dielectric response and partial discharge diagnostics of...
TRANSCRIPT
KTH Electrical Engineering
Dielectric Response and Partial Discharge
Diagnostics of Insulation Systems by Utilizing
High Voltage Impulses
ROYA NIKJOO
Doctoral Thesis
Stockholm, Sweden 2016
ii
TRITA-EE 2016:080 KTH School of Electrical Engineering
ISSN 1653-5146 SE-100 44 Stockholm
ISBN 978-91-7729-033-9 SWEDEN
Akademisk avhandling som med tillstånd av Kungl Tekniska Högskolan framlägges till
offentlig granskning för avläggande av teknologie doktorsexamen onsdagen den 15 Juni
2016 klockan 14.00 i sal E3, Osquars backe 14, Kungl Tekniska Högskolan, Stockholm.
© Roya Nikjoo, June 2016
Tryck: Universitetsservice US AB
iii
Abstract
In this thesis, power system transients are considered as an opportunity for development of on-
line diagnostics of power components and specifically the insulation systems of power transformers
and bushings.
Failures of power components and outages can lead to a huge economical loss in the electric
power grid. Therefore reliable operation of power grid requires on-line monitoring and diagnostics
of power components. Power transformers and their bushings are crucial components in a power
transmission system and their insulation degradation may lead to catastrophic failures. Several
methods exist for diagnostics of these components. However, some of them can only be done off-
line in maintenance periods, and the existing on-line methods generally provide less information,
especially on the internal solid insulation parts.
A new technique for on-line dielectric response measurement of power transformer bushings is
proposed which utilizes natural transients in the power system, such as lightning and switching
surges, as stimuli. Laboratory investigations are done on implementation of the proposed technique.
Measurement considerations, data acquisition and processing involved in achievement of
reasonable accuracy in the Dielectric Response (DR) are presented. Capability of the technique in
tracking of the degradation signatures such as moisture content in the insulation has been evaluated
and it has shown a good level of accuracy by being compared to the Frequency Domain
Spectroscopy (FDS).
The proposed technique is tested on the service-aged 150 kV bushings and feasibility of the
technique for monitoring of dielectric properties of power transformer bushings has been assessed;
the results are promising for the technique to be used in the real application.
Partial Discharges (PD) behavior under transients has been also studied for different materials in
this project. PD behavior of different defects, at different insulation condition, responding to the
overvoltage transients in form of superimposed impulses on ac voltages was investigated and it was
perceived how their distinctive response and the interpretation of that, can be useful for their
identification.
Recordings of the voltage signals and associated PD measurements during and after the
incidence of a high voltage transient are of potential use for condition assessment of the insulation
of power transformers and their bushings, as a form of online diagnostics that considers how the ac
PD is affected by the transient.
Besides the conventional materials, surface ac PD behavior of modified papers under the
superimposed impulses has been studied. New developed dielectric nanocomposites with improved
electrical properties have shown the possibility for enhancement of their PD resistance. Therefore,
surface PD properties of modified paper with silica and zinc oxide nanoparticles under the impulses
have been assessed in this project. Proper type and optimum concentration level of nanoparticles in
the paper are the factors that lead to the improvement of PD behavior in the modified paper under
overvoltage transients.
Keywords: Dielectric Response, Transients, Transformer bushing, Oil-impregnated paper,
Moisture content, Partial discharge, Surface PD, Cavity PD, Superimposed impulse, Nanoparticles
iv
Sammanfattning
Denna avhandling behandlar tillämpningen av spänningstransienter i elkraftsystem som en
möjlig väg till utveckling av on-line diagnostik av kraftkomponenter, speciellt isoleringen i
krafttransformatorer och genomföringar.
Fel på kraftkomponenter kan leda till stora ekonomiska förluster. För att kunna försäkra sig om
en tillförlitlig drift av elnätet krävs on-line övervakning och diagnos av kraftkomponenter.
Krafttransformatorer och deras genomföringar är viktiga komponenter i ett kraftsystem, och
överslag i deras isolering kan leda till katastrofala följder. Det finns flera metoder för diagnostik av
dessa komponenter. En del av dem kan endast utföras off-line i underhållsperioderna; däremot
erbjuder de befintliga on-line metoderna i allmänhet mindre information, särskilt om de inre fasta
delarna av isolersystemet.
En ny metod för on-line dielektrisk respons mätning av krafttransformatorgenomföringar
föreslås som utnyttjar befintliga transienter i kraftsystemet, såsom blixtnedslag och omkopplingar,
för att stimulera komponenten, vars respons då mäts och används för diagnostik.
Laboratorieundersökningar av den föreslagna metoden görs och aspekter av mätning, datainsamling
och bearbetning som är involverade för att uppnå en rimlig grad av noggrannhet i de dielektriska
svaren (DR) presenteras. Metodens förmåga att identifiera föråldringssignaturer såsom fukthalt i
isoleringen har utvärderats och en god nivå av noggrannhet har visats genom jämförelse med
frekvensdomänsspektroskopi (FDS).
Den föreslagna metoden provades på 150 kV genomföringar, vilka har åldrats under drift och
metodens genomförbarhet för övervakning av de dielektriska egenskaperna hos
transformatorgenomföringar verkar lovande för användning i verkliga applikationer.
Partiella urladdningar (PD) vid transienta förlopp har också studerats för olika material i detta
projekt. PD beteendet hos olika defekter vid olika tillstånd för isolationssystemet, har studerats vid
överspänningstransienter i form av impulser överlagrade på växelspänningar; skillnader i beteendet
hos olika defekter undersöktes. Spänningsmätningar och tillhörande PD mätningar under och efter
förekomsten av en högspänningstransient är av potentiell nytta för tillståndsbedömning av isolering
i krafttransformatorer och deras genomföringar, som en form av online-diagnostik.
Nyutvecklade dielektriska nanokompositer med förbättrade elektriska egenskaper har visat på
möjligheten att förbättra deras motståndskraft mot PD. Därför har yturladdningsegenskaper hos
papper, modifierat med nanopartiklar av kiseldioxid och zinkoxid, under impulsspänningar
bedömts i det här projektet. Korrekt typ och optimal koncentrationsnivå av nanopartiklar i papper
är en faktor som leder till en förbättring av PD beteendet i modifierat papper vid
spänningstransienter.
Nyckelord: Dielektrisk respons, transienter, impuls, transformator genomföring,
oljeimpregnerad papper, fukthalt, partiell urladdning, yturladdning, kaviteturladdning,
nanopartiklar
v
Acknowledgements
I am grateful to all those who have helped and supported me during my work.
My deepest appreciation goes to my main supervisor, Assoc. Prof. Hans Edin and my
co-supervisor Dr. Nathaniel Taylor whose patience, encouragement and professional
guidance means a lot to me during this research. This work would not be completed
without their support and guidance. I also express my gratitude to Prof. Rajeev
Thottappillil for motivating me a lot in my career and social area.
I am very fortunate to have a lot of friends in KTH Electric Power Divisions and I’m
greatly indebted to all of them who create a wonderful and friendly atmosphere to work
in and I would like to give my special thanks to Dr. Nadja Jäverberg, who was the first
person I turned to for help and who always supported me in research and daily life.
I extend my sincere thanks to all my friends at Electromagnetic Engineering
Department who have been supportive and helpful a lot during my study at KTH, with
special thanks to: Patrick Janus, Respicius Clemence Kiiza, Mohamad Ghaffarian Niasar,
Xiaolei Wang, Mahsa Ebrahimpouri and Fatemeh Ghasemifard.
Also, thanks to Janne Nilsson and Jesper Freiberg for supporting me in my
measurement setups.
Many thanks to Peter Lönn for his continuous support in maintaining computers and
providing software and Carin Norberg and Ulrika Petterson for financial administration.
Furthermore, I acknowledge ABB for supplying transformer paper and pressboard and
Nynäs AB for supplying transformer oil. Thanks to Fortum Generation and Robert
Bronegård for supplying the transformer bushings. Also, many thanks to Rebecca
Hollertz and her colleagues at KTH Department of Fiber and Polymer Technology for the
collaboration around the new paper materials.
The project is funded by the SweGRIDS, the Swedish Centre for Smart Grids and
Energy Storage. It is a partnership of academia, industry and public utilities, with major
funding from the Swedish Energy Agency as well as the corporate partners, which is
gratefully acknowledged. The project was also part of KIC-InnoEnergy through the
CIPOWER innovation project, and as a part of EIT/InnoEnergy mobility program I had a
chance to spend five months of my PhD at Smart Energy Lab of University of Utah.
Last but not least, I would like to express my gratitude to my parents and my sisters
for all their continuous support, encouragement, and endless love to me.
Thanks to all of you!
Roya Nikjoo
Stockholm, June 2016
vi
vii
List of Publications:
This thesis is based on the following papers:
I. R. Nikjoo, N. Taylor and H. Edin, ''Dielectric Response Measurement with Impulse
Stimulus on AC: Measurement Considerations, and Laboratory Testing on a Bushing'',
submitted to IEEE Transactions on Dielectrics and Electrical Insulation.
II. R. Nikjoo, N. Taylor, R. Kiiza, H. Edin, "Dielectric Response of Oil-impregnated
Paper by Utilizing Lightning and Switching Transients", IEEE Transactions on
Dielectrics and Electrical Insulation, Vol. 22, No. 1, 2015.
III. R. Nikjoo, N. Taylor and H. Edin, ''Insulation Condition Assessment of Power
Transformer Bushings by Utilizing High Voltage Lightning Impulses'', PowerTech,
Eindhoven, Netherlands, 2015.
IV. R. Nikjoo, N. Taylor and H. Edin, ''Effect of Superimposed Impulses on AC Partial
Discharge Characteristics of Oil-Impregnated Paper'', submitted to IEEE Transactions
on Dielectrics and Electrical Insulation.
V. R. Nikjoo, N. Taylor and H. Edin, ''Effect of High Voltage Transients on Surface
Discharge Characteristics of Oil-Paper Insulation'', IEEE Electrical Insulation
Conference (EIC), Seattle, WA, USA, 2015.
VI. R. Nikjoo, N. Taylor and H. Edin, ''Effect of High Voltage Impulses on Surface
Discharge Characteristics of Polyethylene'', Nordic Insulation Symposium (Nord-IS),
Copenhagen, Denmark, 2015.
VII. R. Nikjoo, R. Hollertz, N. Taylor, H. Edin, M. Wåhlander, L. Wågberg, ''Surface
Discharge Characteristics of Oil-impregnated Paper with SiO2 and ZnO nanoparticles
under AC with Superimposed Impulse'', submitted to IEEE Transactions on Dielectrics
and Electrical Insulation.
Licentiate thesis:
R. Nikjoo, ''Diagnostics of Oil-Impregnated Paper Insulation Systems by Utilizing
Lightning and Switching Transients'', Licentiate Thesis, KTH Royal Institute of
Technology, Stockholm, Sweden, 2014.
https://www.diva-portal.org/smash/get/diva2:710964/FULLTEXT01.pdf
viii
Papers related to the project, but not included in this thesis:
VIII. R. Nikjoo, N. Taylor, R. Clemence Kiiza, M. Ghaffarian, X. Wang and H. Edin,
''Insulation Condition Diagnostics of Oil-impregnated Paper by Utilizing Power
System Transients'', 18th International Symposium on High Voltage Engineering
(ISH), Seoul, South Korea, 2013.
IX. R. Nikjoo, N. Taylor, R. Clemence Kiiza, M. Ghaffarian and H. Edin, ''Dielectric
Response of Aged Transformer Bushings Utilizing Power system Transients'', IEEE
Innovative Smart Grid Technologies (ISGT) Europe, Copenhagen, Denmark, 2013.
X. R. Nikjoo and H. Edin, "Effect of Transformer Winding on the Transient Response
of Power Transformer Bushings," International Conference on Lightning Protection
(ICLP), Vienna, Austria, 2012.
XI. R. Nikjoo, N. Taylor, M. Ghaffarian and H. Edin, ''Dielectric Response
Measurement of Power Transformer Bushing by Utilizing High Voltage Transients'',
IEEE Conference on Electrical and Dielectric Phenomena (CEIDP), Montreal,
Canada, 2012.
The author has also contributed to the following papers during her PhD period:
XII. R. Clemence Kiiza, M. Ghaffarian Niasar, R. Nikjoo, X. Wang and H. Edin,
''Change in Partial Discharge Activity as Related to Degradation Level in Oil-
impregnated Paper Insulation'', IEEE Transactions on Dielectrics and Electrical
Insulation Vol. 21, No. 3, 2014.
XIII. R. Clemence Kiiza, M. Ghaffarian Niasar, R. Nikjoo, X. Wang and H. Edin, ''The
effect of PD by-products on the dielectric frequency response of oil-impregnated paper
insulation comprising of a small cavity'', IEEE Transactions on Dielectrics and
Electrical Insulation, Vol. 22, No. 5, 2015.
XIV. R. Clemence Kiiza, R. Nikjoo, M. Ghaffarian Niasar, X. Wang and H. Edin,
''Effect of High Voltage Impulses on Partial Discharge Activity in a Cavity Embedded
in Paper Insulation'', IEEE Conference on Electrical and Dielectric Phenomena
(CEIDP), Montreal, Canada, 2012.
XV. R. Clemence Kiiza, R. Nikjoo and H. Edin, ''Changes in phase resolved partial
discharge patterns in a cavity embedded in paper insulation due to AC voltages and
high voltage impulses'', Proceedings of the 10th Nordic Conference on Electricity
Distribution System management and Development, pp. 93-103, Espoo, Finland, 2012.
ix
XVI. R. Clemence Kiiza, M. Ghaffarian Niasar, R. Nikjoo, X. Wang and H. Edin,
''Effect of High Voltage Impulses on Surface Discharge at the Oil-Paper Interface'',
Proceedings of the 23rd Nordic Insulation Symposium (NORD-IS13), pp. 108-111,
Trondheim, Norway, 2013.
XVII. R. Clemence Kiiza, M. Ghaffarian Niasar, R. Nikjoo, X. Wang and H. Edin,
''Comparison of Phase Resolved Partial Discharge Patterns in Small Test Samples
Bushing Specimen and Aged Transformer Bushing'', IEEE Conference on Electrical
and Dielectric Phenomena (CEIDP), Montreal, Canada, 2012.
XVIII. M. Ghaffarian Niasar, R. Clemence Kiiza, X. Wang, R. Nikjoo, Hans Edin, ''Aging
of Oil Impregnated Paper Due to PD Activity'', 18th International Symposium on High
Voltage Engineering (ISH), Seoul, South Korea, 2013.
XIX. M. Ghaffarian Niasar, R. Clemence, X. Wang, R. Nikjoo, H. Edin, ''Effect of
Temperature on Surface Discharge in Oil'', IEEE Conference on Electrical and
Dielectric Phenomena (CEIDP), Montreal, Canada, 2012.
XX. X. Wang, R.C. Kiiza, M. Ghaffarian Niasar, R. Nikjoo, H. Edin, ''Effect of
dielectric material on decay of surface charge deposited by corona discharge”, 18th
International Symposium on High Voltage Engineering (ISH), Seoul, South Korea,
2013.
x
List of Abbreviations:
Abbreviation Description
ac alternating current
dc direct current
BD breakdown
DR dielectric response
DS dielectric spectroscopy
DGA dissolved gas analysis
DP degree of polymerization
FDS frequency domain spectroscopy
HV high voltage
HVDC high voltage direct current
HDPE high density polyethylene
NFC nanofibrillated cellulose
OIP oil-impregnated paper
PD partial discharge
PDIV partial discharge inception voltage
PRPD phase resolved partial discharge
PDC polarization and depolarization current
PE polyethylene
PC polycarbonate
RVM return voltage measurement
SEM scanning electron microscope
xi
Contents
1 Introduction ......................................................................................................... 1
1.1 Background ................................................................................................... 1
1.2 Literature review ........................................................................................... 2
1.3 Motivation of the study ................................................................................. 3
1.4 Aim ............................................................................................................... 5
1.5 Method .......................................................................................................... 5
1.6 Thesis outline ................................................................................................ 6
1.7 Author’s contributions .................................................................................. 6
2 Theoretical Background ..................................................................................... 7
2.1 Dielectric Response ...................................................................................... 7
2.2 Partial Discharges ......................................................................................... 9
2.2.1 Surface discharges .................................................................................... 9
2.2.2 Cavity discharges ................................................................................... 11
3 Measurement Setups and Sample Preparation ............................................... 13
3.1 AC system with superimposed impulse ...................................................... 13
3.2 PD measurement system ............................................................................. 15
3.3 Recorded data ............................................................................................. 15
3.4 Dielectric response measurement ............................................................... 16
3.5 Materials and sample preparation ............................................................... 17
3.5.1 Study I and II .......................................................................................... 17
3.5.2 Study III ................................................................................................. 18
3.6 PD sample setups ........................................................................................ 19
3.6.1 Surface PD setup, needle electrode ........................................................ 19
3.6.2 Surface PD setup, rod electrode ............................................................. 19
3.6.3 Cavity PD setup ...................................................................................... 20
4 Bushing Dielectric Response Measurement by Utilizing Transients ............. 21
4.1 Diagnostics methods ................................................................................... 21
4.2 Transients in power system ........................................................................ 22
4.3 Dielectric response measurement by transient impulses ............................. 23
xii
4.3.1 Circuit modeling and data analysis ......................................................... 24
4.3.2 Measurement considerations .................................................................. 26
4.3.3 Results and sensitivity analysis .............................................................. 27
1) Sampling rate .............................................................................................. 27
2) Quantisation error ....................................................................................... 28
3) Data acquisition bandwidth ........................................................................ 28
4) Filtering ...................................................................................................... 29
5) Application ................................................................................................. 29
4.4 Moisture content and temperature diagnostics ............................................ 30
4.5 Bushing and partial discharges ................................................................... 32
4.6 Conclusion .................................................................................................. 34
5 Experimental Results on Surface Discharges .................................................. 35
5.1 Surface PD on oil-impregnated pressboard ................................................ 35
5.2 Surface PD on oil-impregnated paper ......................................................... 39
5.3 Surface PD on polyethylene ....................................................................... 41
5.4 Surface PD on polycarbonate ..................................................................... 43
5.5 Discussion ................................................................................................... 45
6 Experimental Results on Cavity Discharges ................................................... 47
6.1 Cavity PD in OIP and impulse .................................................................... 47
6.2 Cavity PD in OIP with moisture content and Impulse ................................ 50
6.3 Discussion ................................................................................................... 52
7 Experimental Results on Surface Discharges of Modified Paper Materials 55
7.1 Surface PD on dry cellulose-based materials .............................................. 55
7.1.1 Surface PD on dry paper ........................................................................ 56
7.1.2 Surface PD on dry pressboard ................................................................ 57
7.1.3 Surface PD on dry NFC .......................................................................... 58
7.1.4 Surface PD on dry K92 .......................................................................... 60
7.2 Surface PD on modified oil-impregnated papers ........................................ 62
7.3 Discussion ................................................................................................... 63
8 Summary of Publications .................................................................................. 65
9 Conclusions and Future Work ......................................................................... 69
xiii
9.1 Conclusions ................................................................................................ 69
9.1.1 Dielectric response measurements by utilizing transients ...................... 69
9.1.2 Partial discharge behaviors under the impulses ...................................... 69
9.2 Future work ................................................................................................. 70
Bibliography ............................................................................................................... 73
xiv
1
Chapter 1
Introduction
1.1 Background
The reliable operation of power components in the electric grid is a key concern for
grid operators and electricity providers. Manufacturing imperfections and aging of
component’s insulation can increase the risk of the component failure. Huge costs of
power equipment like power transformers together with their vital role in the electricity
transmission and distribution system motivates the necessity for equipment condition-
based assessments to avoid failure [1]. Monitoring and non-destructive diagnostic tools
have become favored for avoiding costly and time consuming failures and replacements.
