Académique Documents
Professionnel Documents
Culture Documents
2012
NORHAFIZ AZIS
____________________________________________________________________________
List of contents
LIST OF CONTENTS
LIST OF CONTENTS .......................................................................................................... 3
LIST OF FIGURES ............................................................................................................... 6
LIST OF TABLES .............................................................................................................. 12
ABSTRACT
.............................................................................................................. 15
DECLARATION .............................................................................................................. 16
COPYRIGHT STATEMENT ........................................................................................... 17
ACKNOWLEDGEMENT.................................................................................................. 18
CHAPTER 1.
INTRODUCTION ............................................................................ 19
METHODOLOGY .......................................................................... 62
List of contents
3.4.1.
TAN measurements ..................................................................... 73
3.4.2.
LMA and HMA measurements .................................................... 74
3.5. Pre-processing of materials for ageing experiment ........................................ 79
3.5.1.
Pre-processing of oil .................................................................... 79
3.5.2.
Pre-processing of paper................................................................ 80
3.6. Summary ........................................................................................................ 80
CHAPTER 4.
List of contents
5.5.1.
Calculation procedure ................................................................ 146
5.5.2.
Prediction of TI based on LMA in oil ........................................ 148
5.5.3.
Prediction of transformers life consumption based on LMA ..... 154
5.6. Summary....................................................................................................... 155
CHAPTER 6.
Tensile Index (TI) of New and Service Aged Paper ............... 184
APPENDIX II:
APPENDIX III:
APPENDIX IV:
APPENDIX V:
REFERENCES
............................................................................................................ 207
List of figures
LIST OF FIGURES
Figure 1.1
Figure 1.2
Figure 1.3
Figure 1.4
Figure 1.5
Figure 2.1
Figure 2.2
Figure 2.3
Figure 2.4
Figure 2.5
Figure 2.6
Figure 2.7
Figure 2.8
Figure 2.9
Figure 2.10
Figure 2.11
Figure 2.12
Figure 2.13
Figure 2.14
Figure 2.15
Figure 2.16
Figure 2.17
Figure 2.18
Figure 2.19
Figure 2.20
Figure 2.21
Figure 2.22
List of figures
Figure 2.23
Figure 2.24
Figure 2.25
Figure 2.26
Figure 2.27
Figure 2.28
Figure 2.29
Figure 3.1
Figure 3.2
Figure 3.3
Figure 3.4
Figure 3.5
Figure 3.6
Figure 3.7
Figure 3.8
Figure 3.9
Figure 3.10
Figure 3.11
Figure 3.12
Figure 3.13
Figure 3.14
Figure 4.1
Figure 4.2
Figure 4.3
Figure 4.4
Figure 4.5
Figure 4.6
Figure 4.7
Figure 4.8
List of figures
Figure 4.9
Figure 4.10
Figure 4.11
Figure 4.12
Figure 4.13
Figure 4.14
Figure 4.15
Figure 4.16
Figure 4.17
Figure 4.18
Figure 4.19
Figure 4.20
Figure 4.21
Figure 4.22
Figure 4.23
Figure 4.24
Figure 4.25
Figure 4.26
Figure 4.27
Figure 4.28
Figure 4.29
Figure 4.30
Figure 4.31
Figure 4.32
Figure 5.1
List of figures
Figure 5.2
Figure 5.3
Figure 5.4
Figure 5.5
Figure 5.6
Figure 5.7
Figure 5.8
Figure 5.9
Figure 5.10
Figure 5.11
Figure 5.12
Figure 5.13
Figure 5.14
Figure 5.15
Figure 5.16
Figure 5.17
Figure 5.18
Figure 5.19
Figure 5.20
Figure 5.21
Figure 5.22
List of figures
Figure 5.23
Figure 5.24
Figure 5.25
Figure 5.26
Figure 5.27
Figure 5.28
Figure 5.29
Figure 5.30
Figure 5.31
Figure 6.1
Figure 6.2
Figure 6.3
Figure 6.4
Figure 6.5
Figure 6.6
Figure 6.7
Figure 6.8
Figure 6.9
Figure 6.10
Figure 6.11
LMA and HMA ratio for mineral oil aged under open condition without
copper. LMA, HMA and TAN of oil data is 1 measurement. ............. 142
LMA and HMA ratio for mineral oil aged under sealed condition with
copper. LMA, HMA and TAN of oil data is 1 measurement. ............. 144
LMA and HMA ratio for mineral oil aged under open condition with
copper. LMA, HMA and TAN of oil data is 1 measurement. ............. 145
Calculation procedure for predicting TI and transformer life
consumption. ....................................................................................... 147
Procedure of modelling the relationship between LMA in oil and
transformer age. ................................................................................... 148
TAN versus transformer age. Data is 1 measurement. ........................ 150
LMA in in-service oil versus transformer age. .................................... 151
Modelling procedure for the relationship between LMA in paper and
LMA in oil ........................................................................................... 152
Ratio of LMA in paper and oil versus transformer age ....................... 154
TAN of natural ester and mineral oil aged at 115 C under open
condition without any presence of paper, copper and steel. Data is 1
measurement. ....................................................................................... 157
TAN of natural ester aged at 170 C under open condition with presence
of paper, copper and steel. Data is 1 measurement. ............................ 159
Viscosity of natural ester aged under open condition with presence of
paper, copper and steel. Data are mean of 5 measurements. Mineral oil
was aged under open condition with paper, copper, zinc, iron and
aluminium [107]. ................................................................................. 161
Viscosity versus TAN of natural ester aged under open condition with
presence of paper, copper and steel. Viscosity data are mean of 5
measurements. TAN data is 1 measurement. ...................................... 162
Natural ester aged under open condition with presence of paper, copper,
and steel for single ratio samples. a) Natural ester condition after 50
days of ageing and b) Mixture of aged natural ester and water. ......... 164
TAN of natural ester and mineral oil aged under sealed condition
without any presence of paper, copper and steel. Data is 1 measurement.
............................................................................................................. 165
TAN of natural ester and mineral oil aged under sealed condition with
presence of paper, copper and steel. Data is 1 measurement. For S.
Tenbohlen, mineral oil was aged under sealed condition with paper,
copper, zinc, iron and aluminium [107]. For R. Liao, mineral oil was
aged under sealed condition with paper and copper [2]. ..................... 166
Viscosity of natural ester and mineral oil aged under sealed condition
with presence of paper, copper and steel. Data are mean of 5
measurements. For S. Tenbohlen, mineral oil was aged under sealed
condition with paper, copper, zinc, iron and aluminium [107]. .......... 167
Viscosity versus TAN of natural ester aged under sealed condition with
presence of paper, copper and steel. Viscosity data are mean of 5
measurements. TAN data is 1 measurement. ...................................... 169
LMA and HMA of natural ester and mineral oil aged under sealed
condition without any presence of paper, copper and steel. LMA and
HMA of natural ester data is 1 measurement. ..................................... 170
LMA and HMA of natural ester and mineral oil aged under sealed
condition with presence of paper, copper and steel. LMA and HMA of
natural ester data is 1 measurement. .................................................... 172
10
List of figures
Figure 6.12
Figure 6.13
Figure 6.14
Figure 6.15
Figure 6.16
LMA in paper of natural ester and mineral oil aged under sealed
condition with presence of paper, copper and steel. Data are mean of 2
measurements. ..................................................................................... 174
Influence of LMA on the breakdown voltage of natural ester. Data are
mean of 50 tests and error bars represent the standard deviation. LMA
in natural ester data is 1 measurement. ................................................ 175
Influence of HMA on the breakdown voltage of natural ester. Data are
mean of 50 tests and error bars represent the standard deviation. HMA
in natural ester data is 1 measurement. ................................................ 176
Influence of LMA and HMA on the paper aged under dry condition.
Data are mean of 8 samples and error bars represent the standard
deviation. ............................................................................................. 177
Influence of LMA and HMA on the paper aged under wet condition.
Data are mean of 8 samples and error bars represent the standard
deviation. ............................................................................................. 178
11
List of tables
LIST OF TABLES
Table 2.1
Table 2.2
Table 2.3
Table 3.1
Table 3.2
Table 3.3
Table 3.4
Table 3.5
Table 3.6
Table 3.7
Table 3.8
Table 4.1
Table 4.2
Table 4.3
Table 4.4
Table 4.5
Table 4.6
Table 4.7
Table 4.8
Table 4.9
Table 4.10
Table 4.11
Table 4.12
Table 4.13
Table 4.14
Table 4.15
Table 4.16
Table 4.17
Table 4.18
Table 5.1
Table 5.2
Table 5.3
Table 5.4
Table 5.5
Table 5.6
Table 5.7
Table 5.8
Table 5.9
Table 5.10
Table 5.11
List of tables
Table 5.12
Table 5.13
Table 5.14
Table 5.15
Table 5.16
Table 6.1
Table 6.2
Table 6.3
Table 6.4
Table 6.5
Table 6.6
Table 6.7
Table I.1
Table I.2
Table I.3
Table I.4
Table I.5
Table I.6
Table I.7
Table I.8
Table I.9
Table I.10
Table I.11
Table I.12
Table I.13
Table I.14
Table I.15
Table I.16
Table II.1
Table II.2
Table II.3
Table II.4
Table II.5
Table II.6
Table II.7
Table II.8
Table II.9
Table II.10
Table II.11
LMA and HMA percentage of mineral oil aged under sealed condition
with copper. ......................................................................................... 144
LMA and HMA percentage of mineral oil aged under open condition
with copper. ......................................................................................... 145
Increment rate of TAN and constant of the fitting. ............................. 151
Increment rate of LMA in oil and constant of the fitting. ................... 151
Life consumptions based on average initial TI of 107 Nm/g. ............. 155
Amount of oil, paper, copper and steel ratio for single ratio. .............. 158
Amount of oil, paper, copper and steel ratio for double ratio. ............ 158
LMA and HMA percentages in natural ester and mineral oil aged under
sealed condition without any presence of paper, copper and steel. ..... 170
LMA and HMA percentages in natural ester and mineral oil aged under
sealed condition with presence of paper, copper and steel. ................. 173
Ageing rate of dry condition. ............................................................... 177
Ageing rate of wet condition. .............................................................. 178
Comparison of increment factors between natural ester and mineral oil
under dry ageing conditions. ............................................................... 179
TI of new conductor type paper without oil impregnation at different
lengths. ................................................................................................. 184
TI of new layer type paper without oil impregnation. ......................... 184
TI of new conductor type paper with oil impregnation. ...................... 184
TI of new conductor type paper under folding stress. ......................... 185
TI of new conductor type paper under less paper length in the clamp
area....................................................................................................... 185
TI of Transformer 1. ............................................................................ 185
TI of Transformer 2. ............................................................................ 186
TI of Transformer 3. ............................................................................ 186
TI of Transformer 4. ............................................................................ 187
TI of Transformer 5. ............................................................................ 188
TI of Transformer 6. ............................................................................ 189
TI of Transformer 7. ............................................................................ 189
TI of Transformer 8. ............................................................................ 189
TI of Transformer 9. ............................................................................ 190
TI of different layers for Transformer 10. Paper samples were taken
from B phase, HV bottom.................................................................... 190
TI of different layers for Transformer 6. Paper samples were taken from
B phase, CW top. ................................................................................. 191
LMA in paper of Transformer 1. ......................................................... 192
LMA in paper of Transformer 2. ......................................................... 192
LMA in paper of Transformer 3. ......................................................... 192
LMA in paper of Transformer 4. ........................................................ 192
LMA in paper of Transformer 5. ......................................................... 193
LMA in paper of Transformer 6. ......................................................... 193
LMA in paper of Transformer 7. ......................................................... 194
LMA in paper of Transformer 8. ......................................................... 194
LMA in paper of Transformer 9. ......................................................... 194
LMA in paper of different layers for Transformer 10. Paper samples
were taken from B phase, HV bottom. ................................................ 194
LMA in paper of different layers for Transformer 6. Paper samples were
taken from B phase, CW top. .............................................................. 195
13
List of tables
Table III.1
Table III.2
Table IV.1
Table IV.2
Table IV.3
Table IV.4
Table IV.5
Table IV.6
Table IV.7
Table IV.8
14
Abstract
ABSTRACT
One of the cellulose insulation in high voltage power transformers is paper, and it is
widely used in transformers due to low cost and excellent physical/electrical
properties. However, the performance of paper could be affected by ageing. Ageing of
paper is a complex phenomenon and can be influenced by many factors. This PhD
thesis aimed to examine the ageing of paper in transformers based on Tensile Index
(TI), dielectric strength and Low Molecular Weight Acid (LMA) measurements.
The effect of ageing on the dielectric strength of paper was examined, through
studying the influence of paper and oil ageing by-products such as LMA and HMA on
the AC breakdown voltage. Meanwhile, the end of life for transmission transformer
population in the UK was examined by studying the profile of LMA and TI of paper in
scrapped transformers, with phase, winding, and layer locations taken into
consideration. The relationship between TI, LMA and transformer age was modelled
using previously published formulae in literature. The partitioning of LMA and HMA
between mineral oil and paper was also investigated. Since TI is not directly
measureable when a transformer is in-service, a TI prediction model was developed
based on LMA measured in oil.
The results indicated that there is no reduction effect of ageing (moderate towards
severe) on the dielectric strength of paper, whereas the dielectric strength of paper is
mainly influenced by the oil condition. LMA significantly reduces the breakdown
voltage and accelerates the ageing of paper in mineral oil. The end of life based on TI
is best represented by the 20 % retention of TI. Based on the case studies, the
estimated end of life of transmission transformer population in the UK is around 43
years. The vertical distribution profile of LMA in paper and TI of paper along a
winding is mainly influenced by vertical temperature profile. As paper ages, the
amount of LMA in paper tends to increase as TI reduces. The percentage of LMA in
service aged oil is about one fourth of Total Acid Number (TAN). The partition of
LMA in mineral oil and paper favours its stay in paper and is heavily influenced by the
weight ratio of paper and oil.
The performance of the natural ester under high temperature ageing in open or sealed
condition was evaluated based on TAN and viscosity measurements. The ratio of LMA
and HMA in aged natural ester was examined. The effect of LMA and HMA on the
AC breakdown voltage of natural ester and ageing of paper was also investigated. It
was suggested that natural ester is suitable for application in hermetically sealed
transformers. High temperature oxidation has significant effect on the physical and
chemical properties of natural ester. Due to the polar nature of natural ester, the stay of
LMA in natural ester is favoured over in paper which could be the reason why the
breakdown voltage of natural ester and the ageing of paper are not significantly
affected by LMA as in mineral oil.
15
Declaration
DECLARATION
I declare that no portion of the work referred to in the thesis has been submitted in
support of an application for another degree or qualification of this or any other
university or other institute of learning.
16
Copyright Statement
COPYRIGHT STATEMENT
(i). The author of this thesis (including any appendices and/or schedules to this thesis)
owns certain copyright or related rights in it (the Copyright) and s/he has given The
University of Manchester certain rights to use such Copyright, including for
administrative purposes.
(ii). Copies of this thesis, either in full or in extracts and whether in hard or electronic
copy, may be made only in accordance with the Copyright, Designs and Patents Act
1988 (as amended) and regulations issued under it or, where appropriate, in accordance
with licensing agreements which the University has from time to time. This page must
form part of any such copies made.
(iii). The ownership of certain Copyright, patents, designs, trade marks and other
intellectual property (the Intellectual Property) and any reproductions of copyright
works in the thesis, for example graphs and tables (Reproductions), which may be
described in this thesis, may not be owned by the author and may be owned by third
parties. Such Intellectual Property and Reproductions cannot and must not be made
available for use without the prior written permission of the owner(s) of the relevant
Intellectual Property and/or Reproductions.
(iv). Further information on the conditions under which disclosure, publication and
commercialisation of this thesis, the Copyright and any Intellectual Property and/or
Reproductions described in it may take place is available in the University IP Policy
(see
http://www.campus.manchester.ac.uk/medialibrary/policies/intellectual-
Library,
The
University
Librarys
regulations
(see
17
Acknowledgement
ACKNOWLEDGEMENT
To all my colleagues in the transformer research group and others in Ferranti building,
School of Electrical and Electronic, I appreciate your company and thank you for
providing an enjoyable working environment.
Last but not least, I wish to give my thanks to my parents, my wife Mrs. Siti Nur Sarah
and my daughters Aimi Balqis and Auni Balqis for their patience and selfless support
during my research.
18
Chapter 1 Introduction
CHAPTER 1. INTRODUCTION
1.1.
Background
The majority of power transformers in the power system network are oil-filled and
contain cellulose as an insulator. Apart from cooling transformers, oil also provides
electrical insulation together with cellulose. One of the major cellulose based materials
used in transformers is paper.
As transformers get older, both oil and paper will experience ageing and generate byproducts. These products could affect the performance of transformers and lead to the
risk of failure. Thus, it is important to determine the ageing status of transformers by
monitoring the condition of oil and paper. Oil condition can be determined by different
types of test such as acidity, interfacial tension and oxidation stability. On the other
hand, paper condition is best represented by mechanical strength which can only be
measured directly after transformers are scrapped. There are two types of the test that
can be used to determine the mechanical strength of paper which are Degree of
Polymerization (DP) and Tensile Strength (TS). DP is known as representative
strength measurement of paper while TS is regarded as the true strength measurement
of paper. Both methods can be used as a criterion to determine the end of life of
transformer.
According to [1], it was suggested that the definition of end of life of a transformer
should be made based on measureable mechanical, dielectric and chemical
characteristics. TS or DP is not directly measureable when a transformer is in
operation. Dielectric strength is the characteristic that would relates closely to the
functionality of a transformer, however, it deteriorates extremely slowly as the ageing
progresses, thus, the other alternative to describe the end of life of a transformer is the
chemical characteristic. Through latest knowledge on the paper ageing, the chemical
characteristic could be determined by means of Low Molecular Weight Acid (LMA).
Moreover, LMA could also acts as an ageing indicator provided the partitioning
between LMA in oil and paper could be obtained.
19
Chapter 1 Introduction
Natural ester is considered as the alternative for conventional mineral oil due to the
high fire safety and environmental friendliness. There are several aspects such as
dielectric properties and ageing performance that requires further investigation at
laboratory level before natural ester could be applied in transformers since failure in
the field could be costly. Whilst the ageing performance of natural ester is well
investigated by previous researchers [2-7], the performance of high temperature
oxidation on the natural ester still remains unclear. Moreover, there is less information
regarding the impact of oxidation toward its physical and chemical properties. Apart
from oxidation, there has been no study carried out on the LMA and High Molecular
Weight Acid (HMA) partitioning in natural ester and the effect of these acids on paper
ageing and the dielectric strength.
1.2.
Research objectives
The aim of this research is to examine the in-service ageing of paper through physical,
electrical and chemical analyses by means of scrapping transformers. Ageing
assessment of natural ester is also covered in this PhD thesis. Topics covered in this
thesis are as follows:
Chapter 1 Introduction
Ageing assessment of transformers through chemical analysis
The research on the application of natural ester as the insulation liquid for transformers
is still currently on-going. One of the crucial areas is the oxidation of natural ester
under high temperature. There is not much knowledge of the effect of severe oxidation
on physical and chemical properties of natural ester. Viscosity and Total Acid Number
(TAN) measurements are carried out on the natural ester subjected to high temperature
ageing under either open or sealed condition. It is necessary to examine the ratio of
LMA and HMA in natural ester. Further investigation is also required to determine the
effect of LMA and HMA on the dielectric strength and paper ageing in natural ester.
Based on this study, it is proposed that the reason for the good ageing performance of
paper ageing in natural ester is due to the fact that less LMA stays in the paper whereas
most of them are absorbed by the natural ester.
1.3.
This thesis focuses on the ageing of paper in transformers. However, knowledge on the
ageing of oil is required since it co-exists together with paper in transformers. An
21
Chapter 1 Introduction
overview on the oil ageing mechanisms and how its by-products affect the paper
ageing is given in this section.
1.3.1. Background
Oil is known as an integral part of insulation in a transformer apart from cellulose. The
reliability of a transformer is mainly determined by both oil and paper. The main
function of oil is to dissipate the heat generated by the transformer winding and core,
to insulate between components at different potentials including being able to
withstand system transients due to switching or lightning surges [8].
Mineral oil is the common insulating liquid used in a transformer. It is derived from
crude petroleum formed from buried and decayed vegetable matter or by the action of
water on metal carbides. It is a complex mixture of carbon and hydrogen with a small
proportion of sulphur and nitrogen [8].
Another type of insulating liquid available on the market nowadays is natural ester. It
is derived from vegetable seed oil which is biodegradable and non toxic. Another
advantage of natural ester is the high fire and flash point. However, it has poor
oxidation stability and not suitable to be used in a free breathing transformer.
Chapter 1 Introduction
present as both monoaromates and polyaromates. Polyaromates have several aromatic
rings directly adjacent to each other [10]. Most mineral oils are saturated hydrocarbons
where carbon atoms are connected by single C-C valence bonds. The saturated
compounds display better chemical stability than the corresponding unsaturated
compounds [9].
1.3.2.1.
Ageing of mineral oils are mainly subjected to oxidation. The result of oxidation could
lead to formation of different chemicals species, including aldehydes, ketones,
hydroperoxides and carboxylic acids.
Oxidation
Oxidation of mineral oil starts with the induction period where the hydrocarbon
molecule reacts with oxygen to produces the radical as shown in Figure 1.2. The
radical will react with oxygen to produce peroxide. The next stage is the propagation
period where the peroxide decomposes into hydroperoxide. The instantaneous products
of oxidation include alcohols and ketones, which are subjected to further degradation,
could form carboxylic acids, keto- and hydroxy-carboxylic acids. Other product from
the decomposition of ketones is aldehydes which eventually form resinous deposits by
complex condensation reactions. Generally, the final products of oxidation could be
classified into 2 types. The first type is soluble compounds which are acids and gases.
The second compounds include sludge and x waxes which are insoluble in mineral oil
23
Chapter 1 Introduction
[11]. The oxidation of mineral oil is influenced by fluid composition, temperature,
oxygen and catalyst metals such as copper and iron [9].
Natural ester originated from glycerol and fatty acids which are known as triglycerides.
The fatty acid segments are composed of straight chains having an even number of
carbon atoms. This is the natural result of the biosynthesis of fats, where molecules are
built up to two carbons at one time. Figure 1.3 shows the typical structure of the
triglycerides molecule where the (R, R, R) groups consist of C8-C22 chains. Natural
ester used for transformer application is fatty acid ester triglycerides type. The fatty
acid components are linear chains 14-22 carbons long containing zero to three double
bonds [12].
24
Chapter 1 Introduction
1.3.3.1.
Natural ester could be subjected to either oxidation or hydrolysis due to the structure of
its molecule. In natural ester, the double C-C valence bonds could promote oxidation
while the single COOH valence bonds could introduce hydrolysis.
Oxidation
According to the previous findings [7, 13], oxidation starts with the induction period
where the carbon atom in natural ester reacts with oxygen to create free radical by the
loss of hydrogen atom [14]. The free radical reacts with oxygen and creates peroxide
radicals as shown in Figure 1.4. The propagation period starts once the peroxide
radicals decompose to create hydroperoxide. The final by-products of this reaction
include aldehydes and ketones. The reaction between radicals needs catalysts such as
iron or copper.
Hydrolysis
Hydrolysis of natural ester can take place in three stages where all of them are
reversible which can be seen in Figure 1.5. The first stage is the creation of diglyceride
formed by trglyceride interaction with water [15]. Once the diglyceride reacts with
water, it will create monoglyceride which will form into glycerol as final reaction
product. During all stages, fatty acid will be created as a by-product [7].
25
Chapter 1 Introduction
Oxidation of mineral oil could generate water and acids such as LMA which could
accelerate the paper ageing [16-22]. The insoluble compound such as sludge could
accumulate and might block the flow of the oil which eventually could promote
hotspot and increase the rate of ageing of the paper [11].
On the other hand, it is shown that the ageing of paper in natural ester could be delayed
due to the water scavenging and hydrolytic protection [4]. However, natural ester has
low oxidation stability where, under extreme condition, the physical and chemical
properties could change and affect its performance as an insulating liquid [8].
1.4.
Thesis outlines
The summary of each chapter in this PhD thesis can be seen as follows:
Chapter 1 Introduction
This chapter provides the general background of the work and the motivation of the
research. A general overview on the ageing of oil is summarized in this chapter.
26
Chapter 1 Introduction
Chapter 3 Methodology
This chapter describes the paper sampling procedure from transformers and detail
procedures for TI, LMA and HMA measurements. Several issues regarding the effect
of paper sample condition on TI measurement are also discussed in this chapter.
2.1.
Introduction
Paper and pressboard are one of the basic insulation materials used in the oil filled
transformers. The good electrical and thermal properties are one of the advantages of
these materials. Paper is usually used to wrap the winding conductors. It is known that
under large oil gap, the breakdown could occur at a lower voltage due to the fact that
the probability of finding weakest spot will be high. This effect is also known as
volume effect [23]. Pressboard is used as spacer and barrier to divide the oil volumes
into smaller oil gaps in order to enhance the dielectric strength of oil gaps. The total
breakdown can be avoided if flashover occurs in an oil gap since the voltage is taken
by the rest of oil gaps [24].
