neutral molecules.
Whereas, in high external electric field (Fig. 3.4 (b)) where collisions occur
between neutral molecules and free electrons accelerated in the high electric field.
The collisions energize a valence electron in the neutral molecule promoting it to
the conduction band and generating a positive ion and another free electron [17]
The development of a model that describes the process of field ionization in
dielectric liquids, such as transformer oil, is challenging due to the lack of a
comprehensive liquid-state theory. Consequently, the literature contains very few
publications that propose liquid phase field ionization models. The models that do
exist are based on Zener’s theory of electron tunneling in solids [17, 65-69].
In their ground breaking work, Devins, Rzad and Schwabe [68] applied the
Zener model to dielectric liquids to explain discharge propagation. In this work,
the field ionization charge density generation rate source term is
𝑞 2 𝑛0 𝑎|𝐸⃗⃗ | 𝜋 2 𝑚∗ 𝑎∆2
𝐺𝐹 (|𝐸⃗⃗ |) = 𝑒𝑥𝑝 (− ) (3.15)
ℎ 𝑞ℎ2 |𝐸⃗⃗ |
a is the molecular separation distance, h is Planck’s constant, 𝑚∗ is the effective

electron mass, 𝑛0 is the number density of ionizable species, and ∆ is the
ionization potential. The material parameters for the transformer oil used in Eqn.
(3.15) are shown in Table 3.1. The material parameters should be changed

according to the purity and chemical composition of insulating oil, and the
material information of pure transformer oil used in this paper has been proved by
J. G. Hwang [17, 38, 40]. To apply the ionization ratio by the external electric
field to the insulation oil, it is expressed as Eqn. (3.15) by multiplying the
ionizable density of electrons and molecules.
𝑒 2 𝑛0 𝑎|𝐸⃗⃗ | 𝜋 2 𝑚∗ 𝑎∆2
𝐺𝐼 (|𝐸⃗⃗ |) = 𝑒 2 𝑛0 𝛾(|𝐸⃗⃗ |) = 𝑒𝑥𝑝 (− ) (3.16)
ℎ 𝑒ℎ2 |𝐸⃗⃗ |
𝑛0 is the number density of ionizable molecules. Because commonly used

dielectric liquids such as transformer oil are very complex liquids there will be a
number of individual molecular ionization source terms in play at any one time,
with the dominant one being dependent upon the level of the electric field stress.
In order to very accurately develop a molecular ionization model for such liquids
it is necessary to know the concentrations and ionization potentials of the
individual molecules which make up the liquids. This type of information is
usually unavailable, particularly for commercially used liquids like transformer oil
and so for simulation purposes a reasonable aggregate source term is used [62].
Table. 3.1 Characteristic values of pure transformer oil [1, 17, 38, 67-84]
Parameter Symbol Value Reference
Permittivity of transformer oil 𝜀oil 2.2
Positive ion mobility 𝜇p 5 × 10 𝑚2 𝑠 −1 𝑉 −1
−9 [79]
Negative ion mobility 𝜇n 5 × 10−9 𝑚2 𝑠 −1 𝑉 −1 [79]
Electron mobility 𝜇e 1 × 10−3 𝑚2 𝑠 −1 𝑉 −1 [67, 71, 72]
Ion-ion recombination 𝑅pn 9.908× [73, 83, 84]
10 𝑚3 𝑠 −1 𝑚𝑜𝑙 −1
6
Ion-electron recombination 𝑅pe 9.908× [70, 84]

10 𝑚3 𝑠 −1 𝑚𝑜𝑙 −1
6
Electron attachment time 𝜏a 1.55 × 10−7 𝑠 [70, 84]

Density of ionisable species 𝑁0 0.16603 mol ∙ 𝑚−3
Molecular separation distance 𝑎 3 × 10−10 𝑚
Effective electron mass 𝑚∗ 9.11× 10−32 kg
Molecular ionization energy ∆ 5.12 eV
Specific heat capacity of oil 𝐶p 1870 J ∙ 𝑘𝑔−1 ∙ 𝐾 −1
Thermal conductivity of oil 𝑘T 0.13 W ∙ 𝑚−1 ∙ 𝐾 −1
Mass density of oil 𝜌oil 880 kg ∙ 𝑚−3

3.2.4 DISSOCIATION OF ION BY ELECTRIC FIELD AND

TEMPERATURE
Onsager developed a theory which describes how the conductivity of a weak

electrolyte is dependent upon the level of electric field stress to which the liquid is
subjected [70]. In this theory, highly purified state of transformer oil contains
impurities that can undergo chemical reaction leading to space charge generation.
The neutral molecule of the impurity can dissociate into positive and negative ions
and then free ions can generate electric conductivity by dissociation of oils, and
charges having different polarities are recombined into ion pairs as follows [17,
70-79].
(3.17)
Equation (3.17) describes that 𝐴+ 𝐵 − is neutral ion pairs, 𝐴+ and 𝐵 − are

positive ion and negative ion according to dissociation rate constant, 𝐾𝐷 and
recombination rate constant, 𝐾𝑅 , respectively.
The rate at which neutral ion-pairs dissociate to form free ions and the rate at
which free ions recombine to form neutral ion-pairs is controlled by two kinetic
rate constants, KD for dissociation and KR for recombination. According to
Onsager’s theory, the value of the dissociation constant KD, is dependent upon the
applied electric field, while the recombination constant KR, is electric field
invariant. The electric field dependence of the dissociation constant results in an
increase in the concentration of free ions, and thus the conductivity of the liquid
increases as the electric field increases [62].
With the presence of external electric field, the dissociation rate constant 𝐾𝐷
has the following form:
𝐾𝐷 = 𝐾𝐷0 𝑓(|𝐸⃗⃗ |, 𝑇) (3.18)
Where 𝐾𝑑0 is the zero field dissociation constant (𝑠 −1 ) and 𝑓(|𝐸⃗⃗ |) is the electric
field dependent function, which relates the increase in the dissociation rate to the
electric field level. The electric field dependent function is defined as:
𝐼1 (4𝑏)
𝑓(|𝐸⃗⃗ |) = (3.19)
2𝑏

Where,
𝑞 3 |𝐸⃗⃗ |
b=√ (3.20)
16𝜋𝜀0 𝜀𝑟 𝑘 2 𝑇 2
where, 𝐼1 is modified Bessel function of the first kind, k is the Boltzmann’s

constant and T is the absolute temperature (K). Onsager’s theory of electric field
enhanced ionic dissociation has been used by several authors to describe electrical
conduction in non-polar liquids, particularly transformer oil which is often
classified as a weak electrolyte due to the unavoidable presence of non-
hydrocarbon based impurities in the oil [62, 80-82]. In thermal equilibrium and
when not being stressed by an applied electric field, the following relation
describes the relative concentrations of neutral ion-pairs and free ions, or free
charge carriers in transformer oil:
c𝐾𝐷0 = 𝑛±0
2
𝐾𝑅 (3.21)
where c is the density of neutral ion pairs, 𝑛± 0 is the equilibrium free ion
concentration in the absence of the electric field in the transformer oil. The
recombination rate constant, 𝐾𝑟 is related to the permittivity and mobility by the
Langevin relation as the following [83]:
𝑒0
𝐾𝑅 = (𝜇 + 𝜇− ) (3.22)
𝜀 +
where 𝜇+ and 𝜇− are the mobility of free positive ion and negative ion in
transformer oil, respectively. In order to develop insight into whether or not
electric field enhanced ionic dissociation plays a significant role in the initiation
and growth of streamers in transformer oil, it is necessary to characterize the
parameter cKD, as this term represents the free charge carrier source term in the
transformer oil. This characterization can be accomplished by using the low field
conductivity value of transformer oil. The conductivity of transformer oil is
measured to verify the effect of ion dissociation during discharge inception and
propagation in insulation oil. The initial concentration of ionic species can be
obtained from the medium’s conductivity using the following expression [83]:
σ = e(𝜇+ 𝑛+ 0 − 𝜇− 𝑛− 0 ) (3.23)

where, 𝑛+ 0, 𝑛− 0 are the equilibrium free ion concentration in the absence of an

electric field and oil concentration. 𝜎 is conductivity of insulation oil in the
absence condition of electric field. Assuming that negative ion and positive ion
are at the same energy level, the free ion concentration in the insulation oil
without the influence of electric field condition as following:
𝜎
𝑛± 0 = (3.24)
𝑒(𝜇+ + 𝜇− )
The concentration of the free ions in the absence of the electric field is substituted
into the Eqn. (3.18) as following:
𝜎 2
c𝐾𝐷0 = ( ) 𝐾𝑅 (3.25)
𝑒(𝜇+ + 𝜇− )
The complete charge generation mechanism on ion dissociation by the electric

field is given by equation (3.23) that is combined with Eqns. (3.25) and (3.18).
𝜎 2
c𝐾𝐷 = ( ) 𝐾𝑅 × 𝑓(|𝐸⃗⃗ |, 𝑇) (3.26)
𝑒(𝜇+ + 𝜇− )
The electric field dependent ionic dissociation process, which is described above
is a bulk charge generation mechanism, meaning that when modeling the process,
the term cKD represents the generation term in the charge continuity equations for
both the positive and negative ions. This is contrast to the case of field emission or
Fowler-Nordheim charge injection. Forwler-Nordheim charge injection is a
boundary process and so no bulk generation term exists in the continuity equation
for the injected electron.
(a) Without external electric field (a) With external electric field
Figure 3.5 Ionic dissociation by electric field [amended from 64]

When there is no external electric field (Fig. 3.5 (a)), the positive ion and the
negative ion are in a stable state; there are many neutral ion pairs. Whereas, when
the external energy becomes larger than the bonding energy of neutral ion pairs,
they are detached from each other as shown in Fig. 3.5 (b). On the other hand, the
increase of free ions caused by dissociation increases the conductivity of the fluid
[64].
3.3 CARRIER RECOMBINATION
During the formation and growth of a streamer there will exist concentrations
of free positive ions and electrons in the dielectric liquid. These free charge
carriers are produced due to molecular ionization as discussed in the previous
section. The interaction of these free charge carriers with each other and the
surrounding media can be assumed to open up the possibility of electron-ion
recombination, electron attachment to neutral species, forming negative ions and
ion-ion recombination as shown in Fig. 3.6. When developing an electrodynamic
model for streamer initiation and growth in dielectric liquids it is important to
include, when relevant such processes [62].
By transferring the electron, recombination occurs between positively and
negatively charged particles. A photon may be released from this process [83].
The Langevin-Debye relationship is used to model positive ion-negative ion 𝑅𝑝𝑛
and positive ion-free electron 𝑅𝑝𝑒 recombination rates in the transformer oil (thus
reducing the number of electrons and increasing the number of negative ions in
the liquid dielectric) [1, 17 38, 84-86]. According to the Langevin-Debye
relationship, the recombination rates can be expressed as
𝑞(𝜇𝑝 + 𝜇𝑛 )
𝑅𝑝𝑛 = (3.27)
𝜀
𝑞(𝜇𝑝 + 𝜇𝑒 )
𝑅𝑝𝑒 = (3.28)
𝜀
The ion-electron recombination rate of Eqn. (3.28) is overestimated because

the Langevin-Debye relationship is diffusion limited and valid for situations
where the electric field levels are low to moderate and the recombining species are
of similar physical scale [17, 85].

Figure 3.6 Recombination process in mineral oil
3.4 ELECTRON ATTACHMENT
In addition to recombination, electrons also combine with neutral molecules

to form negative ions. This process is modeled as an electron attachment time
constant. In very pure hydrocarbon liquids, energetic electrons have attenuation
lengths on the order of several centimeters, meaning that an electron can be
expected to travel for several centimeters in the liquid from the point where it was
formed before it attaches to a neutral molecule to form a negative ion [85]. In
commercial dielectric liquids, the attenuation length will be shorter due to higher
levels of impurities in the liquid [17, 87]. The attachment time constant is simply
the quotient of the electron attenuation length and the electron velocity. A
representative set of numbers for such a calculation are: An electron attenuation
length 𝜆𝑎 of 1 mm, an electron mobility 𝜇𝑒 of 1 × 10−4 m2 V/s and an electric
field strength |𝐸⃗⃗ |, of 5 × 107 V/m. These numbers give an attachment time 𝜏𝑎 of:
𝜆𝑎 1 × 10−3
𝜏𝑎 = = 3
= 2 × 10−7 (s) (3.29)
⃗⃗
𝜇𝑒 |𝐸 | 5 × 10
This value for 𝜏𝑎 corresponds well with the attachment time constants used by
other authors in the literature [17, 84-87]. Although attachment undoubtedly takes
place during streamer growth in dielectric liquids, its impact on the overall
streamer propagation process is reasonably small due to the fact that attachment
processes take place on a longer time scale than important dynamics such as the
separation of positive ions and electrons in the ionization zone at a streamer’s tip
[87].

3.5 ION AND ELECTRON MOBILITY VALUES
During the course of the research being presented in this thesis, the ion and
electron mobility values in transformer oil 𝜇± and 𝜇𝑒 , respectively, are used as
input parameters for all electrodynamic simulations. A mobility values of
1 × 10−9 m2 /V ∙ s is used for the positive and negative ions in all simulations.
This value has been verified experimentally for transformer oil in [82]. It is also in
agreement with the theoretical analysis presented in [88]. A mobility value of
1 × 10−4 m2 /V ∙ s is used electrons in all simulations. This value corresponds
with the data presented in [86] and the electron mobility model described in [89].
3.6 SUMMARY
This chapter discussed the electrodynamic processes associated with the

generation and recombination of free charge carriers in transformer oil when the
oil is electrically over stressed. Three charge injection and charge generation
mechanisms, which are commonly suggested as playing a role in streamer
occurrence, development and initial electric field distribution. The dissociation
process of dielectric liquids is explained by Onsager theory that highly purified
state of transformer oil contains impurities and then mineral oil can undergo
chemical reaction leading to space charge generation. The neutral molecule of the
impurity can dissociate into positive and negative ions and then free ions can be
generated by dissociation of oils.
Sections 3.3 and 3.4 discussed the charge recombination and attachment
processes that can take place in transformer oil. The diffusion based Langevin
recombination process for the recombination of positive and negative charge
carriers was described. When developing an electrodynamic model for streamer
initiation and growth in dielectric liquids, it is important process.
The electrohydrodynamics model is directly relevant to numerical simulation
model development that has been undertaken as Chapter 3.


Chapter 4: Creepage Discharge Characteristics at Oil/pressboard Interface by
Nanocomposite Coating Method
CHAPTER 4
Creepage Discharge Characteristics at

Oil/pressboard Interface by Nanocomposite
Coating Method
4.1 INTRODUCTION
This chapter deals with experimental characterization of creepage discharge to

improve the electrical characteristics of oil/pressboard (PB) composite insulation
system by solid layer coating method such as epoxy resin, Teflon and silica/epoxy
nanocomposite on PB surface under positive standard lightning impulse voltage.
Electrode arrangement used in experiment is a rod-plane and needle-bar electrode
arrangement in transformer mineral oil. As well known, the PB surface conditions
influence partial discharge inception voltage (PDIV) and creepage discharge
properties.
Lightning impulse voltage withstanding property is one of important
characteristics in power apparatus. In addition, engineers want to predict the
discharge initiation, development and propagation in composite insulation system
under lightning impulse for upgrading the insulation design of power apparatus.
For these reasons, a lot of researchers have recently been devoted to improve the
dielectric properties of insulating oil by adding specific functional additives into
the oil. However, few studies have been focused on modifications of the
insulating paper in order to improve insulating ability.
From the above view point, creepage discharge properties of conventional PB
are compared with solid layer coated PB, in terms of discharge propagation length
( 𝑙d ) and discharge occurrence probability ( 𝑃d ). As a result, it is found that
discharge occurrence probability of solid layer coated PB increases at all applied
impulse voltage levels depending on coated material and silica filler loading (𝑆l )
compared with that of conventional PB. The result is interpreted in terms of
dielectric property effect, surface modification effect, electron suppression effect
by silica/epoxy nanocomposite coating on PB surface.

4.2 EXPERIMENT SETUP
Figures 4.1 (a) and (b) show experimental setup as type 1 side view and top
view of electrode system, respectively. The diameter of the rod electrode and
plane electrode was 6 and 70 mm, respectively. The rod electrode is placed so that
the tip touches the pressboard surface. The position of pressboard and electrode
configuration in the experimental system determines the electric field distribution
at the oil/pressboard interface. On the other hand, a more directional and intense
electric field is created towards the solid surface from localized and
inhomogeneous electric field at the edge of the rod electrode compared to other
methods. But, the creepage discharge do not always propagate radially in this
electrode configuration. For this reason, it is difficult to evaluate the discharge
propagation length.
(a) Side view (b) Top view

Figure 4.1 Rod-plane electrode configuration (type 1)
In the case of needle-bar electrode configuration as type 2 (Figs. 4.2 (a) and
(b)), the tip radius of the needle electrode was 10 μm and gap distance 50 mm to
the bar electrode. To generate the creepage discharge in one direction, we used
guide electrode beneath a PB. The needle electrode was placed at an angle of 30
degree from the surface of a specimen with the needle tip touching the pressboard
surface. These above two methods have been widely used to study surface
discharge phenomena.
(a) Top view (b) Side view

Figure 4.2 Needle-bar electrode configuration (type 2)

Figures 4.3 (a) to (c) show microscope image of needle electrode before and
after impulse experiment and after breakdown. As can be seen in Fig. 4.3 (b), the
needle electrode tip is not changed after 400 times impulse voltage, whereas
scratches appear nearby needle tip, and the number of scratches and size increased
with increasing the number of the applied voltage voltages (average scratch size:
20 μm). But the results revealed that these kinds of scratches nearby the needle tip
didn’t affect the discharge inception voltage. On the other hand, in Fig. 4.3 (c), the
needle tip is mutated after breakdown. The radius of curvature expanded by times
3, affecting the discharge inception voltage. Thus, when the experiment used the
needle electrode, it is very important to check the needle electrode tip after
experiment. In this work, the needle electrode was checked after every experiment
using a microscope to acquire reliable data.
Figure 4.3 Needle electrode (a) before experiment, (b) after experiment and (c)
after breakdown

4.3 EXPERIMENTAL PROCEDURE
4.3.1 CREEPAGE DISCHARGE OBSERVATION
Figures 4.4 and 4.5 (a) to (d) show schematic diagram and photograph of the
experimental measuring system. Vacuuming was made in the chamber after the
test sample was immersed in the insulating oil for three hours to remove moisture
and gasses from the insulating oil and fill oil molecules into the PB structure.
Discharge image was observed using a digital camera (Fujifilm Finepix S100fs)
and usual digital camera equipped with an image intensifier (Hamamatsu C9016-
2x series-23). Discharge light emission was also detected with a photomultiplier
tube (PMT, Hamamatsu Photonics, H6780-20) to determine PDIV. A 50 Ω
resistor was used to obtain the discharge current signal.
A positive standard lightning impulse voltage was applied to the electrode
varied from 60 to 100 kV by a step with increment of 20 kV (type 1) and from 20
to 45 kV by a step with increment of 5 kV. The interval time between each
voltage level was one minute. The impulse voltage application was repeated 10
times (type 1) and 5 times (type 2) at each voltage level, respectively.
Figure 4.4 Schematic diagram of the experimental setup and measuring system

(a) Oil filled chamber
(b) Oil storage tank and two oil filter

(c) Marx impulse generator and voltage divider
(d) CCD camera, two image intensifier, streak camera, two PMT and three optical
lenses
Figure 4.5 Photograph of the experimental system

(a) FINEPIX S100FS
(b) C9016-2X
Figure 4.6 FINEPIX S100FS and image intensifier (I.I)

Table 4.1 Specification of the C9016-2X
Figure 4.7 Photo multiplier (PMT) Figure 4.8 Power supply

Figure 4.9 Dimensional outlines of PMT
4.3.2 SAMPLE PREPARATION&DIELECTRIC

PROPERTY MEASUREMENT
Five kinds of PB samples were used; namely, neat pressboard (NP), dry
pressboard (DP), epoxy resin coated pressboard (EP-A and EP-F), Teflon coated
pressboard (TP) and pressboard coated by silica/epoxy nanocomposite (SP). Three
samples of each specimen were prepared for reproducibility on experiment results.
Figure 4.10 shows the schematic diagram to prepare for surface coating on the
pressboard surface. In preparing nano silica/epoxy composite, nanoparticles were
dispersed in the base epoxy bisphenol A of F type at the proportion of each weight
percent using a vacuum mixer. The mass ration of epoxy resin to curing agent is
30:10. EP-A, EP-F and SP specimens were prepared by thinly coating PB surface
with bisphenol A type epoxy resin (EP-A), bisphenol F type epoxy resin (EP-F),

silica/bisphenol F type epoxy resin SP, respectively, and then put in a vacuum
oven for 30 minute and then being dried using a dry oven for 3 hours at each 70
and 150℃. DP specimen was also dried under the same condition as EP, TP and
SP specimen. DP is a reference specimen in this research work. SP specimens
with different SiO2 nano filler contents (0, 0.2, 0.5, 1, 3, 5 and 50wt%) were
prepared as shown in Table 4.2. Figures 4.11 (a) and (b) show pressboard samples
used in experiment before and after epoxy coating on the one side surface of PB.
Figure 4.10 Schematic diagrams to prepare for nano silica and epoxy/silica
coating on surface of pressboard
Table 4.2 Properties of each specimen

(a) Dry pressboard (DP) (b) Epoxy resin coated PB (EP-F)

Figure 4.11 pressboard samples used in experiment
Figure 4.12 shows the weight and thickness variation of each sample before
and after drying process. We confirmed that the average decreasing rate of dried
specimen as DP, EP-A, EP-F and SP of weight, thickness and of each specimen
decreased by 5.3% and 3%, respectively.
Dielectric properties of each specimen were measured with an impedance
analyzer (IWATSU Impedance Analysis PSM3750) at room temperature and
frequency from 50 Hz to 2 MHz after the drying process. The number of test
samples of each specimen was two. The size and thickness of a sample was 70
mm by 70 mm and 850 μm, respectively. Prior to measurement, gold electrodes
with diameter of 50 mm were deposited onto both sides of a sample by Magnetron
sputter (MSP-1S) with a thickness of 3 Å for 3 minute to ensure the
reproducibility of measurement and improve the contact area between the
electrode and the surface of PB. The gold sputtering thickness on the PB surface is
54 nm. The dielectric properties were measured after oil immersion process for 3
hours. Note that measurements of dielectric properties were conducted two times
each for two samples with error of 3 %.