Off-line techniques can be done at the planned service period of the component, but long
gaps between service periods can restrict the efficiency of their application. There is a
trend for transition from traditional time based maintenances to condition based
maintenances in smart power grids to achieve more reliable system by frequent and
automated monitoring of important components [2]. Therefore, applicable on-line
methods have attracted much attention. On the other hand, most of the on-line methods
cannot provide as much information as off-line methods due to their limitations in
performance, but they can be used at the operation with normal stress conditions on the
equipment. Decision making based on the on-line condition assessments of power
components (e.g. transformers) can be done according to the majority agreements
between assessments of different on-line techniques [3]- [5].
The two most common diagnostic methods used for insulation systems of power
components are dielectric spectroscopy and partial discharge (PD) measurements.
Dielectric spectroscopy provides dielectric response (DR) of the insulation system over a
wide range of frequency. It is appropriate especially for complex insulation systems like
oil-impregnated paper (OIP). Different insulation defects cause different changes in the
dielectric response of the insulation versus the frequency. This characteristic adds value
to the results obtained by the dielectric spectroscopy for condition assessment. However,
this method can only be used when the component is off-line in its service period, and it
requires an external voltage source.
Partial discharge measurement and analysis can provide information at an early stage
of insulation degradation, before its failure. Partial discharges occur at regions with lower
breakdown strength than surrounding dielectric medium (e.g. cavities) or at overstressed
and sometimes contaminated regions, e.g. insulation interfaces. Cavity and surface
discharges are among the most important defects in the insulation systems that include
solid materials and their prolonged activities can lead to the failure of the insulation.
2
On-line PD measurements of power transformers can be done by acoustic sensors for
detecting and locating partial discharges. However, disturbances can reduce their
accuracy and limit their efficiency. Another technique for on-line PD measurement which
is less affected by the disturbances uses Rogowski-coils placed at the bottom ends of the
bushings that can continuously measure the high frequency currents. However, the
interpretation of PD pulses measured by this method for PD source identification is still
needed [6].
The motivation for this project is to investigate the possibility to further develop these
diagnostic methods with an intention of utilizing them for on-line condition monitoring of
power components and with focus on power transformers and their bushings with
improved analysis tools for interpretation of the on-line measured signals for defects
identification in their insulation system.
1.2 Literature review
A lot of studies have been done on power transformer aging and the involved
degradation mechanisms, e.g. [7]- [11]. Several tools and methods are presented on
diagnostics of power transformers and their bushings [12]- [18] which can be categorized
as off-line and on-line diagnostic methods [19]- [21]. In each group of diagnostic
methods, several chemical and electrical techniques have been used. Dissolved Gas
Analysis (DGA) is the most common chemical technique which can be used in off-line
and on-line condition assessment of the transformers; some defects can be detected by
this analysis [7]- [8]. However, electrical techniques can provide more information about
the inaccessible solid insulations of the component and their condition, e.g. their moisture
content. Dielectric Spectroscopy (DS) is an electrical technique that has been widely
applied in the recent decades [12], [13], [22]. However, it can only be used when the
component is off-line. Loss factor and capacitance measurement at power frequency can
be done on-line [19], [20] but it cannot provide the same rich information as DS.
Therefore, the possibility of measuring dielectric response in an on-line condition [23]
with a new proposed technique for a component such as power transformer bushing is
investigated in this project.
Many experimental works are done to explain the mechanism and behavior of surface
and cavity discharges of different solid materials including cellulose-based
insulations [24]- [29]. PD measurement systems have been developed for monitoring and
defect detection of the insulation parts of the component. Recording of PD activities and
the related analysis can be done in several forms. One of the most common forms of PD
analyses is based on the Phase Resolved Partial Discharge (PRPD) under ac voltages.
Some studies have been done on PRPD patterns of various defects for their
identification [30]- [32]. Different stages of electrical aging can also be interpreted
through PRPD [33]. PD activity of insulation systems under dc voltages has also been
investigated to characterize different defects in HVDC systems [34]- [39].
Power components normally operate under ac or dc voltages. However, transients of
different origins like lightning discharges and switching operations of circuit breakers are
inevitable phenomena that occur in electrical power systems which can stress the
3
components. Consequently, impulse tests are performed on power components before
their application in the grid to assess their impulse withstand ability. Transients usually
can occur during the operation of the component and they will get superimposed on the
ac or dc operating voltages. Although power components are protected from the
transients whose magnitude gets limited through the surge arresters, they can still be
stressed by overvoltages.
The influence of high voltage impulses on insulation materials has been studied from
different point of views. The relation between the degradation level of XLPE cables and
their impulse breakdown strength is investigated in [40] where the results are presented
by showing the correlation of the dielectric loss and the breakdown voltage.
PD inception and breakdown strength of oil-impregnated paper (OIP) with lightning
impulse superimposed on the ac has been studied in [41], [42]. In these articles, the
breakdown (BD) voltage of the OIP under the superimposed impulse is compared to the
BD voltage with single application of the impulse or the single application of the ac
voltage. OIP PD inception voltages under the superimposed impulses are also
investigated and compared to the ac PD inception voltages. In [43], the partial discharge
behavior of air filled cavities with different dimensions in polyethylene under multiple
impulses is investigated and PD inception and extinction during the impulse are studied.
PD properties of corona discharges in air under lightning and switching impulses have
been observed in [44]. It shows that at the rising time of the impulse voltage, there is
always a chance for PD happening and the tail time can affect the number of PDs during
the impulse. However, PD behavior during the impulse has a stochastic characteristic. PD
behavior of cavity in polyethylene (PE) due to an impulse superimposed on ac voltages is
investigated in [45]. The presented results on applied standard lightning impulses
superimposed on different positions in the ac cycle showed that the ac PD got initiated by
the impulse and extinguished after several cycles depending on the ac voltage and the
aging situation of the PE.
In this work, surface and cavity PD behavior of different conventional materials
(mainly celluloses based e.g. OIP) at different insulation condition are investigated before
and after the superimposed impulse on the ac voltage.
In recent years many studies have been done on development of new dielectric
nanocomposites [46]- [50] with improved electrical properties which can lessen the
insulation degradation and failures. PD resistance of nanocomposites and their behavior
in reduction of erosion and aging due to the PD has been studied in [51]- [54]. According
to the previous achievements on nanocomposites, Department of Fiber and Polymer
Technology at KTH are working on improvement of cellulose based insulation materials
of power transformers. The surface PD properties of these modified materials under the
superimposed impulses have been assessed in this project.
1.3 Motivation of the study
Insulation strength of power component can decrease during its life cycle due to
different circumstances and stress factors. Power system transients can be among the
factors that impose stresses to the component and affect their insulation system. They can
4
participate in the aging of the insulation and if the component is in an aged condition,
they may lead to an early failure, Figure 1.1 [55].
Stress
Transient Transient
Transient
Insulation Strength
Life Time
Stress/Stre
ngth
Early failure
Figure 1.1: Effect of transients on early failure of the component
Transient electrical stress can affect the weak points or defects in the insulation either
by activating partial discharges or changing an on-going PD behavior with a specific
character. Figure 1.2 shows different scenarios on how transients can affect the weak
points or defects in the insulation of the component. As it is illustrated, for inactive
defects (no PD activity) impulses may initiate and enhance the PD properties (such PD
numbers and magnitudes), change them temporarily or not affect them at all. For defects
with active PD, in addition to the last three scenarios, reduction of PD properties due to
the impulse is possible.
Impulse
PD properties
Temporary change
Enhancement
TimeBefore transient After transient
No changeNo PD
Impulse
PD properties
Temporary change
Enhancement
TimeBefore transient After transient
No changePD
Reduction
Figure 1.2: Change in PD properties due to the transients
Different defects at different stages of the insulation degradation may respond
differently to the transients, which itself can be used for insulation on-line diagnostics. In
addition, as transients are superimposed on the ac voltage, they can be considered as
natural stimuli to the component and its insulation system, which can yield a lot of
information about the system. Their spectrum contains wide range of frequencies which
makes them interesting to be used for on-line dielectric response measurement of the
5
component. The trend of changes in condition of the insulation can be tracked by this
technique.
Transients can thus have a potential to be used for on-line diagnostics of the
component either by indicating the defects through their PD behavior or by DR
measurements and identifying the aging stage of the insulation. In this way the transients
can be utilized for power system diagnostics.
The major failure of paper insulations has a root in their PD susceptibility which can
also be influenced by the impulses. Therefore, new developed materials with possibility
of improved PD resistance such as modified papers with nanoparticles can be a
motivation for study of their PD behavior under the impulse.
1.4 Aim
The aim of this work was to experimentally evaluate how high voltage impulses that
resemble naturally occurring transients in the power system, can be used to diagnose the
status of commonly utilized electric power equipment such as transformers and bushings,
and in particular the status of their electrical insulation.
The two main objectives were to:
1) develop a method which can measure certain characteristics of the insulation
system from the stimulus of different transients such as transient overvoltages of
the kind occurring from lightning and switching impulses. The objective was to
evaluate how applicable the method is for on-line diagnostics of power
transformer bushing and whether its sensitivity is enough to trace some signatures
of insulation degradation such as increased moisture content in the oil-paper
insulation.
2) investigate the time evolution of partial discharge (PD) activity of the insulation
defects exposed to the overvoltage transients. In-particular to investigate whether
partial discharges can get initiated after a transient or not and what is their
behavior and progression before and after a transient situation for several types of
PD source.
1.5 Method
Laboratory measurements and investigations were done for development of DR
measurement by synthetic transients and it was tested on the real HV transformer
bushings. The results were verified by comparison with the measurement results of
commercial Frequency Domain Spectroscopy (FDS) instruments. A combined system of
ac and impulse generator was implemented for applying the superimposed impulses on ac
voltages. The ac surface and cavity partial discharge behavior of different conventional
materials (mainly cellulose) under the superimposed simulated transients was studied
through a large number of laboratory measurements. Commercial instruments for PD
6
measurements and an oscilloscope for recording signals were used in the study. Also,
new modified paper materials including papers with silica and zinc oxide nanoparticles
were prepared at the Department of Fiber and Polymer Technology and were provided
for investigation of their PD behavior under the impulses.
1.6 Thesis outline
This thesis contains nine chapters followed by Papers I-VII written on most of the
achievements pointed out in the main body of the thesis.
Chapter 2 contains a short theoretical background on dielectric response and partial
discharges.
Chapter 3 describes the experimental setups and measurement systems that are used in
this work.
In Chapter 4, the proposed technique for obtaining dielectric response of power
transformer bushing by utilizing natural transients and its application are explained and
considerations to fulfill the feasibility of the technique are discussed.
Chapter 5 contains the results on surface discharge studies on OIP, oil-pressboard,
polyethylene (PE), and polycarbonate (PC) and their response behavior to the applied
impulses.
Chapter 6 includes the results from the cavity discharge investigation in OIP and its
behavior after the impulse.
In Chapter 7, surface discharges and their behavior under the impulses for paper
materials with nanoparticles are presented.
Chapter 8 summarizes papers I to VII.
In Chapter 9, conclusion of the presented work and suggestions for future work are
mentioned.
1.7 Author’s contributions
The author is responsible for papers I-VII. The results in Paper I to VII were
developed by the author. In paper VII, the materials and all discussions around them were
provided by Rebecca Hollertz and her colleagues at KTH Department of Fiber and
Polymer Technology.
The entire work was supervised by Assoc. Prof. Hans Edin and Dr. Nathaniel Taylor
as co-supervisor.
7
Chapter 2
Theoretical Background
2.1 Dielectric Response
This section provides the theory behind the calculation of the involved parameters in
the dielectric response measurement of insulation systems in power component such as
power transformer bushing.
Dielectric polarization comes about as a dynamic behavior of dielectric material under
the electric field E. In a good insulation, polarization is the dominant factor in the current
measured through the insulation. Electric displacement D increases by Polarization P,
and their relation can bind together by:
0 D E P (2.1)
where, ε0 is permittivity of free space. Several polarization mechanisms can be defined in
dielectrics [56]. In all of these mechanisms, the charges are bound in the material and the
electric field cannot force them to escape. The dielectric response function f(t)
characterizes the linear polarization and relates it to the electric field E for linear and
isotropic material:
0
0
t
t f t d P E (2.2)
Besides the polarization, conductivity of the dielectrics also participates in the current
density of the dielectric under an applied electric field E. The current density J(t) is
defined as the sum of conduction and total displacement current:
dc
d tt t
dt
DJ E (2.3)
where σdc is the dc conductivity of the material. By replacing the electric displacement
and polarization formula in (2.3), the current density can be written as:
8
0
0
c
t
d
d t dt t f t d
dt dt
EJ E E (2.4)
where, ε∞ is the permittivity at high frequencies. By replacing the electric field with the
applied voltage v over the insulation (placed between electrodes) and modifying (2.4), the
total current through the dielectric in time domain can be written as:
0
0 0
( )( ) ( ) ( ) ( )dc
r
dv t di t C v t f v t d
dt dt
(2.5)
where C0 is the geometrical capacitance of the space between the electrodes in absence of
dielectric and εr∞ is the relative permittivity at high frequencies. The first term of the
equation represents the dc conductivity of the material, the second and third terms
respectively define the pure capacitive behavior and the polarization in the dielectric.
By the Fourier transform, the current and voltage in the time domain can be
transformed to the frequency domain and dielectric properties can be represented in the
frequency domain parameters. The relation between current and voltage phasors is
represented by the complex capacitance C :
0
0
' '' ( )dcappI j C j V j C jC j CV
j
(2.6)
where εʹ is dielectric constant, εʹʹ is dielectric loss, the real part of complex capacitance Cʹ
refers to a capacitive effect and the imaginary part Cʺapp refers to the loss effect in the
dielectric. Loss tangent is a geometry-independent parameter which is defined as:
0
'''' ( )
tan' ' ( )
DC
app appC
C
(2.7)
in which δ is the dielectric loss angle. The real part of the complex capacitance and the
loss tangent are useful diagnostic indicators [19] which will be used later in this work.
To obtain the dielectric response of the insulation system, a voltage is applied to the
system, and the current that flows through the insulation is measured. From the measured
current and voltage, insulation properties such as tan δ, Cʹ and Cʺ can be calculated.
9
2.2 Partial Discharges
There are several parameters that play a role in a partial discharge process. In the
presence of adequate applied voltage, there will be a local field enhancement at a
protrusion or in a gas-filled defect in the insulation which drives partial discharges. PD
recurrence and its behavior follow different criteria under ac and dc field. Under ac
voltage, change in the voltage polarity and magnitude can explain the behavior of PD
occurrence. However, under dc voltage, finite resistivity of the dielectric is responsible
for the PD recurrence, Figure 2.1 [34]. At ac voltage, the PD behavior is related to the
magnitude and phase of the voltage and specific phase patterns can be observed for
different defects. In this work, different defects (surface and cavity) under ac and the
superimposed impulse voltages are investigated. A PD behavior under each of these
voltage waveforms has its own features which will be affected by each other.
Figure 2.1: Recurrence of discharges at ac voltage (left) and at dc voltage (right) [57]
2.2.1 Surface discharges
The inhomogeneity of material and strong non-uniform electric field at the interface
can lead to surface PD. The dynamic behavior of surface PD is affected by the electric
field, permittivity and conductivity of the interface materials (which itself is dependent
on moisture, temperature and frequency). In our study, the alternating (ac) electric field is
already present when the impulse is applied to the surface discharge setup, Figure 2.2. If
the ac electric field would be below the surface PD inception voltage, applied impulse
can initiate so-called "streamer" discharges which can propagate towards the ground
electrode and depending on the impulse magnitude their propagation distance will be
different [58] and may lead to a complete flashover. Ionization of gas or liquid caused by
the PD can assist the PD development and wider spread of charges on the surface.
Enhancement of electric field at the ionized zone leads to more ionization which can
assist in PD development. The streamer velocity and space charge density in dielectric
liquid are dependent on the voltage magnitude [58].
A streamer can be stopped by a solid barrier, which will result in a local charge
accumulation. The gas or liquid medium itself can have positive and negative ions;
applied electric field and adsorption of ions can lead to the surface charge accumulation
on the solid. Dissociation of charges from the surface is another possible mechanism
10
leading to the surface charge. Cellulose hydroxyl groups that exist in paper and
pressboard materials, tend to acquire negative charges from the oil, and leave behind the
positive charges [59]. Hydrocarbon molecules of paper material trap electrons and
produce negative ionic space charge. Trapped charges to the surface can be detrapped at
each PD event, and detrapping of charges from the insulation surface affects the PD
statistics.
In the case of an applied positive impulse, electrons move fast to the needle to get
neutralized and leave the slow positive ions behind, forming the ionization zone (positive
net charge), Figure 2.2.
insulation
+
Eb
Figure 2.2: A schematic representation of ions distribution for surface PD
When the ions accumulate in the low field region (ion drift region) they cannot move
fast and they remain there for a longer time. In the next ac half cycle, the needle gets the
negative polarity and there will be a field enhancement between the ionization zone and
the needle which can lead to many partial discharges. As the electric field changes by the
ac cycles, positive ions gradually get a chance to reach the ground electrode and become
neutralized. The continuity of PDs is guaranteed by flux of positive ion toward the
ground electrode.
If a negative impulse is applied to the needle, there will be a strong electric field
around the needle tip, due to the fast movement of electrons; removal of space charges
will be faster than the previous case. There is a time constant for the drift of the ions
which is dependent on the mobility of the ions and the applied electric field.