Both paper and pressboard are made from cellulose. One of the drawbacks of cellulose
insulation is that the ageing is irreversible. As the paper ages, the transformer
performance could be affected. For example, local carbonising of the paper increases
the conductivity to cause overheating which leads to conductor fault and brittle paper
can break away from the winding and block the ducts [25, 26]. Furthermore, the byproducts of paper ageing such as water and acid also could accelerate the degradation
of the paper.
Another type of insulation material in the transformer is oil. Oil can provide dual
functions as cooling and electrical insulation medium within the transformers.
Conventionally, most of transformers are filled with mineral oil. Mineral oil is not
biodegradable and has low fire and flash points. Due to the increase of high fire safety
demand, alternative liquid such as natural esters are becoming available for
transformers.
A review on the mechanisms and factors that could affect paper ageing is given in this
chapter. A study was conducted in order to examine techniques that can be used to
monitor the condition of paper ageing based on ageing by-products. The mechanisms
of oil ageing and its effect to the paper degradation are discussed in the last part of this
chapter.
2.2.
Ageing of cellulose
2.2.1. Background
Cellulose consists of anhydro -D-glucopyranose units linked to each other as shown
in Figure 2.1. The repeating unit of glucose ring in the structure is joined by the
glycosidic oxygen linkage which is one of the parameters that contributed to the
mechanical strength of the paper [27, 28]. This glucose ring is also connected together
through the hydroxyl groups to form a cellulose fibre.
Insulation paper consists of 90% cellulose, 6-7% hemi-cellulose, lignin and pentosans
[26]. Alpha cellulose is normally used as the main cellulose content because it has a
high polymerization level of 2000. The strength of the paper will be affected if the
hemi-cellulose content is higher than 10 %. Lignin provides more rigid fibre bonding
for extra strength and flexibility [30].
Most of the insulation paper is made from softwood because it is more flexible and
stronger than hardwood [29]. Apart from that, softwood has much longer fibre than
hardwood. The structure of the fibre can be seen in Figure 2.2. Fibre is derived from
29
2.2.2.1.
Kraft paper
Kraft paper is made from unbleached softwood pulp by the sulphate process [8]. The
result of process will produce a paper that is slightly alkaline due to the existence of
the sulphate content. The process to produce paper differs according to the
manufacturer. The quality and reliability of the paper depend on the type of the
manufacturing process [8].
The general procedure starts by creating wood pulp mat from dried fibre. Next, large
amount of water was mixed with the wood pulp mat in order to remove any chemical
30
Cotton paper
Cotton paper is another alternative to Kraft paper. The manufacturing of cotton paper
for electrical purpose starts in 1930s and the source of cotton was obtained from waste
and offcuts from cotton cloths. However, this source is not used anymore since most of
cloths in recent years contain synthetic fibres and other materials. Cotton linters
obtained from cotton plant are now used as an alternative source for cotton paper
manufacturing. Cotton linters are obtained from the cotton plant fibre [8].
Cotton paper has longer fibre than Kraft paper but the intrinsic bond is not good. The
production cost of cotton paper is quite high due to the extra work for refining and
crushing processes. The performance of cotton paper could also be improved by
combining with Kraft fibre which has excellent electrical property. The result of this
process creates a paper which has good electrical and mechanical properties as well as
maximum oil absorption capability. [8].
2.2.2.2.
Creped paper
One of the earliest special purpose papers is Creped paper. It is made by irregular close
crimp, which significantly increases its thickness and extensibility in machine
direction. It is usually used at the connection lead or on the electrostatic stress control
31
Highly extensible paper is one of the available alternatives for Creped paper. The
paper is made by adding elastic property during manufacturing process which gives
improved burst, stretch, and cross machines tear properties while retaining its electrical
and mechanical capability. This paper also could withstand rough condition and
wrinkling, which are perfect for winding subjected to high stress [8].
operating temperature of
transformers [8].
Diamond dotted paper is used for application where high mechanical strength is
needed. The paper is made by coating the normal Kraft paper with thermosetting resin
and curing the resin later in the process. The diamond dotted shape is obtained by precoating the paper with two stages resin in a diamond pattern. The resin diamond dotted
creates a large bonding surface whilst ensuring that paper can be efficiently dried and
impregnated with oil [8].
2.2.2.3.
Pressboard
Pressboard with adhesive and high thickness due to the number of individual
boards bonded together and known as laminated paper [34].
33
Calendered pressboard
This type of pressboard initially undergoes pressing at about 55 % water content. The
next step is the dry stage without pressure until it reaches 5 % of moisture level. The
final density of this pressboard is around 0.9 kg/dm to 1.0 kg/dm and further
compression will increase the density up to around 1.15 kg/dm to 1.3 kg/dm [8].
Mouldable pressboard
Mouldable pressboard does not undergo initial pressing such as calendered pressboard.
Instead of pressing, it goes directly for drying until the moisture reaches 5 % and gives
a soft pressboard which is good for oil impregnation and flexibility to be shaped if
necessary. The final density is much lower than calendered pressboard which is around
0.90 kg/dm [8].
34
Precompressed pressboard
As the paper reaches the end of its life, fibres might break from the paper and are
dissolved in the oil. Fibres combined with metallic particles originated from the tank,
could initiate discharge in the oil which eventually led to arcing and sparking. The risk
of electromagnetic force, transient overvoltage, mechanical vibration and short circuit
force to the transformer will also be high once the paper of a transformer is degraded.
Water could reduce the resistivity and increase the dielectric loss of paper while
fibrous materials could align in the oil and provide pathways for short circuit between
conductors.
The aim of obtaining the knowledge of ageing of oil and paper is to prevent the
transformer from failing in-service by determining the current condition state. The
common approach is through analysis of ageing by-products which can be achieved
through DGA and FCA. DGA is mainly used for analyzing gases produced by faults
on oil and paper. On the other hand, FCA is a common method to analyze the ageing
state of paper since it is impossible to obtain paper while the transformer is still inservice.
35
Figure 2.5 Factors that could influence oil and paper ageing in the transformer [26].
Temperature
Type
and type of
of
conclusions
oil/paper
test
Shroff
110 C,
DP,
[35]
120 C,
TS,
130 C and
Comments
DP
is
the
common
FCA,
DP equal to 250
method
140 C
DS
TS equal to 50 % of
determining
(MO, KP and
and
its strength.
the ageing of
TUP)
MC
paper.
Demonstrated the
relationship between DP
for
and water/furanic
TS and it is
compounds.
commonly
used
analyzing
laboratory
36
for
Yoshida
100 C,
[36]
110 C,
TS,
The
120 C and
DP
of this research
160 C
and
is to analyze
(MO, KP and
DS
the ageing of
breakdown
paper obtained
PB)
Demonstrated
aged paper.
the
voltage
of
from scrapped
paper.
approach
Morais
125 C,
DP,
[37]
140 C and
DGA,
relationship between DP
155 C
FCA
and
and
compounds/water
Demonstrated
transformers
the
by TS.
CO/furanic
In addition, the
end of life of
transformer
MC
population
TS,
of
determined
153 C and
DP
represented
166 C
and
measurement.
FCA
Hill [38-
129 C,
MW,
40]
138 C,
paper
could
be
by
Introduced
based on TS.
MW
the
is
paper
the
common non-
on
intrusive
TS
where
the
method
is almost similar to MW
for
(76
the ageing of
kJ/mole
and
79
used
analyzing
kJ/mole).
paper.
In
between
research, it is
furanic
this
propose to use
Suggested
LMA
as
Lessard
150 C,
Sugar
[41]
200 C and
and
additional
250 C
FCA
paper
assessment
tool.
(MO, KP and
TUP)
Levchik
70 C to
FCA,
[42]
160 C
DGA,
that
sugar
that
Demonstrated
significant
amount
MC
furanic
compounds,
and
water,
DP
CO
and
of
CO2
37
an
ageing
Moreover, a TI
prediction
model
on
based
LMA
proposed.
is
Gasser
120 C,
DP
[43]
135 C and
and
150 C
TS
[22],
requires
required
(MO, KP and
TUP)
Emsley
120 C,
DP,
[20, 25,
140 C,
TS
paper in transformers is
44-46]
160 C and
and
mainly
180 C
FCA
hydrolysis
(MO, KP and
dominated
with
common
CP)
energy
by
a
activation
value
of
111
that
fast
kJ/mole.
Demonstrated
increment
of
furanic
compounds occurs at DP
around
400
concentration
and
the
starts
to
Introduced
the
paper
McShane
130 C,
DP,
[3-5, 47]
150 C and
TS
170 C
and
(MO, NE, KP
MC
in mineral oil.
and TUP)
Proposed mechanisms of
protection provided
by
Water scavenging
Water removal from
paper by natural ester.
Water has significant
effect on the paper
ageing.
Hydrolytic protection
Water is consumed
38
hydrolysis
of
natural ester.
Thomas
85 C,
DP,
[48]
100 C and
TS,
on furanic compounds as
115 C
DS
a non-intrusive technique
and
to
FCA
paper.
monitor
ageing
of
Mirzae
140 C,
TS
[49]
150 C and
and
160 C
DGA
on TI according to power
120 C,
[50]
140 C and
Introduced
the
exponential law.
TS
160 C
Demonstrated
that
for
multilayer
insulation
wrapped
around
same
Demonstrated
that
than
electrical
Luundgard
70 C,
DP,
[16-18]
90 C,
TS,
hydrolysis of paper is
110 C and
LMA
catalyzed by dissociation
130 C
and
(MO, KP and
HMA
Demonstrated
that
of acids.
TUP)
Introduced
acids
more
significant
39
90 C,
DP,
110 C and
MC
130 C
and
follows
(MO, NE, KP
FCA
degradation
and TUP)
the
paper
model
MO= Mineral Oil, NE= Natural ester, KP= Kraft Paper, TUP= Thermally Upgraded Paper, PB=
Pressboard, CP= Cotton Paper, DP= Degree of Polymerization, TS= Tensile Strength, TI= Tensile
Index, MC= Moisture Content, MW= Molecular Weight, LMA= Low Molecular Weight Acids, HMA=
High Molecular Weight Acids, DGA= Dissolved Gas Analysis, FCA= Furanic Compound Analysis,
DS= Dielectric Strength, CO= Carbon Monoxide, CO2= Carbon Dioxide
According to several studies [18, 27, 29, 35, 51], paper degradation mechanisms can
be oxidation, hydrolysis and pyrolysis. Oxidation mechanism dominates at
temperatures less than 60 C while hydrolysis initiates at a temperature region from 60
C to 150 C and pyrolysis dominates at high temperature usually over 150 C [52].
Oxidation, hydrolysis and pyrolysis can be differentiated by looking into the activation
energies. Activation energy can be calculated by Equation (2.1) known as the
Arrhenius equation.
-E
1
1
= kt = Aexp RT t
DPt DP0
(2.1)
Where DPt
DP0
Initial DP
Ageing rate
Temperature (K)
40
First, calculate the ageing rate, k for different ageing temperatures based on DP
as shown in Figure 2.6 a).
Next, plot ln k against 1/T. Through regression analysis, the activation energy
and pre exponential factor can be calculated as shown in Figure 2.6 b).
a)
b)
Figure 2.6 Procedures to obtain a) ageing rate, k and b) activation energy, E and pre
exponential factor, A. Calculation made based on Lundgaards data, Kraft paper with 3
% in weight water content (Thermo 70) and mineral oil (Nynas Nytro 10X) are sealed
without air and aged at 70 C, 90 C, 110 C and 130 C.
The typical activation energy obtained from laboratory accelerated ageing experiments
for both oxidation and hydrolysis can be seen in Table 2.2 [27, 53], it was found that
oxidation is promoted by oxygen and copper while hydrolysis is catalyzed by acids
and water. On the other hand, pyrolysis is promoted by high temperature. Based on
Lessards data [41], it was found that the activation energy for pyrolysis is around 82
kJ/mole which is lower than hydrolysis. It is suggested that the ageing temperature is
not high enough to increase the activation energy. According to survey study by
Emsley [27], for cellulose aged under vacuum or air, the pyrolysis activation energies
41
Activation energy
mechanism
(kJ/mole)
Oxidation
74
Hydrolysis
111
Figure 2.7 shows the sketch of ageing rates to different ageing mechanisms. It is
observed that the ageing rate is influenced by both activation energy and
environmental factor for each given temperature.
2.2.3.1.
Oxidation
Oxidation will be started once the hydroxyl groups in cellulose structure are attacked
by oxygen which weakened the glycosidic linkage. Through oxidation, carbonyl and
carboxyl groups will be created as secondary ageing products while moisture as
primary ageing products and these groups will promote hydrolysis in paper and oil as
seen in Figure 2.8 [27, 35, 51]. Carbonyl group (CO) consists of carbon atom (C)
double bonded with oxygen atom (O) while carboxyl group composes of combination
42
2.2.3.2.
Hydrolysis
Hydrolysis is known as the main paper ageing mechanism in transformers. The process
will promote glycosidic scission of cellulose chains and creates free glucose rings as
shown in Figure 2.10. The several stage of hydrolysis of free glucose rings will result
in
formation
of
furanic
compounds
such
as
2-furaldehyde
43
and
5-
2.2.3.3.
Pyrolysis
2.2.3.4.
Metal
cations
(Cu+/Cu2+
Hydrogen
peroxide
or
+
Fe /Fe2+)
Decomposition
of hydrogen
peroxide
HO
radicals
Water
Oxygen
+
High molecular
weight acids
Heat
Acid
Dominant
Oxidation
in oil and
paper
Low molecular
weight acids
Water
Dissociated carboxylic acids
Auto catalytic
process where
hydrolysis will
continue with
dissociation of
LMA
H+ ions
Hydrolysis in paper
Through oxidation, water, carbonyl group and carboxyl group known as carboxylic
acids known as will be generated [16, 17, 19]. There are two types of carboxylic acids
which are Higher Molecular Weight Acid (HMA) and Lower Molecular Weight Acid
(LMA). Dissociation of specific acids such as LMA will generate H+ ions and catalyze
the hydrolysis of the paper. Water does help in dissociation process of LMA by
increasing the number of H+ ions available in the system. It was found that hydrolysis
will become the main mechanism in the system once the environment becomes acidic
due to the reduction in oxidation [18]. The process will be repeated since it is an auto
catalytic process [58].
45
2.2.4.1.
Water
The rate of degradation for paper with 4 % water content in weight is 20 times higher
than that of dry one [20, 21]. An example of the effect of water on ageing rate can be
seen in Figure 2.14. The ageing experiment was conducted at different temperatures
ranging from 60 C to 130 C and the initial paper water content was controlled to 0.47
%, 1.11 % and 1.92 % by weight [59]. It is shown that the ageing rate, k increases as
the initial paper water content increases from 0.47 % to 1.92 %.
Figure 2.14 Effect of water on paper ageing. Kraft paper (Clupak HD75) and mineral
oil (Nynas Nytro 10CX) are sealed under air and aged from 60 C to 120 C [60].
2.2.4.2.
Oxygen
The rate of ageing will be reduced by a factor of 16 if oxygen in the system can be
reduced from 2000 ppm to 300 ppm [20]. Other studies show that the presence of
oxygen can increase the rate of ageing by a factor of 2.5 for paper having 0.3 to 5 %
water content [20, 61]. The rate of ageing will further increase with the existence of
copper which could accelerate the oxidation of the paper [20, 22]. An example of the
46
Figure 2.15 Effect of oxygen on paper ageing. Kraft paper (Thermo 70), mineral oil
(Nynas Nytro 10XN) and copper are sealed with different oxygen pressures and aged
at 90 C, 110 C and 130 C [53].
2.2.4.3.
At low temperature, the combination of water and oxygen could actually reduce the
rate of paper ageing. It was proposed that the hydrogen bonding between water and
oxygen could create shielding effect which reduces the reaction between water and
cellulose [20]. At the working temperatures of a transformer, the inhibiting effect of
the oxygen could be stronger and increasing the oxygen content in the oil could
actually reduce the rate of paper ageing by preventing the water in oil from migrating
to the paper [20].
47
Acids
The rate of paper degradation depends on the dissociation rate of the acids [18]. The
dissociation rate depends on the ability of specific acid such as LMA to dissociate
[17]. The dissociated acid increases the H+ concentration in the system and accelerates
the depolymerization process of the paper. An example of the effect of acids on the
ageing rate is shown in Figure 2.16. Water could help to increase the concentration of
H+ by causing the carboxylic acids to dissociate [18].
Figure 2.16 Effect of LMA on paper ageing under wet condition. Kraft paper (Thermo
70) and mineral oil (Nynas Nytro 10X) containing LMA and HMA are sealed without
air and aged at 70 C, 90 C, 110 C and 130 C [17].
2.2.5.1.
Water
One of the main products of paper ageing is water. According to Figure 2.13, water
could be generated by oxidation and hydrolysis and the production is an auto-catalytic
process. According to the previous finding, it is shown that the main source of water
generation is originated from paper and not from oil [59]. Water is the result of chain
scission of cellulose molecule as shown in y-axis of Figure 2.17 [54]. It was found that
DP will be reduced to half each time water content in paper is increased by 0.5 % due
to the chain scission of cellulose molecule [20-22].
48
Figure 2.17 Relationship between water generation and chain scission of paper [22,
61, 62].
2.2.5.2.
Furanic compounds
Other products of paper ageing are furanic compounds. Furanic compounds are
generated as a result of cellulose chain scission through either pyrolysis or
hydrolysis/oxidation [63]. Furanic compounds are one of the main products of paper
ageing and it can be monitored through oil. Figure 2.18 shows the main furanic
compounds
from
paper
ageing
which
are
2-Furaldehyde
(2FAL),
5-
The origin of furanic compounds is still unclear, but it was suggested that it is mainly
originated from hemicellulose. Others suggested that the main source of furanic
49
Hydrolysis
Cellulose
Hexose
(Glucose)
Oxidation
-CO
-CO2
Pentose
(Arabinose)
2-FAL
2.2.5.3.
Gases
Oil and paper ageing could create different type of gases and two of them are Carbon
Monoxide (CO) and Carbon Dioxide (CO2) [59, 65, 66]. These gases could be formed
through oxidation and pyrolysis [51, 67]. Figure 2.20 shows an example of CO and
CO2 generations with ageing duration at different temperatures. It is shown that except
for 140C, the amount of CO and CO2 generated from the ageing of oil and paper
increase slowly until 6 months and saturate at the later stage of ageing. Yoshida [36]
explained that at the initial stage of ageing, a part of cellulose is easily dissociated and
contributes to the increment of CO and CO2 as shown in Figure 2.21. At the later
stage of ageing, most part of the cellulose left is rigid and dissociates slowly. However
at 140 C, the rigid part of cellulose is also consumed at the initial stage of ageing and
contributes to the high amount of CO and CO2. The less rigid part of cellulose is
known as amorphous and the rigid region is known as crystalline [18].
50
Figure 2.20 Generation of CO and CO2 at different temperatures. Paper, mineral oil,
steel, and aluminium are sealed without air and aged at 100 C, 110 C, 120 C and
140 C [36].
2.2.5.4.
Acids
Acids are known as a common product for oil and paper ageing in a transformer. Acids
could be generated from both oil and paper through oxidation/hydrolysis mechanisms
[16, 17]. An example of acids generated from oil and paper ageing can be seen in
Figure 2.22. Recently it is shown that ageing of oil and paper could generate two types
of acids known as LMA and HMA [16, 17, 19]. These acids can be differentiated
51
Figure 2.22 Acids generation for Kraft and thermally upgraded paper. Kraft paper,
thermally upgraded paper, mineral oil and copper are sealed without air and aged at 90
C, 110 C and 130 C [2].
2.2.5.5.
Methanol
Methanol is another product of paper ageing. It was proposed that methanol originated
from the glycosidic scission of cellulose chain [68]. The result of methanol production
from the cellulose chain scission can be seen in y-axis of Figure 2.23 a). The ageing
experiment was conducted at different temperatures ranging from 60 C to 130 C and
the initial paper water content was controlled to 0.47 %, 1.11 % and 1.92 % in weight
[68]. It was found that the relationship between methanol and chain scissions is almost
linear. The calculation of activation energy shows that the value obtained for methanol
production (106.9 kJ/mole) is in agreement with reduction of DP (103.5 kJ/mole) [60].
Methanol is more sensitive than furanic compounds. Figure 2.23 b) shows that the
furanic compounds increase slowly up to chain scissions equal to 2 (DP equals to 400)
and increases exponentially up to 100 mg/kg [68]. The result indicates that furanic
compounds are less sensitive at a DP range from 1200 to 400 while methanol does not
have this limitation.
52
a)
b)
2.2.6.1.
Degree of polymerization
vs =
53
(2.2)
Intrinsic viscosity, v can be obtained based on the specific viscosity and concentration
of the solution based on Equation (2.3).
v
v = lim s
c 0
c
where c
(2.3)
The last step is to calculate the DP based on intrinsic viscosity using Equation (2.4)
[69].
(2.4)
[v] = KDP v
Where K and
DPv
54
Oxidation of the sample could affect the DP measurement. This can be avoided
by storing the solution under nitrogen.
2.2.6.2.
Tensile strength
Paper derived its strength from the fibre strength and inter-fibre bonding strength [71].
Tensile Strength (TS) can be described by stress and strain curve as shown in Figure
2.25. Machine Direction (MD) is defined as the direction in which the paper moves
during manufacture while Cross Machine Direction (CD) is the direction at right
angles to the machine direction. Paper in MD is stronger but less flexible than in CD.
Paper obtained from scrapped transformers usually is measured in MD. TS can be
measured in two ways which are known as long and zero span measurement. Long
span measurement is also known as wide span measurement in electrical insulation
research paper.
Figure 2.25 Stress and strain behaviour in two directions: MD (Machine Direction)
and CD (Cross Machine Direction) [71].
55
The first method is called long span TS which is known as measurement of paper inter
fibre bonding strength [46]. The test setup can be seen in Figure 2.26 where paper is
placed between two clamps. The upper clamp will move to break the paper and the
force is automatically recorded by the measuring unit.
Force
Upper clamp
(moveable)
Gap between
two clamps
Paper
samples
Lower clamp
(stationary)
According to standards, BS 4415 and BS 1924 [72, 73], the gap distance and width of
sample required for tests are about 180 mm 1 mm and 15 mm 0.1 mm. Tensile
strength, TS can be obtained using Equation (2.5).
TS =
F
w
(2.5)
where F
Width (m).
Long span Tensile Index (TI) is preferred for comparing different types of paper since
it is independent of the grammage of paper. Grammage can be calculated by dividing
the weight with the cross area of the paper. Long span tensile index, TI can be
calculated using Equation (2.6).
56
TI =
TS
g
(2.6)
Where g
Grammage (g/m).
Right clamp
(stationary)
Force
No gap between
two clamps
The calculation of zero span TI is based on failure load, FL which can be obtained
from Equation (2.7).
FL = K(P-P0 )
(2.7)
P0
0.422 (kg/psi).
57
BL =
FL
wg
(2.8)
Finally, zero span tensile index, TI can be calculated using Equation (2.9) below where
9.81 is known as the specific gravity.
(2.9)
TI = 9.81BL
2.2.6.3.
One of the non-intrusive techniques that could be used to monitor the condition of the
paper is furanic compound analysis. According to previous finding, it is shown that
there is a relationship between furanic compounds and DP which is described by
several models [75-77]. Generally, furanic compounds in transformer oil are
determined through High Performance Liquid Chromatography (HPLC). Furanic
compounds could also be measured from paper through extraction with methanol [42,
78]. The basic operation of HPLC system can be seen in Figure 2.28.
Sample will be
eluted here
HPLC Column
PC
Injector
Pump
Sample will be
injected here
UV
detector
Furan compound
will be detected here
Solvent
58
Waste
Observed event
5HMF
Oxidation
2FA
High moisture
2FAL
2AF
5MF
High temperature
Design of transformer
Type of transformer
Type of paper
59
Sample will be
eluted here
GC Column
PC
Injector
Sample will be
injected here
FID and TCD
detector
Gases compound
will be detected here
Carrier gas
Waste
Firstly, gases are extracted from the oil sample using methods suggested in the
standard, BS 60567 [86]. There are four methods that can be used for gases extraction
which are:
Once gases are extracted from the oil sample, it will be eluted by the GC column and
detected by Flame Ionization Detector (FID) and Thermal Conductivity Detector
(TCD). FID is used to detect compounds with high carbon concentration while TCD is
used for all other compounds. There are some difficulties on interpreting the CO and
60
2.3.
Summary
Ageing of cellulose insulation such as paper is mainly influenced by water, acids and
oxygen. In transformers, these products are originally formed from oxidation of
mineral oil and will be produced even further from paper ageing through autocatalytic
process. It is found that the main ageing mechanism of paper in a transformer is
dominated by acid hydrolysis. Through dissociation of specific acid such as LMA, H+
ions are generated and accelerate the hydrolysis of the paper. The rate of LMA
dissociation could be controlled by water.
As paper ages, the risk of transformer failure will increase. Thus, it is important to
monitor the condition of the paper. One of the techniques is by monitoring the paper
ageing by-products. Water, acids, furanic compounds, CO, CO2 and methanol are
among by-products generated by ageing of paper. The common compounds usually
monitored are Furanic compounds and CO/CO2. Recently, it is shown that methanol
also could be a good candidate for paper ageing indicator.