Figure 4.12 Weight and thickness variation of each sample before and after drying
process
4.3.3 SURFACE ROUGHNESS AND NANO SILICA

DISTRIBUTION STATE
Figure 4.13 shows the surface roughness before and after epoxy resin coating.
The result represents an assembly of surface irregularities and constitutes main

surface quality characteristics of each specimen. To characterize the sample

surface roughness, the following parameters were used; Ra average surface
roughness; and Rz 10 point mean surface roughness. Conventional pressboard
have porous, fibrous, protrusions and non-homogenous surface structure. Whereas
epoxy resin coated pressboard is smoother than that of conventional PB after new
coating method on PB. Figure 4.14 shows Ra and Rz of surface roughness of each
type of sample. The surface of NP and DP samples is rougher than that of all the
other kinds of the solid layer coating samples (EP-A, EP-F, TP and SP). Therefore,
it is found that the solid layer coating provides smoother surface on PB than NP
and DP specimen.
Figure 4.15 shown typical SEM image of silica/epoxy nanocomposite. The
average thickness of silica/epoxy nanocomposite coating of each sample was 5.4
to 11.9 μm . Note that the thickness of silica/epoxy coating increased with
increasing nano particle filler contents. Figure 4.16 shows nano silica dispersion
state in base epoxy matrix. As can be seen in Fig. 4.8, low 𝑆l from 0.1 wt% to 5
wt% in the base epoxy matrix exhibit good dispersion of nanoparticles. On the
other hand, the nano silica of 50 wt% 𝑆𝑙 has formed micro size agglomeration
structure in base epoxy resin. In this research, nano silica particles were supplied
from Nissan chemical industries. LTD.
Figure 4.13 Surface roughness before and after epoxy resin coating (a) DP and (b)
EP-A

Figure 4.14 Surface roughness of each type of sample measured using a surface
roughness tester (bottom: Ra, top: Rz)
Figure 4.15 SEM image of the epoxy/silica nanocomposite thickness (Cross

section view), (a) 0.2wt%, (b) 0.5wt%, (c) 1wt%, (d) 3wt%, (e) 5wt% and (f)
50wt%

Figure 4.16 SEM image of SiO2 nanoparticles surrounded by red circle dispersion
in base epoxy matrix (Cross section view) (a) 0.2wt%, (b) 0.5wt%, (c) 1wt%, (d)
3wt%, (e) 5wt% and (f) 50wt%
4.3.4 PARTIAL DISCHARGE INCEPTION PROPERTIES
Figures 4.17 (a) and (b) show examples of discharge images and typical
waveforms of positive standard lightning impulse voltage and those detected with
PMT without and with creepage discharge, respectively. We determined the

partial discharge inception using PMT, image intensifier and discharge current
waveform. In general, the luminous sensitivity of PMT (500 [μA/lm]) is higher
than that of the image intensifier (230 [μA/lm]) as shown in Table. 4.3. Besides
we found that the image intensifier can’t catch the light intensity from creepage
discharge below 50 pC as shown in Figure 4.18. For this reason, we found that
sometime the image intensifier did not perceive the discharge light in spite of
PMTs perceived at low apply voltage. Thus, PMT is more reliable tool to decide
PDIV that is determined when a waveform is detected from PMT instead of the
image intensified camera.
Figures 4.13 (a) and (b) shows general waveforms with/without creepage
discharge. In the case of no PD in Fig. 4.17 (a), there is no signal PMT, discharge
current and image intensifier image. On the other hand, when the creepage
discharge occurred, discharge light emission can be detected clearly by PMTs,
current waveform and image intensifier as shown in Fig. 4.17(b).
(a) (b)
Figure 4.17 Determination of creepage discharge inception voltage (a) without
creepage discharge inception at 25 kV and (b) with creepage discharge inception
at 55 kV

Table 4.3 Specification of PMT and image intensifier [90, 91]
Figure 4.18 Relation between the output of image intensified camera and charge
amount in pC as a function of lightning impulse voltage

4.4 RESULTS (ROD-PALNE ELECTRODE)
4.4.1 DIELECTRIC PROPERTIES
Figures 4.19 (a) and (b) show the real frequency dependence of real part 𝜀r′ and
imaginary part 𝜀r′′ of the complex relative permittivity for different nano silica
filler loading 𝑆l at room temperature. It can be seen in Figures 4.19 (a) that 𝜀r′
decreases stably with increasing frequency from 50 Hz to 2 MHz, whereas the
relative dielectric loss 𝜀r′′ increases with increasing as a function of frequency in
Fig. 4.19 (b). As can be seen in Fig. 4.20, 𝜀r′ of silica/epoxy nanocomposite coated
samples at 100 kHz decreases at 100 kHz corresponding to dominant frequency
component that the standard lightning impulse voltage contains.
It is also found that 𝜀r′ of presents firstly decreases and increases with the
increase of the 𝑆l , and reaches a minimum at 1wt%. The increase of 𝜀r′ with
decrease in frequency is generally known as caused by interfacial polarization at
occuring the interface between epoxy and nanoparticles [53-58]. Q. Wang and G.
Chen et al., [92] has reported same results on decrease 𝜀r′ for silica/epoxy
nanocomposite. When fillers are loaded into the epoxy resin, the changes of
dielectric properties with the presence of the nano size filler and their loading
concentration may be caused by the following reason from Nelson and Forthergill
[93]: decrease in the permittivity 𝜀r′ of nano particle/epoxy resin composites is
related to the restriction of dipole movement at the interaction zone. For example,
the measured real part of the permittivity at 393 K using high permittivity TiO2
filler (≈ 99) is: 9.99 (base resin), 13.8 (micro-composite, size: 1.5 μm), and 8.49
(nanocomposite, size: 23 nm) at 1 kHz. The nanocomposite has a lower
permittivity than the base resin and micro composite. The electrical properties of
nanocomposites are mainly determined by the dispersion and nanoparticle filler
loading characteristics. The physical and chemical characteristics of the interface
can be influenced by these factors, leading to changes in micro scale features such
as traps, carrier mobility and free volume [93, 94]. On the other hand, free volume
in nano size particle/epoxy composite systems will decrease when nanostructured,
and will increase when micro structured. Free volume is demonstrated in
experiment and in theory by Nelson and Forthergill. Obviously, free volume plays
an important role in the reduction of the permittivity 𝜀r′ [93-97].

Figure 4.19 Frequency dependence of complex permittivity at room temperature;

(a) real part of permittivity of each specimen, (b) dielectric loss

Figure 4.20 Relative permittivity 𝜀r′ at 100 kHz
4.4.2 DISCHARGE PATTERNS AND PDIV
Prior to study discharge patterns under 60, 80 and 100 kV of impulse voltage
application, PDIV of each sample was measured first to compare with modified
other samples. Figures 4.21 (a) to (f) show discharge image of each sample at 100
kV. White lines and circles in the figures depict the high voltage electrode and
ground electrode, respectively. Here, let the discharge propagation length (𝑙d ) be
defined as the length from the high voltage electrode to a site of the longest
discharge edge part. It should be noted that creepage discharges under a positive
lightning impulse voltage do not always propagate radially in this electrode
configuration. Such orientation and distribution of branches on the surface of
insulator is considered to be due to distortion of the electric field resulting from
surface charge accumulated on the insulator [98]. The creepage discharge trend
such as discharge light intensity, the number of discharge branches and discharge
length is changed according to dry process and surface coated material on the
surface of PB. The discharges are composed of ramified branches of streamer
propagating on the surface of NP, DP and TP samples, while emitting intense
luminous area. Whereas the discharge image in EP-A and EP-F show weak light

spots at all the voltage levels and propagating discharge shows filamentary
branches. On the other hand, the discharge pattern of SP-0.2 wt% and SP-5 wt%
shows more intense white luminous part at their streamer and branch than that of
EP-F samples and propagating discharge shows filamentary branches. Thus, we
confirmed that the surface conditions and materials on PB influence creepage
discharge properties and tendency.
The total number of discharge branches observed on the surface of each sample
was counted using a binary image analysis program (AZO V2.30) as shown in Fig.
4.22. It is seen in these figures that the The discharges are composed of ramified
branches of streamer propagating on the surface of NP, DP and TP samples. Note
that NP, DP and TP samples have more divided discharge branches from main
streamer than EP-F sample as shown in Fig 4.23.
Figure 4.21 Discharge patterns of each sample at 100 kV using rod-plane

electrodes in mineral oil

Figure 4.22 Discharge binary images of each sample using image analysis
program at 80 kV

Figure 4.23 The number of discharge branches of each sample at 80 kV and 100
kV
Figure 4.24 shows discharge occurrence probability (𝑃d ) at each voltage level
for different kinds of specimen. In NP, the probability (𝑃d ) of 100% of discharge
occurrence is observed at each voltage level from 60 to 100 kV. On the other hand,
it is seen that 𝑃d at 60, 70 and 100 kV is 33.3, 96.6 and 100 % (DP), 13.3, 66.6
and 100% (EP-A), 10, 33.3 and 80% (EP-F) and 36.6, 56.6 and 100% (TP),
respectively. The discharge occurrence probability of DP decreased at only 60 kV.
These results indicate that the drying process and the moisture level in PB should
affect the discharge inception as well as development process. On the other hand,
it should be noticed that 𝑃d of SP (0.2, 0.5, 3 and 5 wt%) are less than 63.5 and
73.3% from 60 to 100 kV, and that 𝑃d of SP with 50 wt% was higher than that of
EP-F at all the voltage levels. These results will be discussed in Section 4.6.
Next, let PDIV ratio 𝑘1 can be defined as the ratio of PD occurrence
probability of SP to that of EP-F. The ratio 𝑘1 of SP specimen was defined as the
voltage of 10% discharge occurrence probability of SP divided by that of EP-F.
Figure 4.25 shows the 10% PD occurrence probability ratio 𝑘1 dependent on filler
loading of SiO2 nano fillers under the impulse voltage. As can be seen in Fig. 4.25,
the ratio 𝑘1 is larger than one, meaning that discharge occurrence suppression
property can be improved with appropriate 𝑆𝑙 compared to that of EP-F. The ratio

𝑘1 presents the highest values at SiO2 filler loading of SP-0.2wt%, while further
increasing 𝑆l results in decrease of 𝑘1 .
Figure 4.24 Discharge occurrence probability 𝑃d at each voltage level with

different coating material
Figure 4.25 PD occurrence probability ratio k1 against EP-F versus filler

loading of nano SiO2, where k1 is defined as the voltage of 10% discharge
occurrence probability of SP divided by that of EP-F.

4.4.3 DISCHARGE PROPAGATION LENGTH
Figures 4.26 and 4.27 show the discharge propagation length ld of each type of
specimen with different SiO2 nano filler loading, respectively, at all the voltage
levels. It is evident from the two figures that ld for each sample increases as the
applied impulse voltage increases from 60 to 100 kV. Note that the average
discharge propagation length ldave of the solid layer coating samples EP-A, EP-F,
TP and DP is shorter than that of NP sample at all the voltage levels. On the other
hand, it should be noticed that ldave of PS-0.2, 0.5 and 5 wt% is shorter than that of
the other samples at all the voltage levels. These results indicate that the solid
layer coating of epoxy, Teflon and nano silica/epoxy composites on PB should
affect not only the discharge inception but also the development process as well.
That might be influenced by dry process and surface modification. These detailed
results will be discussed in Section 4.6 together with the results of the needle-bar
electrode system.
Figure 4.26 Discharge propagation length ld at each lightning impulse voltage

(NP, DP, EP-A, EP-F and TP)

Figure 4.27 Discharge propagation length ld of SP specimens at each lightning

impulse voltage
4.5 RESULTS AND DISCUSSION(NEEDLE-BAR

ELECTRODE)
4.5.1 DISCHARGE PATTERNS AND PDIV
Figures 4.28 (a) to (e) show creepage discharge image of each sample NP, DP,
EP-A, SP-0.2 wt% and SP-3 wt% at 45 kV using the needle-bar electrode system,
respectively. White triangles in the figures depict the high voltage electrode. The
measuring method of discharge propagation length (ld) is the same with Section
4.3.2. The discharges are composed of ramified branches of streamer propagating
on the surface of NP and DP samples, while emitting intense luminous area
appears at the end part of their branch tip and streamer. Note that NP and DP
samples have more discharge branches than EP-A, EP-F and SP specimens and

this phenomenon is similar with the results obtained by the rod-plane electrode
system. Figure 4.29 shows discharge light intensity as a function of the applied
impulse voltage Vi of each kind of sample. The discharge light intensity of EP-A,
EP-F and SP specimen shows weak discharge light intensity at all the voltage
levels than that of NP and DP. Propagating discharge of EP-A, EP-F and SP
samples shows filamentary branches.
Figure 4.30 shows discharge occurrence probability (𝑃d ) at each voltage level
Vi for different kinds of specimen. Each specimen was tested 3 times at least
under the same test conditions. In NP, the probability (𝑃d ) of 100% of discharge
occurrence is observed at each voltage level from 25 to 45 kV. On the other hand,
it is seen that 𝑃d at 20, 25 and 30 kV is 66.6, 86.6 and 100% for DP, 53.3, 73.3
and 86.6% for EP-A and 26.6, 53.3 and 73.3% for EP-F, respectively. These
results indicate that the drying process and moisture level in PB affect both the
discharge inception and development processes. On the other hand, it should be
noticed that 𝑃d of 0.2 and 3 wt% at Vi 40 and 45 kV is 100%. These results
indicate that the SiO2/epoxy composite on the surface of PB should improve the
discharge inception voltage.

Figure 4.28 Discharge patterns of (a) NP, (b) DP, (c) EP-F, (d) SP-0.2wt% and (e)
SP-3wt% at Vi 45 kV

Figure 4.29 Discharge light intensity from PMTs waveforms as a function of Vi

for different kinds of specimen
Figure 4.30 Discharge occurrence probability at each voltage level for different
kinds of specimen

Figure 4.31 shows 𝜀𝑟′ and 30% discharge occurrence probability ratio k2, which
is defined as Vi giving 30% 𝑃d of EP-F divided by that giving 30% 𝑃d of nano
silica/epoxy composite coated PB. The ratio k2 dependent on the filler loading of
silica nano fillers under the Vi. As seen in Figure 4.31 the ratio k2 is larger than
one. Meaning that insulation performance against the creepage discharge
generation can be improved with appropriate filler loadings compared to that of
NP. Note that k2 presents the highest value 1.4 at silica filler loading of 3 wt%,
while decreases with further increasing nano silica filler loading above 5 wt%.
High filler loading sample with 50 wt% of the filler shows still a high ratio k2 than
that of EP-F. Moreover, 𝜀𝑟′ firstly decreases and increases with the increase of the
filler loading, and reaches a minimum at 3 wt%. It can be concluded that the
reduction of 𝜀𝑟′ corresponds to the improved Pd of SP specimens.
Figure 4.31 The ratio k2 of 30% discharge occurrence probability Pd of nano

silica filler loaded to that of EP coated NP and relative permittivity 𝜀r′ at 100 kHz
as a function of nano silica filler loading for epoxy/silica nanocomposite coated
PB

Many researchers conveyed the similar conclusions on the effect of nano filler
loading on the performance of vacuum surface flashover and breakdown
properties, and proposed a variety of models and interpretations. Yu Chen et al.
investigated the vacuum surface flashover characteristic of nano size Al2O3/epoxy
nanocomposite. The flashover voltage shows the highest values at Al2O3 filler
loading of 3 wt%. The experimental results were explained in terms of the effects
of different roles of shallow and deep traps introduced by the nano filler loading
[56]. Shengtao Li et al. also investigated the surface flashover and breakdown
voltages of TiO2/epoxy nanocomposite in SF6 gas. The optimal filler loading of
TiO2 for the normalized flashover voltage k1 was 1 wt% [53-55]. On the other
hand, the present study reveals that the PD occurrence probability of SP-0.2 wt%
SiO2 added epoxy composite insulation (SP) is improved to be 1.3 times as much
as that of EP-F. The decreasing PD occurrence probability with the nano filler
loading can be explained in two ways as follows; 1) SiO2/epoxy nano composites
coating on PB suppresses emission of trapped electron from the surface of PB in
the discharge inception process, and 2) the discharge inception voltage possibly
depends on the surface roughness as well as the filler loadings. More detailed
discussion will be made in Section 4.6
4.5.2 DISCAHRGE PROPAGATION LENGTH
Figures 4.32 and 4.33 show the discharge propagation length ld of each sample
at all the voltage levels. It is evident from the figure that ld for each sample
increases as the applied Vi increases from 20 to 45 kV. Note that the average
propagation length ldave of DP, EP-A, EP-F and SP samples at all Vi is shorter than
that of NP samples by 42%, 64% and 70%, respectively. Figure 4.34 shows
relation between discharge propagation length and relative permittivity. Discharge
propagation length ratio k3 be defined as 45 kV giving ld of neat EP divided by
that giving ld of silica/epoxy nanocomposite. As seen in Fig. 4.34, discharge
propagation length ratio k3 is larger than one, meaning that the insulation
performance against the creepage discharge propagation can be improved with
appropriate 𝑆𝑙 compared to that of EP. Note that k3 presents increases and
decreases with the increase of the 𝑆𝑙 above 3 wt%, respectively. It can be
concluded that the low 𝑆𝑙 improved ld. But note that once discharge occurred, ld of
each sample is becomes the same at each voltage level, besides scattering of the
data overlaps each other, meaning no appreciable effect of the filler loading on ld
in oil condition. These results will be discussed in Section 4.6.