Space charges on the surface changes the distribution of the electric field on the
insulation. Figure 2.3 shows the effect of the surface charge in the change of the applied
electric field simulated by COMSOL Multiphysics 4.3b. For simulation with electrostatic
physics, constant surface charge density (ρs) is introduced on the boundary of oil and
paper between the needle electrode and the ground bar and ac voltage is defined by its
amplitude on the boundaries of the needle.
Figure 2.3: Electric field due to the ac voltage and surface charges a) without surface charge
b) with surface charge of 10 µC/m2 c) with surface charge of 100 µC/m
2
a b c
11
It can be observed that the background electric field gets distorted by accumulation of
surface charges.
Above the ac PD inception, the electric field at each cycle for specific phases is high
enough for the surface PD occurrence, Figure 2.4.
Figure 2.4: Surface ac PD above the inception voltage
If the impulse is applied to the needle superimposed on the ac peak, the same
mechanisms that were explained earlier can affect the PDs and the charge dynamics.
However, in this case, there will be higher ac electric fields involved and the previous ac
PD charges will be supplemented by the charge formed by the impulse effect.
If the impulse is applied at the zero-crossing voltage, there will be no external ac
electric field at the impulse incident and the electric field for creation of new charges will
be lower compared to the applied superimposed impulse at the ac peak voltage.
In general, the influence of superimposed impulses on the ac surface PD must be of a
short duration, from µs to several seconds.
Moisture also plays an important role at the interface [60]. Water is a polar dielectric
which can align easily in the direction of the electric field. It is important to note that
even dynamics of interface charges in fresh oil and aged oil samples can be quite
different, which results from the difference in the moisture content and oil
conductivity [61].
2.2.2 Cavity discharges
In order to have a PD started in a cavity, the electric field inside the cavity must be high
enough for a breakdown, and a free electron must be available to initiate the ionization
process. This seed electron can be supplied through several ways such as an external
source or electron detrapping from the cavity surface. PD in gas-filled cavities follows
Paschen’s law [56]. When a PD gets initiated, it can provide a source of electrons for
later PDs. The cavity surface is affected by the earlier PDs, and the deposited charges on
the surface can facilitate the breakdown in the cavity. Space or surface charges left by PD
activities contribute to the voltage over the defect. Therefore the total electric field will
be defined by the background field and space or surface charges induced electric field,
Figure 2.5. As the size of a defect is often considered to be small over the scale at which
12
the background field changes, the background field of the defect can be assumed to be
uniform [62].
When an impulse is applied to the specimen below the ac inception voltage, it can
initiate a discharge, which assists the development of further discharges. When the PD
occurs, the voltage across the cavity drops to the residual value due to the charge
displacement. In order to have new discharges, the voltage across the cavity should again
exceed the minimum breakdown voltage. As the impulse decays, the voltage across the
cavity builds up in the reverse direction [43]. If by the impulse and the occurred
discharge, the electric field across the cavity gets reduced below the minimum
breakdown voltage, there will be no further discharges until the peak cavity stress
increases again to its minimum breakdown voltage [43]. If the impulse depletes the free
charges in the cavity, there will be a time lag for generation of new seed electron and
recurrence of PDs. This situation may occur when the amount of charges in the cavity are
limited, such as at the PD inception voltage. Above the PD inception voltage, the number
of PDs is quite high in the cavity, and the presence of large number of free charges in the
cavity cannot notably be affected by the impulse.
In general, if the impulse initiates the cavity PD, there is a high chance for the PD to
be continuous, as the extinction voltage for cavity is much lower than the inception
voltage.
HV
E0Eq
Figure 2.5: Charge distribution and electric fields in cavity PD setup after a single PD
13
Chapter 3
Measurement Setups and Sample Preparation
3.1 AC system with superimposed impulse
An ac test transformer system combined with a two-stage Haefely impulse generator
was used to apply a superimposed impulse during ac voltage application to the test
object, Figure 3.1. The impulse generator part is shown in the green block and the ac
system is depicted in the blue block. Each of the two sources (ac and impulse) needs to be
able to supply the test object without adversely affecting the other source. Transient
simulations were done in PSCAD to check the performance of the superimposed system
and to estimate the required values for additional capacitor and resistor elements. The
diagram of the simulated circuit, and the generated superimposed impulse on ac voltage
over the test object, is shown in Figure 3.2. A blocking capacitor (Cb = 0.8 nF) was used
to strongly attenuate the ac voltage that appears at the impulse generator side. According
to the Cb value and other elements in the circuit, it could permit about 30% of the impulse
voltage to appear on the test object. Also to protect the ac transformer from the impulses,
a blocking resistor (Rt) with the high value of 174 kΩ was used. It could also reduce the
influence of the ac source on the impulse waveform.
The superimposed system could be triggered manually or with a phase controlled
triggering circuit, shown in Figure 3.3, consisting of a microcontroller (Arduino) and a
solid state relay, at a specific phase position of the ac voltage. The applied impulse for all
measurements had a standard lightning waveform of 1.2 µs/50 µs.
Cb(0.8 nF)
AC
Zd
Cd(0.5 nF)
(152 nF)
Osc.
ICM systemOsc.
divid
er
Rt(174 kΩ)
Rd (575 Ω)
Rb (50 Ω)
Cc(1 nF)
Osc.
ICM systemOsc.
divid
erDC CS
(15 nF)
CS(15 nF) RP
(3.6 kΩ)
RP(3.6 kΩ)
preamp
Test object
Cd(0.5 nF)
Rhigh
Impulse Generator
PD Meas. Sys.
Figure 3.1: Test circuit for standard lightning impulse superimposed on ac voltage
14
Figure 3.2: PSCAD simulation for impulse superimposed on AC voltage
Figure 3.3: Programmable phase-triggering circuit
A high voltage divider was connected in parallel with the test object to measure the
voltage; it has a 0.5 nF capacitor as the high impedance, and a 152 nF capacitor as the
low impedance element. The divider’s output is connected to a peak-meter, to measure
the ac and impulse voltages. It is also connected to a 30:1 attenuator that provides a
suitable voltage level for the phase-triggering circuit and for capturing the signal on one
channel of a Tektronix DPO4102B oscilloscope.
15
For dielectric response measurement of the bushing, the manual triggering of the
impulse generator was used in the superimposed circuit to avoid the loading effect of the
phase-triggering circuit on the attenuator and imprecision of recording the signals.
3.2 PD measurement system
The PD apparent charges in the test object were measured through a detection
impedance (Zd) of 1 kΩ, connected in series with a coupling capacitor (Cc) of 1 nF,
Figure 3.1. The coupling capacitor was placed in parallel with the test object and the
detection impedance was connected to the preamplifier (preamp). The voltage over the
detection impedance was measured by an Insulation Condition Monitoring (ICM) system
from Power Diagnostix, through its preamplifier with gain of 1, 10 or 100. The ICM
system also has a built-in main amplifier for further amplification. This PD measurement
system can be used to record the PD patterns with respect to time or to the phase of the ac
voltage (Phase Resolved Partial Discharge). To protect the PD measurement system
during the impulses and breakdowns, a gas discharge tube (CG75) and two back-to-back
Zener diodes of 5 V (1N5338B) were connected across the detection impedance. An
inductor of 3.9 mH was also connected across the detection impedance, to carry the 50
Hz current and ensure a low 50 Hz voltage across the detection impedance.
For capturing time-resolved signals in a shorter time window around the impulse
occurrence, the PD signal after low-pass filtering (40 kHz – 800 kHz) and amplification
in the PD measurement system was also recorded on a Tektronix DPO4102B
oscilloscope for duration of 200 ms at a sampling rate of 5 MSa/s.
The calibration of the PD measurement system before each set of experiments was
performed by a PD calibrator from Power Diagnostix. It could inject positive or negative
charge of 10 to 1000 pC. PD patterns in phase or time, recorded from the PD
measurement system, were calibrated in terms of apparent charge. The PDs can induce
charges on the electrodes by which we can measure the PD quantity. The electrodes are
connected to the PD measurement system and the measured induced charge which is
called apparent charge, can quantify the PDs. There are several parameters, which
determine the induced charge, including the field change and the gradient of the scalar
field.
The PD patterns presented in the results show the opposite polarity for charges
compared to their real polarity, due to the measurement of apparent charge through the
coupling capacitor. The PD inception voltages, PDIVs, show the level at which PD
occurs regularly by slowly increasing the voltage amplitude.
3.3 Recorded data
The Tektronix DPO4102B oscilloscope is used for capturing the time-resolved signals
of the applied voltage to the test object and the PD activity in partial discharge
measurements, Figure 3.4. It is important to note that they only indicate the occurrence of
the impulse relative to the ac voltage and its phase, and the relative magnitude of the PD
activity within each ac cycle, without representing the actual values of the voltages.
There are several parameters that could affect the captured signals by the oscilloscope.
The signal representing applied voltage has some high-frequency oscillation during the
16
impulse, and due to the extra loading from the phase-triggering circuit, the attenuator
ratio is not at its nominal value. Also, the PD signal was obtained after filtering and
amplification by a gain that could be changed between the measurements. However, the
correct ac and impulse voltage magnitudes measured by the meters connected to the
voltage divider are written below each figure. The measurements were also checked by
using a HV probe, LeCroy PPE20kV, while generating some superimposed impulses
within the 20 kV range of this probe, to confirm that the voltage at the test object had the
form of a standard lightning impulse.
For recording the time domain signals for the dielectric response measurement of the
bushing by HV impulses, the other channel of the same oscilloscope was used.
Figure 3.4: Tektronix DPO4102B oscilloscope used for recording the signals
3.4 Dielectric response measurement
A Megger IDAX 300 Dielectric Response Analyzer and HP 4284A precision LCR
meter are the commercial instruments that were used for dielectric response
measurements. These instruments measure the dielectric response of the objects by
applying sinusoidal waves at different frequencies. The frequency range of IDAX is from
0.0001- Hz to 10 kHz, and the LCR meter can measure from 20 Hz to 1 MHz.
The measured dielectric responses by applied lightning impulses are mainly verified
by the results from the LCR meter due to the high frequency content of the lightning
impulse.
For surface discharge study of different materials, their permittivity and loss tangent
were measured by Megger IDAX 300 in the frequency range of 0.01 to 10 kHz. A test
cell made of stainless steel electrodes with a 1 mm air gap from the guard electrode was
used for DR measurement of different materials as shown in Figure 3.5 [63]. Figure 3.6
shows the circuit connections for the DR measurement of samples with IDAX 300.
Figure 3.5: Schematic of the test cell used for the dielectric spectroscopy measurements
17
Figure 3.6: Measurement circuit for dielectric spectroscopy measurements
3.5 Materials and sample preparation
The used materials and sample preparations for partial discharge investigations are
described in this section. Several samples of each material were used for each study, to
have a better understanding of their PD behavior under the superimposed impulses.
3.5.1 Study I and II
Study I refers to the surface discharge measurements presented in chapter 5, and study
II refers to the cavity discharge measurements that are presented in chapter 6. For study I,
oil-impregnated paper, oil-impregnated pressboard, PE-300 (HDPE), and natural
polycarbonate are used.
The paper was Munksjö Thermo 70 with thickness of 100 µm, impregnated with Nytro
10XN transformer oil. Transformer paper impregnation has several steps that must be
done according to the guidelines. Different factories may follow different standards for
impregnating the transformer paper. For this project, the impregnation is done according
to a scheme provided by ABB [64]. As the paper is a hygroscopic material, it must be
dried first. The procedure for the impregnation can be found in [63]. After the drying and
the impregnation, the moisture content of the paper is less than 0.5%. This paper is
considered as a dry sample. The oil-impregnation of the pressboard sample was done
with the same procedure as used for the paper. In study I, the paper samples were made
of 10 sheets with total thickness of 1 mm. The pressboard samples had 2 mm thickness
and both the PE and the PC samples were 1 mm thick. In study II, only OIP material was
used for the cavity discharge measurements which will be explained further in the section
3.6.3.
To see the effect of the moisture on the PD behavior of oil-impregnated paper, the
method of closed chemical system by saturated salt solution [65] was used to achieve
different levels of moisture content in the paper. The oil-impregnated papers were placed
in a desiccator with air and a water saturated salt solution in equilibrium and a digital
hygrometer to measure the relative humidity of the air. Each specific salt can change the
relative humidity of the air, up to a specific level which leads to a specific moisture
content in the OIP. The moisture diffusion time for oil-impregnated paper with 100 µm
thickness at room temperature (20 ºC) is about two hours [66], but it takes few days for
the salt solution to reach the equilibrium humidity. Two moisture contents of 3% and
5.5% in the OIP were obtained by using potassium acetate and sodium bromide salts
respectively. More details on the humidity control process of OIP are presented in [67].
18
Table 3.1 shows relative air humidity and the respective equilibrium moisture content of
the oil-impregnated paper, based on Rockland [65] and Cheng [68] for the used salts. The
permittivity and loss tangent of each paper sample were measured before and after the
humidity control process as a function of frequency at 200 V peak, using a Megger IDAX
300 from 0.01 Hz to 10 kHz.
Table 3.1: Relative humidity of air in equilibrium with water saturated salts and
corresponding achieved moisture content of oil-impregnated paper at room temperature.
(20ºC) [68]
Salt
Relative air humidity (%) at 20ºC
Moisture content of paper (%)
Potassium acetate
23 3.0
Sodium bromide
57 5.5
3.5.2 Study III
Study III refers to the PD investigation on different cellulose materials including
modified papers with nanoparticles. This study is done on two groups of materials. The
materials in the first group, are used in the dry condition without being impregnated. This
study was done to investigate the pure behavior of a paper without the effect of the oil.
Ten sheets of dry Munksjö Thermo 70 paper and a dry pressboard were used and their
behavior as conventional materials is compared with the behavior of a dry Nanofibrillated
Cellulose (NFC) paper and a high lignin paper. High lignin paper is denoted as K92 and
contains 14 wt% of lignin. NFC is defined as a cellulose fiber that has been fibrillated to
achieve agglomerates of cellulose microfibril units, see Figure 3.7.
In the second group, Silica (SiO2) and Zinc oxide (ZnO) nanoparticles were used for
surface modification of the Kraft paper. Papers with two different mass concentrations of
silica, 2 wt% and 6 wt%, were used for the surface study. The silica nanoparticles had a
primary particle size of approximately 12 nm.
Papers with silica and ZnO nanoparticles, K92, and the reference paper without
nanoparticles with small amounts of lignin (~3%) were impregnated in the transformer
oil and they were investigated as the second group of materials.
Figure 3.7:NFC SEM image obtained at the Department of Fiber and Polymer Technology
19
All nanofibers used in the investigations, were made at the KTH Department of Fiber
and Polymer Technology.
3.6 PD sample setups
3.6.1 Surface PD setup, needle electrode
For the surface discharge study on paper and pressboard, a setup with a needle
electrode on high potential and a grounded bar on the surface of the insulation was used,
as shown in Figure 3.8. Both needle and ground electrode are placed on the surface, no
electrode was used on the other side of the insulation due to the higher electric field this
could induce and a high risk of paper carbonization or puncture. This arrangement
ensures that the charges get directed on the surface and do not go through the bulk [24].
In this configuration, the PD inception is less sensitive to the distance between the needle
and the ground electrode as the surface discharge is a localized phenomenon compared to
the flashover or breakdown, which are sensitive to the gap distance and are mainly
related to the bulk property of the materials [24].
The needle had a tip radius of 3 µm. The distance between the needle tip and the
ground bar was set to be 3 cm. Both oil-impregnated paper and pressboard samples were
immersed in transformer oil for the experiments. For measurements on dry samples in
air, the needle was placed 5 cm away from the ground electrode. This larger distance was
used to avoid the surface flashover due to the lower breakdown strength of the air than
the oil.
Figure 3.8: Test setup for surface discharge measurements
3.6.2 Surface PD setup, rod electrode
For surface discharge study on polymers in air, an IEC-b electrode (IEC 60343) was
used, Figure 3.9. The IEC-b type electrode consists of a stainless steel rod with 6 mm
diameter and an end curvature radius of 1 mm, and a plane circular electrode on the other
side of the dielectric specimen. As the polymers are electrically stronger materials than
the paper, the high electric field under the rod electrode cannot easily cause a puncture or
breakdown. The surface of the polymer samples in contact with the electrodes could be
painted with silver paint to ensure a good contact.
20
Figure 3.9: Test setup for surface discharge measurements on polymers
3.6.3 Cavity PD setup
Cavity discharge studies were only done on oil-impregnated paper samples. Each
sample had 16 layers of OIP Munksjö Thermo 70 with a total thickness of 1.6 mm placed
between two brass electrodes, Figure 3.10. A cylindrical cavity with diameter of either 8-
mm or 3 mm and a thickness of 0.4 mm was made by punching discs out of the 4 central
layers of paper. The whole set-up of electrodes and paper layers was assembled and
clamped before immersion in the Nytro 10XN transformer oil, to form a gas-filled
cavity [63].
Figure 3.10: Test setup for cavity discharge measurements
21
Chapter 4
Bushing Dielectric Response Measurement by
Utilizing Transients
Power transformer condenser bushings are among the power components with high-
risk, due to their construction and life cycles. As they are connected to the transformers,
their failure can also lead to the failure of the transformers. Therefore, on-line monitoring
of the bushings is of high importance in the transformer monitoring. There are several on-
line and off-line diagnostic techniques, which are currently used in the grid for the
bushings. However, each of the existing off-line and on-line diagnostic methods has some
advantages and disadvantages. Off-line measurement is used when the component is out
of service and usually an external voltage source is needed to be used for the
measurements. On-line measurements or monitoring can be done more often on the
component without disturbing its operation but might not provide enough information
about the condition of the insulation. In this chapter, a new technique proposed in [67]
for on-line diagnostics of power transformer bushing has been assessed on a real bushing
with high voltage impulses. In this technique, simulated natural transients in power
system were used as stimuli for dielectric response measurement of the bushings. A
system for applying a superimposed impulse on an ac voltage was used to simulate the
real situation of the voltages in the power grid.
4.1 Diagnostics methods
Diagnostic methods are the keys for obtaining information on the condition of the
insulation and some features of its state such as conductivity, loss-tangent and humidity
level. A considerable percentage of power transformer failures are due to the failure of
their bushings, which mostly originates from the insulation degradation. Some typical
defects in the oil-impregnated paper power transformer bushings are mentioned in [69].
Early diagnostics of these defects can be used to avoid the bushings failure and increase
the reliability of their operation. The cost of potential failures compared to the cost of
monitoring is an important factor, which must be considered in a reliable grid.
As most of the diagnostic methods and specifically the on-line techniques can only
give a rough estimation of the insulation condition, several techniques are usually applied
in parallel for more reliable condition assessment [3]- [5]. Oil testing, power factor, and
in-service PD measurements are some of these techniques.