Ideally, the best way to analyze the condition of the paper is through DP or TS.
However, it is impossible to obtain paper samples from in-service transformers and the
only opportunity is through scrapping transformers. In this research, paper samples
were obtained from scrapped transformers. Ageing assessment of scrapped transformer
paper based on tests such as TS, LMA and AC breakdown voltage is discussed in this
thesis. Laboratory experiments are also conducted in order to support assessment of
scrapped transformer paper.
61
Chapter 3 Methodology
CHAPTER 3. METHODOLOGY
3.1.
Introduction
Mechanical strength is a common parameter used for assessing the ageing status of
paper. Tensile Strength (TS) can be measured either in long or zero span although in
electrical insulation research papers, long span TS is often written also as TS wide
span measurement. It was reported from previous work [46] that the main factor for
tensile strength reduction of paper is due to the loss of inter-fibre bond strength and
long span TS is known as the best way to represent this parameter. There are several
issues such as paper size and condition that need to be addressed when measuring long
span TS of scrapped transformers paper samples.
The first section of this chapter explains the procedure for sampling paper from
scrapped transformers. A study was conducted to examine the factors that could affect
the TS measurements of scrapped transformers paper samples. The process to measure
LMA and HMA content in paper and oil is presented in the next section. The last
section describes the pre-processing of oil and paper samples for ageing experiments.
3.2.
Most of the transformers in the UK are of the core type design, and they can be further
categorized as three limbs or five limbs. Core type transformers have their core limbs
concentrically wound by winding as shown in Figure 3.1 a). The arrangement of
windings in a transformer is shown in Figure 3.1 b). The closest winding to the core is
normally a tertiary winding (TV) followed by a low voltage winding (LV) and a high
62
Chapter 3 Methodology
voltage winding (HV). For economical reasons, an autotransformer is mainly used for
the high voltage transmission system. Generally, the connection is made by letting the
LV winding to be the common mainly shared with the HV winding. Through this
connection, the LV winding is also called the common winding (CW) and the rest of
the HV winding becomes the series winding (SW). Almost all transformers used for
ageing analysis in this thesis are autotransformers.
HV (SW)
LV (CW)
TOP VIEW
Phase A
TV
Phase C
Phase B
CORE
WINDING
SIDE VIEW
FRONT VIEW
5 LIMBS (CORE
TYPES)
a)
Inner Outer
Top
Centre
Bottom
LV
HV
(CW) (SW)
INSIDE VIEW FROM FRONT VIEW
Core TV
b)
Figure 3.1 Sampling procedure for a) Phase and b) Winding profiles.
The first step of the procedure, once a transformer is taken out from service for
scrapping, is to measure the oil properties such as acids, interfacial tension, furanic
compounds and dielectric properties. The next step is to remove the oil from
63
Chapter 3 Methodology
transformers; the main tank of transformers is cut to reveal the core and windings. At
this stage, usually an engineer will inspect transformers for any irregularities or source
of fault. Finally the windings are transported to the scrapping yard. The paper
sampling is done in the scrapping yard.
For phase profile, paper samples were taken from phase A, phase B and phase C as
shown in Figure 3.1 a). The winding profile was derived from the phase as shown in
Figure 3.1 b) where paper samples were taken from tertiary windings, common
windings and series windings. For both phase and winding profiles, paper samples
were taken from top, centre and bottom positions so that a vertical comparison ageing
status can be made. The scrapped transformer was analyzed vertically because of the
difference in the temperature distribution from bottom to top windings [88]. In general,
the operating temperature at the bottom windings is lower than top windings.
Horizontal comparison is also made among windings in the same phase and among
phases to check the thermal design.
The layer profile was obtained from a single conductor and the sampling procedure is
shown schematically in Figure 3.2. Normally, a number of paper layers were wrapped
around a single conductor varying from 6 layers to 12 layers depending on the design
of a transformer and the thickness of the paper. Each layer is unwrapped carefully and
paper that is in contact with oil was assigned as outer layer paper and inner layer paper
was the paper close to the conductor.
Inner Outer
Outer
Conductor
Core TV
Inner
~~
Insulation
Paper
LV
HV
(CW) (SW)
Layer profile
Winding profile
Oil
Chapter 3 Methodology
3.3.
TS measurements
All measurements of paper strength in this PhD thesis were carried out based on long
span TS. The gap distance and width of sample required for tests are about 180 mm
1 mm. For each location, at least 10 tests were done on paper samples. However, the
amount of paper from a scrapped transformer usually is limited and thus the number of
tests was sometimes reduced to a minimum of 4 tests for each sample in several cases.
The crosshead speed and full scale load range were set to 20 mm/min and 0.5 kN
according to [72, 73]. Tests were carried out in the paper science laboratory with a
condition of 23 C and 50% relative humidity. During the refurbishment of the
laboratory, the tests were conducted in a building having the temperature ranging from
21 C to 23 C and the humidity ranging from 35 % to 50 %.
3.3.1. TS or TI measurements
Paper from scrapped transformers usually comes in different types and grammages.
Tensile Index (TI) is preferred to TS for since it considers grammage in the
calculation.
Conductor and layer types of paper were tested in this section. Table 3.1 shows the
properties of both types of paper. Twenty thickness measurements were carried out for
each type of paper by a digital micrometer. The grammage for the conductor type
paper tested in this thesis range from 46 g/m to 51 g/m with an average value of 48
g/m. On the other hand, the grammage of the layer type paper range from 168 g/m to
171 g/m with an average value of 170 g/m. All grammage measurements were
carried out without any present of oil. The calculated density does match with the
value given by the manufacturer for both types of paper. For each type of paper, 10
tests of long span TS were carried out without any present of the oil. The TS and TI of
both types of paper were calculated by Equation (2.5) and Equation (2.6) in Chapter 2.
65
Chapter 3 Methodology
Table 3.1 Properties of new conductor and layer types paper.
Apparent density
Apparent density
(Manufacturer)
(Calculated)
(g/cm)
(g/cm)
48
0.93
0.92
170
0.76
0.72
Type of
Thickness
Grammage
paper
(mm)
(g/m)
Conductor
0.053
Layer
0.237
It is observed that the conductor type paper has a lower strength than the layer type
paper if comparison is made based on TS as shown in Table 3.2; however the reason
for the high strength of layer type paper is due to the grammage since thickness does
not affect the fibre bonding at a given density and load carrying capability.
The effect of grammage is considered in the TI calculation and it is shown that the
specific strength of the conductor type paper is higher than that of the layer type paper.
Furthermore, TI also could show that the specific strength of the paper is a function of
density where high density paper has higher strength than low density paper depending
on the type of fibre.
Table 3.2 TI and TS comparison between conductor and layer types paper.
Type of paper
TS (kN/m)
TI (Nm/g)
Conductor
5.25 0.21
109 4.58
Layer
15.6 0.53
91.6 3.14
Thus, in this thesis, TI will be used to describe the strength of the paper.
3.3.2.1.
All samples tested for TI in this section were new, conductor paper. The properties of
this type of paper can be seen in Table 3.1. All samples tested were not impregnated in
oil except for the influence of oil impregnation section. All measurements of paper
strength in this section were carried out based on long span TI.
66
Chapter 3 Methodology
3.3.2.2.
Paired t-test
A statistical method known as paired t-test was applied to all data in this section in
order to check whether there is any significant difference between means of samples.
The calculation procedure can be seen in Figure 3.3. The first step is to split the data
into two groups. Each group contains at least 10 TI measurements. The next step is to
subtract data in group 1, x1 with data in group 2, x2. The mean, d and standard
deviation, Sd of the difference is calculated and standard error, Se can be obtained by
dividing Sd with the square root of sample size, n. The T can be calculated based on d
and Se while tcrit at 95 % confidence level can be obtained from T table based on
specified degree of freedom, df. If T (calculated from mean and standard deviation of 2
groups) is higher than tcrit (from T table), it indicates that the mean for 2 groups are
different from each other.
d = x1 -x 2 , Se =
Sd
n
Calculate T
T=
d
Se
If T > tcrit
Mean
difference
No mean difference
Figure 3.3 Paired t-test calculation procedure.
67
Chapter 3 Methodology
3.3.2.3.
Effect of length
The study was carried out since it is difficult to obtain the standard size, 180 mm long,
samples of the service aged paper according to the standard [72, 73]. It can be seen
from Figure 3.4 that the percentage difference between 30 mm and 180 mm is less than
5 % which indicate the lengths within the measurement range has small influence on
the TI measurements. There is a decreasing trend of TI as the length is increased. This
is called scale effect where the probability of finding weakest link will be high if the
size of the sample is increased [89].
Figure 3.4 TI of new conductor paper versus length. Data are mean of 10 samples and
error bars represent the standard deviation.
According to the paired t-test, it is observed that the TI difference between 15 mm and
30 mm is not statistically significant as shown in Table 3.3. Similar observation was
also found for the comparison between 60 mm and 180 mm. On the other hand, it is
shown that there is some significant difference of TI statistically for other length
comparisons.
68
Chapter 3 Methodology
Table 3.3 Means comparison between each length
Mean difference (Based on T test at 95% confidence interval)
3.3.2.4.
15 mm
30 mm
60 mm
180 mm
15 mm
No
Yes
Yes
30 mm
No
Yes
Yes
60 mm
Yes
Yes
No
180 mm
Yes
Yes
No
Service aged paper can be subjected to folding stress due to the wrapping of the paper
around the conductor. A study was carried out on the influence of the folding stress on
TI measurement. In order to study this effect, new paper samples were folded with
moderate force at 2 different locations at an angle of 90 in order to imitate the folding
stress in paper from a service aged transformers as shown in Figure 3.5. It is
anticipated that the folding stress will create a weak point on the fibre. The gap length
for TI measurement was set as 30 mm.
Force
Upper clamp
(moveable)
Gap between
two clamps was
set at 30 mm
Paper
samples
Arrangement
of TI
measurement
Lower clamp
(stationary)
For clamp
Figure 3.5 Paper samples arrangement for the effect of folding stress.
It can be seen that paper under folding stress undergoes 14 % reduction of TI in Figure
3.6. There is a significant difference of TI at the 95 % confidence interval according to
69
Chapter 3 Methodology
the paired t-test. This effect might be more severe for service aged paper since the
folding area condition could be more severe compared to new paper.
Figure 3.6 TI of new conductor paper with and without folding stress. Data are mean
of 10 samples and error bars represent the standard deviation.
3.3.2.5.
The influence of oil impregnation on TI measurements was also examined. Three types
of paper conditions were prepared according to Figure 3.7. The gap length for TI
measurement was set at 30 mm and 180 mm.
Results are shown in Figure 3.8. It is observed that paper impregnated fully with oil,
with or without temperature, experiences a small TI reduction between 3.7 % and 6.1
% as seen in Table 3.4. Statistically, it is found that there is a significant difference of
TI between dry and full impregnation with/without temperature for both 30 mm and
180 mm as shown in Table 3.5. There is no influence of impregnation temperature on
the TI measurement based on paired t-test which suggests that there is no premature
ageing occurring during impregnation process.
A test was also conducted in order to see whether the small reduction is caused by the
slippage of the paper on the clamp by having only the gap between two clamps
70
Chapter 3 Methodology
impregnated with oil. It is found that paper still undergoes a small TI reduction
between 2.6 % to 2.8 %.
Force
Gap between
two clamps was
set at 30 mm
and 180 mm
For clamp
Upper clamp
(moveable)
Paper
samples
Arrangement
of TI
measurement
Lower clamp
(stationary)
Baseline- Dry
Gap between
two clamps
Oil
Sample 1
Sample 2 and 3
Only gap is impregnated with oil
For Sample 2, full impregnation
with oil in vacuum oven without
No vacuum and temperature
temperature
For Sample 3, full impregnation
with oil in vacuum oven at 85 C
Figure 3.7 Preparations of paper samples for the effect of oil impregnation.
Figure 3.8 Influence of oil on TI. Data are mean of 10 samples and error bars
represent the standard deviation.
71
Chapter 3 Methodology
Table 3.4 Percentage difference of TI between baseline and samples.
Difference at
Difference at
30 mm (%)
180 mm (%)
2.6 %
2.8 %
6.1 %
3.7 %
5.2 %
4.6 %
Sample
Sample 1
Sample 2
Sample 3
No
Yes
Yes
Baseline
Sample 1
Sample 2
Sample 3
No
Yes
Yes
Baseline
3.3.2.6.
Another study was carried out in order to study the influence of reducing paper length
under the clamp i.e. clamp area is fully filled with paper or only 50 %. The gap length
for the TI measurement was set at 15 mm. The arrangement of the test can be seen in
Figure 3.9.
Force
Gap between
two clamps was
set at 15 mm
Upper clamp
(moveable)
Paper
samples
Arrangement
of TI
measurement
Lower clamp
(stationary)
For clamp
35 mm
Sample 1
The paper length in
both clamps is 35 mm.
17.5 mm
35 mm
17.5 mm
Sample 2
The paper length in both
clamps is 17.5 mm.
Figure 3.9 Paper samples arrangement for the effect reducing paper length in the
clamp.
72
Chapter 3 Methodology
There is about 35 % reduction of TI for the case where paper has shorter length under
the clamp which can be seen in Figure 3.10. Statistically, it is found that reducing the
length of paper in the clamp can caused significant reduction of TI. The reduction
might be caused by paper slippage on the clamp. It is essential to keep the paper in the
full clamp area with a shorter but fixed gap length to avoid large error in TI
measurement in the case of paper shortage from a service aged scrapped transformer.
Figure 3.10 Effect of reducing paper length in the clamp area of TI. Data are mean of
10 samples and error bars represent the standard deviation.
3.4.
The common practise to measure the Total Acid Number in (TAN) oil is based on
automatic potentiometric measurement [90]. The measurement was done by dissolving
the sample in the solvent (500 ml of toluene, 495 ml of iso-propanol and 5 ml of water)
and titrated potentiometrically with the reagent (0.1 M (Molar Concentration)
Potassium hydroxide in iso-propanol) using a glass indicating electrode and a reference
electrode (Metrohm Solvotrode 6.0229.100). The end point of the measurement was
taken once the volume of reagent corresponds to a PH of 11.5 which was recorded by
Metrohm oil titrino plus. The detail procedure to measure TAN can be seen in Figure
3.11.
73
Chapter 3 Methodology
1st STEP
Titrate the solvent
70 ml of solvent
2nd STEP
The calculation of Total Acid Number (TAN) of oil can be seen in Equation (3.1).
TAN =
where V1
(V1 -CV)mv Cv
m
(3.1)
CV
mv
Cv
The principle of LMA and HMA measurements in oil and paper is based on the
chemical nature of the acid itself. LMA is known as polar compound and HMA is
known as less-polar compound. Thus, by using another polar compound such as water,
the LMA could be extracted from the oil and paper. The measurement of LMA and
HMA was done according to the method developed by Ingebrigtsen [19].
25ml of oil was mixed with an equal volume of distilled water and mixed overnight
using a magnetic stirrer as seen in Figure 3.12. The next step was to leave the mixture
sample still for at least half an hour in order to separate into two layers. Water and oil
74
Chapter 3 Methodology
layer were then extracted and titrated according to BS 62021 [19, 90]. The acidity
result obtained from water layer is defined as LMA while the oil layer as HMA.
Acidity of clean water was measured and subtracted from the measurement of LMA in
oil. The oil layer defined as HMA was measured in the same manner as TAN. The
acidity of the sample was calculated according to the weight of oil used for extraction.
1st STEP
25ml of oil
Mixture of
oil and water
2nd STEP
Stir the mixture overnight
Magnetic stirrer
=
Mixture of
oil and water
3rd STEP
+
Oil extracts
(HMA)
(LMA)
4th STEP
Titrate the oil and water for LMA and HMA using Metrohm Oil
Titrino Plus. LMA and HMA are calculated by Equation (3.2)
The calculation of LMA and HMA in oil can be seen in Equation (3.2).
LMA/HMA in oil =
where Vo
(Vo -CVw )m v C v
mo
(3.2)
CVw
mv
Cv
mo
75
Chapter 3 Methodology
b) Measurement of LMA in paper
A known weight of paper, ranging from 0.5 g to 1.0 g, was prepared and immersed in
25 ml of water for 3 days without stirring as shown in Figure 3.13. After the extraction
period, 10 ml of the extracting water was titrated according to BS 62021 [19, 90]. The
leftover paper was dried in air circulating oven at 70 C for 2 day and acidity was
calculated by dividing with the 40 % of the dry weight of paper since only 10 ml was
used for titration.
1st STEP
Mixture of
paper and
water
25ml of water
2nd STEP
Leave the mixture without any stirring
Mixture of
paper and
water
for 3 days
3rd STEP
10 ml of water
(LMA)
4th STEP
Titrate the water for LMA using Metrohm Oil Titrino Plus. LMA is
calculated by Equation (3.3)
LMA in paper =
(Vp-CVw )m v Cv
mp
76
(3.3)
Chapter 3 Methodology
where Vp
CVw
mv
Cv
mp
3.4.2.1.
A study was conducted to check whether 3 days is sufficient to extract the LMA from
the paper. The test was conducted by having the same paper sample titrated for 3
times and after each titration, new water was added again to the paper for the next
extraction. The age and properties of paper samples tested can be seen in Table 3.6.
Transformers
Age
(years)
Paper
thickness
(mm)
Grammage
(g/cm)
Apparent paper
density
(g/cm)
Transformer a
37
0.13
137
1.08
Transformer b
45
0.09
101
1.17
Transformer c
35
0.11
89
0.79
Transformer d
42
0.04
42
1.07
Two paper samples from scrapped transformers with different LMA in paper values
are shown in Figure 3.14. It is shown that 72.2 % of LMA was extracted on the first,
23.8 % on the second and 4 % on the third extraction for Transformer a. For
Transformer b, 75.5 % of LMA was extracted on the first, 19.7 % on the second, and
4.8 % on the third extraction.
77
Chapter 3 Methodology
Figure 3.14 Low molecular weight acid extraction efficiency using water. Data are
mean of 2 measurements.
The results show that the majority of LMA would be extracted on the first extraction
and only a small amount of LMA was left in the paper on the second and third
extraction. There is no significant difference in the percentage of extraction for paper
having either low or high amount of LMA, therefore extraction once is appropriate and
the value of LMA in paper can be used to conduct cross comparison.
The repeatability of LMA measurements was also checked by measuring twice the
water extract. It can be seen from Table 3.7 that the repeatability of the measurements
is good. The difference between first and second measurements for Transformer b, c
and d are around 18 %, 0.4 % and 6 %. The slight deviation between first and second
measurements for Transformer b might be due to some of LMA is not distributed well
in the water since no stirring is involved during the extraction process. Since the
repeatability of the measurements is good, almost all LMA measurements in this thesis
are measured twice and the average value is taken.
Table 3.7 LMA measurement on service aged paper.
Transformer b
Transformer c
Transformer d
(mg KOH/g)
(mg KOH/g)
(mg KOH/g)
1st
1.30
2.81
3.76
2nd
1.06
2.80
3.98
Measurement
78
Chapter 3 Methodology
3.5.
Mineral oil Gemini X and natural ester FR3 were first filtered using a nylon membrane
filter with a pore size of 0.2 m. The basic properties of these oils based on the
Product Datasheets can be seen in Table 3.8. The important properties are acidity,
viscosity and AC breakdown voltage since it could influence the performance of
transformers after subjected to ageing. The next step is to dry mineral oil in air
circulating oven for 24 hours at 105 C in order to remove the water. On the other
hand, for natural ester, the drying is done in vacuum oven (less than 10 mbar) at 85 C
for 72 hours. The drying of natural ester is done in vacuum oven and at a long duration
because the water saturation is higher than mineral oil. The final water content for
mineral oil and natural ester are 10 ppm and 20 ppm respectively.
Table 3.8 Basic properties of Gemini X and FR3.
Type of paper
Unit
Gemini X
FR3
Density @ 20 C
kg/dm3
0.88
0.92 (25 C)
Viscosity @ 40 C
mm2/s
8.7
34
Flash point
144
316
Pour point
-60
-21
mg KOH/g
<0.01
0.04
10
mg/kg
<20
30
wt%
0.38
-Before treatment
kV
- After treatment
kV
0.03 (100 C)
<0.001
Acidity
Aromatic content
Water content
Antioxidant content
AC breakdown voltage
79
Chapter 3 Methodology
3.5.2. Pre-processing of paper
Conductor paper was used for all ageing experiment. The property of this paper can be
seen in Table 3.2. In order to prepare dry paper sample, the paper was dried in air
circulating oven for 24 hours at 105 C. The next step was to impregnate the paper
with either dry mineral oil or natural ester in vacuum oven for 24 hours at 85 C. The
final water content for paper is less than 0.5 %.
For wet paper sample, the dried paper impregnated with either mineral oil or natural
ester was left in desiccators with a relative humidity of 90 % for 7 days. The final
water content paper is around 5 %.
3.6.
Summary
There are several aspects that need to be taken care when measuring TI of service aged
paper sample. Reducing the length of the paper sample and oil impregnation only has
minor influence on the TI measurements. It is important to eliminate paper samples
which are damaged and mechanically stressed during scrapping since folding effect
show that it could influence the TI measurements. The most crucial practice is to make
sure that the size of the paper sample is sufficient to cover the clamp area in order to
avoid significant errors on the TI measurements of service aged paper.
Three days of extraction period is sufficient to extract most of the LMA from the
paper. Furthermore the repeatability of the LMA measurements of the water extract is
good.
80
4.1.
Introduction
Up until now, there are several studies on the ageing of paper in transformers were
carried out through laboratory ageing experiments. Although much knowledge on the
ageing of paper was obtained through these studies, it is difficult to apply to in-service
transformers. One of the reasons is the ageing mechanisms in transformers is complex
and involves many factors at one time.
This chapter discusses the TI phase, winding and layer profiles of scrapped
transformers. The relationship between TI and a transformer age is modelled using the
Weidmanns equation. Life assessment of transformer population was carried out
based on this relationship. Finally, the effect of ageing on the dielectric strength of
paper is discussed in this chapter.
4.2.
Case study
Rating
240MVA
400/132kV
Age
(years)
11
240MVA
400/132/13kV
11
1000MVA
400/275kV
240MVA
275/132kV
180MVA
275/132kV
120MVA
275/132kV
240MVA
275/132kV
180MVA
400/132/13kV
150MVA
275/132/11kV
35
41
42
45
45
48
50
10
180MVA
275/66kV
37
ageing
- Over wrapping on the HV winding
causes localised heating on certain
areas
82
The sampling of paper for TI analysis was carried out according to the procedure
mentioned in Figure 3.1 and Figure 3.2 of Chapter 3. The oil residue on the paper
samples was wiped using a tissue before being tested for TI. Based on all
measurements, it was found that the TI of paper for scrapped transformers ranges from
84.6 Nm/g to 20.3 Nm/g. The first part of this section describes the profile of TI along
phase, winding and layer.
4.3.1.1.
Three transformers were analyzed for phase and winding profiles which are A, B and
C phase Common Windings (CW) of Transformer 6, A, B and C windings Series
Windings (SW) of Transformer 4 and the CW and SW of B phase of Transformer 8.
These transformers are chosen for analysis because there are enough paper samples
available to give a good representation of the vertical profile of top, centre and bottom
locations. It is known that these locations could show the temperature profile. Paper
properties and Total Acid Number (TAN) of oil before scrapping for each transformer
can be seen in Table 4.2.
Table 4.2 Paper properties and TAN of each transformer for phase and winding
profiles.
Paper
Apparent
thickness
paper density
before scrapped
(mm)
(g/cm)
(mg KOH/g)
Transformer 4
0.15
0.93
0.21
Transformer 6
0.04
1.07
0.16
Transformer 8
0.11
0.79
0.16
Transformers
The vertical profile of TI of each phase can be seen in Figure 4.1. Significant
difference of TI at 95 % confidence interval is observed between A phase, B phase and
C phase at the centre and bottom windings of Transformer 4 according to the paired t83
It is noticed that the TI for a young transformer could be similar to an old transformer
by comparing B phase CW of Transformer 8 and Transformer 4. One of the possible
reasons is the loading of Transformer 8 is higher than Transformer 4. TAN in oil
indicates ageing, and it is shown that the TAN in oil of Transformer 4 is higher than
Transformer 8. It is believed that the chemical age of Transformer 4 is older than
Transformer 8although the usage age is showing oppositely. However, no loading
history could be obtained for both transformers, so the hypothesis cannot be
confirmed. Another reason might be due to the paper thickness of Transformer 4 is
higher than Transformer 8. According to previous literature [43], thick paper degrades
faster than thin paper due to the higher temperature and slower heat conduction.
Based on the phase and winding studies, it is found the TI at the top winding is slightly
lower than bottom winding. The observation is in line with previous findings [91, 92].
One of the main factors that could affect the paper degradation is heat. Since
temperature is usually high at the top winding and hotspots are normally located near
84
a)
b)
Figure 4.2 TI profiles among different windings of Transformer 8. Paper samples were
taken from B phase. Data are mean of 4 to 6 samples and error bars represent the
standard deviation.