Figure 4.32 Discharge propagation length ld of NP, DP, EP-A and EP-F as a
function of Vi
Figure 4.33 Discharge propagation length ld of SP specimens as a function of Vi

for the needle-bar electrode system

Figure 4.34 The relationship between average discharge propagation length ldave
ratio k3 and permittivity 𝜀r′ of SP with different silica filler loading at 45 kV
4.6 DISCUSSION (EFFECT OF SOLID LAYER

COATING)
The effect of the solid layer coating on Pd and ld is discussed. The following
three factors are possibly considered to explain the experimental results on Pd and
ld, respectively:
1) Drying of water content in neat PB and decrease in 𝜀𝑟′ by silica/epoxy

nanocomposite coating
2) Surface modification effect by solid layer coating
3) Silica/epoxy composites coating suppressing the electron emission from the
surface of PB

4.6.1 DRY EFFECT AND PERMITTIVITY EFFECT
Note that cellulose paper placed on a factory floor is said to be able to hold 4-8%
moisture in the relative humidity of 30 to 70% in winter and summer, respectively
as shown in Fig. 4.35 [9]. Besides, PB cellulose fibres form microcapillaries,
which contain insulation oil, gaseous residues and water molecules.
Figure 4.35 Bonds of water molecules in a PB microcapillary [98]
Insulation paper consists of a compound of cellulose fibres linked by hydrogen

and Van-der-Waals bonds. Water is absorbed in the amorphous structures of
cellulose, in the pores of the cell walls and in the polyose-lignin-gel (remains from
wood) [98]. Cellulose fibre form microcapillaries, which contain insulation oil,
gaseous residues in the wet state water. For low water content, the molecules are
bound by strong bonds forming a monolayer. When water content (more than 1-2 %
moisture relative to weight) increases, only relatively weak Van-der-Waals bonds
and capillary forces hold water molecules in the cellulose compound as shown in
Fig. 4.31 [9, 97].
In general, discharge inception voltage is dependent on the inherent
capacitance of solid insulation as well as surface roughness, impurities between
interface of composite insulation and type of materials etc. According to an
empirical formula given in Eqn. (4.1) with increasing the inherent capacitance of
solid insulation, the electric field of triple junction increases. On the other hand,

creepage discharge is more likely to occur and then propagate along the surface of
PB with increase in the C [pF/mm2] of solid insulation, because the electrostatic
energy stored in the C increases [99, 101]. We confirmed that the weight,
thickness after dry process and C after epoxy/silica nanocomposite coating
decreased by 5.3%, 3% and 14%, respectively. Equation 4.1 and figure 4.36 show
relation between capacitance and estimated discharge inception voltage.
𝑉 = K𝐶 −𝑛 (4.1)
C is the specific capacitance of pressboard [pF/mm2]; V is applied voltage [kV];

K and n are constant (estimated n: 0.77)
Estimated discharge inception voltage after dried process and surface coating
on PB from Eqn. 4.1 increases with decreasing specific capacitance as shown in
Fig. 4.32. Besides, in rod-plane electrode results, Pd of DP decreased by 66.7% at
only 60 kV and ld also decreased by and 12.1, 4.2 and 10.2% at each Vi as shown
in Fig. 4.24, 4.26, 4.30 and 4.32. It is because the capacitance decreases, resulting
in less discharge development according to Eqn 4.1. Thus, it can be said that the
moisture level in PB affects the discharge inception and development as well.
Figure 4.36 Relation between estimated discharge inception voltage and the
specific capacitance before and after dry process of pressboard at 70 and 150℃
for 3 hours, respectively

4.6.2 SURFACE MODIFICATION EFFECT
The second possible reason for Pd and ld is a surface modification effect by the
solid layer coating on PB. The surface of NP and DP exhibits numerous
protrusions of cellulose fibrous as shown in Fig. 4.37 [98, 101], whereas the
surface protrusions of EP-A, F and SP samples were covered with epoxy resin.
For that reason, there are no protrusions on the surface of coated PB using a
microscope as shown in Fig. 4.37. We estimated the electric field distribution with
presence of protrusions around the high voltage electrode using COMSOL
Multiphysics 2D MODEL as well as surface roughness measurement results as
shown in Fig. 4.38.
The generalized hydrodynamic drift-diffusion equations combined with
Poisson’s equation were employed in dielectric liquids. The detailed analysis
method using COMSOL MULTIPHYSICS will be dealt in Chapter 6.
Figure 4.39 shows the simulation results, the electric field is disturbed with the
presence of protrusions around the needle electrode simultaneously with discharge
inception over 1 × 108 V/m , and then the creepage discharge is likely to
propagate toward protrusion tip along the pressboard surface. Concentrated
electric field in protrusion tip is more likely to increase with increasing time than
in protrusion pole part as shown in Fig. 4.40. Thus, in the case of NP and DP, it
might be possible that the electric field concentration occurs at the protrusion tip
and it could be starting point of a creepage discharge inception.
Figure 4.37 Surface protrusions on the surface of conventional pressboard (DP)

using microscope and scanning electron microscope image

Figure 4.38 Analysis model with presence of protrusions on the surface of PB

with surface roughness measured in DP
Figure 4.39 The calculated electric field distribution with presence of a

protrusion on the surface of PB around the high voltage needle electrode

Figure 4.40 The electric field results of protrusion tip parts and pole parts
4.6.3 ELECTRON SUPPRESSION EFFECT
The third possible reason for Pd and ld is that epoxy/silica nano composites
coating may suppress initial electron emission contributing to creepage discharge
inception and propagation from the surface of PB, resulting in the decrease of the
Pd at all the Vi as well as the decrease in ld as well.
Let us propose a surface flashover model in the nanocomposite coating PB as
illustrated in Fig. 4.41. In general, oil immersed PB form microcapillaries which
contain insulation oil. Porosity of PB causes oil to penetrate into the PB bulk. The
free charges such as positive ion, negative ion and electron are generated from oil
immersed PB bulk due to molecular ionization and ionic dissociation under a high
electric field [32]. During creepage discharge process, charge carriers emitted
from electron avalanches could excite the electron from the surface of PB, and
newly generated electrons can play a role as the seed of new electron avalanches
and some of these secondary electrons will again strike the surface of PB,
producing tertiary electron [103]. Then, the creepage discharge is more likely to
incept and then can more readily creep along the surface of PB. Deep electron

traps introduced by the epoxy/nanoparticle composites coating can restrain the

emission of the internal secondary electrons and initial electron inside the surface
layer of PB, and the discharge inception voltage can be possibly increased.
Namely, the epoxy/silica nanocomposites coating on PB suppresses the electron
emission and thus extraction of electrons from the surface layer of epoxy/silica
nanocomposites in the discharge inception and propagation process.
Figure 4.41 Surface discharge conceptual model in oil/PB composite insulation

system
4.7 SUMMARY
Chapter 4 focused on surface modifications of the insulating paper by solid

layer coating method to discuss the effect of the solid layer coating at the oil/PB
composite insulation system on the suppression on both creepage discharge
inception and propagation. The creepage discharge behaviour at the oil/pressboard
interface have been discussed with different silica filler loadings, coating
materials and electrode configurations.
The research work reveals that the new coating method on the PB surface can
indeed affect the creepage discharge behaviour as well as dielectric properties.
The experimental results are summarized as follows:
1. 𝜀r′ decreases with the increase of the 𝑆𝑙 , while further increasing 𝑆l results in
increase 𝜀r′ .

2. Silica/epoxy nanocomposite coated PB shows lower Pd and smaller ld with

low silica filler loading up to 3 wt% than that of conventional PB, and should
improve the insulation performance of PB/oil insulation system.
3. Three possibly considered mechanism are proposed to explain the results on
the effect of the coating method on Pd and ld as follows:
1) Drying of water content in neat PB decrease in 𝜀r′ by silica/epoxy

3) Silica/epoxy composites coating suppressing the electron emission from
the surface of PB

Chapter 5: Creepage Discharge Modelling Approach

Chapter 5: Creepage Discharge Characteristics of Silica/Epoxy Nanocomposite Plate
CHAPTER 5
Creepage Discharge Characteristics of

Silica/Epoxy Nanocomposite Plate
5.1 EXPERIMENT
5.1.1 ELECTRODE SETUP
Figures 5.1 (a) and (b) show side view and top views of electrode system,
respectively. The tip radius of the needle electrode was 10 μm and gap distance
was 50 mm to the bar electrode. The needle electrode was touched at an angle of
30 degree from the surface of silica/epoxy nanocomposite. There is no back
electrode at the bottom place of the specimen to consider only the surface effect of
specimens on creepage discharge.
The whole electrode system was placed in a mineral oil immersed chamber as
shown in Fig. 5.2. Vacuuming was made in the chamber for 2 hours to reduce
gasses that influence to PDIV of the oil/solid composite insulation system.
Discharge image was observed using a digital camera (Fujifilm Finepix S100fs)
and a digital camera equipped with an image intensifier (Hamamatsu C9016-2x).
Discharge light emission was also detected with a photomultiplier tube (PMT,
Hamamatsu Photonics, H6780~4) to determine PDIV. A 50 Ω resistor was used to
obtain the discharge current signal.
A positive standard lightning impulse voltage was applied to the needle
electrode varied from 25 kV (PDIV of neat epoxy) to 75 kV by a step with
increment of 5 kV. The interval time between each voltage level was one minute.
The impulse voltage application was repeated 5 times at each voltage level.

(a) Side view
(b) Top view

Figure 5.1 Needle-bar electrode configurations with nano silica/epoxy
composite plate

Figure 5.2 Diagram of experimental setup to measure and observe PDIV and
creepage discharge properties in oil and SiO2 /epoxy composite insulation system
5.1.2 SAMPLE PREPARATION
Table 5.1 lists prepared silica/epoxy nanocomposite with different filler loading
(0.1, 0.5, 1, 3, 5, 10, 20, 45 wt%) of 45 nm and particle size (12, 22, 45, 80, 130,
500 nm and 13 μm) as shown in Table 5.1.
Figures 5.3 and 5.4 show photos and SEM images of prepared with different
nano 𝑆𝑙 . All samples were supplied from Nissan chemical industries., LTD. The
size of each sample for creepage discharge measurement and dielectric property is
70 mm by 100 mm and 3 mm, 70 mm by 70 mm and 1 mm, respectively. Figures
5.3 and 5.4 shows SEM image of both surface layer and middle layer of 12, 22, 45
and 80 nm silica fillers to investigate the nano silica state of dispersion in the base
epoxy matrix. To observe nano silica state of dispersion, samples with pretreated
by Platinum vapor deposition and protective film laminated using focused ion
beam (FIB) with accelerating voltage 1.0 kV. The observation of the particle
dispersion state was investigated by Nissan Chemical Co. Ltd. As can be seen in
Figs. 5.3 and 5.4, white circles in the figures show nano silica particles dispersed
in base epoxy. All the silica/epoxy nanocomposite specimens seem to exhibit
good dispersion of nanoparticles in both surface layer and middle layer.

Table 5.1 Properties of prepared nano silica/EP composites

Sample name Silica Silica Manufacturing Surface Filler loading, x wt%
size purity method treatment (weight percent)
N∙EP Neat
(EP) epoxy - - - -
EP12-3, 10 12 nm normal A A 3, 10 wt%
EP 22-3, 10 22 nm normal A A 3, 10 wt%
EP 45-x 45 nm normal B A 0.1, 0.5, 1, 3, 5, 10,
20, 45 wt%
EP 80-3, 10 80 nm normal A B 3, 10 wt%
EP 500-3 500 nm Other company
normal product A 3 wt%
EP 13000-3 13 μm Other company
normal product A 3 wt%
Figure 5.3 Photos of samples with different nano silica filler loading and neat
epoxy (a) neat epoxy (b) 0.1wt%, (c) 0.5wt%, (d) 1wt%, (e) 3wt%, (f) 5wt%, (g)
10wt%, (h) 20wt% and (i) 45wt%

composites with different nano silica

composites with different nano silica

5.1.3 DIELECTRIC PROPERTY MEASUREMENT
Before dielectric property measurement, all the samples were dried for 1 hour at
100℃ to remove the moisture effect. Figures 5.6 (a) and (b) show the weight and
thickness before and after the drying process of silica/epoxy nanocomposite. As
you can see in the figures, the weight decreases after the dry process and then
increases after 24 hours at room temperature. But there is no variation of thickness
before and after the dry process.
(a) Weight variation before and after dry process of

sample with different filler loading
(b) Thickness variation before and after dry process

Figure 5.6 The variation of weight and thickness of 5wt% silica/epoxy
nanocomposite before and after drying process at 100℃ for 1 hour

Dielectric properties of each specimen were measured with an impedance

analyzer (IWATSU Impedance Analysis PSM3750) at room temperature and
frequency from 50 Hz to 2 MHz after dry process. The number of test samples of
each specimen was two. The size and thickness of a sample was 70 mm by 70 mm
and 1mm, respectively. Prior to measurement, gold electrodes with diameter of 50
mm were deposited onto both sides of a sample by Magnetron Sputter (MSP-1S)
to ensure the reproducibility of measurement and improve the contact area
between the electrode and the surface of epoxy plate. Gold electrode on the
surface of sample is sputtered for 3 minute with a thickness of 3 Å/sec and then
we confirmed that the thickness of sputtering on the surface of epoxy/silica
nanocomposite plate is 54 nm using a thickness gauge as shown in Fig. 5.7. Note
that measurements of dielectric properties were conducted two times each for two
samples with error of 2 %.
Figure 5.7 Silica/epoxy nanocomposites for dielectric property measurement
5.2 RESULTS AND DISCUSSION
5.2.1 DIELECTRIC PROPERTIES
Figures 5.8 (a) to (b) show frequency dependence of real part (permittivity 𝜀r′ )
and imaginary part (dielectric loss 𝜀r′′ ) of the complex dielectric constant 𝜀r∗ for
EP with different nano silica filler loadings at room temperature. It can be seen in
Figure that 𝜀r′ decreases stably with increasing frequency from 50 Hz to 2 MHz.

As can be seen in Fig. 5.9, 𝜀r′ increases with increase in the filler loading at 100
kHz corresponding to dominant frequency component that standard lightning
impulse voltage contains. It is also obvious from Fig. 5.9 that 𝜀r′ decreases with
decreasing the filler loading below 1 wt%, and reaches a minimum at 0.1 and 0.5
wt%, while further increasing nano 𝑆l results in increased 𝜀r′ above 3 wt%. This
reduction in 𝜀r′ was discussed in Chapter 4. As can be seen in Fig. 5.8 (b), 𝜀r′′
decreases with increasing frequency from 50 Hz to 2 MHz, while no significant
difference appears in 𝜀r′′ under 20 wt% loading at 100 kHz.
Figure 5.8 Frequency dependence of complex permittivity at room temperature; (a)

real part of permittivity of each specimen, (b) dielectric loss

Figure 5.9 Relative permittivity at 100 kHz with different nano silica filler loading
Figures 5.10 (a) to (b) and 5.11 (a) to (b) show frequency dependence of real
part (permittivity 𝜀r′ ) and imaginary part (dielectric loss 𝜀r′′ ) of the complex
dielectric constant 𝜀r∗ for EP with different nano silica size of 3 wt% and 10 wt%,
respectively at room temperature. It can be seen in Figure that 𝜀𝑟′ decreases stably
with increasing frequency. As can be seen in Fig. 5.11 (a) and 5.12 (a), 𝜀r′
increases with increase in the filler size at 100 kHz corresponding to dominant
frequency component that standard lightning impulse voltage contains. It is also
obvious from Fig. 5.11 (a) and 5.12 (a) that 𝜀r′ decreases with decreasing the filler
size below 22 nm, while further increasing nano silica filler size results in
increased 𝜀r′ above 22 nm with keeping the filler loading constant. As can be seen
in Fig. 5.10 (b) and 5.11 (b), 𝜀r′′ increases with increasing frequency, while no
significant difference appears in 𝜀𝑟′′ under 20 wt% loading at 100 kHz.
Santanu Singha and M. Joy Thomas et al., [104] have reported dielectric
property with different particle size. Micro size particle/epoxy composites show a
higher permittivity value than the nanocomposites. They suggest the reason on the
high permittivity value in micro composites probably due to the fact that there is
no restriction in the mobility of epoxy chains with micro fillers [104]. Toshikatsu
Tanaka et al., [105-107] have reported electron traps existing inside core quantum
dots (QDs), shell and interface in a core-shell structure embedded in polymer

matrix. The QDs model has formed polymer matrix, shell insulator and core
quantum dot. Following Eqn. 5.1 shows polarizability in core-shell system [105-
107].
(𝜀2 − 𝜀𝑚 )(𝜀1 − 2𝜀2 ) + 𝜌(𝜀1 − 𝜀2 )(𝜀𝑚 − 2𝜀2 )

α = 4π𝑟23 (5.1)
(𝜀2 + 2𝜀𝑚 )(𝜀1 + 2𝜀2 ) + 2𝜌(𝜀2 − 𝜀𝑚 )(𝜀1 − 𝜀2 )
where core with permittivity and shell radius is 𝜀1 and r1, shell with
permittivity , fraction of the total particle volume occupied by the core and
thickness is 𝜀2 , 𝜌 = (r1 /r2 )3 and (r2-r1), respectively.
Eqn 5.1 can be explained in terms of increase in 𝜀r′ with filler loading and size.
On the other hand, α and 𝜀r′ are proportional to each other. The α increase with
increase nano filler size, relation between polarizability and 𝜀r′ shows follows:
𝜀r′ = 𝜀0 + 𝜒
where 𝜀r′ is permittivity [F/m], 𝜀0 is space permittivity [F/m] and 𝜒 is

polarizability.
𝜌 = (r1/r2)3 is the fraction of the total particle volume occupied by the core.
A core-shell nanoparticle can be approximated as an effective homogeneous
sphere with an equivalent permittivity given by the Eqn. 5.2. The 𝜀𝑐 increase with
increase nanoparticle size from Eqn. 5.2.
𝜀1 (1 + 2𝜌) + 2𝜀2 (1 − 𝜌)
𝜀𝑐 = 𝜀2 (5.2)
𝜀1 (1 − 𝜌) + 𝜀2 (2 + 𝜌)

Figure 5.10 Frequency dependence of complex permittivity at room temperature

with different nano silica sizes with keeping the filler loading 3 wt%; (a) real part
of permittivity of each specimen, (b) dielectric loss

Figure 5.11 Frequency dependence of complex permittivity at room temperature

with different nano silica size with 10 wt%; (a) real part of permittivity of each
specimen, (b) dielectric loss

Figure 5.12 Relative permittivity at 100 kHz as a function of nano silica size for
different filler loading
5.2.2 DISCHARGE PATTERNS AND DISCAHRGE

OCCURRENCE PROBABILITY
Figures 5.13 (a) and (b) show creepage discharge image of EP and EP 45-1 wt%
at 75 kV as well as waveforms of two PMT signals, current and impulse voltage.
The decision way of PDIV is same with that given in Chapter 4. Note that white
triangles in the figures depict the high voltage electrode. Here, let the discharge
final length ld be defined as the length from the high voltage electrode to a site of
a longest discharge edge part. Discharge streamer in neat EP exhibits more intense
luminous from the needle tip and higher peak current than that of EP 45-1 wt% at
all the voltage levels as observed from PMT signal and current waveform.
Prior to study creepage discharge pattern under 25-75 kV of impulse voltage
application, PDIV of neat EP was measured to be 30 kV. Three specimens were
used to confirm reproducibility at least under the same conditions. Creepage
discharges of all the samples were found to occur intermittently and randomly at a
given Va. Figure 5.14 shows discharge occurrence probability (Pd) as a function of
Vi with different 𝑆l . As can be seen in the Figure, Pd of 100% is observed over 45
kV for neat EP, while EP 45-45 wt% exhibits higher Pd than neat EP at 30 kV.

Next, let 50% creepage discharge occurrence ratio k4 be defined as Vi giving 50%
Pd of neat EP divided by that giving 50% Pd of silica/epoxy nanocomposite.
(a) Neat epoxy discharge image (b) SP-1wt% discharge image

Figure 5.13 Discharge image of neat EP and SP 45-1wt% at 75 kV
Figure 5.14 Discharge occurrence probability at each voltage level with different
nano silica filler loading

Figure 5.15 shows the relationship between 𝜀𝑟′ and 50% discharge occurrence
probability ratio k4. As seen in figure, the ratio k4 is larger than one, meaning that
the insulation performance against the creepage discharge generation can be
improved with appropriate 𝑆l compared to that of neat EP. Note that k4 presents
the highest value 1.33 at 𝑆l of 1 wt%, while further decreases with increasing nano
𝑆l above 3 wt%. 𝜀𝑟′ presents firstly decreases and increases with the increase of the
𝑆l , and reaches a minimum at 0.5 wt%. It can be concluded that the appropriate
low 𝑆l improved Pd of silica/epoxy nanocomposite specimens.

permittivity 𝜀r′
Figure 5.16 shows Pd as a function of Vi for different nano silica size with
keeping the filler loading constant. As can be seen in Figure, EP with big size
nano particles (130, 500 nm and 13 μm) exhibits higher Pd than neat EP and EP
with small size particles at all the Vi, whereas EP with small size particles as low
as 80 nm and 3 wt% show that creepage discharge occurrence can be improved.
Figure 5.17 shows the relationship between 𝜀r′ and 50% 𝑃d . Note that 50% PD
occurrence probability ratio k4 presents the highest value 1.25 at nano silica size
of 22 nm, while ratio k4 decreases with increasing nano silica size.
Shengtao Li et al., [94] have reported surface flashover performance under
impulse voltage in vacuum with nanoparticles (NPs) and micro size particles
(MPs) in terms of shallow and deep traps. MPs show contrary behavior and MPs
causes a serious low flashover voltage, whereas small amounts (up to 3 wt%) of
NPs improve flashover performance by over 58% compared with neat polystyrene.