22
In off-line measurements, low level of noise, accessibility to some part of the
component and better measurement of some features such as small current are the
advantages. On the other hand, on-line methods [19], [20] are not restricted to the
maintenance periods and give an opportunity to measure the features under in-service
conditions such as operation temperature and voltage stress.
Off-line and on-line measurements are partly complementary, and both can be used to
give the trend of the changes in the insulation for condition monitoring and asset
management. For some components, such as power transformer bushings and windings,
comparison of the results for each phase with other phases or comparison with sister units
that are in the same ambient condition can also be useful in their diagnostics. Considering
the trend of the insulation parameters and their relative values is a useful way for
diagnostics, especially in on-line measurements to decrease the needed sensitivity for
instrument calibration.
Several chemical and electrical diagnostic methods are available for on-line and off-
line monitoring of power transformers and their bushings. Some of the most commonly
used chemical techniques are Dissolved Gas Analysis (DGA), moisture content analysis
and Degree of Polymerization (DP) [7]- [9], [70]. Among electrical methods, Frequency
Domain Spectroscopy (FDS) is commonly used for measurement of dielectric properties
such as loss-tangent as a function of frequency. Time domain methods such as
Polarization and Depolarization Current (PDC) and Return Voltage Measurement
(RVM) [71], [72] are less favorable due to their sensitivity to disturbances [73]. Usually
the electrical methods are considered more reliable for assessing inaccessible solid
insulation part of the components [16], but they cannot be used in an on-line mode.
On-line measurement of capacitance and dielectric loss factor of condenser bushings at
power frequency can reveal some types of internal or external defect through the sum
current method [19] or voltage tap measurement of the bushing [74]. However, the
obtained information is not enough for a good condition assessment. Therefore, an on-
line measurement of dielectric response of the bushing can be of great interest.
4.2 Transients in power system
Power system transients are inevitable phenomena with different origins such as
lightning discharges and switching operations. They can have duration of microseconds
to several seconds. What has made them interesting to be used in our on-line dielectric
response measurement is their wide range of frequencies. A specific categorization
suggested by CIGRE Working Group 33.02.1 classifies the transients in four groups
according to their frequency ranges [75], [76]. Generally, switching impulses are
considered slow transients (simulated by the standard switching impulse), but some types
of circuit breakers and switchgears can cause very fast surges.
These transients are possible choices to be used as stimuli to measure the dielectric
response of the power transformer bushings. Their respective frequency spectrums define
the frequency range in which the dielectric response measurement of the component can
be achieved.
23
4.3 Dielectric response measurement by transient impulses
One potential of using natural transients for dielectric response measurement is in on-
line diagnostics of power transformer bushings. This is the main focus of this work. A
150 kV ASEA oil-impregnated condenser bushing (GEO type from 1962) that had been
in service in the power system for 50 years was investigated by the proposed technique,
Figure 4.1. The results in this part are presented in Papers I.
The power transformer bushing has a test tap that provides access to the outermost
stress-control layer of its condenser part, which is isolated from the flange. The test tap is
a safe, accessible point which can be used for the measurement of the bushing response-
current from the applied voltage to its HV part. The simplest and most suitable way for
measurement of transient response current is using a shunt, which must be connected to
the test tap of the bushing.
Figure 4.1: 150 kV ASEA bushing
The superimposed standard lightning impulse, described in Chapter 3, was applied to
the HV part of the bushing and the impulse response current was measured at the test tap
of the bushing through a parallel capacitive (0.66 µF) and resistive shunt (100 kΩ),
Figure 4.2.
RP(7.2 kΩ) AC
Cd(0.5 nF)
(152 nF)
Osc.
divid
er
Rt(174 kΩ)
Zsh
Osc.
DC
Rhigh
CS(7.5 nF)
Cb(0.8 nF)
Rb (50 Ω)
Cd(0.5 nF)
Rd (575 Ω)
Figure 4.2: Circuit of applied superimposed lightning impulse on ac voltage to the bushing
24
The input voltage was measured through a capacitive high voltage divider and an
attenuator which are explained in section 3.1. For the real application in the field, the HV
measurement can be done through the voltage transformers or through capacitive [77] or
optic sensors [78]. As the condenser bushing itself works as a voltage divider, the output
voltage over the shunt (depending on the shunt value) can be in the permissible input
range of the oscilloscope and therefore can be connected directly. However, a protection
circuit was connected to protect the oscilloscope in case of any breakdown due to the
applied high voltage impulse. The protection circuit includes a series inductance and a
grounded Zener diode to restrict both high current and high voltages. The chosen diode
must be fast enough to be able to operate in the nanosecond timescale. The transfer
function of the protection circuit was quite flat in the frequency range of our
measurement.
4.3.1 Circuit modeling and data analysis
The measured input and output voltages are time domain signals, with information on
dielectric properties of the bushing, which can be defined in the frequency domain. By
the Fourier transform, the time domain signals can be transformed to the frequency
domain. To extract the desirable data (the bushing complex capacitance) from the
measured signals (Vin, Vout), all elements involved in the measurement must be considered
(Figure 4.3) and their respective circuit equations must be solved. The capacitances of the
connected coaxial cables were measured and considered.
The transfer functions of the HV divider and the attenuator were also measured and
modelled in the circuit. The amplitude and phase of the output voltage relative to the
input voltage of the attenuator versus frequency is plotted in Figure 4.4. There is a quite
significant change in phase of the attenuator transfer function which makes it necessary
for this to be modeled. The least required phase accuracy for loss tangent measurement is
in the range of 0.1 degree.
Test Object(bushing)
Vin
Input impedance of the oscilloscope and
capacitance of coaxial cables
Inductance of cables
500 pF
152 nF
Cable
Attenuator
Voltage divider
Vout
L2
L1
R1
ProtectionCircuit
R2
Shunt
Resistance of cables and
connections
shC
Figure 4.3: Circuit model of the measurement system
25
Figure 4.4: Transfer function of the attenuator
The network function H(ω) of the circuit shown in Figure 4.3 can be written in terms
of its elements:
2
2 2 1
|| Z( ) . . .
|| Zatt div
ZH H H
Z Z
ZV Z
V Z ZZ Z
t shout t
in t t sh test object
(4.1)
where Zt=Rt||Ct, Rt is the input resistance of the oscilloscope and Ct is a parallel
combination of the input capacitance of the oscilloscope and the coaxial cable
capacitances. Z1=R1+jωL1 and Z2=R2+jωL2 in which R1, R2, L1, and L2 are the resistances
and inductances of the cable links. Zsh=1/jωshC in which
shC is the complex capacitance
of the shunt. Hatt and Hdiv are the transfer functions of the attenuator and the divider
respectively. By simplifying the formula, the impedance of the test object can be
calculated by:
2
2 11 1
2
. || Z|| Z
( ). . .att div
Z ZZ Z
Z
ZZ Z
H H H Z
t t sh
test object t sh
t
(4.2)
According to the experiment and the investigation that was done in the laboratory, the
contribution of Z1 and Z2 was trivial in the circuit. Therefore, they could be neglected in
the bushing impedance calculation. Then, to identify the complex capacitance of the
bushing, the admittance must be defined:
1 1
1 1
sh t
( ). .Y
|| 1 ( ). .
att divtest object
att div
H H H
Z Z H H H
(4.3)
102
104
106
0.0294
0.0296
0.0298
0.03
0.0302
0.0304
0.0306
0.0308
| V
ch
2 / V
ch
1 |
Frequency [Hz]
102
104
106
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
1.2
Frequency [Hz]P
ha
se
sh
ift [d
eg
] C
h2
w.r
.t. C
h1
26
and the complex capacitance of the bushing can be calculated by:
' ''CY
Cj
jC
test object
test object
(4.4)
4.3.2 Measurement considerations
Dielectric response measurement requires high accuracy as the dielectric losses are
quite small and sensitive to the measurements. In order to have an accurate measurement,
several factors need to be considered. To reduce the complexity of the system, the
number of components in the circuit must be minimized and all parameters in the circuit
must be well known and modelled. Measurement considerations have been described in
more detail in [67].
Shunt selection
A parallel combination of resistive and capacitive shunt is used for the measurements.
Because of the capacitive nature of the test object, a capacitive shunt is of interest, to
maintain the strength of the output signal at all frequencies. In order to control the
leakage current and to avoid a floating potential, a relatively high resistance is used in
parallel with the shunt capacitor [79]. The resistance should not have much lower
impedance than the capacitor, in order to have the dominant current path through the
capacitor, even towards the low end of the frequency range.
In order to choose a proper shunt which covers all transients that can be captured in
the field, the bandwidth of the system, including the shunt must be wide enough to cover
all the available frequency components.
Coupling paths and cable connections
Because of the high accuracy that is needed for dielectric response measurement,
parameters such as cable inductances and stray capacitance at high frequency also
become important. In order to limit these effects, the cables should be as short as possible
and big loops must be avoided in the measurement system. Good shielding is also a key
to reduce the stray capacitances. Inductances must be taken into account, if the
frequencies would be considerably high. At high frequencies or for long cables, the time
delay and attenuation of the cable must be considered.
Recording the signals
The Nyquist criterion must be met for the sampling rate according to the highest
frequency component of the signal. Having high enough sampling rate enables capturing
the detail of fast changes in the signal [80]. Therefore, sampling rates that are several
times above the Nyquist rate are needed for a good practical performance.
The captured signals have quite significant amount of noise particularly at high
frequencies beyond the bandwidth of the impulse. A rough estimation of impulse
bandwidth according to its rise time Tr can be defined by: BW = 0.35 / Tr [81]. For
applied standard lightning impulse, the bandwidth is about 300 kHz. Dividing the noisy
27
voltage signals can lead to even stronger noises in admittance or capacitance. Filtering
can be done either in an analog form within the measurement system or by post
processing digital filtering of the recorded data. However, filtering cannot necessarily
improve our results, which are only considered in the signals bandwidth. This claim was
tested and the results are shown in the next section by looking through the filter effect on
the spectrum. Beside the background noise, harmonics and other distortions can also limit
the accuracy of the results.
Another source of noise can be due to the quantization error from the oscilloscope by
forcing the actual signal values onto the available digital levels. The quantization level
can significantly affect the results, specifically the loss factor. Investigation on how much
the quantization can affect the results is presented in the next section.
4.3.3 Results and sensitivity analysis
The 25 kV superimposed impulses on 3 kV ac voltages were applied to a typical 150-
kV bushing to investigate the effectiveness of the proposed technique for dielectric
response measurements. The voltage signals were captured by the oscilloscope in time,
and by using a Fast Fourier Transform, the measured signals were transformed to
phasors, and formula (4.4) was used to calculate the complex capacitance of the bushing.
In the following part the results for impulse response of the bushing are presented by
going through the sensitivity analysis of the recording data. In all the results, the
reference lightning response, shown in blue with ‘+’ marker, is obtained by the full
bandwidth of the oscilloscope with sampling rate of 500 MSa/s with highest possible
quantization of the signal without filtering.
Sampling rate
As it was noted, a sufficient sampling rate that satisfies the Nyquist criteria is needed
to achieve a good resolution of the signal [80]. The standard lightning impulse has a rise
time of 1.2 µs and its highest frequency component is expected to be a few MHz.
However, oscillations with much higher frequencies appear in the measured signals due
to the existing capacitances and inductances in the circuit, including oscillations resulting
from the measuring devices, such as the voltage divider. Figure 4.5 shows the effect of
oversampling on the accuracy of the results. Sampling rate of 500 MSa/s and 50 MSa/s
are used to capture the lightning (LT) response of the bushing. Although both of these
sampling rates seem to be above the Nyquist rate, oversampling of 500 MSa/s can
improve the accuracy of the results for higher frequencies.
28
Figure 4.5: Effect of sampling rate on lightning response of the bushing
Quantization error
Continuous time domain signals get quantized by the oscilloscope to the relatively
small discrete set of possible levels. The precision of the signal is significantly affected
by the level of quantization. Figure 4.6 shows the lightning response of the bushing with
two levels of quantization. Low quantization can notably reduce the accuracy of the loss
tangent at high frequencies. Even the measured capacitance value (Cʹ) has been affected
by the quantization error. In this case, high quantization represents the voltage amplitude
of the inputs 2 and 2.5 times larger than low quantization level for both channels.
Different quantization is obtained by changing the oscilloscope gain, not the analog-
digital converter (ADC) levels. This shows that making the signal occupy a significant
part of the input range of the oscilloscope can enhance the precision of the results.
Figure 4.6: Effect of quantization on lightning response of the bushing
Data acquisition bandwidth
The oscilloscope bandwidth defines the maximum frequency range at which it can
accurately measure an analogue signal. By limiting the full bandwidth to 20 MHz, it
102
103
104
105
1
1.5
2
2.5
3
x 10-10
Frequency [Hz]
C'
[F]
LCR meter
LT response-500MSa/s
LT response-50MSa/s
102
103
104
105
-0.04
-0.02
0
0.02
0.04
0.06
0.08
0.1
Frequency [Hz]
tan
LCR meter
LT response-500MSa/s
LT response-50MSa/s
102
103
104
105
1
1.5
2
2.5
3
x 10-10
Frequency [Hz]
C'
[F]
LCR meter
LT response-high quantization
LT response-low quantization
102
103
104
105
-0.04
-0.02
0
0.02
0.04
0.06
0.08
0.1
Frequency [Hz]
tan
LCR meter
LT response-high quantization
LT response-low quantization
29
might be possible to avoid high frequency noises and improve the results. However, the
results shown in Figure 4.7 depict a little improvement at low frequency response for tan-
δ but still there is more deviation in the loss at high frequencies. This shows that the
larger bandwidth than 20 MHz is required to accurately measure the signals.
Figure 4.7: Effect of data acquisition bandwidth on lightning response of the bushing
Filtering
As mentioned in section 3, filters do not necessarily improve the results and they can
even worsen the results by changing the precision of the signal. Therefore, choosing a
proper filter with suitable parameters is important for their application. Cmddenoise is a
Matlab function which returns the denoised signal using the wavelets and scaling filters,
and seems to be suitable to be used on our signals. Figure 4.8 compares the results with
and without using cmddenoise on the time domain signal. The result shows only a little
smoothing effect on tan δ at high frequency but in general very little improvement can be
achieved by the filter.
Figure 4.8: Effect of filtering on lightning response of the bushing
Application
Reasonable results (blue responses with ‘+’ marker) on the bushing dielectric response
measurement by lightning impulses, showed their potential to be used for diagnostics of
these components in the power grid. Therefore, it is of interest to know which kind of
102
103
104
105
1
1.5
2
2.5
3
x 10-10
Frequency [Hz]
C'
[F]
LCR meter
LT response-full bandwidth
LT response-20 MHz bandwidth
102
103
104
105
-0.04
-0.02
0
0.02
0.04
0.06
0.08
0.1
Frequency [Hz]
tan
LCR meter
LT response-full bandwidth
LT response-20 MHz bandwidth
102
103
104
105
1
1.5
2
2.5
3
x 10-10
Frequency [Hz]
C'
[F]
LCR meter
LT response
LT response with cmddenoise filter
102
103
104
105
-0.04
-0.02
0
0.02
0.04
0.06
0.08
0.1
Frequency [Hz]
tan
LCR meter
LT response
LT response with cmddenoise filter
30
defects can be identified by this technique.
The main insulation of power transformer bushings is oil-impregnated paper. Moisture
content in the oil-impregnated paper affects the dielectric response at high and low
frequencies. Also the loss peak in tan δ shifts to higher frequencies, when the temperature
increases. Therefore, the moisture content and temperature are two factors that can affect
the dielectric response (specifically the tan δ) of the oil-paper at high frequencies [82]-
[86]. By analyzing the moisture content of the oil, the moisture content of the paper
cannot be interpreted directly; moisture content estimation of oil-paper can be more
reliable through dielectric response measurements. The following section shows how
transient responses can determine the change in the dielectric response of the oil-paper
due to the moisture content and temperature.
4.4 Moisture content and temperature diagnostics
In high voltage OIP bushings, in the presence of moisture content above ~2%, there is
a significant change in tan δ [69]. High moisture content leads to loss increment and
temperature increase which itself may cause gas formation and partial discharges.
Therefore, the early detection of the moisture content of the bushing can be of great
value.
The effectiveness of the proposed technique in diagnostics of moisture content and
change of temperature level in OIP was investigated in [67] and in paper II. The
measurements were done on OIP samples (Munksjö Thermo 70) by applying low voltage
(10 V peak) standard lightning (1.2/50µs) and switching (250/2500µs) impulses [87] and
their respective dielectric response were obtained in almost the same way as explained in
the previous parts. The detailed explanation of the measurements can be found in the
licentiate thesis [67] and paper II.
The switching responses have lower frequency components compared to the lightning
responses. As the frequency range of the reliable results, depends on the bandwidth of the
signal, the results for switching response are plotted up to 2 kHz and for lightning
response up to 100 kHz, according to their bandwidth.
Samples with three different moisture levels (2.4%, 3% and 5.5%) were used and their
corresponding transient responses are shown in Figure 4.9. The real capacitance of the
sample with 3.0% moisture content is lower than the sample with 2.4% moisture content
only due to the different paper samples. It is almost impossible to have two exactly
identical oil-impregnated paper samples. However, loss tangent is geometry independent
and can show the correct sequential change due to the moisture levels (Figure 4.9). The
ability of the proposed technique through lightning and switching responses in following
the trend of the paper insulation condition can be concluded from the figures.
In oil-impregnated paper with moisture contents below 2%, the change in the loss
tangent is negligible in the frequency range of the considered transients. Therefore, it is
expected that the detectable moisture level with this technique would be above 2%.
Often, moisture content above 3% is considered a warning level, but this varies
according to the policy of different companies and power grid operators.
31
a) b)
c) d)
Figure 4.9: Capacitance of samples with different moisture content: a) switching responses
b) lightning responses, loss tangent of samples with different moisture content: c) switching
responses d) lightning responses
Tracking the change of temperature in OIP by the proposed technique was investigated
at three temperature levels. The temperature effect was investigated on dry oil-
impregnated paper with moisture content below 0.5%. The test cell was kept under
controlled temperature in a Weiss Technik climate chamber for a suitable equilibrium
time (2 hours). The temperatures of study were 21ºC (room temperature), 60 ºC and 100-
ºC. The measurement was done by applying low-voltage lightning and switching
impulses from a function generator to the test cell, with the response currents measured
using a shunt. The effect of the temperature on the oil-impregnated paper is a shift of the
loss-tangent in frequency. The change of the dielectric response by only the effect of the
temperature is not as large as by the effect of the moisture, but still it can be seen at high
frequencies.