85
Layer profile
Two transformers were analyzed for layer profile. The paper samples were taken from
B phase, High Voltage (HV) bottom winding of Transformer 10 and B phase, CW top
winding of Transformer 6. These transformers are chosen in order to show the
difference of layer profile between severely aged paper (shown by Transformer 10)
and moderately aged paper (shown by Transformer 6). As mentioned in Table 4.1,
Transformer 10 failed due to the short circuit in turn conductors as a result from severe
solid insulation ageing while Transformer 6 is decommissioned due to the closure of
the substation. Paper properties and TAN of oil of these two transformers can be seen
Table 4.3.
Table 4.3 Age, paper properties and TAN of each transformer for layer profile.
Paper
Paper
thickness
density
before scrapped
(mm)
(g/cm)
(mg KOH/g)
Transformer 10
0.086
1.17
0.42
Transformer 6
0.040
1.07
0.16
Transformers
The profile of TI along layer for Transformer 10 can be seen in Figure 4.3. It is
observed that the 1st layer paper has lower TI than the rest of layers with a value of
22.4 Nm/g. TI remains unchanged from the 2nd layer to 8th layer paper with an average
value of 62.5 Nm/g.
Another example of layer profile for Transformer 6 can be seen in Figure 4.4. It is
observed that the 1st and 11th layers of paper have lower TI than the other layers. TI for
both 1st and 11th layers paper are almost the same, with values of 33.2 Nm/g and 34.4
Nm/g. TI remains unchanged from the 2nd layer to 10th layer paper with an average
value of 44.3 Nm/g
86
Figure 4.3 TI profiles among different layers of Transformer 10. Paper samples were
taken from B phase, HV bottom. Data are mean of 9 to 10 samples and error bars
represent the standard deviation.
Figure 4.4 TI profiles among different layers of Transformer 6. Paper samples were
taken from B phase, CW top. Data are mean of 9 to 10 samples and error bars
represent the standard deviation.
Previous studies have shown similar pattern for multilayer insulation where the outer
layer paper usually degraded more than others [94, 95]. It was proposed that multilayer
insulation in transformers is dominated by two main degradation mechanisms [94].
87
a)
b)
c)
d)
e)
f)
g)
h)
Figure 4.5 Example of paper taken from Transformer 10 a) 1st layer b) 2nd layer c) 3rd
layer d) 4th layer e) 5th layer f) 6th layer g) 7th layer h) 8th layer. Paper samples were
taken from B phase, HV bottom of Transformer 10.
Another reason probably is that the initial TI of 1st layer paper for Transformerh)10 is
already lower than that of Transformer 6 since no information on the initial TI could be
obtained for both transformers. However, based on tests carried out on new conductor
type paper obtained from a transformer manufacturer, it was found that TI is high
ranging from 109 Nm/g to 115 Nm/g as shown in section 3.3.2.3 of Chapter 3. Paper
used in transformers usually has a high initial strength in order to withstand
mechanical stresses such as short circuit [18]. Thus, it is believed that the initial TI of
88
The second mechanism is thermal degradation which concerns the layer of paper close
to the conductor. Thermal degradation generally relies on the conductor temperature. It
can be seen that the effect of thermal degradation is apparent to Transformer 6.
The relationship between the TI of paper and a transformer age for, 9 transformers, can
be seen in Figure 4.6. Transformer 10 was not included in Figure 4.6 since it will be
used as the end of life criterion for next section. The scattering of data for each age is
due to the TI at different locations in a transformer.
It is observed that the TI for transformer population decreases as the age increases. A
similar trend was also observed with the reduction of DP of paper with a transformer
age [96]. However, the trend does not indicate ageing is a function of transformer age,
but a result of accumulative effect from temperature, moisture and acids. Loading is a
major source of heat while moisture might be introduced from the surrounding
atmospheric condition and as a product of ageing generated together with acids.
89
Figure 4.6 TI versus age of 9 transformers. Paper samples were taken from different
locations. Data are mean of 4 to 10 samples.
Figure 4.7 and Figure 4.8 show the daily typical summer and winter loading for some
of the transformers in Figure 4.6. The date for the daily loading of both seasons were
chosen based on the suggestion given by the utility report [97]. The average daily
loading at both seasons for Transformer 1 is high compared to other transformers as
seen in Figure 4.9 and Figure 4.10 which explains why it has the low TI at a young
age. Even though the average daily loading during summer for Transformer 2 is lower
than Transformer 1, it is observed that its TI is almost the same as Transformer 1 and
this might indicate the intrinsic thermal design of Transformer 2 is not as good as
Transformer 1.
On the other hand, it is noticed that TI is low even though the average daily loading of
both season for Transformer 4 and 5 is relatively low. This could be again, a thermal
design issue, however, at this point of research, we assume that all transformers under
study are having normalized thermal design and loading, generic to all transmission
transformers in the UK, so the ageing of paper in a transformer is mainly influenced by
age. As a transformer age increases, paper will degrade which results in loss of
mechanical strength.
90
91
The prominent factor that could affect the reduction trend of transformers under study
is thermal design. An example of bad thermal design is insufficient size of oil duct
which could cause disruption in oil flow and lead to the temperature increase and
accelerate the ageing of paper. However, since all transformers in Table 4.1 are from
different designs, this is also one of the reasons to assume a generic design exists for
this study.
92
4.4.1.1.
Initially, two models were considered to represent the relationship between TI and
transformer age. The model developed by Emsley and Heywood [46] is not considered
since it is difficult to use their model to make a practical estimation from limited
ageing data [43, 98].
TI = TI0exp-CTI t
where
(4.1)
TI
TI0
Initial TI (Nm/g).
CTI
Time (years).
The second model developed by Ding and Wang is shown in Equation (4.2) [98]. The
degradation of paper can be characterized by TI where TI = 0 represents the new
paper with 100 % retention of TI while TI = 1 corresponds to the failure of paper.
TI = *TI (1-exp-kTI t )
TI =1-
TI
TI0
(4.2)
93
TI
Percentage of TI loss.
*TI
k TI
Equation (4.3) can be obtained by re-arranging of Equation (4.2). This equation has a
similar form to Equation (4.1) and can be used to describe the relationship between TI
and transformer age.
TI = TI0 (1-*TI (1-exp-kTI t ))
(4.3)
The main criterion for choosing the suitable model for the relationship between TI and
transformer age is that it can estimate the life when TI reaches the lowest TI of
Transformer 10. The fitting of both models was carried out based on the average of the
measured TI of each transformer. The definition of initial and measured TI is
illustrated in Figure 4.11.
The initial TI was set as a parameter for fitting for both models. It is difficult to obtain
information on the initial TI of paper used in each transformer. Thus, a range of initial
TI values were chosen from both literatures [18, 46] and the measured TI of new
papers obtained from manufacturers. Once the range has been obtained, it is split into
94
TI (Nm/g)
Average
107
Lowest
104
Highest
112
An example of not setting initial TI as a fitting parameter can be seen in Figure 4.12.
At first instance it is observed that there are 2 stages of relationship between TI of
paper and a transformer age. The first stage is the fast reduction of TI of paper from
start until the age reaches 10 years. The second stage is the slow reduction of TI of
paper between age of 10 and 50 years. However, no modelling was carried out to
represent the 2 stage relationship since the paper with age of 10 years could be
subjected to higher loading which in turn caused the low TI. The initial TI based on
fitting is around 96 Nm/g which is lower than any values given in Table 4.4. Thus, the
initial TI needs to be set, in advance, when the fitting is carried out for both models.
Figure 4.12 Fitting for the average of the measured TI according to Weidmann model.
The initial TI was not set as fitting parameter.
95
On the other hand, it is found that the average measured TI of 11 years old
transformers (Transformer 1 and 2) could be modelled by Ding and Wang model as
shown Figure 4.14. As mentioned earlier, the paper samples for both transformers
might be subjected to advanced ageing due to the high loading. Furthermore, Ding and
Wang model cannot reach the criterion of end of life i.e. the final estimated TI remains
at a value around 40 Nm/g, and this indicate that its suitability is more towards the
initial and mid-age trend estimation
Thus, the Weidmann model is selected to represent the relationship between TI and
transformer age since it could meet the criterion defined earlier. Life assessment of
transformer population will be carried out based on this model.
Figure 4.13 Fitting for the average of the measured TI according to Weidmann model.
96
Figure 4.14 Fitting for the average of the measured TI according to Ding and Wang
model.
4.4.1.2.
The end of life criterion of transformer population is determined based on three criteria
as shown in Table 4.5. The first and second criteria are obtained from the IEEE
standard, C57.91-1995 [1] where the first criterion is based on 50 % retention of TI
while the second criterion is based on 25 % retention of TI. The third criterion is based
on the lowest TI of Transformer 10 with a value of 22.44 Nm/g. This criterion was
chosen because Transformer 10 failed due to the reason of ageing.
Criterion 1
50 % retention of TI
Criterion 2
25 % retention of TI
Criterion 3
Lowest TI of Transformer 10
In this section, the relationship between TI and transformer age is represented by the
Weidmann model. The initial TI was set as a fixed parameter where its values are
97
4.4.2.1.
The first step is to assess the life of paper sample taken from top and bottom windings.
Figure 4.15 and Figure 4.16 show the fitting results of the average of the measured TI
at the top and bottom windings.
It is noticed that the fitting lines for top winding reach the end of life criterion faster
than those of bottom winding. For example, the estimated life from the top winding
based on Criterion 3 is around 58 years while that from bottom winding is around 81
years. As mentioned earlier in section 4.3.1.1, high temperature and hotspot usually is
located at the top winding. Since heat could accelerate the ageing of paper, it is
expected that the average measured TI at the top winding is low which in turn lead to
the reduction of paper life.
Figure 4.15 Fitting for the average of the measured TI at the top winding. Each point
represents the average of the measured TI at the top winding of each transformer.
98
Figure 4.16 Fitting for the average of the measured TI at the bottom winding. Each
point represents the average of the measured TI at the bottom winding of each
transformer.
As expected, the ageing rate based on the highest initial TI is the highest for both top
and bottom windings as shown in Table 4.6 and Table 4.7. The r value is high for both
fittings, 0.855. The average ageing rate at the top winding is around 2.7010-2 while
at bottom winding is around 1.9410-2. The ageing rate at the top winding is 1.4 times
higher than that of bottom winding.
Table 4.6 Ageing rate for the average of the measured TI at the top winding.
Fitting
parameters
-1
Based on average
Based on lowest
Based on highest
of the initial TI
of the initial TI
of the initial TI
-2
-2
CTI (years )
2.6810
2.5910
2.8210-2
0.924
0.928
0.905
Table 4.7 Ageing rate for the average of the measured TI at the bottom winding.
Fitting
Based on average
Based on lowest
Based on highest
parameters
of the initial TI
of the initial TI
of the initial TI
CTI (years-1)
1.9310-2
1.8510-2
2.0610-2
0.888
0.899
0.855
99
Generally, the estimated life at the top winding is lower than bottom winding. It is also
observed that the estimated life based on Criterion 1 is around 50 % of estimated life
based on Criterion 2. The finding is expected since the Weidmann model is similar to
the exponential decay equation which is often used to express the half life of
radioactive materials. If Criterion 3 is chosen as the end of life criterion, the estimated
life at the top winding is 58 years with 48 years and 70 years as the lower and upper
limit of 95 % confidence band. On the other hand, the estimated life at the bottom
winding is 81 years with 66 years and 96 years as the lower and upper limit of 95 %
confidence band.
Table 4.8 Estimated life for the average of the measured TI at the top winding.
End of life
criteria
Based on average of
Based on lowest of
Based on highest of
LCL
UCL
LCL
UCL
LCL
UCL
Criterion 1
26
22
31
27
23
32
25
21
30
Criterion 2
51
43
61
53
44
63
50
40
59
Criterion 3
58
48
70
59
49
71
57
47
69
F= Based on fitting, LCL and UCL= Based on lower and upper limit of 95 % confidence band
Table 4.9 Estimated life for the average of the measured TI at the bottom winding.
End of life
criteria
Based on average of
Based on lowest of
Based on highest of
LCL
UCL
LCL
UCL
LCL
UCL
Criterion 1
36
30
43
37
31
45
34
28
41
Criterion 2
72
59
86
75
62
89
68
55
82
Criterion 3
81
67
96
83
69
99
78
63
94
F= Based on fitting, LCL and UCL= Based on lower and upper limit of 95 % confidence band
100
4.4.2.2.
It is shown that if Criterion 3 is taken as the end of life criterion, the life estimated
based on the average of the measured TI is around 68 years while based on the lowest
of the measured TI is around 42 years. It is found that the life estimated based on the
lowest of the measured TI agrees with the current utility end of life definition of 40
years.
Figure 4.17 Fitting for the average of the measured TI. The square point represents
the average of the measured TI from different locations of each transformer.
101
Figure 4.18 Fitting for the lowest of the measured TI. The round point represents the
lowest of the measured TI from different locations of each transformer.
The ageing rates for both the average and lowest of the measured TI are given in Table
4.10 and Table 4.11. The r value is high for both fittings, 0.8. As expected, the
ageing rate for the lowest of the measured TI is higher than the average of the
measured TI. The average ageing rate for the average of the measured TI is around
2.3110-2 while the lowest of the measured TI is around 3.7210-2.
Table 4.10 Ageing rate for the average of the measured TI.
Fitting
Based on average
Based on lowest
Based on highest
parameters
of the initial TI
of the initial TI
of the initial TI
CTI (years-1)
2.3010-2
2.2110-2
2.4310-2
0.831
0.836
0.8
Table 4.11 Ageing rate for the lowest of the measured TI.
Fitting
Based on average
Based on lowest
Based on highest
parameters
of the initial TI
of the initial TI
of the initial TI
CTI (years-1)
3.7110-2
3.6010-2
3.8610-2
0.912
0.918
0.897
102
If Criterion 3 is taken as the end of life criterion, the estimated life from the average of
the measured TI is around 68 years with 55 years and 82 years as the lower and upper
limit of 95 % confidence band. On the other hand, the estimated life from the lowest of
the measured TI is around 42 years with 33 years and 52 years as the lower and upper
limit of 95 % confidence band.
Table 4.12 Estimated life for the average of the measured TI.
End of life
Criteria
Based on average of
Based on lowest of
Based on highest of
LCL
UCL
LCL
UCL
LCL
UCL
Criterion 1
30
25
37
31
26
38
28
24
36
Criterion 2
60
49
73
62
51
75
57
46
70
Criterion 3
68
55
82
70
57
83
65
53
81
F= Based on fitting, LCL and UCL= Based on lower and upper limit of 95 % confidence band
Table 4.13 Estimated life for the lowest of the measured TI.
End of life
Criteria
Based on average of
Based on lowest of
Based on highest of
LCL
UCL
LCL
UCL
LCL
UCL
Criterion 1
19
15
24
20
16
25
18
14
23
Criterion 2
37
29
46
39
30
48
36
28
45
Criterion 3
42
33
52
43
34
53
42
32
52
F= Based on fitting, LCL and UCL= Based on lower and upper limit of 95 % confidence band
The estimated life based on Criterion 2 is slightly lower than Criterion 3 and if
Criterion 1 is chosen as the end of life criterion, the estimated life would be much
lower than Criterion 2 and 3. Thus, it is proposed to re-introduce the 20 % retention of
TI criterion proposed previously by Dakin [99]. Table 4.14 shows the comparison of
the estimated life between Criterion 3 and Criterion 4 (20 % retention of TI). It is
observed that the estimated life is almost similar where the difference between
Criterion 3 and 4 is less than 5 %.
103
measured TI (years)
measured TI (years)
Criterion 3
Criterion 4
Criterion 3
Criterion 4
68
70
42
43
LC
55
57
33
34
UCL
82
83
52
53
F= Based on fitting, LCL and UCL= Based on lower and upper limit of 95 % confidence band
The end of life for transformers is usually defined based on the lowest of the measured
TI or DP. It is found that Criterion 4 is sufficient to represent the end of life of the
transformer population and the estimated life agrees with the current end of life of 40
years.
4.5.
In this section, different experiments were conducted for examining the dielectric
performance of the paper subjected to in-service ageing.
Service aged paper was obtained from Transformer 10 and the TI of the paper is 24.13
Nm/g. New paper, conductor and layer types were used to compare with the service
aged paper. The properties of each paper can be seen in Table 4.15. The density of
conductor and layer types paper was obtained from the manufacturers datasheet. It can
be seen that service aged paper has the highest density followed by conductor and
layer types paper. Aged mineral oil was obtained from previous ageing experiments.
The oil was aged without copper and paper under open condition for 6 months at 105
C.
104
Apparent density
Thickness
(g/m)
(g/cm)
(mm)
Layer paper
170
0.76
0.237
Conductor paper
48
0.93
0.053
101
1.17
0.086
Type of paper
New and aged mineral oils were first filtered and dried according to the procedure in
section 3.5.1 of Chapter 3. The pre-processing procedure of paper is described in
section 3.5.2 of Chapter 3. The acidity of new and aged mineral oil was measured after
the pre-processing procedure. The acidity of new oil is 0.01 mg KOH/g and the acidity
of aged oil is 0.22 mg KOH/g.
AC breakdown test was carried out using sphere electrodes with a diameter of 12.5
mm. All oil and paper samples were tested for breakdown voltage at a same gap
distance. In order to achieve this purpose, different number of paper layers were used
since paper comes in difference thickness as seen in Table 4.16. The gap between
electrodes was determined by the thickness of service aged paper (highest) and the
same gap distance was maintained for conductor and layer type papers. The gap
distance was also maintained when testing breakdown strength of the oil. A 50 Hz AC
voltage with increasing rate of 0.5 kV/s was applied to the paper impregnated with
mineral oil. The breakdown can be detected by the current relay set to 4 mA which
isolates the sample tested from the source in order to avoid the oil and paper
deterioration. A total of 40 measurements of breakdown voltage were carried out on
each sample.
Table 4.16 Number of paper layers used in breakdown tests.
Type of paper
Number of layers
Thickness (mm)
0.212
0.237
0.258
105
All samples were first degassed in a vacuum oven at room temperature before being
tested for breakdown voltage. All paper samples were subjected to 40 breakdowns and
the position of the paper between the electrodes was moved after each breakdown. The
gap distance for all samples is set based on the thickness of service aged paper with a
value of 0.258 mm.
It is observed that the breakdown voltage for service aged paper is higher than both
new conductor and layer types papers as seen Figure 4.19. The results show that the
mechanical strength of paper has no effect on the breakdown voltage of the paper,
however, the density does. Similar observation was noticed for laboratory accelerated
aged paper where the breakdown strength remain almost unchanged even though the
DP of the paper has already reaches 270 after 6 months of ageing at 120 C [100].
Another study for paper aged at different temperatures also shows the same pattern
where the reduction of the mechanical strength will not affect the papers AC dielectric
strength [101].
Figure 4.19 Breakdown voltages of different types of paper impregnated in new oil.
It is observed that density has a significant effect on the breakdown voltage of the
paper as shown in Figure 4.20 a). Service aged paper has the highest breakdown
voltage followed by the conductor and layer types paper. The result is in line with
106
a)
b)
According to previous work, a theory was proposed that the discharge happens in the
oil channel since it is the weak link in oil impregnated paper insulation and takes more
electrical stress than cellulose [102]. Therefore, if the quality of the oil is
compromised, the breakdown voltage of the paper might be affected as well. A study
was conducted in order to look into the effect of oil quality on the breakdown voltage
of the paper. First, a comparison of breakdown voltage is made between new and aged
oil as seen in Figure 4.21. It is shown that the aged oil suffers a 3.8 % reduction of
breakdown voltage and it is believed to be caused by the acids in the oil.
107
For paper impregnated in aged oil, it is observed that the reduction of breakdown
voltage increases ranging from 7 % to 9 % as seen in Figure 4.22, Figure 4.23 and
Figure 4.24. The reduction for paper impregnated in aged oil is almost 2 times higher
than that of the aged oil itself. These results show that the oil quality could affect the
overall performance of breakdown voltage of oil/paper system.
Figure 4.22 Breakdown voltages of layer type paper impregnated in new and aged
oils.
108
Figure 4.23 Breakdown voltages of conductor type paper impregnated in new and
aged oils.
Figure 4.24 Breakdown voltages of service aged paper impregnated in new and aged
oils.
Figure 4.25 shows the comparison between breakdown strengths of paper impregnated
in new and aged oil. Similar trend is observed for all types of paper, that paper
impregnated in aged oil suffers around 7 % to 9 % reduction of breakdown strength.
109
Figure 4.25 Average breakdown strengths of new and aged paper. Data are mean of
50 tests and error bars represent the standard deviation.
Statistical analysis was performed in order to determine the suitable distribution for the
breakdown voltage data. Normal and Weibull distribution were used to fit all
breakdown voltage data as seen in Figure 4.26, Figure 4.27, Figure 4.28 and Figure
4.29. A distribution is considered a good fit if the data points follow a straight line. It is
found that normal distribution is the best fitting for all breakdown voltage data.
0.999
0.99
0.95
0.9
0.75
Probability
0.5
0.25
Layer paper
Weibull fit of layer paper
Conductor paper
Weibull fit of conductor paper
Aged paper
Weibull fit of aged paper
New oil
Weibull fit of new oil
0.1
0.05
0.01
1
10
Breakdown voltage (kV)
Figure 4.26 Weibull probability plots of new and aged paper impregnated in new oil.
110
Layer paper
Weibull fit of layer paper
Conductor paper
Weibull fit of conductor paper
Aged paper
Weibull fit of aged paper
Aged oil
Weibull fit of aged paper
0.95
0.9
0.75
Probability
0.5
0.25
0.1
0.05
0.01
1
10
Breakdown voltage (kV)
Figure 4.27 Weibull probability plots of new and aged paper impregnated in aged oil.
0.999
Layer paper
Normal fit of layer paper
Conductor paper
Normal fit of conductor paper
Aged paper
Normal fit of aged paper
New oil
Normal fit of new oil
0.995
0.99
0.95
Probability
0.9
0.75
0.5
0.25
0.1
0.05
0.01
0
10
15
Breakdown voltage (kV)
20
25
Figure 4.28 Normal probability plots of new and aged paper impregnated in new oil.
111
0.995
0.99
0.95
Probability
0.9
0.75
0.5
0.25
0.1
0.05
0.01
0
10
15
Breakdown voltage (kV)
20
25
Figure 4.29 Normal probability plots of new and aged paper impregnated in aged oil.
Table 4.17 and Table 4.18 show the average breakdown voltage comparison between
normal and Weibull fitting. It is found that the average breakdown voltage obtained
from Weibull fitting is higher than normal fitting. According to the normal fitting, the
oil has the highest value of standard deviation followed by the service aged paper,
conductor and layer types paper. This indicates that oil has a higher variation of
breakdown voltage than paper.
Table 4.17 Average breakdown voltage of new and aged paper impregnated in new
and aged oil based on Weibull fitting.
Layer paper
Conductor
Service aged
(kV)
paper (kV)
paper (kV)
14.1
15.3
17.9
20.1
13.6
14.2
16.3
18.8
Test condition
Oil (kV)
Table 4.18 Average breakdown voltage of new and aged paper impregnated in new
and aged oil based on normal fitting.
Oil
Layer paper
Conductor
Service aged
(kV)
(kV)
paper (kV)
paper (kV)
13.91.57
15.21.33
17.71.16
20.01.43
13.41.87
14.01.37
16.20.91
18.61.58
Test condition
Multiple breakdowns were applied to a single point on paper impregnated in new and
aged oil. Two types of paper were used for the experiment, which are layer type paper
and service aged paper. The properties for both types of paper can be seen in Table
4.15. The gap distance between electrodes was determined by the individual thickness
of each paper. After each breakdown, the oil and paper sample is degassed for 5
minutes in order to reduce the influence of bubbles trap inside the paper pores. Due to
the 4 mA protection relay setting, the energy release during the breakdown test is
small, thus the fibre will not be seriously damaged.
The test results in Figure 4.30 and Figure 4.31 show that the breakdown voltage of
new paper impregnated in either new or aged oil remains almost unchanged even after
100 breakdowns. The degassing process helps to remove bubbles and thus prevents the
reduction of breakdown voltage of the new paper.
However, the situation is different for service aged paper where there is a reduction of
breakdown voltage after a certain number of tests. Point 1 and point 2 represent two
different paper samples. It is observed that for service aged paper impregnated in new
and aged oil, the breakdown voltage suddenly reduces to 0 kV after 32nd, 42nd, 83rd and
90th breakdown all varied but shorter than 100 breakdowns. A 0 kV breakdown voltage
represent the two electrodes is short circuited. The reduction of breakdown voltage for
service aged paper probably due to the carbonization of the fibre. As the paper aged,
the fibre is easily carbonized which could creates a conductive channel between the
two electrodes [100].
According to the visual inspection, service aged paper was easily destroyed when
subjected to breakdown voltage tests as compared to new paper. New paper creates
only carbonized dot while service aged paper create carbonized hole after being
subjected to breakdown tests as shown in Figure 4.32. It is believed that the carbonized
channel in the new paper is not as conductive as that of service aged paper to cause
sudden reduction on the breakdown voltage. In our case study, it is believed that the
energy released by the test equipment is not high and thus requires a numbers of tests
before sudden reduction on the breakdown voltage can be observed.
113
Figure 4.30 Multiple breakdowns at single point of paper impregnated in new oil.