100
MPs can introduce shallow traps while NPs introduce deep traps. The deep traps
are crucial in improving flashover performance by trapping the charge carriers and
thereby weakening secondary electron multiplication processes. In Fig. 5.17, nano
size particle 22 nm can be improved Pd by over 28% and 43% compared with neat
EP and 13 um size particle, respectively. This result shows similar trend with
reference [94]. Thus, it might be possible that the nano size particle improve
creepage discharge inception process, meaning that charge carriers are suppressed
by deep trap from emitted secondary electron of the surface of epoxy/silica
nanocompsoites.
Figure 5.16 Discharge occurrence probability 3 wt% at each voltage level with
different nano silica size

permittivity 𝜀r′ of 3 wt%

101
5.2.3 DISCHARGE PROPAGATION LENGTH
Figures 5.18 to 5.22 show the discharge propagation length ld at Vi and ratio k5
of each kind of sample with different filler loading and size at 75 kV. It is evident
from Figs. 5.18 to 5.22 that ld for each kind of sample increases as the applied
voltage Vi increases from 30 to 75 kV. Note that the average discharge
propagation length ldave of nano silica/epoxy composite samples at all the voltage
levels is shorter than that of neat EP samples. Figs 5.19 and 5.21 show the ldave
ratio k5 at 75 kV. Ratio k5 be defined as 75 kV giving ld of neat EP divided by that
giving ld of silica/epoxy nanocomposite. As seen in figure, the ratio k5 is larger
than one, meaning that the insulation performance against the creepage discharge
propagation can be improved with appropriate 𝑆l and filler size compared to that
of neat EP. Note that k5 presents increases and decreases with the increase of the
𝑆l and filler size above 3 wt% and 80 nm, respectively. It can be concluded that
the low 𝑆l and EP with small size particles improved ld. But as mentioned in
chapter 4, once discharge occurred, 𝑙d of each sample is becoming the same at
each voltage levels, besides scattering of the data overlaps each other, meaning no
appreciable effect of the filler loading on 𝑙d appears under the present condition.
Figure 5.18 Discharge propagation length of EP with different silica filler loading
as a function of Vi with keeping the 45 nm filler size

102
and permittivity 𝜀r′ of EP as a function of silica filler loading at 75 kV
Figure 5.20 Discharge propagation length ld of EP with different silica filler size
as a function of Vi with keeping the 3 wt% filler loading

103
and permittivity 𝜀r′ of EP as a function of silica filler size at 75 kV with keeping
the 3 wt% filler loading
Figure 5.22 Discharge propagation length ld of EP with different silica filler size
as a function of lightning impulse voltage with keeping the 10 wt% filler loading

104
5.2.4 DISCUSSION
Next, the effect of nano silica on PDIV is discussed. The following two reasons
are possibly considered to explain the experimental results of the effect of nano
silica loading on the discharge occurrence probability of EP as shown in Figs.
5.15 and 5.16:
1) Relative permittivity reduction effect by introducing nano size silica particle

in base epoxy
2) Silica/epoxy nanocomposites are more likely to suppress the electron
emission from the surface layer of nanocomposite.
In view of the tradition dielectric theory, it is well known that particles

introduce defects into epoxy resin and such defects act as the charge accumulation
centers, enhancing the local electric field. This causes a decrease in PDIV and
BDV which is confirmed in microcomposites. When the size of the particles in
reduced to nanometer level, the situation is totally different [56, 84, 108-112]. For
instance, the electric strength and surface flashover voltage increase in
particle/epoxy nanocomposites with appropriate filler loading. These
characteristics are associated with the nanoparticle/epoxy interaction zone.
However, the structure and properties of the interaction zone are influenced by
many factors such as the surface modification of particles, dispersion and type of
particles, the physical characteristics of epoxy resin etc. [55].
The electric field distribution at triple junction (mineral oil, pressboard and
electrode) is influenced by the relative permittivity of silica/epoxy nancomposite.
On the other hand, low permittivity may weaken the electric field distortion near
the triple junction and thereby improve the creepage discharge performance.
Following Eqns. 5.1 and 5.2 show random arranged model related with 𝜀r′
dependent filler loading and consider a core-shell system, respectively.
𝑉𝑝 𝑉𝑝
𝜀𝑎𝑘 = ( ) ∗ 𝜀𝑝𝑘 + (1 − 𝑘
) ∗ 𝜀𝑚 (5.1)
100 100
where 𝜀a𝑘 is 𝜀r′ of composites, 𝜀p𝑘 is 𝜀r′ of nano filler, 𝜀m

𝑘
is 𝜀r′ of matrix, 𝑉p is
nano filler loading and k is parallel circuit (k=1), series circuit (k=-1) and random
arranged (k=1/3).
(𝜀2 − 𝜀𝑚 )(𝜀1 − 2𝜀2 ) + 𝜌(𝜀1 − 𝜀2 )(𝜀𝑚 − 2𝜀2 )

α = 4π𝑟23 (5.2)
(𝜀2 + 2𝜀𝑚 )(𝜀1 + 2𝜀2 ) + 2𝜌(𝜀2 − 𝜀𝑚 )(𝜀1 − 𝜀2 )

105
where core with permittivity and shell radius is 𝜀1 and r1, shell with permittivity
and fraction of the total particle volume occupied by the core is 𝜀1 and (r2-r1),
𝜌 = (𝑟1 /𝑟2 )3 and α is polarizability.
The effect of decrease in 𝜀r′ with filler loading and size can not be explained,
but, it can be explained in terms of increase in 𝜀r′ with filler loading and size. The
𝜀r′ reduction effect with low nano filler loading is discussed in Chapter 4. In the
case of Eqn. 5.2, the α increase with increase nano filler size, the relation between
polarizability and 𝜀r′ shows follows:
𝜀r′ = 𝜀0 + 𝜒
where 𝜀r′ is permittivity [F/m], 𝜀0 is space permittivity [F/m] and 𝜒 is

polarizability.
Figure 5.23 shows calculated value of 𝜀r′ for epoxy/silica nanocomposite by

random arranged model as following Eqn. 5.3.
𝑉𝑝 𝑉𝑝
𝜀a𝑘 = ( ) ∗ 𝜀p𝑘 + (1 − 𝑘
) ∗ 𝜀m (5.3)
100 100
where 𝜀a𝑘 is relative permittivity of epoxy/silica nanocomposite, 𝜀p𝑘 is relative

permittivity of nano filler (𝜀r′ : 5.5), 𝜀m
𝑘
is relative permittivity of matrix (𝜀r′ : 3.46)
and k is parallel circuit (k=1), series circuit (k=-1) and random arranged (k=1/3).
The results revealed that the calculated value by random arranged model is good
agreement with experimental measured results and the 𝜀r′ increase with increasing
nano silica filler loading. On the other hand, low filler loading firstly decreases
different from calculated results. This reduction part with low filler loading
considered nanoparticle effect on 𝜀r′ as mentioned in Chapter 4.
Figure 5.24 shows E-d characteristics of small gap between high voltage
electrode and composites surface. The present dielectric property results reveal
that the relative permittivity of low filler loading and small size particle decreased,
reducing the electric field concentration between oil and solid interface as can be
seen in Fig. 5.24. Therefore, the flashover performance in composite insulation
system is superior at low filler loading and small particle size in terms of 𝜀r′ , as
shown in Fig. 5.15 and 5.16. But there is need to explain on detailed mechanism
for influence according to different particle size.

106
Figure 5.23 Calculated value of relative permittivity for epoxy/silica

nanocomposite with different silica filler loading (relative permittivity of matrix
and filler = 3.46 and 5.5)
Figure 5.23 E-d characteristics of each sample with different nano silica filler
loading

107
Second possible discussion is that nano silica/epoxy composites may suppress

electron emission contributing to discharge initiation from the surface layer of
nanocomposites, resulting in the decrease of Pd at all the Vi. Li shengtao et al.,
conducted thermally stimulated current (TSC) and pulsed electroacoustic method
(PEA) measurements in vacuum condition. They indicate that deep electron trap
level with increase in its density is introduced by nanofillers, leading to decrease
in the charge carrier mobility as a result of enhancement of flashover voltage [84,
113]. Yu Chen et al., investigated the vacuum surface flashover characteristics of
nano Al2O3/epoxy composite [114, 115]. The experimental results were explained
in terms of different effects of shallow and deep traps. To the vacuum flashover
process; i.e. deep electron traps can restrain the emission of internal secondary
electrons inside the surface layer of material, and the surface flashover voltage can
be possibly increased by enhancing the density of deeper traps.
With the above considerations taken into account, let us propose a surface
flashover model in the epoxy/silica nanocomposites as illustrated in Fig. 5.25.
Ionized free electron emitted from mineral oil under high electric field, besides
free electrons are emitted from surface layer of epoxy/silica nanocomposites, and
newly generated electrons can play a role as the seed of new electron avalanches.
Then, the discharge is more likely to incept and creep along the interface between
oil and PB. The epoxy/silica nanocomposites suppresses the electron emission
from surface layer and thus extraction of electrons from the surface of
nanocomposites on the discharge inception and propagation process.
Figure 5.25 Surface flashover conceptual model in mineral oil

108
5.3 SUMMARY
Chapter 5 focused on surface flashover inception and propagation of the

silica/epoxy nanocomposite with different silica filler loading and particle size to
verify the nano silica effect in mineral oil under positive standard lightning
impulse voltage.
The experimental results are summarized as follows:
1. Creepage discharge inception process is influenced by nanoparticle filler

loading in base epoxy resin, the nano filler content of 1 wt% produces the
best insulation performance in terms of discharge occurrence probability.
2. Creepage discharge inception is also affected by nano silica size, the nano
filler size of 22 nm, 3 wt% produces the best insulation performance in
mineral oil
3. The nano filler loading and filler size gives does not give appreciable effect
on discharge propagation length under the present condition.
4. To explain our experimental results, we suggested two possibly considered
mechanism on the effect of the coating method on Pd as follows:
1) Relative permittivity reduction effect by introducing nano size silica

particle in base epoxy
Above two mechanisms are generally considered in surface discharge and

surface flashover phenomenon in SF6 gas, further work is needed on creepage
discharge behavior and electrical properties such as streamer velocity, surface
conductivity and resistivity, respectively.

109

110
Chapter 6: Creepage Discharge Modelling Approach
CHAPTER 6
Creepage Discharge Modelling Approach
6.1 INTRODUCTION
An empirical study of creepage discharge behavior at the oil-pressboard

interface using a needle-bar electrode experiment has been presented in Chapters
4 and 5.
A lot of researchers have investigated to understand the creepage discharge
behavior in one insulation system such as oil, gas and air condition using
COMSOL Multiphysics, because it is very difficult to analysis in composite
insulation system. On the other hand, some researchers considered the solid
insulation as ideal case, assuming that the conduction current in solid insulation
for zero space charge. Besides COMSOL Multiphysics have a lot of limitations
for electric field analysis, although they efforts to fit simulation results to
experimental ones. Thus, these problems still remain challenge.
For this reason, we investigated the creepage discharge behavior in
oil/pressboard composite insulation at 5 mm short electrode gap. The composite
insulation model is developed by considering the effective barrier height by
Poole-Frenkel model and relative permittivity of pressboard. Newly considered
parameters are as follows:
1. The introduction of Poole-Frenkel conduction model for solid insulation

2. The introduction of effective barrier height for solid insulation
3. The introduction of dissociation for liquid insulation
The simulation results on Poole-Frenkel model, Effective barrier height and

permittivity of solid insulation are discussed in Section 6.4.2-6.4.3.

111
Chapter 6: Creepage Discharge Modelling Aproach
6.2 GOVERNING EQUATIONS IN OIL
6.2.1 HYDRODYNAMIC DIFFUSION-DRIFT MODEL
The general expression of governing equations that contain the physics to

model streamer development are based on the hydrodynamic diffusion-drift model
for positive ion (𝜌+ ), negative ion (𝜌− ) and electron (𝜌e ) charge densities which
are coupled through Gauss’s Law given in Eqn 3.5. The thermal diffusion
equation (3.6) is included to model temperature variations by the electric field and
Joule heating during the creepage discharge as follows [17, 32, 33, 35-38, 65-67,
116-120]:
−∇ ∙ (𝜀𝑟 ∇V) = 𝜌+ + 𝜌− + 𝜌𝑒 (6.1)
∂𝜌+ 𝜌+ 𝜌− 𝑅± 𝜌+ 𝜌− 𝑅+,𝑒
+ ∇ ∙ (𝜌+ 𝜇+ 𝐸⃗⃗ ) = 𝐺𝐼 (|𝐸|) + 𝐺𝐷 (|𝐸|, 𝑇) + + (6.2)
∂t 𝑞 𝑞
∂𝜌− 𝜌+ 𝜌− 𝑅± 𝜌𝑒
+ ∇ ∙ (𝜌− 𝜇− 𝐸⃗⃗ ) = −𝐺𝐷 (|𝐸|, 𝑇) − + (6.3)
∂t 𝑞 𝜏𝑎
∂𝜌𝑒 𝜌+ 𝜌− 𝑅+,𝑒 𝜌𝑒
+ ∇ ∙ (−𝐷𝑒 ∇𝜌𝑒 𝜌𝑒 𝜇𝑒 𝐸⃗⃗ ) = −𝐺𝐼 (|𝐸|) − − (6.4)
∂t 𝑞 𝜏𝑎
∂T 1
+ 𝐯 ∙ ∇𝑇 = (𝑘 ∇2 𝑇 + 𝐸⃗⃗ ∙ 𝐽⃗) (6.5)
∂t 𝜌𝑙 𝑐𝑣 𝑇
In also (6.1) - (6.5), the parameters 𝜇+ , 𝜇− and 𝜇𝑒 are the mobilities of the positive
ions (1 × 10−9 m2 . V ∙ s [17, 65]), negative ions (1 × 10−9 m2 . V ∙ s [17, 65]) and
electrons (1 × 10−4 m2 . V ∙ s [17, 66, 67]), respectively, 𝜌+ , 𝜌−, 𝜌e are positive
ions, negative ions and electrons, respectively; 𝜀r is the relative permittivity of a
material; E is the total electric field intensity due to the applied voltage and space
charge distributions; T is the temperature; q is the electronic charge; 𝑘T thermal
conductivity (0.13 W/m∙K); 𝑐v specific heat (1.7 × 103 J/kg ∙ K); 𝐽+ , 𝐽− , 𝐽e are
current density of positive ion, negative ion and electron, respectively; 𝐺I (|𝐸|) is
the molecular ionization; 𝐺D (|𝐸|, 𝑇) is the ionic dissociation source term; R is the
recombination rate; 𝜏a is the electron attachment time constant; v is oil’s velocity.

112
The governing equations for solid insulation are composed of Gauss’ law, the
conduction current equation for zero space charge.
−∇ ∙ (𝜀𝑟 ∇𝐸⃗⃗ ) = 0 (6.6)
∇ ∙ 𝐽⃗ = 0 (6.7)
where 𝜀r is relative permittivity of solid insulation; 𝐸⃗⃗ and 𝐽⃗ are electric field
within a solid insulation and current density of the solid insulation that is
following equation (6.8) using Ohm’s law.
𝐽⃗ = 𝜎𝐸⃗⃗ (6.8)
where 𝜎 is the electrical conductivity of the solid insulation, In this simulation,

𝜎 is negligible such that 𝜎 = 0 , i.e. the displacement current density is only
considered.
Ohm’s integration method is a calculating the total current through the surface
integral of the physical quantity across the discharge space. This represents the
total current flowing into integration space as following Eqn. 6.9.
⃗⃗
𝜕𝐷
I = ∫ (𝐽⃗⃗⃗𝑐 + ) ∙ 𝑑𝑎⃗ (6.9)
𝑆 𝜕𝑡
⃗⃗
where 𝐷 is applied voltage and space charge distribution by total flux density and 𝑎⃗
is the surface integration on the electrode or discharge space.
6.2.2 CHARGE EMISSION FROM ELECTRODE AND

SOLID INSULATION SURFACE
To consider the electron emission for the surface of solid insulation and for
ground electrode and back electrode, the charge generation in PB bulk is assumed
to be given by a dominant process among candidates of Poole-Frenke (P-F)
emission from PB surface to dielectric liquid, and field emission from the metal
electrode to dielectric liquid expressed in equations as follows:

113
2 3
𝑒 3 |𝐸⃗⃗ | 8𝜋√2𝑚ΦD 2
𝐽𝐹𝐸 = exp (− ) (6.10)
8𝜋ℎΦD 3ℎ𝑒|𝐸⃗⃗ |
𝑒ΦD
𝐽𝑇𝐻 = 𝐴𝑇 2 𝑒𝑥𝑝 (− ) (6.11)
𝑘𝑇
𝛽𝑃𝐹 √𝐸 − ∅
𝐽𝑃𝐹 = 𝐴𝐸 2 𝑒𝑥𝑝 ( ) (6.12)
𝑘𝑇
4𝜋𝑞𝑘 2 𝑚∗
𝐴=
ℎ3
where 𝐽FE is emitted current density due to the influence of the applied electric
field; 𝐽TH is emitted current density due to the influence of thermal; 𝐽PF is emitted
current density from solid insulation bulk; 𝑒 is the magnitude of electronic charge
( 1.602 × 10−9 𝐶 ); |𝐸⃗⃗ | is the local electric field; h is Planck’s constant; k is
Boltzmann’s constant; m is free electron mass; A = 1.2 × 106 A/(m2 K 2 ) is
Richardson’s constant; ΦD is effective barrier height.
We confirmed that the relation between each current density (𝐽FE , 𝐽TH , 𝐽PF ) and
electric field. The result revealed that the 𝐽FE increases with increasing electric
field linearly. Whereas 𝐽PF shows non-linear relation and high current density than
that of 𝐽FE after 70 kV/mm as shown in Fig 6.1. On the other hand, current density
by temperature variation showed no significant change (about 1E-51 A/m2 ). Thus
we considered 𝐽FE and 𝐽PF from electrode and PB surface, respectively.
Figure 6.1 Relation between each current density (𝐽FE , 𝐽PF ) and electric field

114
ΦD = 𝜙𝑤 − 𝜒, when 𝜙𝑤 is the work function of the metal electrode and 𝜒 is the

electron affinity of the dielectric material in simulation, it is assumed that ΦD
formed at the interface between the ground electrode and oil and between the
pressboard and oil is 4.5 eV and 4.91 eV, respectively. Note that the work
function of tungsten is 4.5 eV, which is assumed to be barrier height in calculation
of JFE into the oil. On the other hand, quantum chemical calculation of molecule
consists of pressboard provides that the vacuum level is located by 1.34 eV below
the lowest unoccupied molecular orbital (LUMO), and the work function is 4.91
eV from Fermi level to vacuum level as shown in Fig. 6.2, 6.3 and 6.4. The reason
for taking ΦD as 4.91 eV arises from the assumption that the carrier located at the
Fermi level can be free into the oil by exceed the energy equivalent to the barrier,
i.e. the energy difference between the vacuum level and Fermi level.
Here, it is assumed that the electron emission by electric field emission and
thermal electron emission is used as boundary conditions. It is also assumed that
the electrode surface and liquid insulation are as ideal surface condition and pure
liquid except for impurities and bubbles in liquid, respectively.
Figure 6.2 Molecule structure of cellulose paper [C6 H10 O5 ]

115
Figure 6.3 Quantum chemical calculation of molecule consists of PB

116
Figure 6.4 Conceptual model of quantum chemical calculation of PB
6.2.3 LIGHTNING IMPULSE VOLTAGE
To simulate the creepage discharge behavior under standard lightning impulse

voltage, we used a source of lightning voltage with 1.2/50 𝜇𝑠 as shown in Fig.
6.5. A convenient analytical representation of the pulse waveform is the double-
exponential expression used here as
V(t) = 𝑉1 [Exp(−𝑡/𝜏2 ) − Exp(−𝑡/𝜏1 )] (6.12)
with 𝜏2 = 1.443𝑇t , 𝜏2 = 0.2𝑇f , 𝑉1 = 𝑉𝑚 exp(𝑇f /1.443𝑇𝑡 )
where 𝑇f and 𝑇t are virtual front time, virtual time of half voltage

117
Figure 6.5 Waveform of standard lightning impulse voltage for numerical

simulation. The rising time, 𝑇𝑓 is 1.2 𝜇𝑠 and the 50% of falling time, 𝑇𝑡 is 50 𝜇𝑠.
(amended from [120])
6.3 NEEDLE-BAR GEOMETRY
Figures 6.6 and 6.7 show needle-bar electrode system and needle-bar geometry
in COMSOL Multiphysics, respectively. The distance between the needle
electrode and bar is 5 mm in simulation. The radius of curvature of needle
electrode is 10 μm. The applied voltage to the needle electrode is 25 kV standard
lightning impulse voltage. Streamer development as well as the related physics
parameters for the needle-bar electrode model was solved using equations (6.1) -
(6.11). Experiment was also performed under the same conditions with the
simulation model.
Many studies of streamers in transformer oil are conducted under a non-
uniform geometry such as needle-sphere or needle plane electrode geometries [17,
62, 118]. The non-uniform geometries generate a highly divergent electric field
distribution, thereby localizing all high field activity, such as streamer initiation,
near the needle electrode tip [17]. Application of the 100 kV impulse voltage to
the needle electrode creates a non-uniform electric field distribution with a large
field enhancement near the sharp needle tip, causing ionization to initiate
streamers.