The results are shown in Figure 4.10 for capacitance and loss tangent. As the loss
tangents at the three temperatures are very close for frequencies less than 1 kHz, the
corresponding switching responses are almost the same. However, for higher frequencies,
101
102
103
104
3
4
5
6
7
x 10-11
Frequency [Hz]
C' [
F]
IDAX-2.4% moisture
Switching response-2.4% moisture
IDAX-3% moisture
Switching response-3% moisture
IDAX-5.5% moisture
Switching response-5.5% moisture
102
103
104
105
106
3
3.5
4
4.5
5
5.5
6x 10
-11
Frequency [Hz]
C'
[F]
LCR meter-2.4% moisture
Lightning response-2.4% moisture
LCR meter-3% moisture
Lightning response-3% moisture
LCR meter-5.5% moisture
Lightning response-5.5% moisture
101
102
103
104
-0.05
0
0.05
0.1
0.15
0.2
0.25
Frequency [Hz]
tan
IDAX-2.4% moisture
Switching response-2.4% moisture
IDAX-3% moisture
Switching response-3% moisture
IDAX-5.5% moisture
Switching response-5.5% moisture
102
103
104
105
106
0
0.05
0.1
0.15
0.2
Frequency [Hz]
tan
LCR meter-2.4% moisture
Lightning response-2.4% moisture
LCR meter-3% moisture
Lightning response-3% moisture
LCR meter-5.5% moisture
Lightning response-5.5% moisture
32
the difference between the results is more visible and can be followed by the lightning
responses.
Figure 4.10: Capacitance and loss tangent measured at different temperatures
4.5 Bushing and partial discharges
For investigating the ability of the proposed technique for diagnostics of defective
bushings, two identical bushings with similar dielectric responses, were used in the
experiment. They had been working in two phases of the same transformer and did not
show any documented problem. Both bushings showed no trace of partial discharges with
100 kV peak ac (which was the maximum possible voltage that could be applied at that
time). An artificial defect was created by removing part of the insulating oil from one of
the bushings (phase B) and exposing it to the several 200 kV HV lightning impulses [88].
After refilling the oil, partial discharges were initiated in the defective bushing at 24 kV
peak ac.
Standard lightning impulses were applied to the two transformer bushings and their
transient responses were measured through the shunt connected to the test tap. Measured
lightning impulses from the HV part of the two bushings are shown in Figure 4.11.
Partial discharges can be detected from the transient response of the bushing of phase B
with observed discharges over the impulse voltage. Their dielectric response could be
obtained with the proposed technique even in presence of discharges within the lightning
response (paper III). However, there was no considerable change in dielectric response of
the bushing B due to the partial discharges.
More investigation was done on the bushing B. Its dielectric response was measured
by IDAX 300 at different voltage levels from 200 V to 27 kV. As can be observed in
Figure 4.12 and Figure 4.13, there is no significant change in the dielectric responses
measured at high voltages compared to at low voltages, and the effect of PD cannot be
observed in change of loss part.
Although the PD can cause power losses and change the loss factor in dielectric
response measurements [63], small voids with high PD cannot contribute to any
101
102
103
104
105
106
3.2
3.3
3.4
3.5
3.6
3.7
3.8
3.9
4x 10
-11
Frequency (Hz)
Ca
pa
cita
nce
(F
)
LCR meter-at 22C
LCR meter-at 60C after 2h
LCR meter-at 100C after 2h
DRA-at 22C
DRA-at 60C after 2h
DRA-at 100C after 2h
Switching response-at 22C
Switching response-at 60C after 2h
Switching response-at 100C after 2h
Lightning response-at 22C
Lightning response-at 60C after 2h
Lightning response-at 100C after 2h
101
102
103
104
105
106
-0.005
0
0.005
0.01
0.015
0.02
0.025
0.03
0.035
Frequency (Hz)L
oss ta
ng
en
t (t
an
)
LCR meter-at 22C
LCR meter-at 60C after 2h
LCR meter-at 100C after 2h
DRA-at 22C
DRA-at 60C after 2h
DRA-at 100C after 2h
Switching response-at 22C
Switching response-at 60C after 2h
Switching response-at 100C after 2h
Lightning response-at 22C
Lightning response-at 60C after 2h
Lightning response-at 100C after 2h
33
measurable change to the power loss of big objects like bushings and transformers.
However, those small voids can significantly affect the strength of the insulation and lead
to the component failure.
Figure 4.11: Lightning impulse voltages measured from the phase B and C bushings
Figure 4.12: Capacitance of the phase B bushing measured at different voltages
0 1 2 3
x 10-4
0
1
2
3
4
5
x 104
Time (s)
Vo
lta
ge
(V
)
Bushing-phase B
Bushing-phase C
1 1.05 1.1 1.15
x 10-4
0
5000
10000
10-1
100
101
102
103
104
10-9.69
10-9.67
10-9.65
10-9.63
10-9.61
Frequency (Hz)
Ca
pa
cita
nce
(F
)
200 V before damage
200 V
300 V
1.2 kV
4.8 kV
14.4 kV
27 kV
34
Figure 4.13: Loss tangent of the phase B bushing measured at different voltages
4.6 Conclusion
The feasibility of dielectric response measurements of power transformer bushing by
superimposed impulses on ac voltage was investigated. The measurements were done to
check the proposed technique in relevance to its realistic application. By considering
factors involved in the measurement and data acquisition that can contribute to their
accuracy, it is possible to achieve reasonable results for the dielectric response of the
bushing measured by applied HV impulses.
The changes in the insulation condition due to the moisture content and the
temperature can be followed by this technique and this approach can be valuable to be
used for on-line assessment of insulation condition of power transformer bushings.
10-1
100
101
102
103
104
10-3
10-2
10-1
Frequency (Hz)
Lo
ss ta
ng
en
t (t
an
)
200 V before damage
200 V
300 V
1.2 kV
4.8 kV
14.4 kV
27 kV
35
Chapter 5
Experimental Results on Surface Discharges
In this chapter, ac surface PD characteristics of different materials under superimposed
impulses have been investigated. The study was mainly focused on oil-impregnated paper
and oil-impregnated pressboard as they are the two major insulation materials of power
transformers and their bushings. Polyethylene (PE) and Polycarbonate (PC) which are
common polymers with electrical applications were also studied. Polyethylene is one of
the insulation materials that are commonly used in cables. PE is among the materials
which are considered to be PD sensitive and long lasting PD can deteriorate it [89].
Polycarbonate is a transparent insulation with good high voltage insulating
characteristics. It absorbs very little moisture and it is used in electrical and electronic
components [90].
It is of interest to know the dielectric property of each of the mentioned materials
which is influential in their PD behavior. Figure 5.1 shows the relative permittivity and
loss tangent of these materials measured by IDAX-300. PE-300 (HDPE) has the lowest
permittivity and the loss, and oil-pressboard has the highest permittivity and the loss,
among the tested materials.
Figure 5.1: Dielectric property of oil-impregnated pressboard, OIP, PC and PE
5.1 Surface PD on oil-impregnated pressboard
Oil-impregnated pressboard of 2 mm thickness, immersed in oil, was used for surface
discharge measurements. The needle electrode placed 3 cm away from the ground
10-2
10-1
100
101
102
103
104
2
2.2
2.4
2.6
2.8
3
3.2
3.4
3.6
3.8
4
Frequency [Hz]
Rela
tive P
erm
ittivity
oil-pressboard
oil-paper
Polycarbonate
Polyethylene 300
10-2
10-1
100
101
102
103
104
10-5
10-4
10-3
10-2
10-1
Frequency [Hz]
tan
oil-pressboard
oil-paper
Polycarbonate
Polyethylene 300
36
electrode bar on the surface of the pressboard. Although pressboard and paper are made
of almost the same material but the surface and bulk of pressboard has a different
structure from the paper. This difference can also lead to their different PD behavior
affected by the impulse. Surface PD behavior of oil-pressboard by applied superimposed
impulse was studied below and above the surface ac inception voltage. When the ac
voltage applied to the samples was kept below their surface PD inception voltage (about
90% of the inception), the applied impulse can initiate surface discharges which decays
after few seconds but the continuous surface PD activity cannot be fulfilled. The results
are presented in paper IV. The charge decay can also be affected by several parameters
related to the properties of both oil and pressboard and applied ac voltage.
Above the surface PD inception, both positive and negative superimposed impulses
were applied respectively at the positive and negative ac peak voltage to the oil-
pressboard samples. The surface PD shows a cluster of PD with their magnitude
increased by the applied positive impulse at the positive ac peak which declines by the
time, Figure 5.2. However, by a negative impulse applied at a negative ac peak the
increased PD behavior is less notable than with the positive impulse, Figure 5.3.
Surface PD behavior of oil-pressboard with time was also investigated. The ac voltage
was set to 1.5 times the PD inception and was run for about 19 hours. Figure 5.4 shows
that at the beginning, the surface PD numbers were high and the number of charges
decreased over time. After the 19 hours, no trace of degradation was observable on the
surface of the pressboard.
Figure 5.5 also shows the PRPD pattern of the oil-pressboard at the beginning and
after 19 hours of PD exposure. It can be observed that the number and magnitude of PD
charges had been decreased.
a) b)
Figure 5.2: Oil-pressboard surface PD at 27 kV (20% above the inception) with 25 kV
positive superimposed impulse, a) a few cycles before and after the impulse, b) PD pattern vs
time for 10 s time window
-0.1 -0.05 0 0.05 0.1-6
-4
-2
0
2
4
6
Time [s]
Vo
lta
ge
[V
]
input voltage
PD
37
a) b) Figure 5.3: Oil-pressboard surface PD at 23 kV (20% above the inception) with 25 kV
negative superimposed impulse, a) a few cycles before and after the impulse, b) PD pattern vs
time for 10 s time window
Figure 5.4: Oil-pressboard surface PD numbers vs time run for about 19 hours at 30 kV
(50% above the inception)
To investigate the effect of the ac voltage phase at which the impulse occurs, on oil-
pressboard surface PD, a superimposed impulse was applied either at the peak or at the
zero-crossing of the ac voltage. Figure 5.6 shows the behavior of surface PD with
superimposed impulse at ac positive peak voltage. The behavior is the same as what was
also observed in Figure 5.2. High magnitude PDs happen following the impulse with
exponential-like decay. When the impulse occurs at the ac zero-crossing, the ac PD
behavior cannot get notably affected by the impulse due to the absence of the ac electric
field, Figure 5.7.
-0.1 -0.05 0 0.05 0.1-6
-4
-2
0
2
4
6
Time [s]
Vo
lta
ge
[V
]
input voltage
PD
0 200 400 600 800 1000 12000
1
2
3
4
5
6
Time (min)
Num
ber
of
PD
per
50 H
z c
ycle
positive charge
negative charge
38
a) b)
Figure 5.5: Oil-pressboard surface PD pattern at 30 kV (50% above the inception), a) first
recorded pattern, b) last recorded pattern after about 19 hours
a) b)
Figure 5.6: Oil-pressboard surface PD at 27 kV (20% above the inception) with 33 kV
superimposed impulse at peak of ac, a) a few cycles before and after the impulse, b) PD
pattern vs time for 10 s time window
a) b)
Figure 5.7: Oil-pressboard surface PD at 27 kV (20% above the inception) with 33 kV
superimposed impulse at zero-crossing of ac, a) a few cycles before and after the impulse, b)
PD pattern vs time for 10 s time window
-0.1 -0.05 0 0.05 0.1-6
-4
-2
0
2
4
6
Time [s]
Vo
lta
ge
[V
]
PD
input voltage
-0.1 -0.05 0 0.05 0.1-6
-4
-2
0
2
4
6
Time [s]
Vo
lta
ge
[V
]
PD
input voltage
39
5.2 Surface PD on oil-impregnated paper
Samples with 10 layers of OIP with a total thickness of 1 mm were used for surface
discharge investigation. The same needle and ground-bar setup was used as mentioned in
the chapter 3 for this investigation. The superimposed impulse effect on OIP surface PD
was studied at voltages below and above the PD inception voltage. The major results on
OIP are presented in paper IV and V. When the ac voltage is below the surface PD
inception, the superimposed impulse can initiate a temporary surface PD on the oil-paper
interface that decays after several ac cycles. Above the PD inception voltage, a
superimposed impulse at the ac peak increases the number of PDs for several ac cycles
which can be observed in Figure 5.8. However, a superimposed impulse applied at the ac
rising zero-crossing couldn’t affect the PD behavior notably, Figure 5.9.
a) b) Figure 5.8: Oil-paper surface PD at 27.6 kV (20% above the inception) with 33 kV
superimposed impulse at peak of ac, a) a few cycles before and after the impulse, b) PD
pattern vs time for 10 s time window
a) b)
Figure 5.9: Oil-paper surface PD at 27.6 kV (20% above the inception) with 33 kV
superimposed impulse at zero-crossing of ac, a) a few cycles before and after the impulse, b)
PD pattern vs time for 10 s time window
-0.1 -0.05 0 0.05 0.1-6
-4
-2
0
2
4
6
Time [s]
Vo
lta
ge
[V
]
PD
input voltage
-0.1 -0.05 0 0.05 0.1-6
-4
-2
0
2
4
6
Time [s]
Vo
lta
ge
[V
]
PD
input voltage
40
The presence of moisture in OIP can significantly affect its surface discharge
characteristics. Therefore, the effect of moisture content in surface PD behavior of OIP
has been also investigated. Two levels of moisture content, 3% and 5.5 % were used in
OIP samples for investigation and the results for this case are presented paper IV. The
OIP with extra moisture content showed a decrease in the PD activity after being exposed
to the superimposed impulse. For the OIP with 3% moisture content this decline is
temporary, Figure 5.10. However, with 5.5% moisture content this decrease will follow
up with an increase in the PD activity, Figure 5.11. The PRPD pattern of OIP surface PD
for higher moisture content is distinctive compared with that of the dry OIP.
a) b)
Figure 5.10: Surface PD activity of oil-paper with 3% moisture content at 24kV ac (20%
above 20 kV inception) with 33 kV superimposed impulse, a) a few cycles before and after the
impulse, b) PD pattern vs time for 10 s time window
a) b) Figure 5.11: Surface PD activity of oil-paper with 5.5% moisture content at 22kV ac (20%
above 18 kV inception) with 33 kV superimposed impulse, a) a few cycles before and after the
impulse, b) PD pattern vs time for 10 s time window
Similar results on surface PD of OIP were obtained by applying the impulse in the
short pause between the ac PD (without the impulse being superimposed on the ac
-0.1 -0.05 0 0.05 0.1-6
-4
-2
0
2
4
6
Time [s]
Vo
lta
ge
[V
]
input voltage
PD
-0.1 -0.05 0 0.05 0.1-6
-4
-2
0
2
4
6
Time [s]
Vo
lta
ge
[V
]
input voltage
PD
41
voltage) which is presented in paper V. In paper V, the effect of carbonization on OIP
and its surface PD behavior due to the impulse was also investigated. The results showed
that, with trace of carbonization on the OIP surface, the PD activity will show a notable
decrease after the impulse.
5.3 Surface PD on polyethylene
The IEC-b type test cell was used for investigation of surface discharges on the
polymers PE and PC, Figure 5.12. A 1 mm thick sample of each material was used for the
surface discharge study. The whole set up was placed in the room environment and
influenced by the room temperature and relative humidity of the air. Surface PD
inception voltage for an air-polymer interface was much lower than for the paper samples
immersed in the oil, and therefore lower magnitude impulses were applied to the PE and
PC samples. Superimposed impulse was applied to the setup with PE 300 at ac voltage
below the inception. Discharges appeared on the surface due to the impulse and decayed
shortly after, Figure 5.13. This behavior was quite similar to the one observed from the
paper and pressboard. Above the surface PD inception, the results from the surface PD
on similar samples at different days with different air relative humidity showed different
behavior due to the impulse. Figure 5.14-16 shows the PD behavior of PE exposed to the
superimposed impulse.
Figure 5.12: Surface discharge measurement set up for PE and PC
a) b)
Figure 5.13: PE surface PD at 3.2 kV (90% of the inception) with 11 kV superimposed
impulse, a) a few cycles before and after the impulse, b) PD pattern vs time for 10 s time
window, 29% RH and 20.2º C
-0.1 -0.05 0 0.05 0.1-5
-4
-3
-2
-1
0
1
2
3
4
5
Time [s]
Vo
lta
ge
[V
]
PD
input voltage
42
a) b)
Figure 5.14: PE surface PD at 4 kV (3.5 kV inception) with 16.5 kV superimposed impulse,
a) PD pattern, b) Maximum magnitude of PD charge vs time for 10 s time window with
impulse effect, 48.1% RH and 21.8º C
a) b)
Figure 5.15: PE surface PD at 4 kV (3.5 kV inception) with 16.5 kV superimposed impulse,
a) PD pattern, b) Maximum magnitude of PD charge vs time for 10 s time window with
impulse effect, 40.1% RH and 23.8º C
a) b)
Figure 5.16: PE surface PD at 4.3 kV (3.6 kV inception) with 16.5 kV superimposed
impulse, a) PD pattern, b) Maximum magnitude of PD charge vs time for 10 s time window
with impulse effect, 32.3% RH and 20.6º C
0 1 2 3 4 5 6 7 8 9 10-0.5
0
0.5
1
1.5
2
2.5x 10
-9
Time [s]
Max C
harg
e [
C]
0 1 2 3 4 5 6 7 8 9 10-0.5
0
0.5
1
1.5
2
2.5x 10
-9
Time [s]
Max C
harg
e [
C]
0 1 2 3 4 5 6 7 8 9 100
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5x 10
-9
Time [s]
Max C
harg
e [
C]
43
It was observed that air relative humidity can play a role in response of surface PD to
the impulse. In the presence of higher humidity, the number and magnitude of discharges
are increased by the impulse and last for several seconds, while at lower relative humidity
in the air, this effect gets less notable. Beside each PD pattern their respective maximum
PD charges was also plotted to clarify the PD behavior. More results on investigation of impulses on surface discharge characteristics of PE is
presented in paper VI. In that paper, impulses were applied separately from the ac and
there was a short time gap between each applied voltage. Almost similar behavior of
surface PD on PE due to the impulse was observed in results of paper VI compared to the
superimposed impulse study. Effect of PE surface flashover on its surface discharge
behavior was also investigated. The results in paper VI shows that surface flashover
caused by the impulse can change the PD activity in either way of being decreased or
increased. Due to the complexity of the flashover mechanism, this cannot be easily
explained.