Figure 4.31 Multiple breakdowns at single point of paper impregnated in aged oil.
a)
b)
Figure 4.32 Paper condition after breakdown tests a) New paper and b) Aged paper.
114
Summary
There are only small differences in TI profiles among different windings and phases.
This is reasonable since most transformers are designed symmetrically among phases
and oil flow distributions considered among windings. A study on multi layer
insulation shows that layer profile does exists and it is influenced by thermal and
chemical degradations.
A relationship between TI and transformer age was found and represented by the
Weidmann model. It is shown that the end of life of transformer population based on
paper TI is best represented by the 20 % retention of TI. Based on this criterion the
estimated life based on the lowest of the measured TI is around 43 years with 34 years
and 53 years as the lower and upper limit of 95 % confidence band.
It is observed that reduction of mechanical strength due to ageing does not bear any
relationship with the breakdown strength of the paper. However, paper density and oil
condition could influence the dielectric strength of the paper. Further study on the
breakdown mechanism shows that new paper could maintain its dielectric strength
even after being subjected to multiple breakdowns, provided that bubbles and other
impurities such as carbonized cellulose particles are avoided. However, as the paper
aged, the fibre strength is damaged and it is easy to carbonize. So, a conductive
channel along the carbonized fibre will be created by this process lead to the reduction
of the dielectric strength of the aged paper.
115
5.1.
Introduction
Ageing of paper and oil could generate different types of by-product such as water,
furans, gases and acids. Some of these by-products were used to determine the ageing
condition of transformers.
Recently it is found that there are two types of acids existing in paper and oil which are
Low Molecular Weight Acid (LMA) and High Molecular Weight Acid (HMA),
whereas most of LMA tend to stay in paper and most of HMA tend to stay in oil [16,
17]. Since a transformers functionality is closely related to its dielectric strength, it is
essential to study the effect of LMA and HMA on this property. It is also important to
examine the effect of these acids on the paper ageing.
So far, there is no publication describing study being conducted on the LMA contents
in scrapped transformers paper samples. LMA in paper samples from scrapped
transformers was measured in this PhD research study. Analysis of the relationship
between LMA and Tensile Index (TI) was performed in order to determine if LMA can
be used to indicate ageing profiles of transformer population.
LMA and HMA tests were carried out on in-service aged oil samples in order to
determine the partitioning between these acids for in-service transformers. Similar tests
were conducted on laboratory aged oil samples in order to study factors that could
affect the generation of LMA in oil. A paper TI prediction model was proposed based
on the measurement of LMA in oil.
5.2.
Effect of LMA and HMA on breakdown voltage of oil and ageing of paper
The effect of LMA and HMA on the breakdown voltage of oil was first examined in
this section. A study was also carried out on the relationship between breakdown
116
5.2.1.1.
Experiment description
Different amounts of either LMA ranging from 0.2 g to 0.4 g or HMA ranging from 1
g to 19 g were mixed with 500 ml dry and filtered mineral oil by magnetic stirrer for 3
hours at 50 C. The oil was dried and filtered according to the procedure described in
section 3.5.1 of Chapter 3. The properties of LMA and HMA can be seen in Table 5.1.
The pKa value describes the extent of an acid can dissociate in solution where smaller
number means the acid has high dissociation rate. After the mixing procedure, the oil
sample was titrated for acidity measurement and then tested for breakdown voltage.
Acid
Melting
Boiling
weight
dissociation
temperature
temperature
M r (g/mol)
constant (pKa)
Tm (C)
Tb (C)
LMA (Formic)
46
3.8
8.3
101
HMA (Stearic)
285
4.9
69.3
232
Types of acid
The AC breakdown test was carried out by mushroom VDE electrodes with a diameter
of 36 mm according to the ASTM standard, ASTM D1816 [103]. A 50 Hz AC voltage
with increasing rate of 0.5 kV/s was applied to the natural ester. The gap distance was
set at 1 mm and a total of 50 measurements of breakdown voltage were carried out on
all samples.
5.2.1.2.
Test results
It is found that LMA has a significant effect on the breakdown voltage of the oil as
seen in Figure 5.1. The oil suffers a rapid reduction of breakdown voltage as LMA
increases up to 0.1 mg KOH/g. As the LMA increases further, the low breakdown
117
Figure 5.1 Influence of LMA on the breakdown voltage of mineral oil. Breakdown
voltage data are mean of 50 tests and error bars represent the standard deviation. LMA
in oil data is 1 measurement.
On the other hand, there is no significant effect of HMA on the breakdown voltage of
the oil when HMA is within the range from 0.1 mg KOH/g to 4 mg KOH/g as shown
in Figure 5.2. Significant reduction of breakdown voltage occurs after HMA reaches 4
mg KOH/g. At HMA around 9 mg KOH/g, the oil suffers 64.2 % reduction of
breakdown voltage. However, for in-service transformers, the acidity of mineral oil
will never reach 4 mg KOH/g and rarely reach 1.0 mg KOH/g. Thus, it is believed that
if there is a reduction of breakdown voltage of an acid oil, it might mainly be
contributed by LMA.
118
5.2.1.3.
A total of 147 transformers excluding scrapped and oil regenerated transformers were
analyzed from a distribution transformer database. Yearly measurements of AC
breakdown tests and TAN were carried out on each transformer and corresponding
relationship is shown in Figure 5.3. The breakdown voltage test was carried out
according to IEC standard, IEC 156 [104] while TAN according to BSI standard, BS
62021[90].
It is observed that there is no clear correlation between TAN and the breakdown
voltage of in-service oil. The scattering of the data is high where the breakdown
voltage can be as low as 7 kV for TAN less than 0.1 mg KOH/g or as high as 60 kV
for TAN around 1.11 mg KOH/g.
Previous finding shows that under laboratory controlled experiment, the breakdown
voltage of oil could be affected by many factors such as water, particles and acidity
[105].
Figure 5.3 Influence of TAN on the breakdown voltage of in-service oil. Breakdown
voltage data are mean of 50 tests. TAN data is 1 measurement.
5.2.2.1.
Experiment description
In order to prepare oil samples with acids, 0.2 g of either LMA or HMA was mixed
with 200 ml dry and filtered mineral oil by magnetic stirrer for 3 hours at 50 C. The
oil was dried and filtered according to the procedure described in section 3.5.1 of
Chapter 3. After the mixing procedure, the oil sample was titrated for acidity. It is
observed that the acidity of samples with LMA has a high deviation ranging from 0.22
mg KOH/g to 0.45 mg KOH/g. On the other hand, the acidity for samples with HMA
is stable ranging from 0.22 mg KOH/g to 0.24 mg KOH/g.
All samples were aged in an air circulating oven at 120 C. Each sample contains 1.75
g of conductor paper and 200 ml of mineral oil containing either LMA or HMA. The
paper was pre-processed according to the procedure described in section 3.5.2 of
Chapter 3. For dry paper sample, the water content is less than 0.5 % while for wet
120
5.2.2.2.
Figure 5.4 shows the affect of LMA and HMA on the paper aged under initial dry
condition. It is observed that the TI of samples with no acid reaches 50 % of its
strength after 57 days of ageing, with HMA after 49 days of ageing and with LMA
after 24 days of ageing.
Figure 5.4 Influence of LMA and HMA on the paper aged under dry condition. Data
are mean of 8 samples and error bars represent the standard deviation.
The relationship between TI and ageing duration was fitted by Equation (4.1). Table
5.2 shows that the ageing rate for samples with LMA is 2.1 and 2.4 times higher than
with no acid and with HMA.
121
5.2.2.3.
Condition
With no acid
1.510-2
0.986
With LMA
3.110-2
0.940
With HMA
1.310-2
0.956
For wet ageing condition, the effect of LMA on the paper ageing is greatly enhanced
as seen in Figure 5.5. The TI of samples with LMA reduces to 50 % of its strength
after 7 days of ageing while with no acid and with HMA after 33 days of ageing.
Figure 5.5 Influence of LMA and HMA on the paper aged under wet condition. Data
are mean of 8 samples and error bars represent the standard deviation.
The ageing rate is further increased for wet ageing condition as seen in Table 5.3. The
ageing rate of samples with LMA under wet ageing condition is around 3 times higher
than under dry ageing condition. It is also observed that the ageing rate of samples
with LMA is 4 times higher than without no acid and with HMA. The r fitting of
samples with LMA is low because the variation of TI measurement will be high, as
paper reaches the brittle state.
122
With no acid
2.210-2
0.793
With LMA
9.710-2
0.700
With HMA
2.210-2
0.911
The ageing rate of samples with no acid is almost similar to with HMA, for both dry
and wet condition, this indicates that there is no effect of HMA on the paper ageing.
The increment factor was calculated by Equation (5.1).
Increment factor =
(5.1)
Table 5.4 shows the increment factors for the combination of acids and water, when
taking dry and new oil and paper condition as reference; the ageing rate for samples
with LMA increases 3.1 times from dry to wet ageing condition. On the other hand, a
low increment of ageing rate is observed for HMA added samples where it increases
1.7 times from dry to wet ageing condition.
Increment factor
With LMA
2.1
6.5
With HMA
0.9
1.5
Condition
It is clear that LMA and water have multiplicative effect on the paper ageing. Similar
observation was also found with DP in previous finding [16]. According to previous
theory, the hydrolysis of paper is catalyzed by H+ originated from dissociated acids
[18]. LMA can be easily dissociated since the pKa value is lower than HMA. By
adding LMA, the rate of acid dissociation will increase and the resultant H+ catalyzes
the hydrolysis of wet paper and thus accelerates the paper degradation.
123
In total, 9 transformers were examined for LMA analysis. Paper samples for LMA
measurement were taken from the same location as in section 4.3 of Chapter 4. The oil
residue on the paper samples were wiped using a tissue before being tested for LMA.
Based on all measurements, it was found that the LMA in paper for scrapped
transformers ranges from 0.2 mg KOH/g to 6.4 mg KOH/g. LMA in paper was first
analyzed by comparing phases, windings and layers where the sampling of paper is
carried out according to the procedure described in Figure 3.1 and Figure 3.2 of
Chapter 3. The second part of this section describes the relationship between LMA in
paper, TI and transformer age.
5.3.1.1.
The same transformers as in section 4.3.1.1 of Chapter 4 were analyzed for the profile
of LMA along windings and phases. The vertical profile of LMA in paper for A, B and
C phases can be seen in Figure 5.6. The age, paper properties and TAN of oil of each
transformer can be seen in Table 4.2 of Chapter 4. It is observed that the LMA in paper
has a big difference between phases at the top and centre winding. For example, the
LMA in paper at the top winding of Transformer 6 for A and C phase are much lower
than B phase.
124
a)
b)
Figure 5.6 LMA in paper profiles among different phases a) Transformer 4 (SW) and
b) Transformer 6 (CW). Data are mean of 2 measurements.
Figure 5.7 LMA in paper profiles among different windings of Transformer 8. Paper
samples were taken from B phase. Data are mean of 2 measurements.
It is suspected that the LMA in paper at the top winding for some of the phases of
Transformer 4 and Transformer 6 might be evaporated due to the high temperature at
this location. Top winding usually has high temperature and hotspot is normally
located near to this location. On the other hand, it is reasonable to expect the LMA in
paper at the centre location between phases has a big difference since the TI for both
transformers also shows similar profiles. Paper ageing will result in the reduction of TI
and at the same time lead to the formation of LMA which mainly stays in paper. Small
125
Figure 5.7 shows the vertical profile of LMA in paper for windings. It is difficult to
draw any conclusion for the winding profile since only one transformer is available for
the analysis.
It can be seen that paper at the top winding yields slightly higher LMA than bottom
winding. The paper at the top winding is usually more degraded than bottom winding
because the temperature is higher at the top. Since most of LMA originates from paper
ageing, it is reasonable to expect that the LMA in paper is slightly higher at the top
winding than bottom winding. However in some cases, LMA in paper at the bottom
winding can be slightly higher than top winding which can be seen from A phase of
Transformer 6. It is suspected that there might be an evaporation of LMA in paper at
the top winding for this transformer.
5.3.1.2.
Layer profile
Same transformers as in section 4.3.1.2 of Chapter 4 were analyzed for the profile of
LMA along layer. The age, paper properties and TAN of oil of each transformer can be
seen in Table 4.3 of Chapter 4. It is observed that between the 8th to 4th layers paper,
the LMA for Transformer 10 ranges from 1.88 mg KOH/g to 3.98 mg KOH/g with an
average value of 2.90 mg KOH/g as shown in Figure 5.8. The highest LMA is
observed at the 2nd layer paper with a value of 7.24 mg KOH/g. The 1st layer paper has
a lower LMA than the 2nd layer paper probably due to the evaporation of LMA.
Similar pattern of LMA in paper for multilayer insulation is observed for Transformer
6, as shown in Figure 5.9. The lowest LMA is observed at the 11th layer paper with a
value of 1.86 mg KOH/g. The LMA slightly increase at the 10th layer paper and remain
almost unchanged up to the 2nd layer paper with an average value of 3.31 mg KOH/g.
The 1st layer paper has the highest LMA with a value of 5.5 mg KOH/g.
126
Figure 5.8 LMA in paper profiles among different layers of Transformer 10. Paper
samples were taken from B phase, HV bottom. Data is 1 measurement.
Figure 5.9 LMA in paper profiles among different layers of Transformer 6. Paper
samples were taken from B phase, CW top. Data are mean of 2 measurements.
Both results suggested that the outer layer paper could be subjected more to chemical
degradations. The paper in direct contact with oil has a higher LMA than the inner
layer. This observation supports the previous assumption that the outer layer paper is
affected by the chemical degradation process [94]. The main mechanism behind this
phenomenon is related to the condition of oil. The ageing of oil and paper produce
127
It can be seen that the LMA in paper of the first three layers of Transformer 10 is
higher than Transformer 6. The high LMA in paper in Transformer 10 is expected
since the outer layer paper suffers more severe degradation than Transformer 6 based
on the mechanical strength reduction. Furthermore, it is noticed that a high amount of
LMA is expected for paper with a low TI as shown in Figure 5.10 and Figure 5.11.
Thus, it is suggested that there is a relationship between TI and LMA in paper.
Figure 5.10 TI and LMA in paper profiles among different layers of Transformer 10.
Paper samples were taken from B phase, HV bottom. LMA in paper data is 1
measurement. TI data are mean of 9 to 10 samples.
128
Figure 5.11 TI and LMA in paper profiles among different layers of Transformer 6.
Paper samples were taken from B phase, CW top. LMA in paper data are mean of 2
measurements. TI data are mean of 9 to 10 samples.
The study on the 1st layer paper of both transformers shows that the LMA in paper for
a young transformer could be higher than an old transformer. On the other hand, the TI
for a young transformer can lower than an old transformer. It is suspected that
Transformer 10 might be subjected to either a higher loading or worse thermal design
than Transformer 6. This is supported by the fact that TAN of the oil for Transformer
10 is higher than Transformer 6. The scrapping report of Transformer 10 briefly
mentioned that the severe degradation of solid insulation is due to the high loading.
However, no loading history could be obtained for both transformers.
5.3.2.1.
Figure 5.12 shows the relationship between LMA in paper and TI based on the
measurements of 9 scrapped transformers. It is shown that there is a link between TI
and LMA; i.e. LMA in paper increases as TI of paper decreases. The correlation
between TI of paper and LMA in paper was calculated based on Pearson correlation
129
r xy =
(x-x) (y-y)
(x-x) (y-y)
2
where rxy
(5.2)
2
= Correlation factor.
x and y
= Variables.
x and y
= Mean of variables.
The correlation factor between TI and LMA in paper is 0.71 which indicates that
there is strong correlation between these parameters. As the paper degrades, LMA will
be generated and it accumulates in the paper which in turn contributes to the further
increment of LMA. The relationship between LMA in paper and TI of paper also
might indicate that the generation of LMA is higher than the consumption of LMA.
The measured LMA is a result from the subtraction of LMA generation with LMA
consumption.
Figure 5.12 LMA in paper versus TI of 9 transformers. Paper samples were taken
from different locations. LMA in paper data are mean of 2 measurements. TI data are
mean of 4 to 10 samples.
130
(5.3)
By expanding Equation (5.3), it will become a linear equation as seen in Equation (5.4)
and the plotting can be seen in Figure 5.13. The form of this equation is quite similar
to the equation representing the relationship between Furanic Compounds (2FAL) and
DP [75]. It can be seen clearly that the scattering of the data is high at the initial stage
of ageing which indicates the LMA in paper varies when TI is still in a good condition.
(5.4)
If the relationship between LMA in paper and LMA in oil could be obtained, the
condition of paper can be known based on the measurement of LMA in oil. In order to
131
5.3.2.2.
The relationship between LMA in paper and transformer age can be seen in Figure
5.14. The LMA in paper data was obtained from the same location as TI in Figure 4.6.
It can be seen that the LMA in paper can be as high as 3.2 mg KOH/g for a young
transformer - Transformer 1. As shown in Figure 5.15 and Figure 5.16, the average
daily loading of typical summer and winter for Transformer 1 is relatively high as
compared to the other transformers, and this could be the reason for the high LMA
observed in the paper at a young age. The effect of loading on the amount of LMA in
paper is also shown by Transformer 5. It is shown that the low amount of LMA in
paper corresponds well with the low average daily loading at both seasons. However, it
is observed that the LMA in paper could be high even though the average daily loading
of both seasons is, low as shown by Transformer 4. Thus, it is believed that age could
also influence the amount of LMA in paper. As transformer age increases, the paper
will degrade with the accelerated help from ageing by-products even at low load which
in turn increase the amount of LMA in paper.
Figure 5.14 LMA in paper versus transformer age of 9 transformers. Data are mean of
2 measurements.
132
Figure 5.15 Average LMA in paper and average daily loading of typical summer.
Figure 5.16 Average LMA in paper and average daily loading of typical winter.
Another possible factor that could influence the amount of LMA left in paper is
design. However, it is difficult to see the effect of design on the amount of LMA in
paper since all transformers under study come from different designs.
An increment trend can be observed between LMA in paper and transformer age by
considering all transformers are from the same design and experienced the same
loading as shown in Figure 5.17. The trend suggested that the LMA tend to accumulate
in the paper with the increment of time. However, there is no evidence to prove the
133
Figure 5.17 Fitting of LMA in paper versus transformer age of 9 transformers. Data
are mean of 2 LMA in paper measurements.
An exponential relationship was used to represent the increment of LMA in paper with
transformer age as shown in Equation (5.5).
A = A0 +BexpCt
where A and A0
(5.5)
Table 5.5 shows the fitting parameter of the relationship between LMA in paper and
transformer age. A0 was set to 0.12. The constant, B was adjusted in order to make sure
the initial LMA in paper obtained from the fitting as close as possible to (A0+B= 0.12
mg KOH/g) since LMA measured from the new conductor paper is around this value.
The r of the fitting is low for both average and highest LMA at a value of 0.38 and
0.294 due to high deviation of LMA in paper for Transformer 1 and 2.
134
B (mg KOH/g)
5.6 10 2
8.0 10 2
C ( years 1 )
8.9 10 2
9.0 10 2
0.380
0.294
5.4.
5.4.1.1.
Experiment description
An experiment was conducted in order to study the partitioning of LMA in paper and
oil. Three different ratios of paper and oil were used as shown in Table 5.6. Around
0.08 g of LMA was mixed with the 75 g of dry and filtered mineral oil by magnetic
stirrer for 3 hours at 50 C in a 250 ml beaker. The oil was dried and filtered according
to procedure described in section 3.5.1 of Chapter 3. After the mixing procedure, the
oil and LMA mixture is titrated for acidity and the value is recorded as initial amount
of LMA in oil. The dry paper is added into oil containing LMA and it is left to stand in
a 250 ml bottle for at least 10 days. The paper was dried according to the procedure
described in section 3.5.2 of Chapter 3. It is observed that there is about 13 %
difference between the highest and lowest value of initial LMA in oil. The difference
might be due to the incomplete mixture of LMA with the oil especially for Sample 2.
Table 5.6 LMA ratio of paper and mineral oil. Data is 1 measurement.
Ratio
LMA added
(g)
(mg KOH/g)
Sample 1
4.67: 75
0.084
0.68
Sample 2
7: 75
0.083
0.62
Sample 3
14: 75
0.082
0.70
Samples
135
Table 5.7 shows that as the paper weight increases, the LMA in oil decreases. On the
other hand, the value of LMA in paper ranges from 12.89 mg KOH/g to 21.17 mg
KOH/g. The results show that most of the LMA tends to stay in the paper. The ratio of
LMA in paper and oil is heavily influenced by the weight ratio of paper and oil. By
increasing the amount of paper in the paper/oil system, more LMA could migrate from
oil to paper and stay in the paper. Furthermore, the percentage ratio of LMA and HMA
in oil might change if the weight ratio of paper and oil changes.
Table 5.7 LMA and HMA ratio of paper and mineral oil. LMA in paper data are mean
of 2 measurements and LMA in oil data is 1 measurement.
Absolute weight
Ratio
LMA in paper
LMA in oil
(mg KOH/g)
(mg KOH/g)
4.67: 70
18.09
0.27
4.47: 1
7: 70
21.17
0.14
15.1:1
14: 70
12.89
0.01
257.8: 1
ratio of LMA
(paper: oil)
The relationship between weight ratio of paper and oil and LMA ratio of paper and oil
can be seen in Figure 5.18. It can be seen that the LMA ratio doubled once the paper
weight is increased by 2.33 g. As the paper weight is increased by 7 g, the LMA ratio
increased almost 8 times.
It is observed that the weight ratio of paper and oil has significant effect on the LMA
ratio of paper and oil. The information could be used for predicting LMA in paper in
transformers based on LMA in oil provided the design information (weight of paper
and oil) is known.
136
Figure 5.18 LMA ratio of paper and oil versus weight ratio of paper and oil. LMA in
paper data are mean of 2 measurements. LMA in oil data is 1 measurement.
5.4.2.1.
Table 5.8 shows the list of transformers used for analysis of LMA and HMA ratio of
oil. All transformers are used at transmission level and still in-service. The oil
sampling was done by the oil maintenance company. All transformers come from
different designs.
Table 5.8 Case study of transformers.
Transformer
Rating
Age (years)
11
240MVA 400/132kV
12
40MVA 275/20.5kV
13
180MVA 275/66kV
16
14
1000MVA 400/275kV
20
15
240MVA 400/132kV
26
16
240MVA 400/132kV
37
17
120MVA 275/66kV
42
Only two loading history could be obtained for this case study. The daily loading of
typical summer and winter of Transformer 15 and 16 can be seen in Figure 5.19 and
137
138
Ideally, the amount of TAN should equal to the total amount of LMA and HMA.
However, in the measurement process, the total amount LMA and HMA could be
slightly lower or higher than the amount of TAN, since it is difficult to precisely
extract all the water that contains LMA from the oil.
Figure 5.21 TAN, LMA and HMA of in-service oil. LMA, HMA and TAN of oil data
is 1 measurement.
LMA percentage
HMA percentage
139
(5.6)
26.6
73.4
33.6
64.9
16
23.4
76.6
20
33.0
67.0
26
24.9
75.1
37
18.4
81.6
42
22.9
77.1
5.4.2.2.
LMA and HMA ratio for ageing of oil without any presence of
paper
A study was conducted on laboratory aged oil sample in order to determine the amount
of LMA generated by oil and factors that could affect the generation. A series of LMA
and HMA measurements were performed on previous laboratory aged mineral oil.
a) Experiment description
Aged mineral oil was obtained from previous ageing experiments. The oil was aged
without paper at 115C under open and sealed condition with or without the presence
of copper. TAN, LMA and HMA were measured for each sample.
It is observed for sealed samples that there is a small increment of LMA in oil and it
reaches to 0.013 mg KOH/g after 45 days of ageing as seen in Figure 5.22. Similar as
140
Figure 5.22 LMA and HMA ratio of mineral oil aged under sealed condition without
copper. LMA, HMA and TAN of oil data is 1 measurement.
It can be seen that the percentage of LMA ranges from 17.5 % to 21.2 % with an
average percentage of 16.8 % and for HMA ranges from 78.8 % to 86.8 % as seen in
Table 5.10.
Table 5.10 LMA and HMA percentages of mineral oil aged under sealed condition
without copper.
Ageing duration (days)
17.5
82.5
17.9
82.1
14.3
85.7
14
13.2
86.8
56
21.2
78.8
Similar to sealed samples, LMA in oil for open samples show increment as the ageing
duration increases and can reach up to 0.074 mg KOH/g as seen in Figure 5.23. It is
believed the actual amount of LMA in oil could be higher than the measured value.
141
Figure 5.23 LMA and HMA ratio for mineral oil aged under open condition without
copper. LMA, HMA and TAN of oil data is 1 measurement.
The percentage ratio of LMA and HMA can be seen in Table 5.11. The variation of
LMA percentage is higher for open samples ranging from 3.5 % to 33.7 % with an
average percentage of 23.4 % ignoring the 3.5 % LMA of 14 days sample, while HMA
ranges from 66.3 % to 96.5 %. The percentage of LMA is low for 14 days sample is
low because the amount of LMA is too low compared to HMA which in turn affect the
calculation of LMA percentage in Equation (5.6).