118
Figure 6.6 Needle-bar electrode system with PB immersed in oil
Figure 6.7 Computer- aided design representation of the needle-bar electrode

geometry used for streamer simulation and conceptual model of electron emission
from ground and PB bulk

119
6.4 SIMULATION RESULTS
6.4.1 CREEPAGE DISCHARGE UNDER HIGH ELECTRIC

FIELD
Figure 6.8 shows result of numerical model to simulate the positive streamer
propagation at mineral oil and pressboard interface at 100 kV. The characteristics
of the electric field dynamics and the charge density dynamics is dependent
molecular ionization model correlate well. As shown in Fig. 6.9, the full
ionization model simulation results show the development of an electric field
wave, which propagates from the needle electrode’s tip towards to the bar
electrode. The initial process of analysis, high electric field region is formed
around needle electrode according to electric field distribution, and then creepage
discharge propagate along oil/pressboard interface at electric field level 5.5 × 108
V/m. The results on electric potential indicate that electric potential drops in the
electric field wave’s tail (the streamer channel). The potential drop in the electric
field wave’s tail given by these results ranges from 3 to 5 × 107 [V/m]. Potential
drops of this level are an order of magnitude greater than 2 to 4 × 106 [V/m]
range for the average drop in the tail of a streamer, which is stated as being
appropriate for the potential drop in a streamer in transformer oil [123, 124].
Figure 6.10 shows space charge density and electron density. The space charge
peak value is the same with electron density and electric field peak value as a
function of time, meaning that the space charge is concentrated, where the peak of
the electric field is formed. The free electron move to the needle electrode at a
high speed than ions, and the ion charge carriers with relatively slow speed remain
in space and then leading to ionization, dissociation, recombination and electron
attachment. In this reason, the level of electron free charge carrier density
decreases as a function of time as shown in Fig. 6.10. Figures 6.11 and 6.12 show
temperature variation at the needle electrode tip by discharge energy and
temperature distribution as a function of time step, meaning that the movement of
the free charge generated by field ionization due to the electric field contributes to
Joule heating that increases the temperature. We found that the high temperature
distribution is formed at creepage discharge streamer head and needle electrode
tip as shown in Fig. 6.12. These results are consistent with other reference works
[17, 40, 62]. Figs. 6.13 (a) and (b) show discharge from in simulation result and
experimental results. Discharge current for simulation is current derived from the
integration of current density during the creepage discharge propagation. The
result revealed that the peak value of simulation calculated discharge current is
around 0.68 A, which is greater than the peak of experimental measured discharge
current about 0.43 A.

120
Figure 6.8 Electric field wave propagation as a function of time at 25 kV

121
Figure 6.9 Electric field distribution and electric potential as a function of

time steps
Figure 6.10 Electron and space charge density distribution as a function

of time steps

122
Figure 6.11 Temperature rise as a function of time at the needle tip
Figure 6.12Temerature distribution as a function of time step

123
(a) Simulation calculated discharge current by Ohm’s integration method
(b) Experimental measured discharge current

Figure 6.13 Discharge current derived from the integration of current density
during the creepage discharge propagation on the surface PB at 25 kV

124
6.4.2 POOLE-FRENKEL MODEL FOR SOLID

INSULATION DEPENDENT ON EFFECTIVE BARRIER
HEIGHT
The experiment results revealed that the BDV is 25 kV at 5 mm gap distance,

BD time is 7 μs. The numerical results on breakdown time (𝑡bd ) and discharge
velocity (𝑣) with considering P-F model are compared without consideration of
the surface charge generation. The results revealed that the P-F model shows fast
streamer propagation (𝑣: 3.28 𝜇𝑠, 𝑡bd : 1.52 𝜇𝑠) with ΦD 4.9 than the case without
consideration of P-F model (𝑣: 3.12 𝜇𝑠, 𝑡bd : 1.6 𝜇𝑠) and BDV of both model is 25
kV.
Lei Zhou and Michael R. Zachariah et al., [124] reported substantial changes of
work function with nanoparticle size dependent. They results show the work
function significant decreasing trend from 4.31 to 4.26 eV as mobility size
increases from 15 to 80 nm. Besides, work function of agglomeration nanoparticle
also decreasing with increasing aggregate of nanoparticles. Low work function is
able to inject charges into the conduction band. On the contrary, higher work
function does not exhibit efficient charge transfer.
J. K. Nelson et al., [115] reported the above similar results by silica/XLPE
nanocomposites, as a result of a change in the work function and activation
energies of nanocomposites by nano size silica and silica surface modification.
Above viewpoint, the creepage discharge characteristics are investigated
dependent on effective barrier height (ΦD ) of solid insulation. The insulation of
ΦD is given 4.9, 5.9, 8.9 and 12 eV assuming PB in simulation and permittivity of
each case is same.
Figure 6.14 shows the discharge velocity and breakdown time dependent on
ΦD of solid insulation at 100 kV. The result revealed that the discharge
propagation velocity and BDV time are influenced by ΦD of solid insulation from
in our simulation results. The breakdown time increases with increasing ΦD of
solid insulation, besides streamer velocity also decreased. Low ΦD has been
shown to dramatically affect streamer propagation and leading to breakdown.

125
Figure 6.14 Discharge velocity and breakdown time dependent on ΦD of solid

insulation under 100 kV lighting impulse voltage
6.4.3 CREEPAGE DISCHARGE VELOCITY AND BDV

DEPENDENT ON PERMITTIVITY EFFECT
Figure 6.15 shows creepage discharge propagation velocity and BDV to

confirm the effect of creepage discharge characteristic depending on permittivity
at oil/pressboard interface. The results revealed that the breakdown time decreases
with increasing permittivity of solid insulation, besides streamer velocity also
increased. The effect of permittivity differences between oil and solid insulation
on streamer propagation has been well documented in the literature extensively.
For example, researchers report that low permittivity insulating solid such as
polyethylene and polypropylene, which have relative permittivity of ~2.3 almost
matching that of oil, result in high flashover voltages for impulse and AC voltage
[126]. The greater the relative permittivity of a material, the greater the
polarization charge that develops in a region where there is a local imbalance of
dipoles. This polarization charge is a source of electric field that affects
electrodynamics in a system such as a composite liquid-solid insulation structure

126
[17, 127]. High permittivity has been shown to dramatically affect streamer
propagation and breakdown time, such that streamers in the oil are attracted
towards the pressboard surface [17, 121, 128]. The result of discharge velocity is
accords with reference papers [46, 128].
Figure 6.15 Creepage discharge propagation velocity and BDV with different
𝜀r′ [127]
6.5 SUMMARY
A composite insulation model has been developed using an FEA software

package to verify the effect of nano silica with considering Poole-Frenkel model
for solid insulation, besides we considered the effective barrier height and relative
permittivity variation of solid insulation at the oil-pressboard interface under
positive standard lightning impulse voltage.
The results of this research work investigated the possibility of the numerical
and experimental approaches used to modeling creepage discharge in dielectric
liquid with a solid insulator with the needle-bar electrode configuration. A fully
coupled finite element model was developed. The numerical results were found to

127
be in good agreement with the experimental results in terms of BD time and BDV.
The BD time of simulated creepage discharge shows similar order with
experimental results. Further work is needed on physical parameters such as
charge generation and conduction mechanism of each dielectric as well as PB
parameters such as surface roughness.

128

129
Chapter 7: Thesis Conclusions and Further Work
CHAPTER 7
CONCLUSIONS AND FURTHER WORK
7.1 CONCLUSIONS
To study the creepage discharge behaviour in composite insulation system, we

considered the experiment and simulation analysis. In experiment, a creepage
discharge using a rod-plane electrode and needle-bar electrode configuration has
been developed to study the creepage discharge characteristics compared with
conventional pressboard at the oil/pressboard interface. We have focused on
surface modifications of the insulating paper by solid layer coating method to
improve the insulation property. Five different pressboard samples, i.e. neat PB,
dry PB, epoxy coated PB, Teflon coated PB and epoxy/silica nanocomposite
coated PB have been tested to verify the nano silica effect on discharge inception
and discharge propagation with different surface condition. The research work
reveals that the new coating method on the PB surface can indeed affect the
creepage discharge behaviour as well as dielectric properties. We suggested three
possibly considered mechanism to explain the experimental results on the effect of
the coating method on improvement of Pd and ld.
1) Drying of water content in neat PB and decrease in 𝜀𝑟′ by silica/epoxy

3) Silica/epoxy composites coating suppressing the electron emission from the
surface of PB
The dominant mechanism to explain the results on Pd is silica/epoxy

composites coating effect (3), but the mechanisms (1) and (2) are enough to
explain the results on ld than that of (3).
To verify the effect of nano silica particles on discharge inception and

propagation, creepage discharge behaviour have been tested with different filler
loading and nano particle size. The results suggest that creepage discharge

130
inception process is influenced by appropriate nanoparticle filler loading and filler

size in base epoxy resin, the low silica filler loading and small size produces the
best insulation performance. To explain our experimental results, we suggested
two possibly considered mechanism on the effect of the coating method on Pd as
follows:
1) Relative permittivity reduction effect by introducing nano size silica particle

Above two mechanisms all contributed to the discharge inception process,

further work is needed on creepage discharge behavior and electrical properties
such as streamer velocity, surface conductivity and resistivity, respectively.
The work presented in this thesis has also made a significant contribution to the
development of a computational simulation model of surface discharge at the
oil/pressboard interface. A 2-D axial symmetry model has been developed using
COMSOL Multiphysics, an FEA software package. In general, the electric field
dependent molecular ionization mechanism plays an important role in the
discharge inception and propagation at the oil-pressboard interface. On the other
hand, the composite insulation model is developed by considering Poole-Frenkel
model for solid insulation. The numerical results were found to be in good
agreement with the experimental results in terms of BD time and BDV. The BD
time of simulated creepage discharge shows similar order with experimental
results.
7.2 FURTHER WORK

For further work, experiments on the nano silica effect of creepage discharge
compare with conventional pressboard under AC voltage, switching impulse
waveforms as well as negative impulse voltage should be performed. To
investigate the applicability of pressboard by epoxy/silica nanocomposite in oil
filled transformer, long term life test and degradation behaviour at oil/pressboard
composite insulation system should be performed.
The experiment of the epoxy/silica nanocomposite plate under different

condition such as SF6 gas and air condition to compare with mineral oil results

131
and to verify the nanoparticle effect under AC voltage or/and impulse polarity
should be performed.
Further computational work to study the creepage discharge behavior due to

negative streamer should also be considered. Besides, the comparison with
experiment and simulation results should be performed. But there is needed some
parameters such as electron, positive and negative ion mobility by electric field
or/and temperature, besides further work is needed on physical parameters of
pressboard such as surface roughness of conventional pressboard should be
considered.

132
Appendix A
APPENDIX A
List of Publications
A.1 Refereed Conference Papers
The following papers have been peer reviewed and presented at International
Conferences:
(No. 6 and 7 are scheduled to present a paper at International Conference on
Dielectric Liquids at Manchester Univ. in 2017)
1. Kyunghoon Jang, Shigeyoshi Yoshida, Masahiro Kozako and Masayuki Hikita,

“Effect of Thin Solid Layer Coating on Creepage Discharge Characteristics in
Oil/Pressboard Composite Insulation System”, Proceeding of Electrical Insulation
and Dielectric Phenomena (CEIDP), IEEE Conference (4-pages), pp. 832-835,
2015
2. Kyunghoon Jang, Takao Akahoshi, Masahiro Kozako and Masayuki Hikita,

“Impulse Creepage Discharge Properties of Solid Coated Pressboard/Oil
Composite Insulation with Rod-Plane Electrode System”, Proceeding of Korea-
Japan Joint Symposium, (4-pages), pp.11-14, 2015

“Nano SiO2/Epoxy Coating Effect on Creepage Discharge Characteristics in
Oil/Pressboard Composite Insulation System”, International Conference on
Dielectric (ICD), (4-pages), Vol. 1, pp.394-397, 2016
4. Takao Akahoshi, Kyunghoon Jang, Masahiro Kozako and Masayuki Hikita,

“Effect of Discharge Trace Formation on Creepage Discharge Characteristics in
Oil/Pressboard Composite Insulation system under Impulse Voltage Application”
International Conference on Dielectric (ICD), (4-pages), Vol. 1, pp.342-345,
2016

“Dependence of Silica Nano Filler added to Epoxy Resin on Creepage Discharge
Properties in Mineral Oil”, International Work on High Voltage Engineering
IWHV, (4-pages), pp.73-76, 2016

133
Appendix A
6. Masaki Kurachi, Kyunghoon Jang, Takao Akahoshi, Masahiro Kozako and

Masayuki Hikita, “Dielectric Properties and Creepage Discharge of Epoxy/silica
Nanocomposite in Mineral oil”, International Conference on Dielectric Liquids,
June, 2017
7. Soichiro Kainaga, Shigeyoshi Yoshida, Takahiro Umemoto, Manabu

Yoshimura and Takao Tsurimoto (Mitsubishi Electric Corporation), Takao
Akahoshi, Kyunghoon Jang, Masaki Kurachi, Masahiro Kozako and Masayuki
Hikita (Kyushu Institute of Technology), “Effect of Discharge Trace on Creeping
Discharge Characteristics in Oil/Pressboard Composite Insulation System under
Impulse Voltage Application”, International Conference on Dielectric Liquids,
June, 2017
8. Choi Jin-Hyun, Kim Su-Hun, Jang Kyunghoon, Hikita Masayuki, Lee Se-Hee,
“Finite Element Analysis for Surface Discharge Due to Interfacial Polarization at
the Oil-Nanocomposite Interface”, 21st International Conference on the
Computation of Electromagnetic Fields, June 18-22, 2017
A.2 Peer Reviewed Journal Papers

Composite Insulation with Rod-Plane Electrode System”, IEEJ Transaction on
Fundamentals and Materials (A), 2017
2. Takao Akahoshi, Kyunghoon Jang, Masahiro Kozako and Masayuki Hikita,

“Effect of White discharge Trace on Impulse Creeping Discharge Characteristics
in Oil/Pressboard Composite Insulation System”, IEEJ Transaction on
Fundamentals and Materials (A), 2017
A.3 Refereed Domestic Conference Papers
1. Shigeyoshi Yoshida, Kyunghoon Jang, Masahiro Kozako, Masayuki Hikita,

“Lightning Impulse Creepage Discharge Characteristics in Oil with Presence of
Pressboard Barrier”, Kyushu branch annual conference, pp.4, 2014

134
Appendix A
2. Kyunghoon Jang, Shigeyoshi Yoshida, Masahiro Kozako and Masayuki Hikita,

“Effect of Epoxy Resin Coating on Creepage Discharge Properties of
Oil/Pressboard Composite Insulation System”, Domestic Annual Conference,
no.1-138, 2015
3. Takao Akahoshi, Kyunghoon Jang, Masahiro Kozako, Masayuki Hikita,

“Characteristics of Creepage Discharge Propagation after Formation of White
Discharge Trace under Impulse Voltage Application in Oil/Pressboard Composite
Insulation System”, The Institute of Electrical Engineers of Japan, pp.234-239,
2015

Composite Insulation with Needle-Bar Electrode System”, Domestic Annual
Conference, no.1-170, 2016
5. Takao Akahoshi, Kyunghoon Jang, Masahiro Kozako, Masayuki Hikita, “State

Change of White discharge Trace by Pressure in Oil/Pressboard Composite
Insulation System”, Domestic Annual Conference, no.1-169, 2016
6. Takao Akahoshi, Kyunghoon Jang, Masahiro Kozako, Masayuki Hikita,

“Effect of White Discharge Trace on Impulse Creeping Discharge Characteristics
in Oil/Pressboard Composite Insulation system”, Fundamentals and Materials
Society IEE Japan, pp.234-239, no.5-E-a2-3, 2016
7. Kyunghoon Jang, Takao Akahoshi, Masahiro Kozako and Masayuki Hikita”,

Creepage Discharge and Dielectric Characteristics of Nano Silica/Epoxy
Composite Coated Pressboard in Mineral Oil”, Annual Conference in Korea,
pp.49-50, 2016
8. Jin-Hyung Choi, Kyunghoon Jang, Takao Akahoshi, Masayuki Hikita, Se-Hee

Lee, “Finite Element Analysis using Surface Charge on Liquid-Solid Interface in
Insulating Liquid”, Annual Conference in Korea , pp.118-119, 2016
8. Takao Akahoshi, Masaki Kurachi, Kyunghoon Jang, Masahiro Kozako,

Masayuki Hikita, “Characteristics of discharge on Oil Gap Subdivided by Barrier
PBs in the Oil.PB Composite Insulation System under Lightning Impulse Voltage
Application”, The Institute of Electrical Engineers of Japan, 2016
9. Jang Kyunghoon, Takao Akahoshi, Masaki Kurachi, Masahiro Kozako,

Masayuki Hikita, “Numerical Model Analysis on Creepage Discharge Behavior at
the Oil/Pressboard Interface”, Domestic Annual Conference, 2017

135
Appendix A
10. Masaki Kurachi, Kyunghoon Jang, Takao Akahoshi, Masahiro Kozako,

Masayuki Hikita, “Impulse Creepage Discharge Inception Voltage of Epoxy
Silica Nanocomposite in Mineral Oil” Domestic Annual Conference, 2017
11. Takao Akahoshi, Masaki Kurachi, Kyunghoon Jang, Masahiro Kozako,

Masayuki Hikita, “Impulse Creeping Discharge Characteristics in Oil/Pressboard
Composite Insulation system after Formation of the White Trace”, Domestic
Annual Conference, 2017
A.4 Patent
1. “沿面放電の発生を抑制する方法” with Nissan Chemical Industries., LTD
Number:10778 Patent application: 2016-132011 (Proof)

Submission date:2016.07.01
【Patent certificate】
【Document name】
【Number】10778
【Recipient】Commissioner of the patent and trademark office
【International patent classification】C01B 33/145
【Inventor】
【Address】千葉県袖ヶ浦市北袖１１番１日産化学工業株式会社材料科学
研究所内
【Name】末村尚彦
【Inventor】
【Address】千葉県袖ヶ浦市北袖１１番１日産化学工業株式会社材料科学
研究所内
【Name】石田智久
【Inventor】
【Address】福岡県北九州市戸畑区仙水町１番１号国立大学法人九州工業
大学内
【Name】小迫雅裕
【Inventor】
大学内
【Name】匹田政幸
【Inventor】