Figure 5.17 shows a typical surface PD pattern of PE (a) and its PD number variation
over time (b). Unlike pressboard, the surface PD activity of PE has a slightly increased
rate during the 1 hour test.
a) b) Figure 5.17: PE surface PD pattern at 4.3 kV (20% above the inception), a) typical
recorded pattern, b) PD numbers vs time run for about 1 hour at 4.3 kV
5.4 Surface PD on polycarbonate
Plates of 1 mm thick PC were investigated with the IEC-b electrode setup. Below the
surface PD inception, superimposed impulse can initiate the PD which decays after
several seconds, Figure 5.18. Above the PD inception, impulses cannot affect the surface
PD behavior notably and depending on the impulse magnitude, the intensity of PDs,
abruptly following the impulse, can differ (Figure 5.19). A typical PRPD pattern of PC
and the time trend of its PD charges are shown in Figure 5.20. The result shows that by
time, the PD activity on PC surface has a slight increasing rate (similar to PE). However,
a different PRPD pattern can be observed for PC due to its different chemical structure
and electrical properties.
0 10 20 30 40 50 603
3.5
4
4.5
5
5.5
6
6.5
Time (min)
Num
ber
of
PD
per
50 H
z c
ycle
positive charge
negative charge
44
Figure 5.18: PC surface PD pattern vs time for 10 s time window at 3.2 kV (90% of the
inception) with 11 kV superimposed impulse, 34.8% RH and 19.7 º C
Figure 5.19: PC surface PD at 4.6 kV (20% above the inception) a few cycles before and
after the impulse, a) with 16.5 kV superimposed impulse, b) with 11 kV superimposed
impulse, 34.8% RH and 19.7 º C
a) b) Figure 5.20: PC surface PD pattern at 4.6 kV (20% above the inception), a) typical
recorded pattern, b) PD numbers vs time run for about 1 hour at 4.6 kV
-0.1 -0.05 0 0.05 0.1-8
-6
-4
-2
0
2
4
6
8
Time [s]
Vo
lta
ge
[V
]
PD
input voltage
-0.1 -0.05 0 0.05 0.1-8
-6
-4
-2
0
2
4
6
8
Time [s]
Vo
lta
ge
[V
]
PD
input voltage
0 5 10 15 20 25 30 35 40 452.5
3
3.5
4
4.5
5
5.5
6
6.5
Time (min)
Num
ber
of
PD
per
50 H
z c
ycle
positive charge
negative charge
45
5.5 Discussion
The effect of superimposed impulse on surface discharge characteristics of OIP, oil-
pressboard, PE and PC showed some similar and some distinctive behavior. An applied
impulse can initiate the ac partial discharges. Formation of space charges on the surface
due to the impulse voltage can change the electric field, and depending on the polarity of
the impulse the field distribution will be different. Space charges remain on the surface of
materials for a certain time which is affected by the ac electric field and ion mobility.
Surface PD initiation below the inception voltage due to the impulse was a temporary
effect with different decay time for each material. Both paper and pressboard showed
increase in their PD activity above the inception due to the impulse. However, their PD
patterns were different from each other. Behavior of PD with impulse effect could also be
affected by the impulse polarity and the properties of the interface materials in dealing
with charges. Although the effect of negative impulse was only investigated for
pressboard, it can be of interest to know the behavior of other materials responding to
negative charges injected by the impulse. The electrode configuration and the type of
electric field stresses that it applies to the insulation can also affect the PD behavior. OIP
samples with moisture content showed different surface PD behavior with impulse
compared to the corresponding dry samples. Water is a polar dielectric and its specific
electrical characteristic can change the PD behavior. The effect of humidity on surface
discharge behavior of PE with impulse showed its important role affecting the surface PD
behavior. The mechanisms involved in each observed pattern have many routes in the
chemical and physical reaction of the materials. A detailed identification would require
further modeling and experimental work for each case, outside the scope of this thesis.
Surface charges can change the conductivity of boundary layer [62] and PD behavior
evaluation gets more difficult. However, the distinctive PD behavior of how the ac PD is
affected by the impulse for each material can provide interesting information about the
insulation and can be helpful in its condition assessment.
46
47
Chapter 6
Experimental Results on Cavity Discharges
Cavities are important PD sources in oil-impregnated paper insulation. Inclusion of
gas bubbles or voids in cellulose can weaken its dielectric properties and make it
susceptible to PD activity. This chapter presents the results on investigation of impulse
effect on the cavity PD behavior of OIP. The investigation is done on several
characteristics of cavity PD including the PD initiation by the impulses. Moisture content,
as a challenging factor for OIP insulation performance, can affect the PD activity.
Therefore, its influence on OIP cavity PD exposed to the impulse has also been
investigated.
6.1 Cavity PD in OIP and impulse
The OIP introduced in the sample preparation section were used for the cavity PD
studies. 16 layers of OIP with an embedded cylindrical cavity of 8 mm diameter and 400-
µm height in the middle were used. The samples were placed between two brass
electrodes and assembled before being immersed in the transformer oil (Figure 6.1) to
form a gas filled cavity [63].
Figure 6.1: OIP cavity measurement set up
When a voltage is applied to the sample, the electric field in the air-filled void will be
higher than surrounding OIP. Therefore, by applying the impulse, it is likely that the
electric field in the cavity goes beyond its critical level for breakdown into PD. The
experiment was done to check whether the impulse can initiate the cavity PD below the
48
inception voltage. The ac voltage was applied to a sample at several steps below the
inception voltage, and a superimposed impulse was applied at the positive peak of the ac.
The Figure 6.2 shows that at voltages close to the cavity inception voltage, the impulse
voltage can initiate continuous ac PD activity. If the applied ac voltage level was lower
than 80% of the inception, the PD initiation by the impulse had more temporary behavior.
Therefore, an impulse cannot necessarily initiate the cavity PD at any ac voltage level
below the inception. The effect of impulse amplitude on PD initiation at voltages close to
the inception was also investigated in paper IV. The results show that at ac voltages
close to the inception, the PD can get initiated even with low magnitude impulse.
However, there can be more involved parameters such as insulation condition which can
participate in the PD behavior.
a) b)
c) d)
Figure 6.2: Cavity PD pattern vs time for 10 s time window with impulse effect, a)with ac
at 98% of inception b) with ac at 90% of inception c) with ac at 80% of inception d) with ac at
75% of inception of 6.7 kV, All measured with a 22 kV impulse.
Another study was done on the impulse effect at cavity PD inception voltage. An
interesting phenomenon was observed on the PD behavior at its inception voltage with
applied impulse. When the electric field in the cavity is just at its PD inception level, the
applied superimposed impulse could extinguish the PD for several cycles following the
impulse, Figure 6.3. There may be several reasons leading to this behavior including
49
reduced electric field, ion neutralization, charge depletion or high conducting void
surface that would shield the void interior from the electric field [62]. This pattern was
observed on all samples used for this study.
a) b)
Figure 6.3: Oil-paper cavity PD at 5.2 kV inception with 20 kV superimposed impulse, a)
a few cycles before and after the impulse, b) PD pattern vs time for 10 s time window
Above the PD inception voltage of a cavity, quite many discharges occur and the
electric field in the cavity is high enough for continuous PD activity and supply of seed
electrons are more than sufficient. When the impulse applies to the cavity above its
inception, it is less likely to be affected by the impulse. Figure 6.4 shows the PD behavior
at ac voltage of 20% above the PD inception before and after the impulse. It can be
inferred that only for one cycle the PD gets affected by the impulse and for the following
cycles the effect is not notable.
Figure 6.4: Oil-paper cavity PD a few cycles before and after the impulse, at 6.4 kV (20%
above the inception), with 20 kV superimposed impulse
-0.1 -0.05 0 0.05 0.1-1.5
-1
-0.5
0
0.5
1
1.5
Time [s]
Vo
lta
ge
[V
]
PD
input voltage
-0.1 -0.05 0 0.05 0.1-1.5
-1
-0.5
0
0.5
1
1.5
Time [s]
Vo
lta
ge
[V
]
PD
input voltage
50
In some situations, there can be several cavities with different sizes within the solid
insulation. This can lead to a different PD inception condition for each of them. Another
interesting behavior was observed for samples with different cavity sizes imposed to the
impulse above the inception voltage. Samples of 16 layers of OIP were used. One 8 mm
and four 3 mm diameter disc-shaped cavities were made at the middle layers with a
thickness of 400 µm. An ac voltage applied above the PD inception level was applied,
and a superimposed impulse was triggered at the ac positive peak voltage. The result
shows that by an applied impulse, a cluster of larger magnitude PD can be observed in
the PD pattern for several cycles after the impulse, Figure 6.5. This pattern can be due to
the presence of cavities with different sizes in the OIP.
Figure 6.5: PD pattern of oil- paper with several cavities vs time for 10 s time window with
impulse effect, at 6 kV (20% above the 5 kV inception) with 22 kV impulse
6.2 Cavity PD in OIP with moisture content and Impulse
As it was already mentioned, moisture can affect the electrical properties of the paper
insulation and changes its PD behavior. Therefore, the OIP with two levels of moisture
content, prepared by the procedure mentioned in chapter 3, was used to study the OIP
cavity PD behavior in presence of the moisture and the imposed impulse. The same size
samples with a single cavity were used for this study as used in section 6.1. The PD
inception voltage of OIP with 3% and 5.5% moisture contents was measured lower than
the dry OIP (less than 0.5% moisture content) [91], [92].
The same investigation done on dry paper was performed on the cavity PD behavior of
OIP with moisture content. The cavity PD initiation by an impulse was examined on OIP
with 3% and 5.5% moisture contents. The OIP with 3 % moisture content showed the
same behavior as dry paper for impulses below the ac inception voltage. Figure 6.6 shows
that with ac voltage 10% below the inception, the applied impulse could initiate the PD.
51
This PD activity below the inception can be either continuous or temporary for several
seconds, for the OIP with 3 % moisture content. However, The PD did not get initiated
for the OIP sample with higher moisture content (5.5%) below the inception voltage.
a) b)
Figure 6.6: Oil-paper with 3% moisture content, cavity PD at 2.7 kV (90% of the
inception) with 22 kV superimposed impulse, a) a few cycles before and after the impulse, b)
PD pattern vs time for 10 s time window
With ac voltage at the cavity PD inception, OIP with 3% moisture content showed the
same pattern as the dry OIP, and impulse could extinguish PDs for several following
cycles, Figure 6.7. However, for OIP with 5.5% moisture content, applied impulse could
increase the number and magnitude of charges for several following cycles after the
impulse, Figure 6.8. It is good to note that the superimposed impulse triggering circuit
couldn’t work with ac voltages below 3.5 kV and due to the low inception voltage of OIP
with moisture content the impulses were triggered manually for those voltages.
Therefore, as can be observed in Figure 6.7, the impulse is applied at the negative peak of
the ac but still it showed the same effect as when it was applied at positive cycle.
a) b)
Figure 6.7: Oil-paper with 3% moisture content, cavity PD at 3 kV inception with 22 kV
superimposed impulse, a) a few cycles before and after the impulse, b) PD pattern vs time for
10 s time window
-0.1 -0.05 0 0.05 0.1-1.5
-1
-0.5
0
0.5
1
1.5
Time [s]
Vo
lta
ge
[V
]
PD
input voltage
-0.1 -0.05 0 0.05 0.1-1.5
-1
-0.5
0
0.5
1
1.5
Time [s]
Vo
lta
ge
[V
]
PD
input voltage
52
Above the PD inception voltage, the impulse could not affect the cavity PD in OIP
with moisture content, Figure 6.9, as it was observed for the dry paper.
a) b)
Figure 6.8: Oil-paper with 5.5% moisture content, cavity PD at 3.8 kV inception with 22
kV superimposed impulse, a) a few cycles before and after the impulse, b) PD pattern vs time
for 10 s time window
Figure 6.9: Oil-paper with 5.5% moisture content, cavity PD a few cycles before and after
the impulse, at 4.8 kV (20% above the inception), with 22 kV superimposed impulse
6.3 Discussion
The results obtained by the investigation of impulse on the ac cavity PD in OIP
showed quite interesting and distinctive behavior for different situations. The
superimposed impulse can initiate the PD, when the ac voltage over the sample is close to
the PD inception voltage. An impulse can provide a suitable condition for electron
-0.1 -0.05 0 0.05 0.1-1.5
-1
-0.5
0
0.5
1
1.5
Time [s]
Vo
lta
ge
[V
]
PD
input voltage
-0.1 -0.05 0 0.05 0.1-1.5
-1
-0.5
0
0.5
1
1.5
Time [s]
Vo
lta
ge
[V
]
PD
input voltage
53
generation. A cavity may show continued PD at a voltage below the inception voltage,
after initiation of PD activity by an impulse. Trapping and detrapping of charges in paper
can also affect the PD behavior after the impulse.
This PD initiation by the impulse can be observed also for paper with 3% moisture
content. However, with higher moisture content (5.5%) the PD initiation couldn’t be
observed due to the impulse. At the ac inception voltage, the PD could extinguish by the
imposed impulse for several cycles after its incident. This behavior can be due to the
change of the electric field over the cavity, decrease of ions or any phenomenon that can
involve in PD extinction. This behavior was also observable for OIP with 3% moisture
content, whereas for OIP with 5.5% moisture content the PD showed increase in
magnitude after the impulse applied at the PD inception which can be due to the high
moisture content. Above the cavity PD inception, the impulse did not have any notable
effect for any of the cases. It is important to consider that clear explanation of each
behavior requires more fundamental investigations which are out of the scope of this
thesis.
54
55
Chapter 7
Experimental Results on Surface Discharges
of Modified Paper Materials
Cellulose or wood-based materials have been widely used in high voltage power
equipment, particularly in power transformers and their bushings. Despite their good
insulation properties, they can be exposed to tangential electrical stresses and be affected
by surface discharges which may lead to their degradation and failure. Therefore, it is of
great interest to inspect whether their properties can be improved by modifying their
surface with nanoparticles.
Silica is used for modification of paper due to its low dielectric constant, low dielectric
loss and resistance to surface degradation. ZnO nanoparticles have a semiconducting
property, which is believed to enable them to trap electrons in the dielectric. Beside
papers with nanoparticle treatment, the paper with higher lignin content was also studied.
Two sets of measurement were done on the modified paper materials. One set of
measurements was done only on dry NFC (Nanofibrillated Cellulose), K92 (high lignin
paper), transformer paper and pressboard in air atmosphere without being impregnated.
Another set of measurements were done on oil-impregnated reference paper provided by
KTH department of Fiber and Polymer Technology, and the modified ones with silica
and ZnO nanoparticles and high lignin K92. The results from the dry unimpregnated
papers are illustrated in this chapter. However, major results on impregnated samples are
presented in paper VII.
7.1 Surface PD on dry cellulose-based materials
Surface PD behavior of NFC and high lignin K92 paper were investigated at their
interface with air without impregnating them. This study was done to purely investigate
the paper behavior without the influence of the transformer oil. Their behavior under
surface PD has been compared to the behavior of unimpregnated paper (Munksjö Thermo
70) and pressboard. Dielectric response of paper, pressboard, NFC and K92 measured by
IDAX-300 is plotted in Figure 7.1.
56
Figure 7.1: Dielectric property of paper, pressboard, NFC and K92
The surface PD inception voltage of each material, air relative humidity and
temperature at which the measurements were done is shown in table 7.1. The air relative
humidity and temperature for all cases are almost the same which makes their
comparison more relevant. Their surface PD was measured 1 kV above the inception
voltage of each case.
Table 7.1: Surface PD measured and applied voltages
Material Inception
Voltage
Voltage of
Measured PD
Superimposed
Lightning Impulse
Air Relative Humidity
and Temperature
paper 4.5 kV 5.5 kV 22 kV 41% RH, 23.9ºC
pressboard 4.3 kV 5.3 kV 22 kV 42.3%RH,24.6ºC
NFC 4.2 kV 5.2 kV 22 kV 39.8% RH, 24ºC
K92 6.2 kV 7.2 kV 22 kV 43.3%RH,24.2ºC
7.1.1 Surface PD on dry paper
Paper materials are mainly used in an oil-impregnated condition in power components.
However, the surface PD behavior of pure paper without impregnation in oil can be
studied for investigation of the paper material effect. 10 layers of paper with 1 mm total
thickness were used and a needle electrode was placed 5 mm away from ground bar
electrode on the surface of the paper. The surface PD inception of paper with air interface
is much lower than the impregnated paper in the oil and it shows how much impregnating
the paper and immersing it in oil can improve its PD resistance property. The ac voltage
was kept 1 kV above the surface PD inception of paper. The superimposed voltage of 22-
kV magnitude was applied to the samples. The impulse increased the surface PD on
paper for few cycles (Figure 7.2) but the effect was less than the effect that was observed
from the impulse on OIP. Figure 7.3 (a) shows a typical surface PRPD pattern on paper
without impregnation and (b) shows the trend of PD numbers changing by the time. It can
be seen that the number of PD charges decreases by time on the paper.
10-2
10-1
100
101
102
103
104
0
2
4
6
8
10
12
14
Frequency (Hz)
Rela
tive P
erm
ittivity
pressboard
paper
NFC
K92
10-2
10-1
100
101
102
103
104
10-3
10-2
10-1
100
101
Frequency (Hz)
tan
pressboard
paper
NFC
K92
57
a) b) Figure 7.2: dry paper surface PD a) a few cycles before and after the impulse, b) PD
pattern vs time for 10 s time window
a) b) Figure 7.3: dry paper surface PD pattern at 5.5 kV (above the inception), a) typical
recorded pattern, b) PD numbers vs time run for about half an hour
7.1.2 Surface PD on dry pressboard
The surface PD on dry pressboard of 2 mm thick with the same needle electrode
configuration as the paper was investigated. With almost the same inception and applied
voltage as paper, the pressboard shows higher magnitude and number of charges rather
than paper. After the impulse, the number of negative charges increases for several
seconds and gets back to the previous level, Figure 7.4. Figure 7.5 (b) shows the number
of PD charges on pressboard run for about half an hour. Increase in negative charges by
the impulse can be observed with sharp peaks shown in red dashed ovals in Figure 7.5
(b). Those are the moments that the superimposed impulse was applied to the sample.
The surface PD on pressboard has quite distinctive behavior due to the impulse.
-0.1 -0.05 0 0.05 0.1-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
Time [s]
Vo
lta
ge
[V
]
input voltage
PD
0 5 10 15 20 25 300
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Time (min)
Num
ber
of
PD
per
50 H
z c
ycle
positive charge
negative charge
58
a) b) Figure 7.4: dry pressboard surface PD at 5.3 kV a) a few cycles before and after the
impulse, b) PD pattern vs time for 10 s time window
a) b)
Figure 7.5: dry pressboard surface PD pattern at 5.3 kV (above the inception), a) typical
recorded pattern, b) PD numbers vs time run for about half an hour
7.1.3 Surface PD on dry NFC
Three layers of NFC papers with 130 µm thickness were placed on 7 layers of normal
paper for surface discharge study of NFC. As both the needle and ground electrode are
placed on the surface of the sample, only NFC paper would be involved in the surface
discharges. The same setup configuration used for NFC as it was used for the paper.