142
33.7
66.3
28.6
71.4
15.8
84.2
14.2
85.8
14
3.5
96.5
56
24.6
75.4
According to both sealed and open samples, it is shown that oil could also generate
LMA with a percentage among TAN from 16.8 % to 23.4 %.
It can be seen in Figure 5.24 that the LMA in oil is almost at the same level as HMA
in oil for sealed samples aged with copper. There is a slight increment pattern of LMA
in oil observed and the highest recorded is 0.052 mg KOH/g. There is no significant
increment of TAN observed for all ageing durations.
143
Figure 5.24 LMA and HMA ratio for mineral oil aged under sealed condition with
copper. LMA, HMA and TAN of oil data is 1 measurement.
It is observed that the LMA percentage of sealed samples aged with copper is higher
than without copper as shown in Table 5.12. The percentage of LMA ranges from 44.2
% to 46.3 % with an average percentage of 45.4 % and HMA ranges from 53.7 % to
54.1 %. It is found that copper could affect the percentage ratio of LMA and HMA
ratio.
Table 5.12 LMA and HMA percentage of mineral oil aged under sealed condition with
copper.
Ageing duration (days)
14
45.9
54.1
21
44.2
55.8
36
46.3
53.7
Only 2 aged oil samples could be obtained for the case study as seen in Figure 5.25. As
expected, the presence of both copper and oxygen increases the TAN of the oil
significantly and the highest LMA in oil observed is 0.098 mg KOH/g. It is also
noticed that the LMA in oil for both samples is almost similar. It is believed that the
actual amount of LMA in oil for 56 days sample should be higher than the 12 days
sample. Thus, it is suspected that the LMA in oil for 56 days sample might be
144
Figure 5.25 LMA and HMA ratio for mineral oil aged under open condition with
copper. LMA, HMA and TAN of oil data is 1 measurement.
Table 5.13 LMA and HMA percentage of mineral oil aged under open condition with
copper.
Ageing duration (days)
12
43.7
56.3
56
15.1
84.9
d) Summary on LMA and HMA ratio of oil ageing without any presence of paper
It is found that LMA could also be generated by oil and its amount could be increased
under oxidation and presence of copper. The result supports the assumption that the
ageing mechanisms in transformers starts from the oxidation of oil which generates
LMA and initiates the hydrolysis of paper as shown Figure 2.13 of Chapter 2.
Once paper is introduced in the oil, the percentage ratio of LMA in oil might change
since this acid prefers to stay in the paper. Thus, the LMA measured in paper might be
a combination of LMA obtained from oil oxidation and paper ageing.
145
It is proposed that LMA could be used as an ageing marker to determine the ageing
condition of the paper. Ideally, the LMA in paper could be predicted if the relationship
between LMA in paper and LMA in oil is known. Through LMA in paper, the TI can
be calculated based on Equation (5.4). Furthermore, life assessment of transformers
could be carried out based on the predicted TI.
The calculation can be divided into two stages. The first stage calculates the TI and the
second stage computes the transformers life consumption. LMA in paper and TI are
predicted in the first stage of the calculation. The second stage calculates the life
consumption of a transformer. TI and a transformers life consumption will be
computed as the outputs for the calculation.
146
INPUT
Equation to represent LMA in oil versus
transformer age
Equation to represent LMA in paper
versus transformer age
OUTPUT 1
TI
PREDICTIONOFTRANSFORMERSLIFE
CONSUMPTION BASED ON LMA
Prediction of transformers life consumption
Calculate transformer life consumption based on Equation (4.1)
OUTPUT 2
Transformer life
consumption
Equation (5.4) = LMA in paper versus TI
Equation (4.1) = TI versus transformer age
Figure 5.26 Calculation procedure for predicting TI and transformer life consumption.
147
5.5.2.1.
The first step is to determine the relationship between LMA in paper/LMA in oil and
transformer age. Two inputs are required for the model as follows:
Input 1
Input 2
The procedures to determine the equation that can represent LMA in oil versus
transformer age are shown in Figure 5.27.
PRO 1
Assumptions
PRO 2
Fitting of TAN versus transformer age (based on distribution
transformers)
PRO 3
Apply the same equation of TAN versus transformer age
to LMA in oil versus transformer age
Figure 5.27 Procedure of modelling the relationship between LMA in oil and
transformer age.
PRO 1 Assumptions
There are two limitations in determination of the equation that can represent the
relationship between LMA in oil and transformer age, and they are:
148
Lim 2
Thus, three assumptions are made in order to simplify the fitting process and they are
listed as follows:
Asm 1
Based on the fact that both scrapped and in-service transformers operate
at transmission level where the loading profiles of all transformers do
not vary significantly.
Asm 2
The yearly LMA in oil follow the same increment pattern of yearly
TAN.
Based on the fact that the LMA percentage for transformers with age
ranging from 1 year to 42 years varies slightly with the average value of
26.1 % as shown in Table 5.9.
Asm 3
Due to Lim 2, it is difficult to model the relationship between LMA in oil versus
transformer age. Thus, an approach based on Asm 2 was employed which required
fitting of TAN versus transformer age. Further assumption is made based on Asm 3
since the available TAN data are from distribution transformers.
The same case study as in section 5.2.1.3 was used to determine the relationship
between TAN and transformer age as shown in Figure 5.28. It is observed that there is
a scattering of the data in the early age which is believed to be caused by abnormal
149
Through filtering the data, it was found that there is a general pattern of TAN
increment with age. Thus, it is assumed that the LMA in oil would also follows the
same pattern since the LMA percentage varies slightly at a value around 26.1 % and
the percentage is similar for both distribution and transmission transformers based on
Asm 2 and Asm 3. It is found that the relationship between TAN and transformer age
could be fit by Equation (5.5). The fitting was done according to the average, highest
and lowest TAN of each transformer age. The fitting parameters are shown in Table
5.14. The initial TAN was set to be 0.01 mg KOH/g. The r for fitting based on highest
TAN is low, 0.446, due to the fact that less data is available when the transformer age
is exceed 55 years. However, the r for fitting based on the average and lowest TAN is
quite high, 0.886 and 0.886, which indicates that the equation is suitable to represent
the relationship between TAN and transformer age.
150
Based on
Based on
Based on lowest
parameters
average TAN
highest TAN
TAN
-3
-3
B (mg KOH/g)
4.310
9.310
0.1410-3
C( years 1 )
7.710-2
7.110-2
1.310-2
0.886
0.446
0.886
PRO 3 Apply the same equation of TAN versus transformer age to LMA in oil
To model the LMA in oil versus transformer age, by applying the same equation as
TAN versus transformer age, is shown in Figure 5.29. The result of the fitting can be
seen in Table 5.15. The initial LMA in oil was set to 0.001 mg KOH/g. The low r,
0.364, is due to less data available for the study.
Value
B (mg KOH/g)
2.010-3
C (years-1)
6.310-2
0.364
151
(5.7)
+
LMA IN OIL VERSUS TRANSFORMER AGE
0.06Age
(5.8)
=
LMA IN PAPER VERSUS LMA IN OIL
Figure 5.30 Modelling procedure for the relationship between LMA in paper and
LMA in oil
5.5.2.2.
Once the model for LMA in paper versus LMA in oil is obtained, the TI could be
predicted based on Equation (5.4). An example of the TI calculation based on LMA in
oil is shown as follows:
For examples, the LMA in oil measured from two transformers are 0.01 mg
KOH/g and 0.03 mg KOH/g.
152
Based on Equation (5.9), the LMA in paper for both transformers are 0.59 mg
KOH/g and 2.55 mg KOH/g.
Step 3 Calculate TI based on Equation (5.4)
Through Equation (5.4), the predicted TI for both transformers are 97.9 Nm/g
and 49.1 Nm/g.
Since the prediction is made based on a few reasonable assumptions, an error on the
prediction of TI of the paper could be expected. The model could be improved if the
oil of a transformer which is planned to be scrapped can be obtained for the
measurement of LMA. Through this procedure, the relationship between LMA in
paper and oil could be refined and thus improve the prediction of TI.
5.5.2.3.
A study was conducted to examine how the LMA in paper and oil ratio changes over a
transformers life. The ratio shown in Equation (5.10) was calculated based on the data
of the same age obtained from both Equation (5.7) and Equation (5.8).
(5.10)
The relationship between LMA in paper/oil ratio and transformer age can be seen in
Figure 5.31. It is observed that initially, the ratio reduces slightly and starts to increase
after 10 years in-service. It is believed that at the initial stage of ageing, less LMA is
generated in transformers and most of them are consumed to accelerate the ageing of
the paper. As the paper aged, more LMA will be generated in the system and stay in
the paper which contributes to the increment of the ratio. It is suspected that at this
point, the generation of LMA is higher than the consumption of LMA since the
measured LMA is a result from the subtraction of LMA generation with LMA
consumption.
153
Figure 5.31 Ratio of LMA in paper and oil versus transformer age
5.5.3.1.
t=
where
lnTI 0 -lnTI
CTI
(5.11)
TI
Predicted TI (Nm/g).
TI0
Initial TI (Nm/g).
CTI
Life (years).
Ageing rate based on average initial TI and lowest measured TI are chosen in order to
illustrate the life consumption calculation. The ageing rate based on these criteria is
3.7110-2 as given in Table 4.11 of Chapter 4 while according to Table 4.4 of Chapter
4, the average initial TI is 107 Nm/g. By using the predicted TI from the previous
154
5.6.
Predicted TI (Nm/g)
97.9 Nm/g
49.1 Nm/g
21
Summary
LMA could affect the performance of the breakdown voltage of oil. It is shown that the
breakdown voltage of oil could be reduced to half when LMA in oil is as low as 0.06
mg KOH/g. There is no effect of HMA on this property. LMA has a significant impact
to the ageing of paper. The impact is greatly enhanced with moisture as the ageing rate
can be increased by a factor of 6. On the other hand, it is observed that there is little
effect of HMA on ageing of paper.
There is no significant difference of LMA in paper profile along phases and windings
of scrapped transformers. The profile of LMA in paper for multilayer insulation shows
that there is an existence of chemical degradation which mainly involves the paper
layer in contact with oil. Two relationships were found between LMA in paper, TI and
transformer age. The first relationship shows that as LMA in paper increases, TI
decreases. The shape of Chendongs model between furanic compounds and degree of
polymerization [75] was found suitable to represent the relationship between LMA in
paper and TI. It is also found that LMA in paper increases as transformer age
increases.
The LMA ratio in paper and oil is strongly influenced by the weight ratio of paper and
oil. A study on the in-service oil shows that there is a slight increment of LMA in oil
as age increases. The average percentage of LMA in in-service aged oil is around 26.1
%. Further study on laboratory aged oils shows that LMA could also be generated
from the oil. Using information obtained from scrapped and in-service transformers, a
general TI prediction model based on LMA in oil was developed with a few reasonable
assumptions.
155
6.1.
Introduction
Acidity and viscosity are common methods used to assess the oxidation level of
natural ester. Acidity is usually expressed as Total Acid Number (TAN) and viscosity
is defined as fluids internal resistance to flow, also known as a measure of fluid
friction. Acids accelerate the ageing process of oil and paper. Recently it was
suggested that the current measurement technique of TAN is unsatisfactory since it
could not distinguish the types of acid in oil [16, 19]. According to these studies, it
was found that acids generated from ageing of paper and mineral oil consist of Low
Molecular Weight Acid (LMA) and High Molecular Weight Acid (HMA), where
LMA was found to be more dangerous since it could accelerate the paper degradation
[16]. However, there is no study being conducted on the LMA and HMA of natural
ester and its impact toward paper ageing.
The performance of natural ester under oxidation is examined through TAN and
viscosity measurements. Next, the behaviour of natural ester aged under sealed
condition is examined. TAN, viscosity, LMA and HMA measurements were carried
out throughout the ageing duration. Finally, the effect of LMA and HMA on the paper
ageing and the breakdown voltage of natural ester are discussed in this chapter.
6.2.
The first part of this section investigates the oxidation process of natural ester. The
second part discusses the behaviour of natural ester under oxidation and pyrolysis with
the presence of paper, copper and steel.
156
6.2.1.1.
Experiment description
Aged natural ester and mineral oil were obtained from previous ageing experiments.
The natural ester and mineral oil were aged without paper and copper at 115 C under
open and sealed conditions. TAN was measured for each sample.
6.2.1.2.
Test results
The comparison of TAN between natural ester and mineral oil aged under open
condition without any presence of paper, copper and steel can be seen in Figure 6.1. It
is observed that the TAN of natural ester is higher than mineral for all ageing
durations. The TAN of natural ester starts to increase rapidly after 7 days of ageing. On
the other hand, the increment of TAN for mineral oil is almost linear. The results show
that under oxidation, the generation of TAN for natural ester is much higher than
mineral oil.
Figure 6.1 TAN of natural ester and mineral oil aged at 115 C under open condition
without any presence of paper, copper and steel. Data is 1 measurement.
157
6.2.2.1.
Experiment description
Table 6.1 Amount of oil, paper, copper and steel ratio for single ratio.
Natural ester (ml)
Copper (cm2)
Steel (cm2)
900
7.86
0.10
3.31
The second ratio can be seen in Table 6.2 and it is defined as double ratio. Natural
ester and mineral oil were used for the ageing experiments of double ratio for
comparative purpose.
Table 6.2 Amount of oil, paper, copper and steel ratio for double ratio.
Natural ester/mineral oil (ml) Kraft Paper (g) Copper (cm) Steel (cm)
900
7.86
0.20
6.62
Kraft paper with a density of 0.93 g/cm was used for all ageing experiments. The oil
and paper were dried and filtered according to the procedure described in section 3.5.1
and 3.5.2 of Chapter 3. On the other hand, copper and steel were de-coated by a sand
paper in order to remove the surface oxidation layer.
Open and sealed ageing experiments were conducted for both ratios of natural ester
and only sealed ageing experiment was carried out for mineral oil. For sealed samples,
PTFE screw cap with a gasket ring was used and the sealing tape was wrapped around
the bottle neck in order to reduce the migration of oxygen from outside.
The ageing temperature was set at 170 C and only 6 samples were placed in the air
circulating oven at any given time in order to obtain a good and even temperature
158
6.2.2.2.
Test results
a) TAN
Figure 6.2 shows the comparison of TAN for natural ester and mineral oil aged under
open condition with presence of paper, copper and steel between single and double
ratios.
Figure 6.2 TAN of natural ester aged at 170 C under open condition with presence of
paper, copper and steel. Data is 1 measurement.
It is observed that for natural ester, the initial increment of TAN is slow for both single
and double ratio samples. After 10 days of ageing, the TAN increases rapidly and
saturates after 30 days of ageing for single ratio samples and 23 days of ageing for
double ratio samples. The high amount of TAN for natural ester might be caused by
severe oxidation and pyrolysis. The oxidation is contributed by the free oxygen of
open samples while pyrolysis from the high temperature of the ageing experiment.
159
b) Viscosity
The viscosity of natural ester aged under open condition with the presence of paper,
copper and steel can be seen in Figure 6.3. It is observed that the increment of
viscosity for mineral oil is not as rapid as natural ester. The viscosity increases from 10
cP to 23 cP after 60 days of ageing.
It was found that the viscosity of natural ester starts to increase rapidly after 10 days of
ageing and reaches the plateau after 30 days of ageing. The high viscosity of natural
ester indicates that both oxidation and pyrolysis could significantly change the
molecular structure of natural ester. According to the previous literature [106], the
mechanism of viscosity change can be described by the molecular weight change.
Under the presence of air, the triglycerides molecules of natural ester are exposed to
thermal polymerization. Through this process, the molecules will react with oxygen
and lead to chain scission or cross linking. Chain scission and cross linking will result
in the production of large molecules which can change the molecular weight and
increase the viscosity.
160
Figure 6.3 Viscosity of natural ester aged under open condition with presence of
paper, copper and steel. Data are mean of 5 measurements. Mineral oil was aged under
open condition with paper, copper, zinc, iron and aluminium [107].
It is shown that there is no clear effect of doubling the surface area of copper and steel
on the viscosity of natural ester. The only difference of viscosity between single and
double ratio samples is observed at 10 days of ageing where the viscosity of double
ratio samples is twice higher than that of single ratio samples. The highest viscosity for
single and double ratio samples is 417.6 cP and 518.6 cP.
It is generally known that even severely aged mineral oil in an in-service transformer
can seldom have a significant increase of viscosity, therefore viscosity is not normally
used as the indicator of ageing for mineral oil. To show the comparison, some
viscosity results of mineral oil are also plotted in Figure 6.3. The viscosity data of
mineral oil was obtained from previous literature [107]. The materials ratio of the
ageing experiment is 30 g of paper/pressboard, 300 ml of mineral oil, 0.15 g of
aluminium, 0.15 g of zinc, 0.75 g of iron and 0.75 g of copper. The sample was aged at
130 C under open condition for 60 days [107].
161
Figure 6.4 shows that there are 2 stages which can be observed from the relationship
between viscosity and TAN of natural ester aged under open condition with the
presence of paper, copper and steel. In the first stage, the viscosity increases linearly as
TAN increases to 8 mg KOH/g which indicates that the molecular weight change
occurs at the same time as the generation of acids. The molecular weight change of
natural ester occurs as a result of chain scission and cross linking of triglycerides
molecules [106]. The second stage starts once the chain scission and cross linking of
triglycerides molecules reduces and TAN continues to increase. It is shown that the
viscosity remains unchanged as TAN increases beyond 8 mg KOH/g.
Figure 6.4 Viscosity versus TAN of natural ester aged under open condition with
presence of paper, copper and steel. Viscosity data are mean of 5 measurements. TAN
data is 1 measurement.
6.2.2.4.
162
It is believed that the ageing of natural ester under severe oxidation and pyrolysis can
be described by 3 stages. The first stage is the consumption of oxidation inhibitor. At
this period, the TAN and viscosity increases slowly with time because the oxidation
inhibitor is still available to protect the natural ester from oxidation as shown in Figure
6.1, Figure 6.2 and Figure 6.3.
The second stage starts once the oxidation inhibitor is depleted. It is shown by the
rapid increment of TAN and viscosity, as shown in Figure 6.2 and Figure 6.3 after 10
days of ageing for natural ester. Since the ageing temperature is high in these case
studies, it is believed that pyrolysis might exist together with oxidation and helped to
accelerate the degradation of natural ester.
The third stage initiates once the TAN saturates at a high value and it is observed that
at this stage the natural ester changes to a dark coloured and gelling substance which
first appears on the top part of the bottle as shown in Figure 6.5 a). The bulk natural
ester is not gelled yet because it has a small surface area to access oxygen i.e. large
volume/surface ratio. Exposure of natural ester with a large surface area to access
oxygen i.e. small volume/surface ratio, will cause it to oxidize more rapidly.
Moreover, it was also found that the LMA and HMA content of natural ester at this
stage could not be determined since the mixture of water and oil could not be separated
and a yellowish solution was created as shown in Figure 6.5 b). It is believed that the
chemical structure of the sample has changed significantly after subjected to oxidation
and pyrolysis.
163
Gelling
substance
a)
b)
Figure 6.5 Natural ester aged under open condition with presence of paper, copper,
and steel for single ratio samples. a) Natural ester condition after 50 days of ageing and
b) Mixture of aged natural ester and water.
6.3.
6.3.1.1.
Experiment description
The description of the ageing experiment can be seen in section 6.2.1.1. TAN, LMA
and HMA were measured for each sample.
6.3.1.2.
Test results
Figure 6.6 shows the TAN of natural ester and mineral oil aged without any presence
of paper, copper and steel. It was found that the TAN generated by natural ester aged
under sealed condition is much lower than open condition as seen in Figure 6.1. The
highest TAN for natural ester aged under sealed condition is 0.21 mg KOH/g while
under open condition is 0.72 mg KOH/g. There is no rapid increment of TAN and it is
164
Figure 6.6 TAN of natural ester and mineral oil aged under sealed condition without
any presence of paper, copper and steel. Data is 1 measurement.
6.3.2. Ageing of natural ester with presence of paper, copper and steel
6.3.2.1.
Experiment description
6.3.2.2.
Test results
a) TAN
The comparison of TAN between natural ester and mineral oil aged under sealed
condition with the presence of paper, copper and steel can be seen in Figure 6.7. For
double ratio samples, the TAN of mineral oil is available for comparison with natural
ester.
It was found that the TAN of natural ester for both ratios is much lower than in open
ageing condition as seen in Figure 6.2 and fluctuate at a value between 1 mg KOH/g to
165
Figure 6.7 TAN of natural ester and mineral oil aged under sealed condition with
presence of paper, copper and steel. Data is 1 measurement. For S. Tenbohlen, mineral
oil was aged under sealed condition with paper, copper, zinc, iron and aluminium
[107]. For R. Liao, mineral oil was aged under sealed condition with paper and copper
[2].
It is observed that the effect of doubling the surface area of steel and copper on the
TAN of natural ester is slightly different in sealed condition from open condition. The
TAN of double ratio samples are lower than those of single ratio samples. The highest
TAN of single ratio samples is 2.06 mg KOH/g while that of double ratio samples is
1.37 mg KOH/g.
Additional TAN of mineral oil was also obtained from literatures [2, 107] and plotted
together in Figure 6.7. The ageing condition for S. Tenbohlens experiment is
described in section 6.2.2.2 b). The ageing condition for R. Liaos experiment is
described in [2]. The materials ratio of this ageing experiment is 5 g of paper, 50 g of
166
b) Viscosity
The comparison of viscosity between natural ester and mineral oil aged under sealed
condition with paper, copper and steel can be seen in Figure 6.8. For double ratio
samples, the viscosity of mineral oil is available for comparison with natural ester.
The viscosity of natural ester for both ratios remains unchanged at a value around 32
cP until the end of ageing period. Based on double ratio samples, it was found that the
pattern of viscosity for mineral oil is similar to natural ester, it remains unchanged at a
value around 8 cP. Since little oxygen and water exist in the sample initially, the
polymerization of tryiglycerides molecules is slow which results in the little change of
molecular weight. Thus, since the change in the molecular weight is less, the viscosity
of natural ester remains unchanged during the ageing period.
Figure 6.8 Viscosity of natural ester and mineral oil aged under sealed condition with
presence of paper, copper and steel. Data are mean of 5 measurements. For S.
Tenbohlen, mineral oil was aged under sealed condition with paper, copper, zinc, iron
and aluminium [107].
167
On the contrary, there is a slight change in the viscosity of mineral oil observed by S.
Tenbohlen. The ageing condition for S. Tenbohlen is described in section 6.2.2.2 b)
[107]. Since the viscosity measurements were only done at the beginning and the end
of S. Tenbohlen experiment, his results are inconclusive.
6.3.2.3.
The relationship between viscosity and TAN of natural ester aged under sealed
condition with presence of paper, copper and steel can be seen in Figure 6.9. It is
observed that the pattern is different from open samples as seen in Figure 6.4. The
viscosity increases only slightly and remains unchanged when acidity is increased
from 1.03 mg KOH/g to 2.06 mg KOH/g. The result indicates that by minimising the
contact of natural ester with oxygen, the chain scission and cross linking of natural
ester triglycerides molecules could be reduced significantly.
In sealed condition, it is envisaged that reaching stage 2 as shown in Figure 6.4 will
take considerably long time hence it is absolutely right for the natural ester
manufacturers to recommend natural ester to be used in sealed transformers.
168
Figure 6.9 Viscosity versus TAN of natural ester aged under sealed condition with
presence of paper, copper and steel. Viscosity data are mean of 5 measurements. TAN
data is 1 measurement.
6.3.3.1.
a) Experiment description
b) Test results
Figure 6.10 shows the comparison of LMA and HMA for natural ester and mineral oil
aged under sealed ageing condition without any presence of paper, copper and steel. It
was found that there is a small increment of LMA in natural ester. The highest LMA
and HMA in natural ester are 0.048 mg KOH/g and 0.19 mg KOH/g. On the other
hand, the increment trend of LMA in mineral oil is not significant as in natural ester.
The highest LMA and HMA in mineral oil are 0.0048 mg KOH/g and 0.032 mg
KOH/g.
169
Figure 6.10 LMA and HMA of natural ester and mineral oil aged under sealed
condition without any presence of paper, copper and steel. LMA and HMA of natural
ester data is 1 measurement.
The LMA/HMA percentages versus ageing duration for both natural ester and mineral
oil without any presence of paper, copper and steel are given in Table 6.3. The
calculation of LMA/HMA percentage was carried out according to Equation (5.5) of
Chapter 5. For natural ester, it is observed the LMA percentage ranges from 7.5 % and
26.8 % with an average percentage of 16.9 %. For mineral oil, the LMA percentage is
slightly lower than natural ester and ranges from 13.2 % to 17.9 % with an average
percentage of 15.8 %.
Table 6.3 LMA and HMA percentages in natural ester and mineral oil aged under
sealed condition without any presence of paper, copper and steel.
Ageing time
(days)
Natural ester
Mineral oil
(%)
(%)
LMA
HMA
LMA
HMA
17.1
82.9
17.5
82.5
7.5
92.5
17.9
82.1
26.8
73.2
14.3
85.7
20.2
79.8
14
12.8
87.2
13.2
86.8
170
6.3.3.2.
