323

AGEING OF CELLULOSE IN
MINERAL-OIL INSULATED
TRANSFORMERS

Task Force
D1.01.10

October 2007
 AGEING OF CELLULOSE
IN MINERAL-OIL INSULATED
TRANSFORMERS

Task Force
D1.01.10

Members:
Lars E. Lundgaard – Norway (Convenor)
Dave Allan - Australia
Ivanka Atanasova Höhlein - Germany
Regine Clavreul - France
Mats O. Dahlund - Sweden
Hans-Peter Gasser - Switzerland
Richard Heywood – United Kingdom
Christoph Krause - Switzerland
Marie-Claude Lessard - Canada
Tapan Kumar Saha - Australia
Victor Sokolov – Ukraine
Alphonso de Pablo - Spain

Copyright © 2007
“Ownership of a CIGRE publication, whether in paper form or on electronic support only infers right of use for
personal purposes. Are prohibited, except if explicitly agreed by CIGRE, total or partial reproduction of the
publication for use other than personal and transfer to a third party; hence circulation on any intranet or other
company network is forbidden”.
Disclaimer notice
“CIGRE gives no warranty or assurance about the contents of this publication, nor does it accept any
responsibility, as to the accuracy or exhaustiveness of the information. All implied warranties and conditions are
excluded to the maximum extent permitted by law”.

n° ISBN : 978-2-85873-018-6
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

TABLE OF CONTENTS
Page

1 Executive summary ............................................................................................................... 5

1.1 Background...................................................................................................................... 5
1.2 Major failure scenario...................................................................................................... 5
1.3 Materials .......................................................................................................................... 5
1.4 Ageing of cellulose .......................................................................................................... 6
1.5 Interaction with other ageing processes........................................................................... 8
1.6 Diagnostics....................................................................................................................... 8
1.7 Maintenance aspects ........................................................................................................ 9
1.8 Test techniques for material ageing performance.......................................................... 10
2 Introduction ......................................................................................................................... 11
2.1 Background and aim of work......................................................................................... 11
2.2 Paper ageing and transformer risk assessment .............................................................. 12
2.3 Reinvestments, maintenance and condition control ...................................................... 13
2.4 IEC, IEEE and Cigre publications related to overall situation ...................................... 14
3 Materials............................................................................................................................... 15
3.1 Introduction.................................................................................................................... 15
3.2 Composition of cellulose ............................................................................................... 15
3.2.1 Source...............................................................................................................................................15
3.2.2 Processing/Treatment .......................................................................................................................15
3.2.3 Composition of sulphate wood pulp .................................................................................................16
3.3 Products for electrical insulation ................................................................................... 17
3.3.1 Manufacturing ..................................................................................................................................17
3.3.2 Special treatments.............................................................................................................................19
3.3.3 Products and application (summary) ................................................................................................20
3.4 Ageing effect on material properties ............................................................................. 22
3.4.1 Degree of polymerization .................................................................................................................22
3.4.2 Tensile strength ................................................................................................................................22
3.4.3 Burst strength....................................................................................................................................22
3.4.4 Elongation.........................................................................................................................................22
3.4.5 Folding endurance ............................................................................................................................22
3.4.6 Interaction with water.......................................................................................................................22
3.5 IEC publications related to cellulose materials.............................................................. 25
4 Ageing kinetics – influence from materials and condition .............................................. 26
4.1 Background; design and service conditions................................................................... 26
4.2 Transformer insulation ageing ....................................................................................... 26
4.3 Paper degradation; the general kinetic model................................................................ 29
4.4 Structure of Cellulose and the reaction rate................................................................... 30
4.5 Degradation processes ................................................................................................... 31
4.5.1 Hydrolysis.........................................................................................................................................33
4.5.2 Oxidation ..........................................................................................................................................33
4.5.3 Pyrolysis ...........................................................................................................................................33
4.5.4 Influence from oil .............................................................................................................................33
4.6 Ageing of cellulosic insulation in transformers............................................................. 34
4.6.1 Kraft paper........................................................................................................................................34
4.6.2 Thermally Upgraded Paper...............................................................................................................36
4.6.3 Ageing of wood ................................................................................................................................37
4.7 IEC, IEEE and Cigre publications related to ageing kinetics ........................................ 38