136
Appendix A
大学内
【Name】チャンギョンフン
【Inventor】
大学内
【Name】赤星卓勇
【An applicant for a patent】
【Identification number】000003986
【The name or title】日産化学工業株式会社
【An applicant for a patent】
【The name or title】国立大学法人九州工業大学
【Agent】
【Patent attorney】
【The name or title】栗原浩之
【Erected agent】
【Patent attorney】
【The name or title】山▲崎▼ 雄一郎
【Patent attorney fee】
【予納台帳番号】042309
【Amount of payment】14,000円
【List of submission】
【物件名】明細書1
【物件名】特許請求の範囲1
【物件名】要約書1
【物件名】図面 1
【Comprehensive mandate number】1212778
【Document name】明細書
【Title of invention】沿面放電の発生を抑制する方法
【Technical field】
【０００１】
本発明は、高電圧（電力）機器の絶縁材料の沿面放電発生を抑制する技術
に関する。
【背景技術】
【０００２】
高電圧（電力）機器において、固体絶縁材料の表面は高電圧により沿面放
電が起きやすい箇所である。ここで表面とは、液体／固体、気体／固体、
真空／固体などの異質の界面を意味する。この場合、沿面放電は固体中で

137
Appendix A
はなく、液体、気体、真空中のいずれかで発
生する。
【０００３】
従来、樹脂表面の沿面放電を抑制する技術としては、沿面放電が生じる可
能性のある固
体絶縁材料の誘電率制御による電界緩和（この場合、液体・気体・真空の
誘電率に合わせるべく、固体の低誘電率化を行うこと）、誘電率の空間的
な傾斜分布による電界緩和、導電率の非線形性による電界緩和、表面の凹
凸による沿面距離の増加、などが報告されている。つまり、沿面放電を防
止する対策としては、電界緩和あるいは物理的な沿面距離を取る方策しか
実施されていない。
【先行技術文献】
【特許文献】
【０００４】
【特許文献１】特開２０１２－１１０２０６号公報
【特許文献２】特開２０１６－０３１８４５号公報
【発明の概要】
【発明が解決しようとする課題】
【０００５】
本発明は、固体絶縁材料である樹脂硬化体自体の電界緩和（即ち樹脂の低
誘電率化）に頼ることなく、樹脂硬化体の表面の沿面放電の発生を抑制す
る方法を確立することを課題
とする。
【課題を解決するための手段】
【０００６】
本発明は、固体絶縁材料である樹脂中にナノサイズの無機微粒子を分散さ
せたナノコンポジット絶縁材料を用いることで、樹脂表面の沿面放電の発
生を抑制するという知見に基づいて完成されたものである。
【０００７】
上記目的を達成する本発明は、ナノサイズの無機微粒子を樹脂中に分散さ
せて樹脂硬化体の表面の沿面放電の発生を抑制することを特徴とする、沿
面放電の発生を抑制する方法
にある。
【０００８】
ここで、無機微粒子としては、シリカ、アルミナ、チタニア、ジルコニア、
酸化亜鉛、酸化錫、酸化アンチモン、酸化鉄、酸化マグネシウム、又はこ
れらを主成分とする複合酸化物微粒子を用いることができる。
【０００９】
ナノサイズの粒子とは、一般的には、ナノメートル（ｎｍ）オーダーの大
きさを持つ粒子のことで、一般的には１～数百ｎｍの大きさの粒子をいう
が、本発明では、平均粒子径が１～４００ｎｍのものとする。なお、平均

138
Appendix A
粒子径は、後述するとおり、比表面積による
ものとする。
【００１０】
本発明で用いる無機微粒子は、その平均粒子径が５～２００ｎｍであるの
が好ましい。樹脂としては、エポキシ樹脂などの一般的な熱硬化性樹脂を
用いる。ナノコンポジットにおける無機微粒子の添加率は０．１～５０質
量％である。
【発明の効果】
【００１１】
本発明を用いることで、固体絶縁材料の電界緩和や物理的な沿面距離増加
を施さずとも、樹脂硬化体の固体表面の沿面放電発生電圧を向上すること
ができ、ＧＩＳ（ガス絶縁開閉装置）の絶縁スペーサや油入り変圧器の
プレスボード用コーティング樹脂等へ好適に利用さ
れうる。
【図面の簡単な説明】
【００１２】
【図１】沿面放電の測定系の概略構成を示す図である。
【発明を実施するための形態】
【００１３】
本発明に用いられる絶縁樹脂としては、例えば、エポキシ樹脂、ポリイミ
ド樹脂、ポリアミドイミド樹脂、シリコーン等の熱硬化性樹脂が挙げられ
る。また、ポリエチレン、ポリプロピレン等の熱可塑性樹脂も挙げられる。
【００１４】
本発明に用いられるエポキシ樹脂としては特に限定されるものではないが、
ビスフェノールＡ型エポキシ樹脂、ビスフェノールＦ型エポキシ樹脂、脂
環式エポキシ樹脂等が挙げられる。通常、これらエポキシ樹脂と硬化剤等
を組み合わせて配合された後、注型・熱硬化して所望の形状に成形される。
【００１５】
本発明に用いるエポキシ樹脂の硬化剤としては、フェノール樹脂、アミン
類、ポリアミド樹脂、イミダゾール類、ポリメルカプタン、酸無水物等が
挙げられる。
【００１６】
フェノール樹脂としては、例えばフェノールノボラック樹脂、クレゾール
ノボラック樹脂等が挙げられる。
【００１７】
アミン類としては、例えばピペリジン、Ｎ，Ｎ－ジメチルピペラジン、ト
リエチレンジアミン、２，４，６－トリス（ジメチルアミノメチル）フェ
ノール、ベンジルジメチルアミン、２－（ジメチルアミノメチル）フェノ
ール、ジエチレントリアミン、トリエチレンテトラミン、テトラエチレン
ペンタミン、ジエチルアミノプロピルアミン、Ｎ－アミノエチルピペラジ
ン、ジ（１－メチル－２－アミノシクロヘキシル）メタン、メンセンジア

139
Appendix A
ミン、イソフオロンジアミン、ジアミノジシクロヘキシルメタン、１，３
－ジアミノメチルシクロヘキサン、キシレンジアミン、メタフェニレンジ
アミン、ジアミノジフェニルメタン、ジアミノジフェニルスルホン等が挙
げられる。これらの中で液状であるジエチレントリアミン、トリエチレン
テトラミン、テトラエチレンペンタミン、ジエチルアミノプロピルアミン、
Ｎ－アミノエチルピペラジン、ジ（１－メチル－２－アミノシクロヘキシ
ル）メタン、メンセンジアミン、イソフオロンジアミン、ジアミノジシク
ロヘキシルメタン等は好ましく用いることができる。
【００１８】
ポリアミド樹脂としては、ダイマー酸とポリアミンの縮合により生成する
もので、分子中に一級アミンと二級アミンを有するポリアミドアミンであ
る。
【００１９】
イミダゾール類としては、２－メチルイミダゾール、２－エチル－４－メ
チルイミダゾール、１－シアノエチル－２－ウンデシルイミダゾリウムト
リメリテート、エポキシイミダゾールアダクト等が挙げられる。
【００２０】
ポリメルカプタンは、例えばポリプロピレングリコール鎖の末端にメルカ
プタン基が存在するものや、ポリエチレングリコール鎖の末端にメルカプ
タン基が存在するものであり、液状のものが好ましい。
【００２１】
本発明に用いる酸無水物としては一分子中に複数のカルボキシル基を有す
る化合物の無水物が好ましい。これらの酸無水物としては、無水フタル酸、
無水トリメリット酸、無水ピロメリット酸、無水ベンゾフェノンテトラカ
ルボン酸、エチレングリコールビストリメリテート、グリセロールトリス
トリメリテート、無水マレイン酸、テトラヒドロ無水フタル酸、メチルテ
トラヒドロ無水フタル酸、エンドメチレンテトラヒドロ無水フタル酸、メ
チルエンドメチレンテトラヒドロ無水フタル酸、メチルブテニルテトラヒ
ドロ無水フタル酸、ドデセニル無水コハク酸、ヘキサヒドロ無水フタル酸、
メチルヘキサヒドロ無水フタル酸、無水コハク酸、メチルシクロヘキセン
ジカルボン酸無水物、クロレンド酸無水物等
が挙げられる。
【００２２】
これらの中でも常温、常圧で液状であるメチルテトラヒドロ無水フタル酸、
メチルエンドメチレンテトラヒドロ無水フタル酸、メチルブテニルテトラ
ヒドロ無水フタル酸、ドデセニル無水コハク酸、メチルヘキサヒドロ無水
フタル酸が好ましい。これら液状の酸無水物は粘度が２５℃での測定で１
０ｍＰａｓ～１０００ｍＰａｓ程度である。
【００２３】
また、上記硬化物を得る際、適宜、硬化促進剤が併用されても良い。硬化
促進剤としてはトリフェニルホスフィンやトリブチルホスフィンなどの有

140
Appendix A
機リン化合物、エチルトリフェニルホスフォニウムブロマイド、メチルト
リフェニルホスホニウムリン酸ジエチル等の第４級ホスフォニウム塩、１，
８－ジアザビシクロ（５，４，０）ウンデカン－７－エン、１，８－ジア
ザビシクロ（５，４，０）ウンデカン－７－エンとオクチル酸の塩、オク
チル酸亜鉛、テトラブチルアンモニウムブロミド等の第４級アンモニウム
塩、ジメチルベンジルアミン等の３級アミン等が挙げられる。これらの硬
化促進剤は、硬化剤１質量部に
対して、０．００１～０．１質量部の割合で含有することができる。
【００２４】
本発明に用いられるナノサイズの無機微粒子としては、絶縁性や分散性の
観点から、金属酸化物微粒子が好ましい。金属酸化物微粒子の好ましい例
としては、シリカ、アルミナ、チタニア、ジルコニア、酸化亜鉛、酸化錫、
酸化アンチモン、酸化鉄、酸化マグネシウム又はこれらを主成分とする複
合酸化物微粒子が挙げられる。このうち、シリカは誘電率が４程度と低く、
熱膨張率も低いため、本発明に用いられる無機微粒子として好適に用い
られる。
【００２５】
本発明に用いられるシリカは、粒子形態のシリカであり、平均粒子径が２
００ｎｍ以下、例えば５ｎｍ～２００ｎｍ、好ましくは１０ｎｍ～１００
ｎｍであり、より樹脂の沿面放電を抑制する観点から、好ましくは２０ｎ
ｍ～８０ｎｍである。粒子径が２００ｎｍよりも大きい、または５ｎｍよ
りも小さいと、沿面放電抑制効果が小さくなる。
【００２６】
本発明に用いられるシリカ粒子の平均粒子径とは、窒素吸着法により測定
された比表面積値から算出される平均粒子径値である。
【００２７】
特に本発明では、上記平均粒子径の値を有するコロイダルシリカを好適に
使用でき、該コロイダルシリカとしては、シリカゾルを用いることができ
る。シリカゾルとしては、ケイ酸ナトリウム水溶液を原料として公知の方
法により製造される水性シリカゾル及び該水性シリカゾルの分散媒である
水を有機溶媒に置換して得られるオルガノシリカゾルを原料として使用す
る事が出来る。また、メチルシリケートやエチルシリケート等のアルコキ
シシランを、アルコール等の有機溶媒中で触媒（例えば、アンモニア、有
機アミン化合物、水酸化ナトリウム等のアルカリ触媒）の存在下において
加水分解し、縮合して得られるシリカゾル、又はそのシリカゾルを他の有
機溶媒に溶媒置換したオルガノシリカゾルも原料として用いることができ
る。
【００２８】
上述のオルガノシリカゾルにおける有機溶媒の例としては、アルコール類、
エーテル類、ケトン類、エステル類、アミド類、炭化水素類、ニトリル類
等が挙げられる。

141
Appendix A
【００２９】
アルコール類としては、メタノール、エタノール、１－プロパノール、２
－プロパノール、１－ブタノール、イソブチルアルコール、２－ブタノー
ル、エチレングリコール、グリセリン、プリピレングリコール、トリエチ
レングリコール、ポリエチレングリコール、ベンジルアルコール、１，５
－ペンタンジオール、ジアセトンアルコール等が挙げられる
。
【００３０】
エーテル類としては、ジエチルエーテル、ジブチルエーテル、テトラヒド
ロフラン、ジオキサン、エチレングリコールモノメチルエーテル、エチレ
ングリコールモノプロピルエーテル、プロピレングリコールモノメチルエ
ーテル、プロピレングリコールモノエチルエーテル、ジエチレングリコー
ルモノメチルエーテル、ジエチレングリコールモノエチルエーテル、ジエ
チレングリコールモノブチルエーテル等が挙げられる。
【００３１】
ケトン類としては、アセトン、メチルエチルケトン、２－ペンタノン、３
－ペンタノン、メチルイソブチルケトン、２－ヘプタノン、シクロヘキサ
ノン等が挙げられる。
【００３２】
エステル類としては、ギ酸エチル、酢酸メチル、酢酸エチル、酢酸プロピ
ル、酢酸ブチル、エチレングリコールモノエチルエーテルアセタート、エ
チレングリコールモノブチルエーテルアセタート、プロピレングリコール
モノメチルエーテルアセタート等が挙げられ
る。
【００３３】
アミド類としては、アセトアミド、Ｎ，Ｎ－ジメチルホルムアミド、Ｎ，
Ｎ－ジメチルアセトアミド、Ｎ－メチルピロリドン等が挙げられる。
【００３４】
炭化水素類としては、ｎ－ヘキサン、シクロヘキサン、ベンゼン、トルエ
ン、キシレン、ソルベントナフサ、スチレン等があげられ、更にハロゲン
化炭化水素類としてはジクロロメタン、トリクロロエチレン等が挙げられ
る。
【００３５】
ニトリル類としては、アセトニトリル、グルタロニトリル、メトキシアセ
トニトリル、プロピオニトリル、ベンゾニトリル等が挙げられる。
【００３６】
上記オルガノシリカゾルの市販品の例としては、例えば商品名ＭＡ－ＳＴ
－Ｓ（メタノール分散シリカゾル、日産化学工業（株）製）、商品名ＭＴ
－ＳＴ（メタノール分散シリカゾル、日産化学工業（株）製）、商品名Ｍ
Ａ－ＳＴ－ＵＰ（メタノール分散シリカゾル、日産化学工業（株）製）、
商品名ＭＡ－ＳＴ－Ｍ（メタノール分散シリカゾル、日産化学工業（株）

142
Appendix A
製）、商品名ＭＡ－ＳＴ－Ｌ（メタノール分散シリカゾル、日産化学工業
（株）製）、商品名ＩＰＡ－ＳＴ－Ｓ（イソプロパノール分散シリカゾル、
日産化学工業（株）製）、商品名ＩＰＡ－ＳＴ（イソプロパノール分散シ
リカゾル、日産化学工業（株）製）、商品名ＩＰＡ－ＳＴ－ＵＰ（イソプ
ロパノール分散シリカゾル、日産化学工業（株）製）、商品名ＩＰＡ－Ｓ
Ｔ－Ｌ（イソプロパノール分散シリカゾル、日産化学工業（株）製）、商
品名ＩＰＡ－ＳＴ－ＺＬ（イソプロパノール分散シリカゾル、日産化学工
業（株）製）、商品名ＮＰＣ－ＳＴ－３０（ｎ－プロピルセロソルブ分散
シリカゾル、日産化学工業（株）製）、商品名ＰＧＭ－ＳＴ（１－メトキ
シ－２－プロパノール分散シリカゾル、日産化学工業（株）製）、商品名
ＤＭＡＣ－ＳＴ（ジメチルアセトアミド分散シリカゾル、日産化学工業
（株）製）、商品名ＸＢＡ－ＳＴ（キシレン・ｎ－ブタノール混合溶媒分
散シリカゾル、日産化学工業（株）製）、商品名ＥＡＣ－ＳＴ（酢酸エチ
ル分散シリカゾル、日産化学工業（株）製）、商品名ＰＭＡ－ＳＴ（プロ
ピレングリコールモノメチルエーテルアセテート分散シリカゾル、日産化
学工業（株）製）、商品名ＭＥＫ－ＳＴ（メチルエチルケトン分散シリカ
ゾル、日産化学工業（株）製）、商品名ＭＥＫ－ＳＴ－ＵＰ（メチルエチ
ルケトン分散シリカゾル、日産化学工業（株）製）、商品名ＭＥＫ－ＳＴ
－Ｌ（メチルエチルケトン分散シリカゾル、日産化学工業（株）製）及び
商品名ＭＩＢＫ－ＳＴ（メチルイソブチルケトン分散シリカゾル、日産化
学工業（株）製）等を挙げることができるが、これらに限定されない。
【００３７】
本発明に使用されるシリカ粒子の表面は疎水化処理されていても良い。
疎水化処理剤としては、シラザン、シロキサン又はアルコキシシラン及び
その部分加水分
解物若しくはその重合した２量体～５量体のオリゴマーが挙げられる。
【００３８】
シラザンとしては、例えばヘキサメチルジシラザン、及びヘキサエチルジ
シラザンが挙げられる。
【００３９】
シロキサンとしては、例えばヘキサメチルジシロキサン、１，３－ジブチ
ルテトラメチルジシロキサン、１，３－ジフェニルテトラメチルジシロキ
サン、１，３－ジビニルテトラメチルジシロキサン、ヘキサエチルジシロ
キサン及び３－グリシドキシプロピルペンタメチルジシロキサンが挙げら
れる。
【００４０】
アルコキシシランとしては、例えばトリメチルメトキシシラン、トリメチ
ルエトキシシラン、トリメチルプロポキシシラン、フェニルジメチルメト
キシシラン、クロロプロピルジメチルメトキシシラン、ジメチルジメトキ
シシラン、メチルトリメトキシシラン、テトラメトキシシラン、テトラエ
トキシシラン、テトラプロポキシシラン、テトラブトキシシラン、エチル

143
Appendix A
トリメトキシシラン、ジメチルジエトキシシラン、プロピルトリエトキシ
シラン、ｎ－ブチルトリメトキシシラン、ｎ－ヘキシルトリメトキシシラ
ン、ｎ－オクチルトリエトキシシラン、ｎ－オクチルメチルジエトキシシ
ラン、ｎ－オクタデシルトリメトキシシラン、フェニルトリメトキシシラ
ン、フェニルメチルジメトキシシラン、フェネチルトリメトキシシラン、
ドデシルトリメトキシシラン、ｎ－オクタデシルトリエトキシシラン、フ
ェニルトリメトキシシラン、ジフェニルジメトキシシラン、ビニルトリメ
トキシシラン、ビニルトリエトキシシラン、ビニルトリス（βメトキシエ
トキシ）シラン、γ－メタアクリルオキシプロピルトリメトキシシラン、
γ－アクリルオキシプロピルトリメトキシシラン、γ－（メタアクリルオ
キシプロピル）メチルジメトキシシラン、γ－メタアクリルオキシプロピ
ルメチルジエトキシシラン、γ－メタアクリルオキシプロピルトリエトキ
シシラン、β－（３，４－エポキシシクロヘキシル）エチルトリメトキシ
シラン、γ－グリシドキシプロピルトリメトキシシラン、γ－グリシドキ
シプロピルメチルジエトキシシラン、γ－グリシドキシプロピルトリエト
キシシラン、Ｎ－β（アミノエチル）γ－（アミノプロピル）メチルジメ
トキシシラン、Ｎ－β（アミノエチル）γ－（アミノプロピル）トリメト
キシシラン、Ｎ－β（アミノエチル）γ－（アミノプロピル）トリエトキ
シシラン、γ－アミノプロピルトリメトキシシラン、γ－アミノプロピル
トリエトキシシラン、Ｎ－フェニル－γ－アミノプロピルトリメトキシシ
ラン、γ－メルカプトプロピルトリメトキシシラン、３－イソシアネート
プロピルトリエトキシシラン、トリフルオロプ
ロピルトリメトキシシラン、ヘプタデカトリフルオロプロピルトリメトキ
シシラン、ｎ－デシルトリメトキシシラン、ジメトキシジエトキシシラン、
ビス（トリエトキシシリル）エタン及びヘキサエトキシジシロキサンが挙
げられる。
【００４１】
本発明における樹脂中のナノシリカの配合量としては、例えば０．１～５
０質量％であり、好ましくは0.2～30質量％であり、より樹脂の沿面放電
を抑制する観点から、好ましくは0.5～20質量％、より好ましくは1～１５
質量％である。
【００４２】
ナノシリカの配合量が５０質量％よりも多いと樹脂硬化物の誘電率が上昇
し、放電抑制効果が小さくなる。また、ナノシリカの配合量が０．１質
量％よりも少ないとシリカを添加した効果が小さくなり、樹脂の放電抑制
効果が得られない。
【００４３】
上述のオルガノシリカゾルを、例えばエポキシ樹脂と混合して脱溶媒する
ことで、ナノシリカが分散したエポキシ樹脂を得ることが出来る。この樹
脂に適宜、硬化剤を添加して、注型・加熱等によって硬化反応を行い、所
望の絶縁樹脂成型物が得られる。