LMA and HMA ratio of natural ester aged with presence of paper,
a) Experiment description
b) Test results
The comparison of LMA and HMA in natural ester and mineral oil aged under sealed
condition with the presence of paper, copper and steel can be seen in Figure 6.11.
There is no clear increment pattern of LMA in natural ester observed for single ratio
samples. On the contrary, it is observed that there is an increment trend of LMA in
natural ester for double ratio samples.
It was found the amount of LMA for ageing of natural ester with presence of paper,
copper and steel is higher than ageing of natural ester only shown in Figure 6.10. One
of the possible reasons is due to the polar nature of natural ester. As paper aged, more
LMA will be produced in the system. Since natural ester is much more polar, therefore
the partition between paper and natural ester favours the stay of LMA in natural ester.
It is observed that the LMA in natural ester for single and double ratio samples is
higher than mineral oil. From section 6.3.2.2 a), it was shown that the TAN is much
higher in natural ester and so is the amount of LMA in natural ester is higher. It is
noticed that there is no clear increment pattern of LMA in mineral oil. The highest
LMA and HMA in mineral oil are 0.038 mg KOH/g and 0.015 mg KOH/g.
It was found that there is an effect of doubling the surface area of steel and copper on
the LMA of natural ester. The LMA of double ratio samples is slightly higher than that
171
Figure 6.11 LMA and HMA of natural ester and mineral oil aged under sealed
condition with presence of paper, copper and steel. LMA and HMA of natural ester
data is 1 measurement.
The LMA and HMA percentages of natural ester and mineral oil aged under sealed
condition with the presence of paper, copper and steel can be seen in Table 6.4. It was
found that mineral oil has a higher LMA percentage than natural ester. Most of acids in
mineral oil are LMA which percentages are ranging from 51.1 % to 71.3 %. On the
other hand, natural ester has a much lower LMA percentage than mineral oil ranging
from 1.5 % to 8.6 % for single ratio and from 4.6 % to 11.6 % for double ratio
samples. Most of acids in natural ester are HMA ranging from 91.4 % to 98.5 % for
single ratio samples and 88.3 % to 95.4 % for double ratio samples.
It was found that the average percentage of LMA for ageing of natural ester with the
presence of paper, copper and steel is lower than ageing of natural ester only as seen in
Table 6.3.
172
Natural ester -
Mineral oil -
Ageing time
single ratio
double ratio
double ratio
(days)
(%)
(%)
(%)
LMA
HMA
LMA
HMA
LMA
HMA
10
1.5
98.5
4.6
95.4
68.4
31.6
20
4.3
95.7
7.5
92.5
70.4
29.6
23
71.3
28.7
27
11.7
88.3
51.1
48.9
30
8.6
91.4
40
3.4
96.6
50
2.7
97.3
A comparison was made for LMA in paper between natural ester and mineral ester as
seen in Figure 6.12. It is observed that LMA in paper for mineral oil is higher than
natural ester. The increment of LMA in paper for mineral oil is almost linear where the
concentration starts to slightly declined after 23 days of ageing. On the other hand,
LMA in paper for natural ester for both ratios remains almost unchanged after 10 days
of ageing at much lower value than mineral oil.
It was found that there is a significant effect of doubling the surface area of steel and
copper on the LMA in paper of natural ester. The LMA in paper of double ratio
samples saturates at much higher value than single ratio samples. The highest LMA in
paper of double ratio and single ratio samples are 4.1 mg KOH/g and1.7 mg KOH/g.
173
Figure 6.12 LMA in paper of natural ester and mineral oil aged under sealed condition
with presence of paper, copper and steel. Data are mean of 2 measurements.
6.4.
ageing of paper
The first part of this section examined the effect of LMA and HMA on the breakdown
voltage of natural ester. The effect of LMA and HMA on the paper aged under dry and
wet condition was studied in the second part of this section.
6.4.1.1.
Experiment description
Different amounts of either LMA ranging from 0.2 g to 1.6 g or HMA ranging from 1
g to 19 g were mixed with 500 ml dry and filtered natural ester by magnetic stirrer for
3 hours at 50 C. The properties of these acids can be seen in Table 5.1 of Chapter 5.
The natural ester was dried and filtered according to the procedure described in section
3.5.1 of Chapter 3. After the mixing procedure, the natural ester sample was titrated for
acidity and tested for breakdown voltage. The breakdown voltage test was carried out
using mushroom VDE electrodes with a diameter of 36 mm according to ASTM
standard, ASTM D1816 [103]. A 50 Hz AC voltage with increasing rate of 0.5 kV/s
174
6.4.1.2.
Test results
It was found that the breakdown voltage remains almost unchanged even after the
LMA reaches to 4 mg KOH/g as shown in Figure 6.13. The reason behind this
phenomenon might be due to the polar nature of natural ester. It is known that the
breakdown voltage of oils could be influenced by polar compounds such as water.
Once the saturation limit of water in oils is exceeded, the breakdown voltage will be
significantly reduced. Since LMA is a polar compound, it is suspected that its
saturation limit in natural ester is high, which might explain the reason why the
breakdown voltage is not affected.
Figure 6.13 Influence of LMA on the breakdown voltage of natural ester. Data are
mean of 50 tests and error bars represent the standard deviation. LMA in natural ester
data is 1 measurement.
It is observed that there is no effect of HMA on the breakdown of natural ester. The
breakdown voltage remains unchanged even after the HMA increases up to 9 mg
KOH/g as seen in Figure 6.14.
175
Figure 6.14 Influence of HMA on the breakdown voltage of natural ester. Data are
mean of 50 tests and error bars represent the standard deviation. HMA in natural ester
data is 1 measurement.
6.4.2.1.
Experiment description
In order to prepare natural ester sample with a different type of acids, 0.2 g of either
LMA or HMA was mixed with 200 ml dried and filtered natural ester by magnetic
stirrer for at least 3 hours at 50 C. The natural ester was dried and filtered according
to the procedure described in section 3.5.1 of Chapter 3. After the mixing procedure,
the oil sample was titrated for acidity. For samples with LMA, the acidity ranges from
1.17 mg KOH/g to 1.26 mg KOH/g while with HMA the acidity around 0.26 mg
KOH/g. Each sample contains 1.75 g of dry or wet conductor paper and 200 ml of
mineral oil containing either LMA or HMA. The paper was pre-processed according to
the procedure described in section 3.5.2 of Chapter 3. The properties of paper and
mineral oil can be seen in Table 3.1 and Table 3.8 of Chapter 3. All samples were
sealed by PTFE screw cap with a gasket ring in order to reduce the migration of
oxygen from outside and aged in an air circulating oven at 120 C for up to 64 days. A
total of 8 measurements of TI were carried out on each sample since the amount of
paper used in the ageing experiment is rather limited.
176
It was found that LMA has a significant effect on the paper aged under dry condition
in natural ester as seen Figure 6.15. There is no effect of HMA on the paper ageing and
it is observed that the effect is almost similar to with no acid. It is observed that the TI
of samples with LMA reach 50 % of its strength after 37 days of ageing while with
HMA and no acid after 64 days of ageing.
Figure 6.15 Influence of LMA and HMA on the paper aged under dry condition. Data
are mean of 8 samples and error bars represent the standard deviation.
The fitting of relationship between TI and ageing duration was carried out by Equation
(4.1). It was found that the rate of ageing for samples with HMA is similar to with no
acid as seen in Table 6.5. The rate of ageing of samples with LMA is 2 times higher
than with HMA and no acid.
With no acid
1.010-2
0.928
With LMA
1.910
-2
0.909
With HMA
0.910-2
0.984
177
Figure 6.16 shows the effect of LMA and HMA on the paper aged under wet condition
in natural ester. For all ageing conditions, the TI of 14 days samples are lower than
those of 28 days and 63 days samples and this could be caused by the systematic flaw
of sample preparation due to time limit, no confirmation test have been done so far.
Figure 6.16 Influence of LMA and HMA on the paper aged under wet condition. Data
are mean of 8 samples and error bars represent the standard deviation.
The fitting parameters for paper aged under wet condition in natural ester can be seen
in Table 6.6. It is observed that the fitting of samples with LMA and HMA is not good
due to the high scattering of the data. The r for both samples are quite low compared
to with no acid.
With no acid
1.110-2
0.809
With LMA
2.410-2
0.388
With HMA
1.210-2
0.295
Table 6.7 shows the increments factors for natural ester and mineral oil under dry
ageing condition. The increment factor is calculated by Equation (5.1) of Chapter 5.
The increment factor of samples with LMA for natural ester aged under wet condition
178
Table 6.7 Comparison of increment factors between natural ester and mineral oil under
dry ageing conditions.
6.5.
Condition
Natural ester
Mineral oil
With LMA
1.9
2.1
With HMA
0.9
0.9
Summary
Oxidation can significantly change the chemical and physical properties of natural
ester. The mechanism of the natural ester degradation under severe oxidation could be
divided into 3 stages. The first stage is the consumption of oxidation inhibitor. The
second stage is the fast oxidation phase of natural ester where the oxidation inhibitor
has already been depleted. The third stage is the saturation stage where natural ester
starts to experience severe physical and chemical changes. Thus, the cooling
performance of natural ester might be affected if it is used in free breathing
transformers.
On the other hand, it was found that natural ester is suitable for hermetically sealed
transformers. The TAN and viscosity remain stable even under high temperature of
ageing for a considerably long period. It is also noticed that more LMA stays in natural
ester which could be the direct reason why there is less severe degradation to the paper
in natural ester than in mineral oil, apart from water scavenging and hydrolytic
protections.
There is no effect of LMA and HMA on the breakdown voltage of natural ester within
the case study range. It is suspected that the saturation limit of LMA in natural ester is
high which prevents any reduction on the breakdown voltage.
179
7.1.
Conclusions
7.1.1. General
This thesis described an extensive study on ageing assessment of the scrapped
transformers through Tensile Index (TI) and Low Molecular Weight Acid (LMA)
measurements as well as high temperature ageing assessment of paper/natural ester
insulation system. Through experimental examinations and data analyses, the
objectives of this thesis are accomplished and thus useful findings and conclusions are
made.
180
It was found that LMA has a significant reduction effect on the breakdown voltage of
mineral oil. Investigation on the effect of LMA on paper ageing reveals that the ageing
rate is greatly enhanced if LMA is dissociated by water. The LMA in paper profile of
scrapped transformers was examined according to phases, windings and layer
locations. Similar as TI profile, the LMA in paper profile among different windings
and phases shows no major difference. The examination shows that the chemical
degradation mechanism does exist in multilayer insulation and it mainly affects the
outer layer paper. Based on the population analysis, it was found that LMA in paper
increases exponentially as the transformer age increases. The end of life criterion based
on LMA in paper could be determined if the analogy of the relationship between LMA
in paper and TI of paper is described the same as Chendongs equation of furanic
compounds and degree of polymerization. The study on the LMA partitioning in paper
and oil shows that it is strongly influenced by the weight ratio of paper and oil. The
181
The effect of high temperature oxidation on natural ester was studied and compared
with mineral oil. It was found that the oxidation of natural ester under high temperature
can be classified into three stages. The first stage is the fast consumption of oxidation
inhibitor. The second stage initiates once the oxidation inhibitor is depleted and the
third stage is the gelling and solidifying stage. It was found that natural ester is suitable
for application in sealed transformers since the ageing performance is good and
comparable to mineral oil. It is observed that most LMA tend to stay in natural ester
and this could be the reason for the good ageing property of paper in natural ester apart
from water scavenging and hydrolytic protection. Furthermore, it is shown that the
effect of LMA on the paper aged in natural ester is not as severe as in mineral oil. It is
suspected that the LMA migrates into natural ester more since it is more polar and thus
prevents severe degradation to the paper. It is also suspected that the saturation of
LMA in natural ester is high which prevents the natural ester from suffering the
reduction of AC breakdown voltage.
7.2.
Future work
The LMA in paper profiles along phases, windings and layers can be further examined
by studying the migration of LMA through oil circulation. Taking layer profile as an
example, the layer profile can be further studied by determining the extent of LMA
migration under different layers. The chemical and thermal degradation mechanisms of
layer profile could also be verified by laboratory experiments. Apart from migration of
LMA, the TI prediction model can be improved by obtaining LMA in oil before the
transformer is scrapped.
182
183
Table I.1 TI of new conductor type paper without oil impregnation at different
lengths.
Test No.
15 mm (Nm/g)
30 mm (Nm/g)
60 mm (Nm/g)
1
2
3
4
5
6
7
8
9
10
Average (Nm/g)
Standard
deviation (Nm/g)
117.3
108.7
108.8
115.5
108.1
112.8
117.7
110.3
116.3
114.3
113.0
117.1
113.2
119.7
106.9
108.8
115.6
111.7
122.8
119.4
112.0
114.7
105.6
104.3
102.2
105.4
111.3
107.2
113.1
111.1
104.4
117.6
108.2
180 mm
(Nm/g)
112.8
104.7
109.5
105.9
104.7
114.0
114.6
112.7
112.6
102.3
109.4
3.8
5.1
4.9
4.6
184
101.6
111.2
104.6
109.1
109.2
108.7
104.2
96.6
104.0
4.0
2.6
4.4
Parameter
1
2
3
L1
64.0
67.6
59.9
185
L1
56.8
54.7
42.7
SW
L2
81.4
63.4
61.4
L3
81.3
71.8
63.9
Parameter
1
2
3
4
5
6
7
8
Average (Nm/g)
Standard
deviation
(Nm/g)
L1= Top outside.
L2= Middle outside.
L3= Bottom outside.
L4= Top inside
L1
66.5
58.1
75.2
59.6
69.8
69.7
67.3
66.6
79.8
82.1
80.7
86.3
76.3
73.5
84.4
81.3
77.1
69.7
73.7
70.8
67.6
72.4
73.4
72.1
29.6
61.9
59.9
33.1
60.0
49.8
63.2
74.3
74.0
87.7
65.1
77.0
72.0
51.1
43.3
61.5
47.6
60.1
4.1
3.8
12.9
9.2
13.7
6.0
5.3
5.4
4.0
7.8
HV
L1
60.0
65.9
53.5
78.0
43.7
62.3
59.3
56.7
HV
L2
81.6
85.7
94.5
64.1
87.4
89.4
78.8
92.7
L1
28.1
37.0
34.5
42.1
24.4
36.3
47.6
45.3
L3
56.3
94.1
60.7
59.5
61.0
74.1
53.3
41.5
186
3.6
C phase
LV
L1
24.0
65.9
47.8
73.0
52.8
18.1
68.7
54.9
L3
68.4
72.7
63.3
67.6
63.3
69.0
67.4
TV
L4
25.1
37.3
59.4
32.1
47.4
44.1
41.7
32.5
L1
32.6
46.6
51.3
25.3
61.6
29.5
46.2
29.4
HV top
L1
28.1
20.5
46.8
28.4
29.7
9.5
30.3
25.5
L4
39.5
28.7
54.3
77.0
60.9
66.7
67.3
56.7
Parameter
1
2
3
4
5
6
7
8
9
10
Average (Nm/g)
Standard
deviation
(Nm/g)
L1
21.9
25.0
42.6
40.6
57.6
38.6
35.1
34.0
27.3
47.9
37.1
38.9
43.6
85.9
57.4
60.0
21.1
47.6
29.5
23.4
21.8
28.5
16.5
60.8
60.8
37.8
64.4
48.6
39.7
36.8
26.4
57.3
7.4
15.8
20.5
10.4
13.6
9.8
14.0
10.4
14.1
6.0
6.7
L1
52.5
30.8
58.4
39.6
50.0
39.2
45.1
A phase
SW
L2
L3
70.8
55.4
51.2
38.6
46.2
55.2
76.8
52.3
65.1
40.9
53.7
45.2
57.4
47.7
52.0
65.4
60.6
51.0
L1
46.1
41.2
48.1
52.2
40.7
45.2
43.4
56.1
58.1
56.1
48.7
L1
34.7
35.9
33.4
30.8
41.7
35.3
12.1
6.5
4.0
1
2
3
4
5
6
7
8
9
10
Average (Nm/g)
Standard
10.3
deviation
(Nm/g)
L1= Top outside.
L2= Middle outside.
L3= Bottom outside.
8.1
187
7.0
L3
24.1
19.3
55.5
59.8
50.6
48.7
30.2
40.3
25.7
40.6
39.5
Tap
L3
34.1
41.3
35.3
26.5
30.1
42.3
29.1
34.1
10.9
Parameter
3.3
SW
L2
49.9
51.2
30.3
42.1
32.0
35.2
39.0
19.2
49.6
30.9
37.9
C phase
SW
L2
48.6
54.0
69.0
42.8
55.8
43.4
51.7
34.1
43.1
50.4
49.3
9.5
L3
42.1
38.6
49.1
34.7
36.4
54.0
42.5
7.6
L3
57.3
43.0
49.9
44.1
56.8
50.4
73.4
53.5
L9
51.9
48.1
57.9
61.0
56.3
59.6
44.9
57.8
63.6
58.9
56.0
L10
50.9
56.3
29.5
45.7
48.8
60.4
37.7
67.2
46.0
49.2
L11
31.7
50.0
44.7
51.5
54.8
57.3
39.0
35.6
45.6
L12
40.0
26.0
46.5
35.7
44.5
29.4
49.7
46.8
39.4
48.8
40.7
10.3
11.2
5.9
11.4
9.4
8.2
Parameter
1
2
3
4
5
6
7
8
9
10
Average (Nm/g)
Standard
deviation
(Nm/g)
Parameter
B phase
Not stated
L1
L2
44.9 55.4
57.0 44.2
52.4 29.2
57.7 53.3
50.5 44.9
52.1 37.0
52.6 45.4
59.6 24.0
51.7 50.6
51.6
53.2 43.6
9.8
5.4
C phase
HV
L1
L2
9.9 67.9
33.3 66.3
31.0 57.3
38.9 51.5
36.8 56.2
46.5 57.3
24.9 62.8
37.9 60.3
35.4 57.2
46.7
32.7 58.4
1
2
3
4
5
6
7
8
9
10
Average (Nm/g)
Standard
4.4 10.5 10.4
deviation
(Nm/g)
L1= Top outside.
L2= Bottom outside.
L3= 3rd discs from top outside.
L4= 3rd discs from bottom outside.
L5= 4th discs from top outside.
L6= 4th discs from bottom outside.
L7= 5th discs from top outside.
L8= 5th discs from bottom outside.
L9= 6th discs from top outside.
L10= 6th discs from bottom outside.
L11= 7th discs from top outside.
L12= 7th discs from bottom outside.
6.4
188
7.5
4.9
Parameter
L1
32.3
41.4
38.5
29.4
23.8
15.8
27.4
32.3
30.8
21.4
29.3
1
2
3
4
5
6
7
8
9
10
Average (Nm/g)
Standard
7.7
deviation
(Nm/g)
L1= Top outside.
L2= Middle outside
L3= Bottom outside.
6.1
10.9
14.1
13.1
L3
22.1
61.2
46.0
71.7
47.7
39.0
52.2
45.7
61.0
43.8
49.0
L1
15.6
18.9
30.5
15.6
32.5
23.9
24.3
52.7
13.6
25.3
C phase
CW
L2
32.0
33.0
49.7
38.6
27.6
39.6
42.5
57.0
33.2
30.6
38.4
13.7
12.2
9.2
Parameter
1
2
3
4
5
6
7
8
9
10
Average (Nm/g)
Standard
deviation
(Nm/g)
L1= Top outside.
L2= Bottom outside.
7.5
Parameter
1
2
L1
19.7
26.8
9.2
189
15.3
SW
L2
28.7
40.4
41.0
45.3
45.1
46.7
60.8
68.6
64.0
48.9
10.2
L3
37.3
49.7
27.4
27.6
52.8
69.4
56.4
69.7
55.5
38.9
48.5
12.9
L1
31.0
24.5
SW
L2
40.9
39.7
L3
35.4
52.1
Parameter
HV
L1
9.7
12.1
21.5
36.1
28.3
32.0
40.3
28.0
31.3
29.5
1
2
3
4
5
6
7
8
9
10
Average
26.9
(Nm/g)
Standard
9.8
deviation
(Nm/g)
L1= Top outside.
L2= Middle outside.
L3= Bottom outside.
36.3
38.7
45.5
36.3
15.6
25.6
20.7
17.7
20.8
22.1
21.8
21.6
11.0
20.1
20.3
32.4
45.0
39.8
24.5
23.9
33.1
33.5
28.4
36.8
38.3
31.7
37.5
51.4
36.9
43.9
9.3
11.8
2.3
7.8
6.8
4.6
3.6
9.0
LV
L1
20.2
23.8
20.2
25.0
-
C phase
HV
L1
L2
37.1
45.7
41.1
58.5
26.4
65.9
37.7
51.7
35.7
52.0
41.0
47.3
45.3
50.6
40.5
-
39.6
46.0
44.1
52.3
22.3
38.1
53.1
7.1
10.9
4.3
7.2
2.5
5.6
7.0
Table I.15 TI of different layers for Transformer 10. Paper samples were taken from B phase,
HV bottom.
Number of layer
Parameter
1
2
3
4
5
6
7
8
1
11.3
52.3
65.9
50.0
76.5
75.1
76.1
61.9
2
29.0
44.6
61.5
71.6
74.1
80.3
39.3
78.9
3
16.9
56.7
70.4
63.4
71.8
54.1
76.3
68.6
4
16.4
56.7
57.4
40.9
51.0
77.6
71.9
81.9
5
54.6
42.4
70.8
52.9
84.9
55.9
39.9
53.7
6
10.4
59.8
60.6
40.4
55.4
70.1
73.1
49.5
7
36.8
32.9
72.0
71.1
45.9
83.4
59.1
83.4
8
5.0
83.6
62.7
80.6
67.3
75.2
37.6
56.9
190
28.3
15.8
22.4
59.6
58.0
54.6
45.1
62.9
72.9
69.7
61.3
77.2
86.6
69.1
33.3
85.9
69.1
70.3
24.4
56.8
64.8
38.9
63.9
14.9
13.5
8.4
14.3
14.0
16.5
19.5
14.7
Table I.16 TI of different layers for Transformer 6. Paper samples were taken from B phase,
CW top.
Number of layer
Parameter
1
2
3
4
5
6
7
8
16.7
60.2
25.9
46.7
42.4
40.7
45.4
50.7
1
40.2
42.7
54.9
50.0
35.0
39.1
51.8
35.5
2
31.8
36.2
40.5
54.9
61.3
45.9
53.2
56.8
3
15.8
40.1
38.3
33.6
51.4
36.5
46.1
53.1
4
24.4
62.5
42.4
35.3
48.8
40.9
38.7
41.4
5
27.9
52.1
48.7
57.8
49.8
47.9
54.6
26.1
6
33.1
50.1
52.0
52.4
43.4
59.5
52.0
23.6
7
62.9
62.0
43.0
32.1
35.0
53.8
42.7
46.5
8
46.5
32.6
46.6
18.4
45.3
38.0
24.5
33.9
9
32.8
21.3
45.5
49.2
54.2
38.9
10
33.2
46.0
43.8
42.4
46.2
44.7
46.3
40.6
Average (Nm/g)
Standard
14.1
13.8
8.1
13.1
7.9
7.8
9.4
11.2
deviation
(Nm/g)
Parameter
9
10
11
47.8
46.2
14.0
1
45.5
47.9
24.0
2
20.8
55.2
31.0
3
32.0
38.9
27.3
4
45.2
52.4
41.6
5
39.5
40.3
39.7
6
17.9
50.8
35.1
7
56.7
69.5
50.6
8
24.4
48.8
46.2
9
47.1
56.9
10
37.7
50.7
34.4
Average (Nm/g)
Standard
13.2
8.8
11.6
deviation
(Nm/g)
191
Appendix II: Low Molecular Weight Acid (LMA) in Service Aged Paper
APPENDIX II:
Parameter
L1
2.08
2.24
L2
1.88
3.19
1
2
Average (mg
2.16
2.53
KOH/g)
L1= Top outside
L2= 72nd discs from top outside.
L3= 144th discs from top outside.
L4= Bottom outside.
Parameter
1
2
Average (mg
KOH/g)
L1= Top outside.
L2= Middle outside.
L3= Bottom outside.
L4= Top inside
L3
0.38
0.11
L4
1.77
2.32
L1
3.23
3.19
SW
L2
1.67
1.55
0.24
2.05
3.21
1.61
L3
0.40
1.24
1.79
0.82
1.91
HV
L1
2.81
2.79
1.97
0.97
HV
L2
2.11
1.51
1
2
Average (mg
2.80
1.81
KOH/g)
L1= Top outside.
L2= Top outside (2nd layer paper)
L3= Bottom outside.
L4= Top outside (near to conductor).
Parameter
1
2
Average (mg
KOH/g)
0.50
L3
0.43
0.57
LV
C phase
TV
HV top
L1
4.02
3.58
L3
2.59
3.13
L1
3.79
-
L4
2.84
-
L1
2.38
-
L1
2.44
2.06
L4
1.92
1.30
3.80
2.86
3.79
2.84
2.38
2.25
1.61
L3
1.92
3.63
Tap
L3
2.24
2.64
3.30
2.77
2.44
2.44
3.08
192
2.82
2.87
Appendix II: Low Molecular Weight Acid (LMA) in Service Aged Paper
Parameter
L1
3.19
3.48
1
2
Average (mg
3.33
KOH/g)
L1= Top outside.