From Table 7.1, it can be seen that the PD inception of NFC is a bit lower than other
samples but in general at almost the same voltage level as paper and pressboard it showed
lower PD activity. Both positive impulses at positive ac peak and negative impulses at
negative ac peak were applied to the NFC sample. Positive impulse could increase the
negative PD for several cycles, which is observable in Figure 7.6. In contrast, negative
impulse could reduce negative PD charges for several cycles following the impulse,
Figure 7.7. Figure 7.8 (b) shows the surface PD trend on NFC by time. It can be
observed that the PD numbers on NFC decreases significantly by the time.
-0.1 -0.05 0 0.05 0.1-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
Time [s]
Vo
lta
ge
[V
]
input voltage
PD
0 5 10 15 20 25 30 350
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
Time (min)
Num
ber
of
PD
per
50 H
z c
ycle
positive charge
negative charge
59
a) b)
Figure 7.6: dry NFC surface PD at 5.2 kV a) a few cycles before and after the positive
impulse, b) PD pattern vs time for 10 s time window
a) b) Figure 7.7: dry NFC surface PD at 6 kV a) a few cycles before and after the negative
impulse, b) PD pattern vs time for 10 s time window
a) b) Figure 7.8: dry paper surface PD pattern at 5.2 kV (above the inception), a) typical
recorded pattern, b) PD numbers vs time run for about half an hour
-0.1 -0.05 0 0.05 0.1-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
Time [s]
Vo
lta
ge
[V
]
input voltage
PD
-0.1 -0.05 0 0.05 0.1-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
Time [s]
Vo
lta
ge
[V
]
input voltage
PD
0 2 4 6 8 10 12 14 16 18 200
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
Time (min)
Num
ber
of
PD
per
50 H
z c
ycle
positive charge
negative charge
60
7.1.4 Surface PD on dry K92
High lignin paper K92, showed higher surface PD inception than the others. Applied
superimposed impulse to K92 didn’t show any specific change in the surface discharges,
Figure 7.9-12, except a little increase in number of charges after the impulse. However, in
a long run of half an hour, the surface PD pattern of K92 had a change. The magnitude of
positive charges increased, Figure 7.9-12, and two distinct clusters of charges were
visible in its PRPD pattern, Figure 7.12. Also, the number of negative charges after being
decreasing, it started to increase again, Figure 7.13. This change might be due to the
moisture absorbing of the K92 paper.
a) b) Figure 7.9: dry K92 surface PD at 7.2 kV a) a few cycles before and after the impulse, b)
PD pattern vs time for 10 s time window, in 3 min
a) b) Figure 7.10: dry K92 surface PD at 7.2 kV a) a few cycles before and after the impulse, b)
PD pattern vs time for 10 s time window, in 15 min
-0.1 -0.05 0 0.05 0.1-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
Time [s]
Vo
lta
ge
[V
]
input voltage
PD
-0.1 -0.05 0 0.05 0.1-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
Time [s]
Vo
lta
ge
[V
]
input voltage
PD
61
a) b) Figure 7.11: dry K92 surface PD at 7.2 kV a) a few cycles before and after the impulse, b)
PD pattern vs time for 10 s time window, in 27 min
a) b) Figure 7.12: dry K92 surface PD pattern at 7.2 kV (above the inception), a) at the
beginning of the measurement b) at the end of the measurement (after 30 min)
Figure 7.13: dry K92 surface PD numbers vs time run for about half an hour at 7.2 kV
-0.1 -0.05 0 0.05 0.1-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
Time [s]
Vo
lta
ge
[V
]
input voltage
PD
0 5 10 15 20 25 300
0.5
1
1.5
2
2.5
Time (min)
Num
ber
of
PD
per
50 H
z c
ycle
positive charge
negative charge
62
7.2 Surface PD on modified oil-impregnated papers
The paper samples with ZnO and silica nanoparticles were impregnated in the
transformer oil and their surface PD behavior was compared to the impregnated reference
paper and K92. The main results of this study are presented in paper VII. All the papers
including the ones with ZnO and silica nanoparticles and K92 showed lower permittivity
(measured by Dielectric Response Analyzer, IDAX 300) than the reference paper at
frequencies below 0.1 Hz and their surface PD inception were higher than the reference.
Below and close to the inception voltage, superimposed impulse couldn’t initiate the
surface PD on papers with silica nanoparticles and K92. The reference paper behaved the
same as ordinary OIP used in chapter 5 and showed a temporary surface PD activity
following the impulse. The surface PD on samples with ZnO nanoparticles also got
initiated by the impulse and decayed after several cycles, but it showed a bit different
pattern of partial discharges compared to the reference (paper VII).
Above the surface PD inception, again the reference paper and the paper with ZnO
nanoparticles had an increase in their PD activity following the impulse, but the impulse
effect on paper with ZnO particles are less than its effect on the reference paper and the
injected charges by the impulse decayed faster. Three types of paper with silica
nanoparticles were used. Two types with 2wt% silica, one well washed sample (W-SiO2
2 wt%) and another with more ions attached. The washed silica particle filled papers, had
the lowest dielectric loss compare to other materials and had a flat low permittivity at all
frequencies. It did not get affected by the impulse and its PD activity remained
unchanged after the impulse. However, the other one showed occurrence of high
magnitude PDs after the impulse, similar to the pattern observed from the oil-pressboard,
Figure 7.14. The paper with 6 wt% silica had quite high amount of PD above the
inception but didn’t show any change in its PD activity with applied superimposed
impulse. The K92 paper had the lowest permittivity at frequencies above 1 Hz and
highest surface PD inception. It had quite low PD activity above the inception compared
to the reference with lower lignin content and its PD behavior remained unchanged with
the impulse, Figure 7.15.
a) b)
Figure 7.14: Paper with SiO2 2 wt% nanoparticles surface PD a) a few cycles before and
after the impulse, b) PD pattern vs time for 10 s time window at 28 kV (20% above 26 kV
inception), with 28 kV impulse
-0.1 -0.05 0 0.05 0.1-6
-4
-2
0
2
4
6
Time [s]
Vo
lta
ge
[V
]
input voltage
PD
63
a) b)
Figure 7.15: Paper with high lignin (K92), surface PD a) a few cycles before and after
the impulse, b) PD pattern vs time for 10 s time window at 30 kV ( 20% above 28 kV
inception), with 28 kV impulse
7.3 Discussion
Modifying the paper with nanoparticles such as silica can reduce the permittivity of the
paper to the value closer to the permittivity of the oil. The factor that cause the space
charges form at the oil-paper interface is the permittivity difference between the oil and
the paper. Therefore, reducing the permittivity of paper close to the oil permittivity with
nanoparticles can improve their surface PD behavior. However, the sample preparation
and nanoparticles concentration in the paper can affect their surface PD, and the
permittivity reduction cannot necessarily improve their PD behavior. Applied impulses
generate space charges at the oil-paper interface. As already mentioned in chapter 5, drift
of charges through the oil and trapping and detrapping of charges from the paper surface,
can describe the PD behavior after the impulse. Nanoparticles characteristic in dealing
with charges can somewhat explain the modified paper surface PD behavior with applied
impulse. Papers with treatment of added nanoparticles can show good behavior for
surface PD resistance but they are not necessarily suitable as bulk insulation and their ac
and impulse breakdown strength must be well investigated. As moisture is an important
factor in affecting the dielectric behavior of the paper and significantly influences their
surface PD behavior, the hygroscopic behavior of paper with nanoparticle is another
important factor that must be investigated.
-0.1 -0.05 0 0.05 0.1-6
-4
-2
0
2
4
6
Time [s]
Vo
lta
ge
[V
]
input voltage
PD
64
65
Chapter 8
Summary of Publications
Paper I
In this paper, the possibility of utilizing power system transients as stimuli for on-line
dielectric response measurements of bushings has been studied. This new proposed
technique is tested in the lab by using a HV superimposed impulse on an ac voltage, on a
150 kV in-service aged power-transformer bushing. The challenges and measurement
considerations which must be taken into account to improve the performance of the
technique is addressed in this paper. Analysis is done on processing of measured signals
to achieve a reasonable accuracy in the results. It shows that the measurement
considerations and a wise choice of data acquisition, and data processing parameters, can
lead to a suitable accuracy in DR measurement by using superimposed impulse on the ac
voltage, and there is a high potential for this technique to be used for frequent monitoring
of dielectric properties of power transformer bushings.
Paper II
This paper investigates the same technique that is proposed in Paper I. Dielectric
response measurement by utilizing lightning and switching impulses is presented. The
experiments were performed on oil-impregnated paper as a major insulation of power
transformers and their bushings. The capability of the proposed technique in detection of
moisture content and temperature as responsible parameters in insulation failure of power
transformers and their bushings is investigated. The study is done by using both low
voltage and high voltage impulses. Different circuit configuration and considerations,
according to these two voltage levels are discussed. The results show a good agreement
with the results obtained from off-line commercial instruments for dielectric
spectroscopy.
Paper III
Diagnostics of a power transformer bushing with the technique proposed in Paper I
and II is presented. Measurements were done on two 150 kV power transformer
bushings. In order to investigate the ability of the technique for diagnosis of the bushings,
an artificial defective state was created in one of the bushings by removing part of the
insulating oil and exposing it to 200 kV HV lightning impulses. After refilling the oil, the
66
degradation state initiated partial discharges in that bushing. Standard HV lightning
impulses were applied to the power transformer bushings and their transient responses
were measured through their test taps. The dielectric response of the damaged bushing
did not show any notable change and could be obtained by the new technique. However,
partial discharges could be observed from the transient response voltage of the damaged
bushing, with many discharges superimposed on its impulse response voltage.
Paper IV
This paper investigates the effect of superimposed impulses on ac voltage, on surface
and cavity partial discharges of oil-impregnated paper and oil-impregnated pressboard.
The PD behaviors before and after the impulse are compared for ac voltages below and
above the ac PD inception voltage. The influence of moisture content in the OIP is
studied on papers with moisture content of <0.5%, 3.0% and 5.5%. Distinctive behavior
of PD due to the impulse was observed for each defect and insulation condition either by
change in the PD rate and magnitude or by change in PD inception and extinction voltage
level. This investigation showed the importance of recording the voltage signals and
associated PD measurements during and after the incidence of a high voltage transient.
This can be used for condition assessment of the insulation of power transformers and
their bushings, as a form of online diagnostics.
Paper V
This paper investigates the effect of transients such as lightning impulses on the
surface discharge characteristics of oil-impregnated paper. In this study impulses were
applied to the samples in subsequent applications of ac stresses. Two different electrode
configurations, rod (IEC-b) and needle-ground bar on surface were used for the study. In
the needle electrode configuration, surface discharges couldn’t notably damage the paper
due to the low normal field component, and the applied impulses caused a prompt change
in the total number of discharges which decayed fast. However, with the rod electrode,
the paper gets carbonized and the total number of discharges decreases under the applied
impulse. The influence of moisture content (5.5%) in oil-impregnated paper was also
studied. For this case, the total number of discharges increases considerably after an
applied impulse.
Paper VI
This paper investigates the influence of high voltage impulses on the surface discharge
characteristics of polyethylene (PE) during subsequent application of 50 Hz ac voltage.
Two types of PE are used for this study. Surface discharge features of PE due to the
impulse were investigated in the case of having either a streamer or flashover on the
surface. A prompt change in surface PD numbers was observed after the applied impulse,
and they decreased with time. However, under surface flashover due to the impulse, the
PD behavior was not consistent. Flashover could either increase the PD numbers or leave
them unchanged
67
Paper VII
In this paper, newly developed paper materials with nanoparticles were used for
surface discharge investigation of oil-impregnated paper under ac voltage with a
superimposed impulse. Two different concentrations (2 wt% and 6 wt %) of silica (SiO2)
and zinc oxide (ZnO) nanoparticles (1 wt%) were used in modification of Kraft paper
samples. The modified paper with nanoparticles could improve the partial discharge
behavior of OIP at its interface with oil. Paper samples with low concentration of silica
with rigorous preparation showed an improvement in surface PD resistance under the
impulse. Papers with ZnO and 6 wt% silica could not perform as well as papers with 2
wt% silica due to the surface PD and the impulse. This shows the importance of the type
and concentration of nanoparticles in improvement of surface PD behavior of the
modified OIP. The effect of high lignin content in Kraft paper is also studied which
showed a better surface PD characteristic under the impulse, compared to the paper with
low lignin content.
68
69
Chapter 9
Conclusions and Future Work
9.1 Conclusions
9.1.1 Dielectric response measurements by utilizing transients
In this thesis a novel technique for on-line diagnostics of power components,
specifically power transformer bushings, has been proposed. Impulses can be the stimuli
from which, the measurement of dielectric response is made in the proposed technique.
The use of natural impulses such as lightning and switching is suggested as a way of
simplifying the method.
The usefulness of the technique has been investigated through laboratory
measurements on real power transformer bushings. Measurement requirements,
considerations and limits are discussed and analysis is done to evaluate the technique in
achieving on-line DR with reasonable accuracy.
The feasibility of the technique for diagnostics of moisture content and temperature in
insulation of power transformer bushings was investigated through DR measurement of
oil-impregnated paper samples with lightning and switching impulses. The good
agreement of the results with FDS results shows its capability to detect the defects that
change the DR at high frequencies.
In general, according to the obtained results, the proposed technique seems promising
in monitoring of the insulation of power transformer bushings by utilizing transients and
it can be used as an on-line dielectric spectroscopy without interrupting the operation of
the component or requiring an external source.
9.1.2 Partial discharge behaviors under the impulses
The effect of power system transients, simulated as standard lightning impulses, on
partial discharge behavior of several types of insulation, was presented. Oil-impregnated
paper, as a major insulation of power transformers and bushings, is investigated in the
major part of the study. OIP and pressboard are conventional insulations that have been
used for decades. The behavior of surface and cavity insulation defects as two major PD
sources is studied under ac and HV impulses.
The effect of impulses in initiation of PD under ac inception voltages was investigated
for surface and cavity defects in OIP to test the hypothesis presented in Figure 1.2. The
results showed that there is a high probability for PD inception of OIP under an impulse,
70
and its continuous activity depends on the defect type, ac voltage, and impulse magnitude
and insulation degradation level. The surface PD initiation by the superimposed impulse,
at ac voltages close to the ac PD inception, had a temporary activity for several seconds.
However, for the cavity defect, the PD initiation had more continuous behavior.
Surface and cavity ac PD behavior, before and after the impulse, with ac voltages above
the ac PD inception voltage, have their own feature for each defect and insulation
condition. The distinctive behavior of PD under the impulse for each condition can be
used for further condition assessment of the insulation and supports the importance of the
study. The effect of moisture content of the OIP on its PD behavior under the impulse,
also showed distinctive characteristics for each defects; e.g. the effect of moisture content
(3% and higher) in OIP for surface discharges, above the PD inception voltage, was a
temporary decrease in the PD activity following the impulse, whereas the surface PD
activity on a dry OIP, showed an increase after the impulse.
Surface PD characteristics of polyethylene and polycarbonate, as commonly used
polymers, under superimposed impulse were also investigated.
In general, according to the obtained results, recordings of the voltage signals and
associated PD measurements, during and after the incidence of a high voltage transient,
are of potential use for condition assessment of the insulation of power components as a
form of on-line diagnostics. For the transformers, this can be done through the Rogowski-
coils installed at the bottom of the bushings which measures high frequency currents and
recording of the data can be triggered by the transients.
Surface PD behavior of new developed paper insulations with nanoparticles (silica and
zinc oxide) was also studied under the impulse. The silica nanoparticles with low
concentration showed PD resistive behavior under the impulse. Paper with high lignin
content as another investigated material also showed good PD resistant behavior, both
under the ac and superimposed impulse. This shows a possibility to improve the
performance of paper material under PD degradation and impulse stresses.
9.2 Future work
Further work that can be developed as a continuation of this project is described as
follows:
The on-line dielectric response technique by transients can be further developed
and installed in the field on real power transformer bushings, in order to evaluate
its performance in the field.
Artificially added moisture or hotspots can be defined in the bushing for more
investigation of DR measurement by transients.
Application of on-line DR can be investigated for other power components such
as power cables.
Electromagnetic simulation can be done for both cavity and surface discharges
under the superimposed impulse on ac voltage for further investigation of the
physical phenomena.
71
PD measurements can be done in conditions closer to the real application of OIP
at operation temperature of the transformer and with more controlled conditions
of moisture and gas content of the oil.
Further investigations can be done on the effect of negative superimposed
impulses applied to the different materials at different phases of the ac cycle.
The effect of impulses on thermally aged samples, or aged samples by PD, can
also be investigated.
Further studies on the physics of PD behavior under the impulse for different
materials and insulation condition are desirable.
72
73
Bibliography
[1] R.E. James and Q. Su, Condition Assessment of High Voltage Insulation in Power System
Equipment, The Institution of Engineering and Technology (IET), UK, ISBN: 978-0-86341-
737-5, 2008.
[2] J. Barker, "High Voltage Asset Monitoring for Today’s Smart Grid", ABB Inc. report,
Monitoring and Smart grid, June 2013.
[3] M. Puetter, C. Enk, I. Hong, M. Kruger, A. Kraetge, "On Site Measurement on Power
Transformers Using the whole Testing Toolbox", 18th International Symposium on High
Voltage Engineering (ISH), Seoul, South Korea, 2013.
[4] T. McGrail, K.Elkinson, "Asset Management: Power Transformer Fleet Condition Assessment
and Risk Management", 18th International Symposium on High Voltage Engineering (ISH),
Seoul, South Korea, 2013.
[5] M. Wang and A. J. Vandermaar, "Review of Condition Assessment of Power Transformers in
Service", IEEE Electrical Insulation Magazine, Vol. 18, No. 6, pp. 12-25,
November/December, 2002.
[6] Dieter König and Y. Narayana Ha, Partial Discharges in Electrical Power Apparatus, Chapter
7, vde-verlag gmbh, Berlin, Offenbach, 1993.
[7] A. M. Emsley and G. C. Stevens, "Review of Chemical Indicators of Degradation of
Cellulosic Electrical Paper Insulation in Oil-filled Transformers", IEE Proceedings Science,
Measurement and Technology, Vol.141, pp. 324-334, 1994.
[8] "IEEE Guide for the Interpretation of Gases Generated in Oil-immersed Transformers", IEEE
Standard C57.104-1991, pp. 1-30, 1991.
[9] B. Pahlavanpour, M. A. Martins, and A. DePablo, "Experimental Investigation into the
Thermal-ageing of Kraft Paper and Mineral Insulating Oil", IEEE International Symposium on
Electrical Insulation, Boston, MA, USA, pp. 341-345, 2002.