144
Appendix A
【実施例】
【００４４】
（材料の準備）
ビスフェノールＡジグリシジルエーテルＪＥＲ８２８（三菱化学（株）製、
エポキシ価１８５ｇ／ｅｑ．）に、平均粒子径１２～５００ｎｍのシリカ
粒子を分散させたエポキシモノマー分散シリカゾルをシリカ濃度３０．５
質量％となるよう調整した。平均粒子径５００ｎｍのシリカ粒子には、ア
ドマファインＳＯ－Ｃ２（アドマテックス（株）製）を用いた。平均粒子
径は、シリカの３００℃乾燥粉末の比表面積を比表面積測定装置モノソー
ブ（登録商標）ＭＳ－１６（ユアサアイオニクス（株）製）を用いて測定
し、平均粒子径（ｎｍ）＝２７２０／比表面積（ｍ2／g）の換算式により
算出した。
【００４５】
【表１】
【００４６】
実施例１～実施例１０表１で準備したエポキシモノマー分散シリカゾルと、
酸無水物（リカシッドＭＨ－７００、新日本理化（株）製）、反応促進剤
（ジメチルベンジルアミン、東京化成）を表２に記載の配合比で混合しエ
ポキシ樹脂硬化用組成物を得た。得られたエポキシ樹脂硬化用組成物を注
型板（離型フィルムＰＬ＃４００（フタムラ化学（株）製）で被覆された
ガラス板３ｍｍ厚）に流し込み、７０℃で２時間、９０℃で２時間、１５
０℃で８時間の硬化条件で加熱処理を行い、エポキシ樹脂硬化体を得た。
【００４７】
【表２】

145
Appendix A
【００４８】
（沿面放電の測定方法）
沿面放電の測定系の概略構成を図１に示す。この装置は、各印加電圧時の
時間波形及び発光像を取得し、沿面放電発生確率を算出するものである。
【００４９】
試料１としては、実施例１～１０で作製したエポキシ樹脂硬化体を用いた。
試料１の厚さは、３ｍｍである。
【００５０】
図１に示すように、試料１の表面に針電極２と、平板電極３とをギャップ
長ｇ＝５０ｍｍで配置した。これらの試料１及び針電極２と平板電極３と
の電極系は、試験容器４内に配置した。試験容器４内に鉱油５を充填し、
試料１及び電極系を鉱油５内に配置した。
【００５１】
針電極２の直径は１ｍｍで、先端曲率半径は１０μｍであり、試料１の表
面に対してθ＝３０°傾けて配置した。針電極２をインパルス電圧発生装
置６に接続し、平板電極３は５０Ωの抵抗を介して接地した。針電極２に、
分圧器７を介して、正極性標準雷インパル電圧（１．２μｓ／５０μｓ）
を、３５ｋＶ～７５ｋＶまで、５ｋＶステップで、印加間隔１分間で印加
し、沿面放電の発生をイメージインテンシファイア付きのＣＣＤカメラ
８を用いて確認した。試料１の数は３枚、試験回数は各電圧で５回実施し、
計１５回の実験で沿面放電発生確率を算出した。
【００５２】
また、試料１の比誘電率は誘電率計測器を用いて評価した。
【００５３】
試料条件、試料の比誘電率の測定結果、および沿面放電試験結果を表３に
示した。表３からシリカ微粒子の充填率が増加すると比誘電率が増加する
傾向がわかる。一方で、シリカ微粒子の粒径が５００ｎｍの試料以外は放
電発生確率が低下し、放電発生電圧が増加することがわかる。また、放電
発生確率が１００％に到達する電圧も、粒径が５００ｎｍの試料以外は５
～１５ｋＶ上昇した。つまり、本提案の手法を用いることで、樹脂の比誘
電率はほぼ不変あるいは増加するにも関わらず、樹脂表面の沿面放電の発
生を抑制できるこ
とになる。
【００５４】
【表３】

146
Appendix A
平均粒子径１３０ｎｍのシリカ粒子を分散させたエポキシモノマー分散シリカゾルを
シリカ濃度３０．５質量％となるよう調整した。
【表４】
粒子径エポキシ樹脂 SiO2(%)
モノマーゾル６ 130nm ビスフェノールA型 30.5
実施例１～１０と同様の条件で、下記組成にて実施例１１及び比較例３の硬化物を作成
した。
【表５】
モノマーゾル(g) エポキシ樹脂(g) 酸無水物(g) 促進剤(g) SiO2(%)
実施例１1 モノマーゾル３ 0.81 26.3 21.7 0.3 0.5
比較例３モノマーゾル６ 5.0 23.5 21.7 0.3 3.0
上記と同様の装置を用い、沿面放電の発生を光電子増倍管９を用いて確認した。光電
子増倍管９は、信号用９ａと、ノイズ用９ｂとを用いた。
【表６】
シリカ粒子径シリカ添加率比誘電率３５ｋＶにおける放電発生確率が
[nm] [wt%] (@100 kHz) 放電発生確率[%] 100%の電圧[kV]
実施例１ 12 3 3.46 31 50
実施例２ 22 3 3.50 20 50
実施例１１ 0.5 3.35 40 45

45
実施例４ 1 3.48 30 45
実施例９ 80 3 3.53 31 45
比較例１シリカなし 0 3.47 71 40
比較例２ 500 3 3.58 71 40
比較例３ 130 3 3.59 80 45
試料条件、試料の比誘電率の測定結果、および沿面放電試験結果を表４に示した。シリ
カ微粒子の粒径が１３０ｎｍ以上の試料以外は放電発生確率が低下し、放電発生電圧が
増加することがわかる。また、放電発生確率が１００％に到達する電圧も、粒径が５０
０ｎｍ以上の試料以外は５～１０ｋＶ上昇した。つまり、本提案の手法を用いることで、

147
Appendix A
樹脂の比誘電率はほぼ不変あるいは増加するにも関わらず、樹脂表面の沿面放電の発生
を抑制できることになる。
【産業上の利用可能性】
【００５５】
本発明によって、本発明を用いることで、固体絶縁材料の電界緩和や物理
的な沿面距離増加を施さずとも、固体表面の沿面放電発生電圧を向上する
ことができ、GIS(ガス絶縁開閉装置)の絶縁スペーサや油入り変圧器のプ
レスボード用コーティング樹脂等へ好適に利用されうる。
【符号の説明】
【００５６】
１試料
２針電極
３平板電極
４試験容器
５鉱油
６インパルス電圧発生装置
７分圧器
８カメラ
９光電子増倍管
９ａ信号用
９ｂノイズ用
【書類名】特許請求の範囲
【請求項１】
ナノサイズの無機微粒子を樹脂中に分散させて樹脂硬化体の表面の沿面放
電の発生を抑制することを特徴とする、沿面放電の発生を抑制する方法
【請求項２】
前記無機微粒子がシリカであることを特徴とする、請求項１に記載の方法。
【請求項３】
前記樹脂が熱硬化性樹脂である請求項１又は２に記載の方法。
【請求項４】
前記無機微粒子の添加率が０．１～５０質量％である請求項１～３のいず
れか一項に記載の方法
【請求項５】
前記無機微粒子の平均粒子径が５～２００ｎｍである請求項１～４のいず
れか一項に記載の方法。
【書類名】要約書
【要約】
【課題】固体絶縁材料である樹脂自体の電界緩和（即ち樹脂の低誘電率化）
に頼ることなく、樹脂表面の沿面放電の発生を抑制する手法を確立するこ

148
Appendix A
と。
【解決手段】ナノサイズの無機微粒子を樹脂中に分散させて樹脂硬化体の
表面の沿面放
電の発生を抑制することを特徴とする、樹脂表面の沿面放電の発生を抑制
する方法による。
【選択図】なし
【書類名】図面
【図１】

149
Appendix A

150
Appendix B
APPENDIX B
B.1 Experiment at 5 mm Short Gap at 25 kV
Figure B1. Experimental setup at 5mm gap distance in mineral oil

151
Appendix B
Figure B2. Experimental waveform of OSC
B.2 Snapshots of Electric Field Distribution using Rod-Plane

Electrode Configuration
Figure B3. Simulation model of rod-plane electrode configuration

152
Appendix B
Figure B4. Simulation results of rod-plane electrode configuration under positive

lightning impulse voltage

153
Appendix C
APPENDIX C
COMSOL Multiphysics
C.1 How to setup FEM mesh for simulation
Figure C1 shows distribution of FEM mesh on the needle-bar electrodes. In

general, mesh is a crucial step in obtaining a reliable simulation result. Dense
mesh size is formed to produce accurate simulation results for complicated
Multiphysics finite element analysis combined with multiple systems, while
calculation needs a lot of analysis time. In some cases, calculation is not possible
due to a lack of system resources. On the other hand, if the mesh size is not
densely formed to obtain the analysis result within a short time, accurate
calculation results can’t be obtained. In such problems, the sharp geometries as
needle electrode requires a fine mesh in these regions. In addition, discharge
dynamics often take place in a small volume with respect to the entire simulation
space. A fine mesh is hence needed to resolve the solution at such region. Thus, a
proper mesh size is required to reflect the Multiphysics analysis of interested
region as shown in Fig. C1.
There are two types of mesh element that can be used in COMSOL: triangular
and quadrilateral elements. Triangular meshing was preferred since it meshes
sharp electrodes more efficiently. COMSOL also offers a tool to specify the size
of mesh elements along boundaries and subdomains ensuring better accuracy at
regions of interest. However, investigations revealed that the element type does
not affect the simulation results significantly. All the simulation results are
compared with Kyungpook University’s simulation results of Professor Se-Hee
Lee to verify our results [36, 37, 64, 114, 121].

154
Appendix C
Figure C1 Distribution of FEM mesh on the needle-bar electrode

155
References
REFERENCES
[1] Hidayat Zainuddin, “Study of Surface Discharge Behaviour at the Oil-

pressboard Interface”, Thesis for the degree of Doctor of Philosophy,
university of southampton, 2013.
[2] Forensic Science newsletter, 2005,
http://www.forensic.cc/newsletter/transformer-failures
[3] P. Przybylek, Z. Nadolny and H. Moscicka-Grzesiak, “Bubble Effect as a
Consequence of dielectric Losses in Cellulose Insulation”, IEEE
Transactions on Dielectrics and Electrical Insulation, Vol. 17, no. 3, 2010
[4] A. J. Kachler, I. Hohlein, “Aging of Cellulose at Transformer Service
Temperatures. Part 1: Influence of Type of Oil and Air on the Degree of
Polymerization of Pressboard, Dissolved Gases, and Furanic Compounds in
Oil”, Feature article, Vol. 21, Issue 2, 2005
[5] S. V. Kulkarni and S. A. Khaparde, “Transformer Engineering Design and
Practice”, CRC Press, pp. 496, 2004.
[6] V. Sokolov, Z. Berler and V. Rashkes, “Effective methods of assessment of
insulation system conditions in power transformer: a view based on practical
experience”, Proceedings of the Electrical Insulation Conference and
Electrical Manufacturing & Coil Winding Conference, USA, pp. 659-667,
papers (39), 1999
[7] Yangchun Cheng, Jinqing Wei, Chongzhi Zhao and Haoyong Song,
“Experimental Research on Creepage Discharge Between oil-impregnated
Pressboard Layers”, Proc. 17th Conference CEFC, pp. 1-10, 2009
[8] CIGRE Working Group 12.18: Life Management of Transformers,
“Guidelines for life management techniques for power transformers”, Draft
Final Report Rev. 2, 22, 2002
[9] Thomas A. Prevost and T. V. Oommen, “Cellulose Insulation in Oil-Filled
Power Transformers: Part 1-History and Development”, IEEE Electrical
Insulation Magazine, Vol. 22, Issue: 1, 2006
[10] J. A. Lapworth and A. Wilson, “Transformer Internal Over-Voltages Caused
by Remote Energisation”, IEEE PES Power Africa Conference and
Exposition, pp. 1-6, 2007
[11] DU Yue-fan, LV Yu-zhen, WANG Fo-chi, Ki Xiao-xin and Li Cheng-rong,
“Effect of TiO2 Nanoparticles on the Breakdown Strength of Transformer
Oil”, IEEE International Symposium on Electrical Insulation (ISEI),6-9
June, 2010
[12] LV Yu-zhen, Le-feng Wang, Xiao-xin Li, Yue-fan Du, Jian-quan Zhou and
Cheng-rong Li, “Experimental Investigation of Breakdown Strength of

156
References
Mineral Oil-Based Nanofluids”, IEEE International Conference on

Dielectric Liquids, pp. 1-3, 2011
[13] R. Hollertz, L. Wagberg and Claire Pitois, “Novel Cellulose Nanomaterials”,
IEEE International Conference on Liquid Dielectrics, Bled, Slovenia, pp.1-
4, 2014
[14] Rebecca Hollertz, David Ariza, Claire Pitois and Lars Wagberg, “Dielectric
Response of Kraft Paper from Fibres Modified by Silica Nanoparticles”,
Annual Report Conference on Electrical Insulation and Dielectric
Phenomena, pp.459-462, 2015
[15] Min Chen, Limin Wu, Shuxue Zhou and bo You, “A Method for the
Fabrication of Monodisperse Hollow Silica Spheres”, Advanced materials,
Vol. 18, pp. 801-806, 2006
[16] Ruijin Liao, Fuzhou Zhang, Yuan Yuan, Tuan Liu and Chao Tang,
“Preparation and Electrical Properties of Insulation Paper Composed of
SiO2 Hollow Spheres”, MDPI open access Journal, Vol. 5, pp. 2943-2951,
2012
[17] Jae-Won George Hwang, “Elucidating the Mechanisms Behind Pre-
Breakdown Phenomena in Transformer Oil Systems”, Thesis for the degree
of Doctor of Philosophy, Massachusetts Institute of technology, 2010
[18] A. Kara, O. Kalenderli and K. Mardikyan, “Effect of Dielectric Barriers to
the Electric Field of Rod-Plane Air Gap”, proceedings of the COMSOL
users Conference, 2006
[19] R. Hanaoka, T. Kohrin, T. Miyagawa and T. Nishi, “ Creepage Discharge
Characteristics over Solid-Liquid Interfaces with Grounded Side Electrode”,
International Conference on Conduction and Breakdowns in Dielectric
Liquids, pp.308-315, Vol. 9, 2002
[20] Fri Murdiya, Ryoichi Hanaoka, Hideki Akiyama and Katsunori Miyagi,
“Creeping Discharge Developing on Vegetable-Based Oil/Pressboard
Interface under AC Voltage”, IEEE Trans. Dielectr. Electr. Insul., Vol. 21,
pp.2102-2110, 2014
[21] F. Murdiya, R. Hanaoka, R. Hashi and K. Miyagi, “Properties of Creeping
Streamer Progressed Dielectric Barrier with Narrow Gap in PFAE Oil”,
Journal of Energy and Power Engineering, Vol. 7, pp.1-4, 2013
[22] Fri Murdiya, “Research on Creeping Discharge Phenomena in Insulation
Oils-Vegetable based Oils as Substitute of Mineral Oil”, Thesis for the
degree of Doctor of Philosophy, UNIVERSITY OF KANAZAWA, 2015
[23] M. J. Heathcote, J & P Transformer Book, 12th ed., Butterworth-Heinemann,
1998
[24] R. Bartnikas, “Engineering Dielectrics, Volume 3: electrical Insulating
Liquids”, American Society for Testing and Materials, 1994

157
References
[25] J. Bidlack, M. Malone and B. Russel, “Molecular Structure and Component

Integration of Secondary Cell Walls in Plants”, Proc. Okla. Acad. Sci., Vol.
72, pp.51-56, 1992
[26] P. M. Mitchinson, “Surface Tracking in the Interphase Region of Large
Trasnformer”, Ph,D. thesis, University of Southamton, 2008
[27] “The Effect of Moisture on the Breakdown Voltage of Transformer Oil”
www.vaisala.com
[28] M. C. Silva, “Characterization of three non-product materials from a
bleached eucalyptus kraft pulp mill, in view of valorizing them as a source
of cellulose fibres”, Industrial Crops and Products, Vol. 27, Issue 3,
pp.288-295, 2008
[29] T. V. Oommen and Thomas A. Prevost, “Cellulose Insulation in Oil-Filled
Power Transformers: Part 2-Maintaining Insulation Integrity and Life, IEEE
Electrical Insulation Magazine, Vol. 22, issue: 2, 2006
[30] K. Giese, “The Effects of Cellulose Insulation Quality on Electrical Intrinsic
Strength”, IEEE Electrical Insulation Magazine, Vol. 10, pp.38-42, Issue 5,
1994
[31] Y. Suzuki and M. Takagi, “Oil Impregnation in Transformer Boards
Theoretical Analysis of Changes in Impregnation Depth”, IEEE
Transactions on Electrical Insulation, Vol. EI-19, 1984
[32] H. Zainuddin, P. L. Lewin and P. M. Mitchinson, “Modeling the Inter-phase
Region of High Voltage Transformers”, IEEE International Conference on
Dielectric Liquids, pp.1-4, 2011
[33] Ho-Young Lee and Se-Hee Lee, “Numerical Analysis for Surface Discharge
on Solid Insulation in the Dielectric Liquid and Experimental Validation”,
Korea-Japan Joint Symposium, 2015
[34] F. M. Clark, “Insulating Materials for Design and Engineering Practice”,
John Wiley & Sons, Inc., 1962
[35] J. Dai, Z. D. Wang and P. Jarman, “Moisture and ageing effect on the
creepage discharge characteristics at the oil/transformer-board interface
under divergent field”, Annual report conference on electrical insulation
and dielectric phenomena (CEIDP), pp. 662-665, 2008
[36] Ho-Young Lee, Jae-Seung Jung, Hong-Kyu Kim and Se-Hee Lee,
“Numerical and Experimental Validation of Discharge Current with
Generalized Energy Method and Integral Ohm’s Law in Transformer Oil”,
OEEE Transactions on Magnetics, Vol. 50, no. 2, 2014
[37] Ho-Young Lee and Se-Hee Lee, “Finite Element Analysis of Positive
Streamer Propagation with Lightning Impulse Voltage”, International
Conference on Electrical Machines and Systems (ICEMS), pp.1836-1839,
2010

158
References
[38] Hwang. J. G. et al., “Modeling streamers in transformer oil: The transitional

fast 3rd mode streamer”, Properties and Applications of Dielectric Materials
(ICPADM), pp.573-578, 2009
[39] Jadidian, Jouya, Markus Zahn, Nils Lavesson and Karl Borg, “Surface
Flashover Breakdown Mechanisms on Liquid Immersed Dielectrics”,
Applied Physics Letters 100, no. 17, 2012
[40] Francis O’Sullivan, J. George Hwang and Markus Zahn, “A Model for the
Initiation and Propagation of Positive Streamers in Transformer Oil”, IEEE
International symposium on Electrical Insulation (ISEI), pp.210-214, 2008
[41] R. Liu, C. Tornkvbist, V. Chandramouli, O. Girlanda and L. A. A.
Pettersson, “Ester fluids as alternative for mineral oil: the difference in
streamer velocity and pre breakdown voltages”, Annual Report conference
on Electrical Insulation and Dielectric Phenomena CEIDP, pages 543-548,
2009
[42] X. Yi and Z. D. Wang, “The Influences of Solid Surface on the Propagation
of Creepage Discharge in Insulating Liquids”, IEEE Transaction on
Dielectric and Electrical Insulation, Vol. 22, no. 1, 2015
[43] M. Akyuz, L. Gao and V. Cooray, “Positive Streamer Discharges along
Insulating Surface”, IEEE Transaction on Dielectric and Electrical
Insulation, Vol. 8, no. 6, 2001
[44] P. M. Mitchinson, P. L. Lewin, G. Chen and P. N. Jarman, “A new approach
to the study of surface discharge on the oil-pressboard interface”, IEEE
International Conference on Dielectric Liquid (ICDL), pp.1-4, 2008
[45] A. Beroual, “Pre-breakdown Mechanisms in Dielectric Liquids and
Predicting Models”, Electrical Insulation Conference (EIC), pp.117-128,
2016
[46] O. Lesaint and G. Massala, “Positive Streamer Propagation in Large Oil
Gaps: Experimental characterization of propagation modes”, IEEE
Transactions on Dielectric and Electrical Insulation, Vol. 5, no. 3, 1998
[47] Q. Liu and Z. D. Wang, “Streamer Characteristic and Breakdown in
Syhnthetic and Natural Ester Transformer Liquids under Standard Lightning
Impulse Voltage”, IEEE Transactions on Dielectrics and Electrical
[48] O. Lesaint and G. Massala, “Transition to fast streamers in mineral oil in the
presence of insulating solids”, IEEE International Symposium on Electrical
Insulation, Mntreal, Quebec, Canada, June 16-19, pp.737-740, 1996
[49] Lars Lundgaard, Dag Linhjjell and Gunnar Berg, “Propagation of Positive
and Negative Streamers in Oil with and without Pressboard Interfaces”,
IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 5, no. 3,
1998