L2= Middle outside
L3= Bottom outside tap.
Parameter
1
2
Average (mg
KOH/g)
Parameter
A phase
SW
L2
L3
1.92
2.73
1.82
2.88
L1
1.99
-
L1
3.85
4.19
1.87
1.99
4.02
2.81
2.70
L3
1.80
3.81
2.80
L10
1.80
2.14
L11
3.04
1.88
L12
2.29
2.33
1.52
1.97
2.46
2.31
0.39
B phase
Not stated
L1
L2
2.55 1.53
3.14 1.18
2.42
1.34
2.19
1.01
1.18
C phase
HV
L1
L2
2.84 1.42
4.48 0.10
1
2
Average (mg
2.85 1.35 3.66
KOH/g)
L1= Top outside.
L2= Bottom outside.
L3= 3rd discs from top outside.
L4= 3rd discs from bottom outside.
L5= 4th discs from top outside.
L6= 4th discs from bottom outside.
L7= 5th discs from top outside.
L8= 5th discs from bottom outside.
L9= 6th discs from top outside.
L10= 6th discs from bottom outside.
L11= 7th discs from top outside.
L12= 7th discs from bottom outside.
Parameter
C phase
SW
L2
2.03
3.38
0.76
1
2
Average (mg
3.59
KOH/g)
L1= Top outside.
L2= Middle outside
L3= Bottom outside.
4.52
3.71
5.53
193
4.75
3.42
C phase
CW
L1
L2
4.44 3.76
3.63 3.97
L3
4.10
3.97
4.03
4.04
3.87
Appendix II: Low Molecular Weight Acid (LMA) in Service Aged Paper
Parameter
1
2
Average (mg
KOH/g)
L1= Top outside.
L2= Bottom outside.
Parameter
3.70
6.39
2.99
5.97
5.01
2.95
1
2
Average (mg
2.59
KOH/g)
L1= Top outside.
L2= Middle outside.
L3= Bottom outside.
5.83
3.74
L2
2.67
3.31
3.01
1
2
Average (mg
5.79 5.26
KOH/g)
L1= Top outside.
L2= Middle outside.
L3= Bottom outside.
L4= Top inside
L5= Top 2nd strand from inside
Parameter
1.53
SW
2.65
1.93
5.62
L1
4.23
4.42
SW
L2
3.72
4.14
L3
3.54
3.54
4.33
3.93
3.54
C phase
HV
L1
L2
6.84
1.57
5.86
2.09
6.35
1.83
Table II.10 LMA in paper of different layers for Transformer 10. Paper samples were taken
from B phase, HV bottom.
Number of layer
Parameter
1
2
3
4
5
6
7
8
5.51
7.24
4.67
2.89
3.83
1.88
3.98
1.94
1
2
Average (mg
5.51
7.24
4.67
2.89
3.83
1.88
3.98
1.94
KOH/g)
194
Appendix II: Low Molecular Weight Acid (LMA) in Service Aged Paper
Table II.11 LMA in paper of different layers for Transformer 6. Paper samples were taken
from B phase, CW top.
Number of layer
Parameter
1
2
3
4
5
6
7
8
4.27
2.92
2.13
3.89
2.72
3.34
3.22
2.94
1
6.80
2.86
3.74
3.76
4.34
3.30
2
Average (mg
5.53
2.89
2.94
3.89
3.24
3.84
3.22
3.12
KOH/g)
Parameter
9
10
11
3.26
2.81
1.92
1
4.01
1.80
2
Average (mg
3.26
3.41
1.86
KOH/g)
195
Table III.1 AC breakdown voltages of different types of paper impregnated in new oil. The
gap distance between two sphere electrodes was set at 0.258 mm.
Breakdown voltages of different types of paper and new
oil (kV)
Test No.
Conductor
Service
Layer type
New oil
type
aged paper
14.3
18.6
20.2
16.7
1
14.3
17.2
18.1
17.1
2
16.9
20.4
19.0
14.6
3
15.3
20.4
17.8
14.1
4
15.8
16.7
21.5
14.2
5
16.7
18.6
21.7
14.9
6
15.2
18.6
21.1
16.8
7
13.8
18.9
24.2
15.0
8
14.3
17.9
20.5
17.1
9
14.8
19.3
20.4
11.8
10
12.2
16.7
19.7
15.9
11
13.7
17.1
20.4
12.7
12
13.5
16.8
19.7
14.4
13
16.4
16.9
19.4
13.9
14
17.1
18.0
18.4
14.8
15
17.3
18.4
22.8
15.0
16
15.7
17.5
18.3
13.1
17
17.1
19.3
20.2
14.1
18
15.7
18.9
18.6
14.3
19
13.8
17.5
21.4
15.6
20
16.2
17.1
20.8
16.3
21
14.5
17.6
18.4
12.7
22
16.5
16.7
18.4
14.8
23
14.8
16.7
20.7
12.9
24
16.1
16.5
20.1
13.5
25
16.3
16.2
20.2
11.8
26
16.9
16.1
19.0
13.6
27
15.7
18.4
18.6
12.7
28
13.1
19.0
21.3
12.6
29
14.3
17.0
21.1
13.5
30
13.5
16.2
20.7
11.5
31
15.2
18.6
18.3
12.7
32
17.2
17.0
18.9
12.2
33
14.2
19.5
18.8
11.5
34
15.7
18.3
19.6
13.7
35
15.3
18.1
20.2
12.1
36
14.4
16.0
18.0
14.2
37
12.8
17.1
21.0
12.4
38
14.4
17.2
20.2
12.8
39
196
15.6
15.2
1.3
16.7
17.7
1.2
22.2
20.0
1.4
12.9
13.9
1.6
Table III.2 AC breakdown voltages of different types of paper impregnated in aged oil. The
gap distance between two sphere electrodes was set at 0.258 mm.
Breakdown voltages of different types of paper and aged
oil (kV)
Test No.
Conductor
Service
Layer type
Aged oil
type
aged paper
14.9
17.7
18.0
16.4
1
15.7
16.6
17.7
14.4
2
15.0
15.2
16.6
15.8
3
14.9
15.5
19.8
16.1
4
14.6
16.1
18.7
10.5
5
15.5
16.7
20.1
12.7
6
13.5
18.3
19.0
11.8
7
16.0
17.3
19.4
9.4
8
14.9
16.0
16.0
9.4
9
14.9
17.7
17.7
13.4
10
12.4
15.5
20.2
13.2
11
13.1
16.8
15.5
14.3
12
12.7
16.1
18.9
14.0
13
15.1
15.3
19.0
14.5
14
15.0
17.4
20.7
14.0
15
13.8
15.9
19.7
14.4
16
12.6
16.7
19.8
11.8
17
16.0
17.1
20.4
11.7
18
14.9
16.2
16.6
15.5
19
11.8
16.7
17.8
14.3
20
12.7
17.5
19.8
14.4
21
14.9
16.7
17.9
11.8
22
16.9
16.5
17.5
11.4
23
13.1
15.3
18.9
12.4
24
13.3
17.3
20.2
14.6
25
13.5
15.9
15.5
13.0
26
15.0
15.6
15.2
12.1
27
13.5
15.3
20.2
15.3
28
13.5
14.7
19.6
14.3
29
14.8
15.5
17.5
15.0
30
12.1
15.8
19.2
16.8
31
15.8
15.1
18.4
14.1
32
15.7
15.9
20.3
12.3
33
12.6
15.3
18.4
9.4
34
13.1
14.8
16.9
14.1
35
13.3
17.3
21.6
14.0
36
11.6
15.8
18.5
11.3
37
197
11.4
13.4
13.5
14.0
1.4
15.7
15.0
16.7
16.2
0.9
198
18.8
17.0
20.2
18.6
1.6
14.2
12.9
14.5
13.4
1.9
Appendix IV: Effect of Low Molecular Weight Acid (LMA) and High Molecular Weight
Acid (HMA) on Breakdown Voltage of Oil and Ageing of Paper
APPENDIX IV:
Effect of Low Molecular Weight Acid (LMA) and High
Molecular Weight Acid (HMA) on Breakdown Voltage of Oil and Ageing of
Paper
Table IV.1 AC breakdown voltages of new mineral oil with different amount of LMA. The
gap distance between two VDE electrodes was set at 1 mm.
Breakdown voltage of new mineral oil with different
amount of LMA (kV)
Test No.
(Clean) 0.02
0.05mg
0.4 mg
0.6 mg
mg KOH/g
KOH/g
KOH/g
KOH/g
49.4
25.5
13.1
7.2
1
47.5
21.5
12.2
9.4
2
49.0
21.2
11.6
8.7
3
46.0
27.1
11.8
7.0
4
49.4
25.9
12.8
7.8
5
54.6
25.8
13.1
9.0
6
51.9
17.0
13.1
8.7
7
44.9
26.6
11.3
8.2
8
52.2
25.1
10.0
8.7
9
49.2
22.4
12.5
10.0
10
45.7
26.6
11.8
8.7
11
53.0
18.4
10.7
10.0
12
51.5
29.5
11.6
8.7
13
37.3
26.7
11.9
10.0
14
46.0
22.8
12.8
8.7
15
41.7
18.9
10.8
10.0
16
56.9
21.2
12.0
8.7
17
49.5
21.9
9.4
8.2
18
54.1
26.0
11.9
9.2
19
53.5
28.2
11.1
9.4
20
39.9
27.2
12.5
8.7
21
56.5
19.0
12.7
6.4
22
42.9
19.3
11.7
8.1
23
50.5
17.2
11.0
7.1
24
44.2
27.0
11.8
8.6
25
51.8
20.3
12.6
6.9
26
55.0
18.2
9.3
7.3
27
44.6
18.9
10.8
7.3
28
53.1
14.4
11.5
8.3
29
50.0
25.4
10.7
10.4
30
48.5
19.6
12.7
8.4
31
47.2
26.9
12.1
8.5
32
51.0
26.2
10.9
8.7
33
39.7
19.5
11.4
9.1
34
55.6
24.7
11.9
7.8
35
54.3
21.5
9.9
8.5
36
199
Appendix IV: Effect of Low Molecular Weight Acid (LMA) and High Molecular Weight
Acid (HMA) on Breakdown Voltage of Oil and Ageing of Paper
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Average (kV)
Standard deviation (kV)
47.8
45.7
47.9
51.2
50.3
45.8
53.6
53.9
50.1
56.3
54.9
52.7
45.7
51.1
49.5
4.6
16.8
23.8
22.9
20.8
24.4
23.6
26.4
24.3
20.7
21.1
23.8
18.1
17.5
19.8
22.6
3.6
11.5
2.5
11.0
9.3
9.5
12.7
10.3
11.3
10.4
10.0
11.4
11.3
10.2
11.4
11.2
1.6
7.3
6.8
7.9
7.5
7.5
7.6
6.7
6.5
8.1
6.5
6.9
5.6
6.4
8.0
8.1
1.1
Table IV.2 AC breakdown voltages of new mineral oil with different amount of HMA. The
gap distance between two VDE electrodes was set at 1 mm.
Breakdown voltage of new mineral oil with different
amount of HMA (kV)
Test No.
0.6 mg
1.6 mg
3.1 mg
9.0 mg
KOH/g
KOH/g
KOH/g
KOH/g
45.0
59.5
50.3
19.4
1
45.2
58.0
48.8
15.7
2
45.3
55.3
52.3
23.3
3
47.5
53.2
50.2
21.4
4
38.1
53.8
50.3
23.0
5
42.2
53.6
49.6
22.3
6
45.6
53.8
43.2
7.1
7
46.9
47.0
50.5
9.1
8
39.7
50.4
47.8
18.5
9
38.3
48.2
44.2
19.5
10
39.6
51.0
47.3
17.9
11
41.2
51.5
42.9
18.4
12
44.2
47.9
40.6
14.7
13
37.7
57.4
42.6
22.3
14
42.0
46.4
45.6
29.1
15
41.5
52.6
49.2
27.2
16
43.3
48.0
41.1
20.1
17
39.5
52.5
43.2
17.7
18
41.5
50.0
40.2
18.7
19
39.1
55.4
41.5
27.4
20
46.8
53.2
40.3
9.5
21
44.3
44.5
37.5
13.2
22
200
Appendix IV: Effect of Low Molecular Weight Acid (LMA) and High Molecular Weight
Acid (HMA) on Breakdown Voltage of Oil and Ageing of Paper
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Average (kV)
Standard deviation (kV)
43.8
43.3
37.1
43.8
37.5
44.9
41.4
36.3
40.2
39.0
41.8
40.2
41.5
41.9
41.1
39.8
41.5
40.9
40.0
45.9
39.5
43.7
46.4
32.6
39.2
44.5
47.0
42.4
41.8
3.2
51.5
47.5
49.4
46.3
42.2
47.9
36.6
42.7
42.9
43.3
45.8
46.8
46.8
45.3
47.3
44.1
35.7
43.7
42.4
41.5
47.8
45.5
43.7
40.9
48.6
44.7
41.7
39.3
47.7
5.3
45.1
43.5
44.6
43.8
37.0
48.5
47.8
49.1
45.9
38.1
35.2
36.2
44.0
46.3
43.5
38.5
52.5
47.8
42.8
45.5
47.5
39.1
56.3
49.2
46.4
39.3
51.8
51.1
45.1
4.8
25.1
18.5
12.0
9.3
9.7
5.8
15.1
11.4
11.9
26.2
40.4
6.5
16.0
17.1
7.2
5.7
15.1
13.4
19.1
15.9
17.1
24.1
15.1
15.5
22.6
28.8
32.2
11.9
17.7
7.3
Table IV.3 AC breakdown voltages of new natural ester with different amount of LMA. The
gap distance between two VDE electrodes was set at 1 mm.
Breakdown voltage of new natural ester with different
amount of LMA (kV)
Test No.
(Clean)
0.6 mg
1.2 mg
2.0 mg
4.0 mg
0.04 mg
KOH/g
KOH/g
KOH/g
KOH/g
KOH/g
52.6
47.7
42.1
48.8
47.3
1
51.9
48.4
40.6
46.9
38.3
2
49.9
46.3
46.4
49.5
44.2
3
45.4
46.8
42.7
49.0
43.5
4
32.7
48.7
40.8
57.5
42.0
5
54.4
53.5
46.3
51.3
41.5
6
39.9
47.1
28.2
53.0
45.1
7
48.6
45.9
44.0
50.3
27.1
8
201
Appendix IV: Effect of Low Molecular Weight Acid (LMA) and High Molecular Weight
Acid (HMA) on Breakdown Voltage of Oil and Ageing of Paper
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Average (kV)
Standard deviation (kV)
48.2
46.9
50.8
51.5
42.6
49.1
48.4
50.1
44.6
36.4
56.3
50.3
41.0
50.5
33.7
49.5
48.7
42.5
48.0
44.2
40.2
50.9
44.9
49.1
49.5
43.1
47.6
40.7
42.0
36.5
45.9
46.9
47.4
47.6
47.2
24.2
38.6
35.8
44.5
34.4
49.0
46.4
45.2
6.3
48.1
48.5
49.0
45.5
47.5
48.8
50.9
45.6
49.1
46.1
40.8
50.0
27.9
48.6
51.6
50.3
47.7
45.2
53.1
49.4
46.2
40.0
42.9
51.3
53.1
50.9
38.7
48.1
45.3
45.3
51.3
51.9
48.5
45.8
51.4
49.7
50.1
48.6
49.5
43.3
51.7
48.1
47.6
4.3
202
44.2
45.5
30.8
45.1
39.1
46.6
44.2
41.9
38.5
43.8
44.4
46.3
44.2
42.8
40.2
43.6
43.3
43.2
47.8
39.0
44.8
36.9
35.8
43.3
37.9
33.5
30.0
46.4
44.3
44.8
33.2
30.0
32.0
32.1
35.5
39.8
35.2
38.1
31.8
35.3
35.3
31.4
39.9
5.5
58.7
53.7
53.1
55.2
50.0
54.2
58.2
54.4
50.4
52.1
47.6
50.9
59.0
58.7
47.1
52.1
54.0
47.3
46.6
49.7
58.0
51.0
58.9
58.3
50.4
50.4
53.7
48.7
55.3
60.9
50.7
36.2
43.3
45.6
46.4
56.6
47.8
46.4
46.3
51.3
50.0
49.0
51.5
4.8
37.4
38.3
34.3
39.4
34.2
38.5
34.8
42.8
39.2
30.8
37.2
29.1
29.2
40.5
41.0
35.6
41.5
42.0
39.6
38.3
32.2
38.3
32.2
36.9
39.0
37.1
37.9
38.5
37.6
36.9
40.3
37.1
36.0
38.0
28.5
32.6
34.7
36.5
33.5
37.4
34.0
31.5
37.2
4.4
Appendix IV: Effect of Low Molecular Weight Acid (LMA) and High Molecular Weight
Acid (HMA) on Breakdown Voltage of Oil and Ageing of Paper
Table IV.4 AC breakdown voltages of new natural ester with different amount of HMA. The
gap distance between two VDE electrodes was set at 1 mm.
Breakdown voltage of new natural ester with different
amount of HMA (kV)
Test No.
9.0 mg KOH/g
51.0
1
53.1
2
48.4
3
49.4
4
54.2
5
45.8
6
49.3
7
45.8
8
48.4
9
42.9
10
54.9
11
43.8
12
51.3
13
51.9
14
49.3
15
44.0
16
44.2
17
45.0
18
52.6
19
41.5
20
50.4
21
45.0
22
45.3
23
30.8
24
55.6
25
46.2
26
43.8
27
46.1
28
51.3
29
53.9
30
41.5
31
36.2
32
53.1
33
53.8
34
39.5
35
39.6
36
50.4
37
48.4
38
45.1
39
51.8
40
203
Appendix IV: Effect of Low Molecular Weight Acid (LMA) and High Molecular Weight
Acid (HMA) on Breakdown Voltage of Oil and Ageing of Paper
45.9
38.4
53.3
38.2
54.1
54.1
45.3
47.5
45.1
47.9
47.3
5.4
41
42
43
44
45
46
47
48
49
50
Average (kV)
Standard deviation (kV)
Table IV.5 TI of conductor type paper ageing in mineral oil under dry condition. Data are
mean of 8 samples.
Ageing under dry condition
Parameter
Start
With no acid
With LMA
Ageing duration
Average (Nm/g)
Standard deviation (Nm/g)
Parameter
Ageing duration
Average (Nm/g)
Standard deviation (Nm/g)
0
14
28
117.4
94.3 73.6
8.8
11.3 10.3
Ageing under dry
condition
With HMA
14
28
63
103.9
74.7 52.1
8.7
10.1 14.0
63
51.7
16.5
14
58.9
10.5
28
64.1
8.2
63
13.5
5.7
Table IV.6 TI of conductor type paper ageing in mineral oil under wet condition. Data are
mean of 8 samples.
Ageing under wet condition
Parameter
Start
With no acid
With LMA
Ageing duration
Average (Nm/g)
Standard deviation (Nm/g)
Parameter
Ageing duration
Average (Nm/g)
Standard deviation (Nm/g)
0
14
28
117.4
66.9 60.5
8.8
12.5
9.3
Ageing under wet
condition
With HMA
14
28
63
78.4
74.0 39.5
8.7
13.8 21.6
204
63
54.3
14.3
14
12.6
5.0
28
18.7
2.9
63
14.6
4.2
Appendix IV: Effect of Low Molecular Weight Acid (LMA) and High Molecular Weight
Acid (HMA) on Breakdown Voltage of Oil and Ageing of Paper
Table IV.7 TI of conductor type paper ageing in natural ester under dry condition. Data are
mean of 8 samples.
Ageing under dry condition
Parameter
Start
With no acid
With LMA
Ageing duration
Average (Nm/g)
Standard deviation (Nm/g)
Parameter
Ageing duration
Average (Nm/g)
Standard deviation (Nm/g)
0
14
28
120.0
94.4 83.7
7.7
14.3 10.0
Ageing under dry
condition
With HMA
14
28
63
104.0
94.3 61.7
7.7
6.5
14.5
63
69.5
11.8
14
72.4
11.8
28
78.1
6.8
63
32.2
11.1
Table IV.8 TI of conductor type paper ageing in natural ester under wet condition. Data are
mean of 8 samples.
Ageing under wet condition
Parameter
Start
With no acid
With LMA
Ageing duration
Average (Nm/g)
Standard deviation (Nm/g)
Parameter
Ageing duration
Average (Nm/g)
Standard deviation (Nm/g)
0
14
28
120.0
79.5 79.2
7.7
12.8 11.4
Ageing under wet
condition
With HMA
14
28
63
77.6
83.2 76.0
7.5
6.3
9.1
205
63
66.6
8.7
14
40.5
9.2
28
58.0
4.3
63
52.3
6.9
List of Publications
[3]
206
References
REFERENCES
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
"IEEE guide for loading mineral oil immersed transformers," IEEE Std
C57.91-1995, 1996.
R. Liao, S. Liang, C. Sun, L. Yang and H. Sun, "A comparative study of
thermal aging of transformer insulation paper impregnated in natural ester and
in mineral oil," European Transactions on Electrical Power, vol. 20, pp. 518533, 2010.
C. P. McShane, J. L. Corkran, K. J. Rapp and J. Luksich, "Aging of paper
insulation retrofilled with natural ester dielectric fluid," IEEE Conference on
Electrical Insulation and Dielectric Phenomena (CEIDP), Albuquerque, NM,
USA, pp. 124-127, 19-22 October, 2003.
C. P. McShane, K. J. Rapp, J. L. Corkran, G. A. Gauger and J. Luksich, "Aging
of paper insulation in natural ester dielectric fluid," IEEE Conference on Power
Engineering Society (PES) Transmission and Distribution, Atlanta, GA, USA,
vol. 2, pp. 675-679, 28 October - 2 November, 2001.
C. P. McShane, K. J. Rapp, J. L. Corkran, G. A. Gauger and J. Luksich, "Aging
of kraft paper in natural ester dielectric fluid," IEEE International Conference
on Dielectric Liquids (ICDL), Graz, Austria pp. 173-177, 7 - 12 July, 2002.
K. J. Rapp, C. P. McShane and J. Luksich, "Interaction mechanisms of natural
ester dielectric fluid and kraft paper," IEEE International Conference on
Dielectric Liquids (ICDL), Coimbra, Portugal pp. 393-396, 26 June - 1 July,
2005.
H. M. Wilhelm, L. Tulio, R. Jasinski and G. Almeida, "Aging markers for inservice natural ester based insulating fluids," Dielectrics and Electrical
Insulation, IEEE Transactions on, vol. 18, pp. 714-719, 2011.
M. J. Heathcote, "The J & P Transformer Book," Elsevier 13 Ed., 2007.
P. K. B. Hodges, "Hydraulic Fluids," Elsevier, 1996
NynasNapthenic, "Base oil handbook," Nynas Napthenic AB, 2001.
J. Sabau, "On line reclamation of aged transformer oils," Electrical Insulation
Conference and Electrical Manufacturing & Coil Winding Conference, Ohio,
USA, pp. 557-565, 16 - 18 October, 2001.
C. P. McShane, "Vegetable oil based dielectric coolants," IEEE Electrical
Insulation Magazine, vol. 8, pp. 34-41, 2002.
E. Sherwin, "Oxidation and antioxidants in fat and oil processing," Journal of
the American Oil Chemists' Society, vol. 55, pp. 809-814, 1978.
R. Kumarathasan, A. B. Rajkumar, N. R. Hunter and H. D. Gesser,
"Autoxidation and yellowing of methyl linolenate," Progress in Lipid Research,
vol. 31, pp. 109-126, 1992.
D. Rooney and L. R. Weatherley, "The effect of reaction conditions upon lipase
catalysed hydrolysis of high oleate sunflower oil in a stirred liquid - liquid
reactor," Process Biochemistry, vol. 36, pp. 947-953, 2001.
L. E. Lundgaard, W. Hansen and S. Ingebrigtsen, "Ageing of mineral oil
impregnated cellulose by acid catalysis," IEEE Transactions on Dielectrics and
Electrical Insulation, vol. 15, pp. 540-546, 2008.
L. E. Lundgaard, W. Hansen, S. Ingebrigtsen, D. Linhjell and M. Dahlund,
"Aging of kraft paper by acid catalyzed hydrolysis," IEEE International
Conference on Dielectric Liquids (ICDL), Coimbra, Portugal pp. 381-384, 26
June - 1 July, 2005.
207
References
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33]
[34]
References
[35]
[36]
[37]
[38]
[39]
[40]
[41]
[42]
[43]
[44]
[45]
[46]
[47]
[48]
References
[49]
[50]
[51]
[52]
[53]
[54]
[55]
[56]
[57]
[58]
[59]
[60]
[61]
[62]
[63]
[64]
[65]
210
References
[66]
[67]
[68]
[69]
[70]
[71]
[72]
[73]
[74]
[75]
[76]
[77]
[78]
[79]
[80]
[81]
[82]
References
[83]
[84]
[85]
[86]
[87]
[88]
[89]
[90]
[91]
[92]
[93]
[94]
[95]
[96]
[97]
[98]
212
References
[99]
[100]
[101]
[102]
[103]
[104]
[105]
[106]
[107]
[108]
213