[10] A. M. Emsley, X. Xiao, R. J. Heywood, and M. Ali, "Degradation of Cellulosic Insulation in
Power Transformers. Part 2: Formation of Furan Products in Insulating Oil", IEE Proceedings
Science, Measurement and Technology, Vol. 147, pp. 110-114, 2000.
[11] L. E. Lundgaard, W. Hansen, D. Linhjell and T. J. Painter, "Aging of Oil-impregnated Paper
in Power Transformers", IEEE Transactions on Power Delivery, Vol. 19, No. 1, pp. 230-239,
2004.
74
[12] Walter S. Zaengl, "Dielectric Spectroscopy in Time and Frequency Domain for HV Power
Equipment, Part I: Theoretical Considerations", IEEE Electrical Insulation Magazine, Vol. 19,
No. 5, pp. 5-19, September/October 2003.
[13] U. Gäfvert, L. Adeen, M. Tapper, P. Ghasemi and B. Jonsson, "Dielectric Spectroscopy in
Time and Frequency Domain Applied to Diagnostics of Power Transformers", Proceedings of
the 6th International Conference on Properties and Applications of Dielectric Materials, Vol.
2, pp. 825-830, China, 2000.
[14] R. Neimanis, L. Lennholm, and R. Eriksson, "Determination of Moisture Content in
Impregnated Paper Using Near Infrared Spectroscopy", IEEE Conference on Electrical
Insulation and Dielectric Phenomena (CEIDP), Vol. 1, pp. 162-165, Austin, TX, 1999.
[15] T. K. Saha, "Review of Time Domain Polarisation Measurements for Assessing Insulation Condition in Aged Transformers", IEEE Transactions on Power Delivery, Vol. 18, No. 4, pp. 1293 - 1301, 2003.
[16] T. K. Saha, "Review of Modern Diagnostic Techniques for Assessing Insulation Condition in Aged Transformers", IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 10, No. 5, 2003.
[17] M. Jaya, D. Geißler and T. Leibfried, "Accelerating Dielectric Response Measurements on Power Transformers—Part I: A Frequency-Domain Approach", IEEE Transactions on Power Delivery, Vol. 28, No. 3, pp. 1469- 1473, 2013.
[18] K. G. N. B. Abeywickrama, Y. V. Serdyuk and S. M. Gubanski, "Exploring Possibilities for
Characterization of Power Transformer Insulation by Frequency Response Analysis (FRA) ",
IEEE Transactions on Power Delivery, Vol. 21, No. 3, pp. 1375-1382, 2006.
[19] D. Chu and A. Lux, "On-Line Monitoring of Power Transformers and Components: A Review
of Key Parameters", IEEE Transformer On-Line Monitoring Task Force, 2000.
[20] R. Schwarz and M. Muhr, "Measurement Techniques for Transformer Diagnostic",
International Symposium on Electrical Insulating Material, pp. 326-329, Mie, Japan, 2008.
[21] A. Setayeshmehr, A. Akbari, H. Borsi and E. Gockenbach, "On-line Monitoring and Diagnosis of Power Transformer Bushings", IEEE Tranactions on Dielectrics and Electrical Insulation, Vol. 13, No. 3, 2006.
[22] M. Jaya, T. Leibfried and M. Koch, "Information within the dielectric Response of Power
Transformers for Wide Frequency Ranges", IEEE International Symposium on Electrical
Insulation, pp. 1-5, San Diego, 2010.
[23] B. Sonerud, T. Bengtsson, J. Blennow and S. M. Gubanski, "Dielectric Response
Measurements Utilizing Semi-Square Voltage Waveforms", IEEE Transactions on Dielectrics
and Electrical Insulation, Vol. 15, No. 4, pp. 920-926, 2008.
[24] 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, No.
2, March/April 2010.
[25] R.A. Fouracre E. Santos, I. Timoshkin, M. J. Given and S. J. Macgregor, "Surface Discharge
Propagation: The Influence of Surface Charge", Twenty-Seventh International Power
Modulator Symposium, pp. 39-42, Arlington, VA, 2006.
75
[26] R. Liu, C. Tornkvist, and K. Johansson, "Space charge distribution in composite oil/cellulose
insulation", IEEE Conference on Electrical Insulation and Dielectric Phenomena (CEIDP), pp.
316–321, Arlington, TX, 1994.
[27] J. Dai, Z.D. Wang and P. Jarman, "Moisture and Aging Effect on the Creepage Discharge
Characteristics at the oil\Transformer-board Interface under Divergent Field", IEEE
Conference on Electrical Insulation Dielectric Phenomena (CEIDP), pp. 662-665, Quebec,
2008.
[28] C. Forssén and H. Edin, "Partial discharges in a cavity at variable applied frequency part 1:
measurements", IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 15, No. 6, pp.
1601-1609, 2008.
[29] M. Ghaffarian Niasar, et al. "Partial discharges in a cavity embedded in oil-impregnated paper:
effect of electrical and thermal aging", IEEE Transactions on Dielectrics and Electrical
Insulation, Vol. 22, No.2, pp. 1071-1079, 2015.
[30] A. Cavallini, G. C. Montanari and F. Ciani, "Analysis of partial discharge phenomena in
paper-oil insulation systems as a basis for risk assessment evaluation", IEEE International
Conference on Dielectric Liquids (ICDL), pp. 241-244, 2005.
[31] R. Altenburger, C. Heitz and J. Timmer, "Analysis of phase-resolved partial discharge patterns
of voids based on a stochastic process approach", Journal of Physics D: Applied Physics,
Vol. 35, pp. 1149-1163, 2002.
[32] C. Hudon and M. Belec, "Partial discharge signal interpretation for generator
diagnostics", IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 12, No. 2, pp.
297-319, 2005.
[33] J. Li, W. Si, X. Yao and Y. Li, "Measurement and simulation of partial discharge in oil
impregnated pressboard with an electrical aging process", Measurement Science and
Technology Journal, Vol. 20, No. 10, 2009.
[34] P. H. F. Morshuis and J. J. Smit. "Partial discharges at DC voltage: their mechanism, detection
and analysis", IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 12, No. 2,
pp.328-340, 2005.
[35] L. Bao, J. Li, J. Zhang, T. Jiang and X. Li, "Partial Discharge Process and Characteristics of
Oil-Paper Insulation under Pulsating DC Voltage", Journal of Electrical Engineering and
Technology, Vol. 11, No. 2, pp. 436-444, 2016.
[36] Elisabeth Lindell, Partial Discharges at Repetitive Rapidly Changing Voltages, Doctoral
thesis, Chalmers University of Technology, 2009.
[37] E. Lindell, T. Bengtsson, J. Blennow and S. M. Gubanski, "Measurement of Partial Discharges
at Rapidly Changing Voltages", IEEE Transactions on Dielectrics and Electrical Insulation,
Vol. 15, No. 3, pp.823-831, 2008.
[38] Thomas Hammarström, Partial Discharges at Fast Rising Voltages, Doctoral thesis, Chalmers
University of Technology, 2014.
[39] T. Hammarström, T. Bengtsson, J. Blennow and S. M. Gubanski, "Evidence for Changing PD
Properties at Short Voltage Rise Times", IEEE Transactions on Dielectrics and Electrical
Insulation, Vol. 18, No. 5, pp.823-831, 2011.
76
[40] B. Holmgren, Dielectric response, breakdown strength and water tree content of medium
voltage XLPE cables, KTH Department of Power Engineering, Stockholm, Sweden, 1997.
[41] Y. Kamata, et al., "Dielectric strength of oil-immersed transformer insulation with
superimposed AC and lightning impulse voltage", IEEE Transactions on Dielectrics and
Electrical Insulation, Vol. 25, No. 4, pp.683-687, 1990.
[42] G. Ueta, T. Tsuboi, J. Takami and S. Okabe, "Insulation Characteristics of Oil-immersed
Power Transformer under Lightning Impulse and AC Superimposed Voltage", IEEE
Transactions on Dielectrics and Electrical Insulation, Vol. 21, No. 3, pp. 1384-1392, 2014.
[43] R. J. Densley and B. Salvage, "Partial discharges in gaseous cavities in solid dielectrics under
impulse voltage conditions", IEEE Transactions on Dielectrics and Electrical Insulation, Vol.
EI-6, No. 2, pp.54-62, 1971.
[44] Z. Sun, X. Zhao, J. Li, and Y. Li, "Experiment investigation of partial discharges under
impulse voltage", Proceedings of the 9th International Conference on Properties and
Applications of Dielectric Materials, paper No. E-31, Harbin, China, July 2009.
[45] R. J. Densley, "Partial discharges in electrical insulation under combined alternating and
impulse stresses", IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 5, pp. 96-
103, 1970.
[46] M. Roy, J. K. Nelson, R. K. MacCrone, Linda S. Schadler, C. W. Reed and Richard Keefe,
"Polymer nanocomposite dielectrics-the role of the interface", IEEE Transactions on
Dielectrics and Electrical Insulation, Vol. 12, no. 4, pp. 629-643, 2005.
[47] T. J. Lewis, "Nanometric dielectrics", IEEE Transactions on Dielectrics and Electrical
Insulation, Vol. 1, No.5, pp. 812-825, 1994.
[48] J. K. Nelson, J. C. Fothergill, L. A. Dissado and W. Peasgood, "Towards an understanding of
nanometric dielectrics", Annual Report Conference on Electrical Insulation and Dielectric
Phenomena (CEIDP), pp. 295-298, 2002.
[49] T. Tanaka, "Dielectric nanocomposites with insulating properties", IEEE Transactions on
Dielectrics and Electrical Insulation, Vol. 12, No.5, pp. 914-928, 2005.
[50] T. Tanaka, G. C. Montanari and R. Mulhaupt, "Polymer nanocomposites as dielectrics and
electrical insulation perspectives for processing technologies, material characterization and
future applications", IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 11, No.5,
pp. 763-784, 2004.
[51] C. Meichsner, T. Clark and P. Groeppel, "Protective layer of AL 2 O 3 nanocomposites and
surface composition after electrical stress", IEEE Transactions on Dielectrics and Electrical
Insulation, Vol. 22, No. 5, pp. 2944-2950, 2015.
[52] C. Hoffmann, and F. Jenau, "On surface degradation mechanisms of epoxy filled with Silica
nanoparticles caused by partial discharges", 10th IEEE International Conference on Solid
Dielectrics (ICSD), pp. 1-4, Potsdam, Germany, 2010.
[53] D. Fabiani, A. Cavallini, G. C. Montanari and L. Testa, "Extraction of aging markers for
nanostructured epoxy resin aged under surface discharges", IEEE Electrical Insulation
Conference (EIC), pp. 387-391, Montreal, Canada, 2009.
77
[54] H. Jin, P.Morshuis, A. Rodrigo Mor and T. Andritsch, "An investigation into dynamics of
partial discharge propagation in mineral oil based nanofluids", IEEE International Conference
on Liquid Dielectrics, pp. 1-4, Slovenia, 2014.
[55] C. Pinto, "Life expectancy analysis program for electrical machine insulation", ABB Group,
Global R&D Manager Machines Service, March 4, 2009.
[56] Kuffel et al, High Voltage Engineering, 2nd ed., Newnes 2000.
[57] U. Fromm, Partial Discharge and Breakdown Testing at High DC Voltage, Ph.D. thesis TU
Delft, ISBN 90-407-1155-0, 1995.
[58] Y. Li, H. B. Mu, Y. H. Wei, G. J. Zhang, S. H. Wang, W. Z. Zhang and M. Zahn, "Sub-
microsecond streamer breakdown in transformer oil-filled short gaps", IEEE Transactions on
Dielectrics and Electrical Insulation, Vol. 21, No. 4, pp. 1616-1626, 2014.
[59] I.A. Metwally, "Influence of the solid insulating phase on streaming electrification of
transformer oil", IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 4, No. 3, pp.
327-340, 1997.
[60] 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, No. 2,
2010.
[61] M. Hao, et al., "Space charge behaviour in oil and impregnated pressboard combined
insulation system", IEEE 18th International Conference on Dielectric Liquids (ICDL), 2014.
[62] L. Niemeyer, "A Generalized Approach to Partial Discharge Modeling", IEEE Transaction on
Dielectrics and Electrical Insulation, Vol. 9, No. 4, pp. 510-528, 1995.
[63] M. Ghaffarian Niasar, Mechanisms of Electrical Ageing of Oil impregnated Paper due to
Partial Discharges, Doctoral thesis, KTH Royal Institute of Technology, 2015.
[64] Private communication with ABB Corporate Research, Västerås, Sweden, 2011.
[65] L. B. Rockland, "Saturated Salt Solutions for Static Control of Relative Humidity between 5º
and 40º C", Analytical Chemistry Journal, Vol. 32, No. 10, 1960.
[66] A. Akbari, S. DehPahlevan, and H. Borsi, "Analyzing dynamic of moisture equilibrium in oil-
paper insulation in power transformers for efficient drying", IEEE Conference on Electrical
Insulation and Dielectric Phenomena (CEIDP), pp. 545–548, 2006.
[67] Roya Nikjoo, "Diagnostics of Oil-Impregnated Paper Insulation Systems by Utilizing
Lightning and Switching Transients", Licentiate thesis, KTH Royal Institute of Technology,
2014.
[68] J. Cheng, Broadband Dielectric Properties of Impregnated Transformer Paper Insulation at
Various Moisture Contents, Master of Science thesis in Electromagnetic Engineering, KTH,
Stockholm, Sweden, 2011.
[69] V. Sokolov, B. Vanin, "Evaluation and Identification of Typical Defects and Failure-Modes of
110-750 KV Bushings", ZTZ-Service Co. report, 2011.
[70] Y. Du, M. Zahn, B. Lesieutre, A. Mamishev, and S. Lindgren, "Moisture equilibrium in
transformer paper-oil systems", IEEE Electrical Insulation Magazine, Vol. 15, No. 1, pp. 11–
20, 1999.
78
[71] N. A. Muhamad, B. T. Phung, T. R. Blackburn and K.X. Lai, "Polarization and Depolarization
Current (PDC) Tests on Biodegradable and Mineral Transformer Oils at Different Moisture
Levels", Power Engineering Conference AUPEC, pp.1-6, Australia, 2009.
[72] V. D. Houhanessian and W. S. Zaengl, "Time Domain Measurements of Dielectric Response in Oil-Paper Insulation Systems", IEEE International Symposium on Electrical Insulation, Vol. 1, pp.47-52, Montreal, Canada, 1996.
[73] A. Helgeson, Analysis of Dielectric Response Measurement Methods and Dielectric Properties of Resin-Rich Insulation During Processing, PhD thesis, KTH Royal Institute of Technology, Stockholm, 2000.
[74] A. Setayeshmehr, A. Akbari, H. Borsi and E. Gockenbach, "On-line Monitoring and Diagnoses of Power Transformer Bushings", IEEE Tranactions. on Dielectrics and Electrical Insulation, Vol. 13, No. 3, 2006.
[75] J. C. Das, Transients in Electrical Systems (Analysis, Recognition, and Mitigation), McGraw-
Hill, 2010.
[76] Lou van der Sluis, Transients in Power System, John Wiley & Sons Ltd, 2001.
[77] J. Törnqvist, Non-contact High Voltage Measurements: Modeling and On-site Evaluation,
Master of Science thesis, Department of Physics, Umeå University, Sweden, 2012.
[78] A. Kumada and K. Hidaka, "Directly High-Voltage Measuring System Based on Pockels
Effect", IEEE transactions on power delivery, Vol. 28, No. 3, 2013.
[79] X. Xu, Enhancement in Dielectric Response Characterization of Insulation Materials,
Licentiate thesis in High Voltage Engineering, Chalmers University of Technology, Göteborg,
Sweden, 2013.
[80] B. Sonerud, T. Bengtsson, J. Blennow, S. M. Gubanski, "Noise and Aliasing Aspects in a
Multiharmonic-Dielectric-Response-Measurement System", IEEE Transactions on
Instrumentation and Measurement, Vol. 60, No. 12, pp.3875-3882, 2011.
[81] N. S. Nise, Control Systems Engineering, Fifth ed., John Wiley & Sons, 2008.
[82] X. Chen, P. H. F. Morshuis, B. Lennon and J.J. Smit, "Experimental Investigation on
Dielectric Spectroscopy of Insulating Paper and Oil", Annual Report Conference on Electrical
Insulation and Dielectric Phenomena (CEIDP), pp.1-4, USA, 2010.
[83] I. Fofana, H. Hemmatjou, F. Meghnefi, M. Farzaneh, A. Setayeshmehr, H. Borsi and E.
Gockenbach, "On the Frequency Domain Dielectric Response of Oil-paper Insulation at Low
Temperatures", IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 17, No. 3, pp.
779-806, 2010.
[84] J. Gielniak, A. Graczkowski, S. Gubanski, H. Moranda, H. Moscicka, K. Walczak, "Influence
of Thermal Aging on Dielectric Response of Oil-paper Insulation", Materials Science-Poland
Journal, Vol. 27, No. 4/2, 2009.
[85] J. Hai, Z. Ma, R. Liao, G. Chen and Lijun Yang, "A Comparative Study of Moisture and
Temperature Effect on the Frequency Dielectric Response Behavior of Pressboard Immersed
in Natural Ester and Mineral Oil", 17th International Symposium on High Voltage
Engineering (ISH), Hannover, Germany, 2011.
79
[86] C. Ekanayake, S. M. Gubanski, A. Graczkowski and K. Walczak, "Frequency Response of Oil
Impregnated Pressboard and Paper Samples for Estimating Moisture in Transformer
Insulation", IEEE Transactions on Power Delivery, Vol. 21, No. 3, pp. 1309-1317, 2006.
[87] High-voltage test techniques – Part 1: General definitions and test requirements, IEC standard
60060-1, ed3.0, 2010.
[88] Respicius C. Kiiza, The Effect of HV Impulses on Partial Discharge Activity and on the
Dielectric Response in Oil-impregnated paper Insulation, Doctoral thesis, KTH Royal
Institute of Technology, 2014.
[89] T. Blythe and D. Bloor, Electrical properties of polymers, Cambridge University Press, 2008.
[90] "Polycarbonate compound solutions for electrical applications", www.rittal.co.uk, May 2013.
[91] Krause, P. Brupbacher, A. Fehlmann, and B. Heinrich, "Moisture effects on the electric
strength of oil/pressboard insulation used in power transformers", IEEE International
Conference on Dielectric Liquids, ICDL, pp. 369-372. 2005.
[92] R. H. Khawaja, W. G. Ariastina and T. R. Blackburn, "Partial discharge behaviour in oil-
impregnated insulation", Proceedings of the 7th International Conference on Properties and
Applications of Dielectric Materials, Vol. 3, pp. 1166-1169, 2003.