159
References
[50] Y. Kamata, E. Ohe, K. Endoh, S. Furukawa, “Development of Low-

permittivity Pressboard and its Evaluation for Insulation of Oil-immersed
EHV Power Transformers”, IEEE Transactions on Electrical Insulation,
Vol. 26, no. 4, 1991
[51] David Ariza, Marley Becerra, Rebecca Hollertz and Claire Pitois, “On the
initiation of negative streamers at mineral oil-solid interfaces”, Annual
Report Conference on Electrical Insulation and Dielectric Phenomena
(CEIDP), pp.563-565, 2015
[52] David Ariza, Marley Becerra, Rebecca Hollertz and Claire Pitois,
“Propagation of negative streamers along mineral oil-solid interfaces”,
Phenomena (CEIDP)”, pp.566-569, 2015
[53] Weiwang Wang and Shengtao Li, “Impulse and DC surface flashover of
LDPE/TiO2 nanocomposites in Vacuum”, International Conference on
Condition Monitoring and Diagnosis, pp.497-500, 2016
[54] MaBO, Daomin Min, Jiasheng Ru, Shaoming Pan, Huang Yin, Yuanwei
Zhu and Shengtao Li, “Investigating the Influence of Temperature and DC
Electric Field on the Electric Conduction of Oil Impregnated Paper
Insulation”, International Conference on Condition Monitoring and
Diagnosis, pp.514-517, 2016
[55] Shengtao Li, Weiwang Wang, Fengyan Ni and Guilai Yin, “Surface
Flashover in Vacuum and Bulk Breakdown in Polystyrene
Nanocomposites”, Conference Proceedings of ISEIM, pp.486-490, 2011
[56] Yu Chen, Yonghong Cheng, Kai Wu, J. Leith Nelson, “Flashover
Characteristic of Epoxy Composites Filled with Different Micro Inorganic
Oxide Particles under Nanosecond Pulse in Vacuum”, IEEE Transactions
on Plasma Science, vol. 37, Issue 1, pp. 195-203, 2008
[57] Y. H. Cheng, Y. Chen, P. J. Tang, K. Wu, L. Shao, Q. Chen, “Study on the
Vacuum Surface Flashover Characteristics of Epoxy Composites with
Different Fillers under Steep High Voltage Impulse”, International
Conference on Solid Dielectrics, Winchester, UK, pp.349-352, 2007
[58] Yu Chen, “Flashover Characteristic of Epoxy Composite Filled with
Different Micro-Inorganic Oxide Particles under Nanosecond Pulse in
Vacuum”, IEEE Transactions on Plasma Science, Vol. 37, no. 1, 2009
[59] Kenneth Wechsler and Michael Riccitiello, “Electric Breakdown of a
Parallel Solid and Liquid Dielectric System”, AIEE Electrical Insulation
Committee, paper 61-245, 1960
[60] K. Weehsler and M. Riccitiello, “Electric Breakdown of a Parallel Solid and
Liquid Dielectric System”, AIEE Trans. Part 3, Power Application System,
Vol. 80, pp. 365-368, 1961

160
References
[61] Jadidian, Jouya, Markus Zahn, Nils Lavesson, Ola Widlund and Karl Borg,
“Surface Flashover Breakdown Mechanisms on Liquid Immersed
Dielectrics”, Applied Physics Letters 100, no. 17, 2012
[62] Francis M. O’Sullivan, “A Model for the Initiation and Propagation of
Electrical Streamers in Transformer Oil and Transformer Oil Based
Nanofluids” Doctor of Philosophy, Massachusetts institute of technology,
2006
[63] R. H. Fowler and L. Nordheim, “Electron Emission in Intense Electric
Field”, Proc. Roy. Soc. A, Vol. 119, no. 781, pp. 173-181, 1928
[64] Ho-Young Lee, “Multiphysics Analysis of Electrical Discharge and
ELectrogydrodynamics in Dielectric Liquid”, Thesis for the degree of
Doctor of Philosophy, Kyungpook National University, 2014
[65] Uno G 𝑎̈ fvert, Albert Jaksts, Christer T 𝑜̈ rnkvist and Lars Walfridsson,
“Electrical Field Distribution in Transformer Oil”, IEEE Transactions on
Electrical Insulation, Vol. 27, no.3, 1992
[66] Werner F. Schmidt, “Electronic Conduction Processes in Dielectric Liquids”,
IEEE Transactions on Electrical Insulation, Vol. 19, no. 5, 1984
[67] A. O. Allen, “Drift Mobilities and Conduction Band Energies of Excess
Electrons in Dielectric Liquids, NSRDS-NBS, 1976
[68] J. C. Devins, S. J. Rzad and R. J. Schwabe, “Breakdown and prebreakdown
phenomena in liquids”, J. Appl. Phys., 1981
[69] C. Zener, “A theory of the electrical breakdown of solid dielectrics”, Proc.
Roy. Soc. A, 1934
[70] J. Qian, “Analysis of Polarity Effects in the Electrical Breakdown of
Liquids”, J. Phys. D. Appl., Vol. 39, pp. 359-369, 2006
[71] H. Akiyama, “Streamer discharges in liquids and their application”, IEEE
Transaction Dielectric Insulation, Vol. 7, pp.646-653, Issue 5, 2000
[72] W. F. Schmidt, “Electronic Conduction Processes in Dielectric Liquids”,
IEEE Trans. Electr. Insul., pp.389-418, Issue 5, 1984
[73] A. Denat, B. Gosse and J. P. Gosse, “Electrical Conduction of Solutions of
an ionic Surfactant in Hydrocarbons”, IEEE International Conference on
dielectric Liquids, Vol. 12, pp.197-205, 1981
[74] L. Onsager, “Deviations from Ohm’s Law in Weak electrolytes”, J. Chem,
Phys., Vol. 2, pp. 599-615, 1934
[75] L. Onsager, “Deviations from Ohm’s Law in Weak Electrolytes”, J. Chem.
Phys., Vol. 2, pp. 599-615, 1934
[76] H. Zainuddin, P. L. Lewin and P. M. Mitchinson, “Modeling the Inter-phase
Region of High Voltage Transformer”, IEEE International Conference on
Dielectric Liquids, pp.1-4, 2011

161
References
[77] P. M. Mitchinson, P. L. Lewin, B. D. Strawbridge and P. Jarman, “Tracking

and surface discharge at the oil-pressboard interface”, IEEE Electrical
Insulation Magazine, Vol. 26, pp. 35-41, 2010
[78] G. G. Touchard, T. W. Patzek and C. J. Kadke, “A physicochemical
explanation for flow electrification in low-conductivity liquids in contact
with a corroding wall”, IEEE Transactions on Industry Applications, Vol.
32, pp. 1015-1057, 1996
[79] A. P. Washabaugh and M. Zahn, “A chemical reaction based boundary
condition for flow electrification”, IEEE Transactions on Dielectrics and
Electrical Insulation, Vol. 4, pp. 688-709, 1997
[80] A. Alj, A. Denat, J. P. Gosse and B. Gosse, “Greation of Charge Carriers in
Nonpolar Liquids”, IEEE Transaction Electrical Insulation, Vol. EI-20, pp.
221-231, 1985
[81] F. Pontiga and A. Castellanos, “Electrical Conduction of Electrolyte
Solutions in Non-polar Liquids”, IEEE Transaction Indus. Appl., Vol. 32,
no. 4, 1996
[82] U. Gafvert, A. Jaksts, C. Tornkvist and L. Walfridsson, “Electrical Field
Distribution in Transformer Oil”, IEEE Transaction Electrical Insulation,
Vol. 22, pp. 647-660, 1992
[83] J. M. Cabaleiro, T. Paillat and O. Moreau, “Modelling of static development
and dynamic behavior of the electrical double layer at oil-pressboard
interface”, Electrostatics Joint Conference, 2009
[84] Trung Nam Tran, “Surface Discharge Dynamics: Theory, Experiment and
Simulation”, Ph.D. thesis, University of Southamton, 2010
[85] U. Gafvert, A. Jakes, C. Tornkvist and L. Walfridsson, “Electrical field
distribution in transformer oil”, IEEE Transaction Electrical Insulation, Vol.
22, pp. 647-660, 1992
[86] W. F. Schmidt, “Liquid State Electronics of Insulating Liquids”, CRC Press,
1997
[87] F. M. O’Sullivan, “A model for the initiation and propagation of electrical
streamers in transformer oil and transformer oil based nanofluids”, Ph.D.
thesis, Massachusetts Institute of Technology, 2007
[88] I. Adamczewski, “Ionization, Conductivity and Breakdown in Dielectric
Liqudis”, Tay-lor & Francis, pp. 185-228, 1996
[89] A. Mozumder, “ Electron Mobility on Liquid Hydrocarbons: Application of
the Quasi-Ballistic Model”, Chem. Phys. Letts., no. 233, pp. 167-172, 1995
[90] Photomultiplier Tube datasheet has been download from:
www.datasheetcatalog.com
[91] Image Intensifier units datasheet has been download from:
www.datasheetcatalog.com

162
References
[92] Q. Wang and G. Chen, “Effect of nanofillers on the dielectric properties of

epoxy nanocomposite”, Advances in Materials Research, Vol. 1, No. 1,
2012
[93] J. Keith Nelson, “Dielectric Polymer Nanocomposites”, Chapter 7, pp. 216-
220, 2010
[94] Toshikatsu Tanaka and Alun S. Vaughan, “Tailoring of Nanocomposite
Dielectrics”, from fundamentals to devices and application Chapter 9, 2016
[95] Toshikatsu Tanaka and Masahiro Kozako, “Proposal of a Multi-core Model
for Polymer Nanocomposite Dielectrics”, IEEE Transaction on Dielectric
and Electrical Isulation, Vol. 12, no. 4, Issue 1, 2005
[96] J. K. Nelson and Y. Hu, “The impact of nanocomposite formulations on
electrical voltage endurance”, International Conference on Solid Dielectric,
Toulouse, France, pp.832-835, Vol. 2, 2004
[97] F. Sadaoui and A. Beroual, “DC Creeping Discharges over Insulating
Surface in Different Gases and Mixtures”, IEEE Transaction Dielectric
Electric Insulation, Vol. 21, no. 5, 2014
[98] M. Koch, S. Tenbohlen, “Evolution of bubbles in oil-paper insulation
influenced by material quality and ageing”, Published in IET Electric Power
Application, Vol. 5, Issue 1, pp.168-174, 2010
[99] Kyunghoon Jang, Takao Akahoshi, Masahiro Kozako and Masayuki Hikita,
“Nano SiO2/Epoxy Coating Effect on Creepage Discharge Characteristics in
Oil/Pressboard Composite Insulation System”, International Conference on
Dielectric (ICD), pp.394-397, 2016
[100] Y. Yokoi and Y. Suzuki, “Empirical Formula and Polarity Effect of the
Impulse Surface Discharge in Air”, IEEJ, Vol.90, no.4
[101] Hitoshi Okubo, “High Electric Field Theory”, Ohmasha, 2013
[102] Kyunghoon Jang, Takao Akahoshi, Masahiro Kozako and Masayuki Hikita,
Composite Insulation with Rod-plane Electrode System”, Conference on
Electrical Insulation & Dielectric Phenomena (CEIDP), pp.832-835, 2015
[103] T. Tanaka, “Analysis of electric Charge Accumulation in Polymeric
Insulation by Quantum Chemical Calculation,”, Annual Meeting of The
Institute of Engineers of Japan, 2015
[104] Santanu Singha and M. Joy Thomas, “Dielectric Properties of Epoxy
Nanocomposite”, IEEE Trnasactions on Dielectric Electrical Insulation,
Vol. 15, no. 1, 2008
[105] Toshikatsu Tanak, “Analysis of Nano-fillers as Quantum Dots”, The
papers of Technical Meeting on Dielectric and electrical Insulation, IEEJ,
pp. 33-38, 2016

163
References
[106] T. Tanaka, “A Novel Concept for Electronic Transport in Nanoscale

Spaces Formed by Islandic Multi-cored Nanoparticles”, IEEE
International Conference on Dielectric (ICD), vol. 1, pp.23-26, 2016
[107] T. Tanaka, “A Quantum Dot Model for Permittivity of Polymer
Nanocomposites”, IEEE Conference on electrical Insulation and Dielectric
Phenomena (CEIDP), pp. 40-43, 2016
[108] Yong-Hong Cheng, Zeng Bin Wang and Kai Wu, “Pulsed Vacuum
Surface Flashover Characteristics of TiO2/Epoxy Nano-Micro Composites”,
IEEE Transaction on Plasma Science, Vol. 40, no. 1, 2012
[109] Shengtao Lo, Guilai Yin, Suna Bai and Jianying Li, “A New Potential
Barrier Model in Epoxy Resin Nanodielectrics”, IEEE Transactions on
Dielectric and Electrical Insualtion, Vol. 18, no. 5, 2011
[110] Shengtao Li, Guilai Yin, G. Ghen, Jianying Li, Suna Bai, “Short-term
Breakdown and Long-term Failure in Nanodielectrics: A Review, IEEE
Transactions on Dielectric and Electrical Insulation, Vol. 17, no. 5, 2010
[111] Hulya Kirkici, Mert Serkan and Kalyan Koppisetty, “Nano/Micro
Dielectric Surface Flashover in Partial Vacuum”, IEEE Transactions on
Dielectric and Electrical Insulation, Vol. 14, no. 4, 2007
[112] Yujie Hy, Robert C. Smith, J. Keith Nelson and Kinda S. Schadler, “Some
mechanistic understanding of the impulse strength of nanocomposite”,
Phenomena (CEIDP), pp.31-34, 2006
[113] Shengtao Li, Guilai Yin, “Surface Flashover in Vacuum and Bulk
Breakdown in Polystyrene Nanocomposites”, IEEE Trans. Dielectr. Insul.,
pp.486-490, 2011
[114] Y. Chen, Y. Cheng, “Vacuum flashover properties of epoxy composite
coating filled with different micro/nanometer inorganic oxide particles
under nanosecond pulse”, Prog. Natural Sci., vol. 18, 2008
[115] Mihir Roy, J. Keith Nelson and R. K. MacCrone, “Candidate mechanisms
controlling the electrical characteristics of silica/XLPE nano dielectrics”,
Journal of Materials Science, Vol. 42, Issue 11, pp. 3789-3799, 2007
[116] Francis O’Sullivan, J. George Hwang and Markus Zahn, “A Model for the
Initiation and Propagation of Positive Streamers in Transformer Oil”,
IEEE International Symposium on Electrical Insulation (ISEI), pp.210-214,
2008
[117] Olivier Ducasse, Liberis Papageorghiou, Olivier Eichwald, Nicols Spyrou
and Mohammed Yousfi”, Critical Analysis on Two-Dimensional Point to
Plane Streamer Simulations using the Finite Element and Finite Volume
Methods”, IEEE Transactions on Plasma Science, Vol. 35, no. 5, 2007

164
References
[118] Jouya Jadidian, “Charge Transport and Breakdown Physics in

Liquid/Solid Insulation System”, Doctor of Philosophy, Massachusetts
Institute of Technology, 2013
[119] Se-Hee Lee, Se-Yeon Lee, Young-Ki Chung and Il-Han Park, “Finite
Element Analysis of Corona Discharge Onset in Air with Artificial
Diffusion Scheme and under Fowler-Nordheim Electron Emission”, IEEE
Transaction on Magnetics, Vol. 43, no. 4, 2007
[120] Shuji Sato, Koji Kato, Tomoaki Ito and Sumiko Sakaguchi, “Double-
Exponential Parameter Determination for the Lightning Impulse Voltage”,
IEEJ, Vol. 120-A, no. 11, 2000 (In Japaness)
[121] COMSOL AB., COMSOL Multiphysics Modeling Guide, 2008
[122] Y. V. Torshin, “On the Existence of Leader Discharges in Mineral Oil”,
IEEE Transaction dielectric Electrical Insulation, Vol. 2, no. 1, 1995
[123] Y. V. Torshin, “Prediction of Breakdown Voltage of Transformer Oil from
Predischarge Phenomena”, IEEE Trnasaction Dielectric Electrical
[124] Lei Zhou and Michael R. Zachariah, “Size resolved particle work function
measurement of free nanoparticles: Aggregates vs. spheres”, Chemical
Physics Letters, Vol. 525-526, pp.77-81, 2012
[125] M. U. Anker, “Effect of Test Geometry, Permittivity Matching and Metal
Particles on the Flashover Voltage of Oil/solid Interfaces”, IEEE
Transaction Power App. System, Vol. PER-3, Issue 12, pp.33-33, 1983
[126] S. Ingebrigtsen, H. S. Smalo and L. E. Lundgaard, “Effects of electron
attaching and electron releasing additives on streamers in liquid
cyclohexane”, IEEE Transaction Dielectric Electrical Insulation, Vol. 16,
Issue 6, pp.1524-1535, 2009
[127] Jin-Hyun Choi, Kyunghoon Jang, Takao Akahoshi and Se-Hee Lee,
“Finite Element Analysis using Surface Charge on Liquid-Solid Interface in
Insulating Liquid”, Annual Conference, 2016 (in Korean)
[128] G. Massala and O. Lesaint, “Positive streamer propagation in large oil
gaps: Electrical properties of streamers,” IEEE Tran. Diel. Elec. Insu., vol. 5,
no. 3, pp. 371-381, 1998.

165

Kou K 435

Uploaded by

Copyright:

Available Formats

Kou K 435

Uploaded by

Document Information

Original Description:

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

Kou K 435

Uploaded by

Copyright:

Available Formats

Creepage Discharge Behaviour and Numerical

Model Analysis at the Oil/Pressboard Interface

Supervisor: Prof. Masayuki Hikita

Creepage Discharge Behaviour and

Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering i

Creepage Discharge Behaviour and Numerical Model

Electrical insulation has been described as the most important component of

Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering ii

Thesis Supervisor: Prof. Masayuki Hikita

Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering iii

Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering iv

Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering v

Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering vi

Figure 1.1 Structure of oil filled transformer ........................................................ 2

Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering vii

Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering

Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering ix

Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering x

Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering xi

Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering xii

Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering

Table 2.1 Creepage discharge on different pressboard condition...................... 14

Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering

First and foremost, I would like to express my deepest gratitude to Prof.

Secondly, I would like to convey my honest appreciation to Prof. June-Ho Lee

Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering xv

Symbols used in thesis

∆ Molecular ionization energy [J]

Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering

Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering

Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering

1.1 OIL FILLED TRANSFORMER

Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering 1

(a) Oil filled power transformer

(b) Schematic of inner structure of oil filled transformer

Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering 2

1.2 RESEARCH MOTIVATION AND PROBLEM

Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering 3

1.3 RESEARCH OBJECTIVES AND

Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering 4

1. The experimental work presented in this thesis contributes to a further

2. This experiment results may assist in the prediction of discharge initiation

1.4 THESIS STRUCTURE

Chapter 3 presents general electrohydrodynamics model in mineral oil in terms

Chapter 4 explains about creepage discharge properties at the interface between

Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering 5

Chapter 5 explains about the creepage discharge and dielectric properties of

Chapter 6 describes the numerical analysis method in composite insulation

Chapter 7 is the conclusions of this work, which is summarizing an important

Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering 6

Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering 7

Insulations of Oil Filled Transformer &

2.1 MINERAL OIL

 Paraffinic – straight and branched saturated hydrocarbons

Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering 8

The paraffinic hydrocarbons are composed of straight and branched chains of

Figure 2.1 Molecular structures of hydrocarbon groups (paraffinic, naphthenic and

Kyushu Institute of Technology, Dept. of Electrical and Electronics Engineering 9