Page 2 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

5 Degradation products in impregnated insulation systems .............................................. 39

5.1 Introduction.................................................................................................................... 39
5.2 Degradation products from cellulosic insulation materials ........................................... 39
5.2.1 Water ................................................................................................................................................39
5.2.2 Acids.................................................................................................................................................40
5.2.3 Furanic compounds...........................................................................................................................41
5.2.4 Carbon oxides...................................................................................................................................41
5.3 Important degradation products of oil ........................................................................... 42
5.3.1 Acids and peroxides..........................................................................................................................42
5.3.2 Sludge...............................................................................................................................................42
5.4 Other degradation products............................................................................................ 43
5.4.1 Hydrocarbons ...................................................................................................................................43
5.4.2 Hydrogen ..........................................................................................................................................43
5.4.3 Miscellaneous ...................................................................................................................................43
5.5 IEC, and Cigre publications related to degradation products in impregnated
insulation systems ................................................................................................................. 44
6 Diagnosis............................................................................................................................... 45
6.1 Established chemical diagnosis from oil samples.......................................................... 45
6.1.1 Water ................................................................................................................................................45
6.1.2 DGA 47
6.1.3 Furanic compounds...........................................................................................................................48
6.2 Potential new chemical diagnosis from oil samples ...................................................... 53
6.2.1 Sugars ...............................................................................................................................................53
6.2.2 Organic acids ....................................................................................................................................54
6.2.3 Volatile compounds..........................................................................................................................54
6.2.4 Furanic analogous compounds .........................................................................................................55
6.2.5 Phenolic resin degradation................................................................................................................55
6.3 Measurement of the degree of polymerisation of cellulosic insulation ......................... 57
6.3.1 Context .............................................................................................................................................57
6.3.2 Sampling...........................................................................................................................................57
6.3.3 Test procedure ..................................................................................................................................59
6.3.4 Results from post mortem analysis...................................................................................................59
6.4 Dielectric diagnosis of solid insulation.......................................................................... 62
7 Life management of transformers ..................................................................................... 64
7.1 Main questions regarding cellulose insulation............................................................... 64
7.2 Remaining life estimation.............................................................................................. 64
7.3 New transformers........................................................................................................... 66
7.3.1 Quality of materials. .........................................................................................................................66
7.3.2 Transformer design...........................................................................................................................66
7.3.3 Condition control..............................................................................................................................67
7.4 Treatment of service aged transformers......................................................................... 68
7.4.1 General .............................................................................................................................................68
7.4.2 Transport (diffusion) of contaminants ..............................................................................................69
7.4.3 Removal of contaminants from cellulose .........................................................................................70
7.5 Effect of oil treatment on cellulose ageing .................................................................... 71
7.5.1 Role and characteristics of the oil.....................................................................................................71
7.5.2 Oil reconditioning and reclamation ..................................................................................................72
7.6 Methods for cellulose reconditioning ............................................................................ 74
7.7 IEC and Cigre publications related to life management of transformers ...................... 76
8 Procedures for verifying ageing performance of cellulosic insulation ........................... 77
8.1 Proof testing................................................................................................................... 77
8.1.1 IEC 60216-1 to 60216-6 "Electrical insulating materials - Properties of thermal endurance" .........77
8.1.2 IEEE C57.100-1999 "Standard test procedure for thermal evaluation of liquid-immersed

Page 3 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

distribution and power transformers" ........................................................................................................77

8.1.3 TAPPI T 573pm-03 "Accelerated temperature ageing of printing and writing paper by dry oven
exposure apparatus"...................................................................................................................................77
8.2 Laboratory scale experiments ........................................................................................ 78
8.3 Simulation of real transformers e.g. Weidmann test apparatus ..................................... 78
8.3.1 Principle............................................................................................................................................78
8.3.2 Expansion systems............................................................................................................................79
8.3.3 Temperature control .........................................................................................................................79
8.4 Slow thermal ramping.................................................................................................... 79
References.................................................................................................................................... 80

Page 4 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

1 EXECUTIVE SUMMARY

1.1 BACKGROUND
Ageing of power transformers is a growing concern of utilities around the world as power
systems grow older and demands on power quality become ever more important. Oil-
cellulose transformer insulation will irreversibly age, and the cellulose insulation cannot be
renewed. The question of transformer ageing and renewal is further accentuated by power
grids and stations operating at increased loads, higher potential overloading and larger and
more frequent load changes. The loading guides describe how loss of life is influenced by
time and temperature.

The ageing of the oil-impregnated cellulose insulation and the associated loss of mechanical
strength is one of the important factors that limit the life expectancy of a transformer. It is
commonly acknowledged that the ageing is governed by temperature. Transformers, using
cellulose insulation, are normally assumed to operate for decades at nominal load at a
specified ambient temperature. Research shows that contaminants like water, acids and
oxygen influence the ageing of solid and liquid insulation significantly. The role of these
contaminants is not considered in the loading guides.

From a utility perspective the important questions are:

− What are the ageing mechanisms?
− How do temperature, water, dissolved by-products etc. act on insulation ageing?
− Which diagnostic indicators can be used to assess ageing?
− How can ageing be slowed down?
− Which maintenance practices can be recommended to extend transformer life?

TF D1.01.10 was established to report on the normal ageing of cellulose; i.e. at design
temperatures for continuous service and accepted overloads. Different practices concerning
transformer materials and design (e.g. sealing) have to be considered. Important issues
discussed are how ageing proceeds depending on the transformer materials, transformer
condition and service conditions etc. Furthermore, what are the possibilities for diagnosis, life
assessment and maintenance? The report examines additional aspects of material ageing (i.e.
role of oxygen, water and acids) not reflected in the loading guides.

1.2 MAJOR FAILURE SCENARIO

For cellulose insulation, the important issue is how ageing influences the mechanical rigidity
of a winding; the major failure scenario being winding movement resulting from high
mechanical stresses during short circuit. What is important is the rigidity of the paper, its
ability to withstand shear stresses between conductors and spacers, and the changing of the
winding compression resulting from changes in the shape and elasticity from plastic
deformation of paper and pressboard.

1.3 MATERIALS
In power transformers, cellulose is used in various forms as a solid insulation material.
Insulation made from pure "kraft" pulp provides outstanding oil impregnation characteristics
and thus excellent performance in electrical fields. Cellulose materials also have good
geometric stability in oil, are easy to mould, wrap, bend, saw, grind, mill (depending on

Page 5 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

thickness) and are still by far the most economic solid insulation for oil cooled power
transformers.

For the manufacturing of paper and pressboard for electrical insulation, mainly unbleached
softwood kraft pulp is used. The cellulose is refined from the tree by the so-called "sulphate"
or "kraft" process. After processing the typical composition of unbleached kraft pulp is 78-
80% cellulose, 10-20 % hemicellulose and 2-6 % lignin.

Cellulose is a linear condensation polymer consisting of anhydroglucose joined together by

glycosidic bonds, Figure 1. The degree of polymerization (DP) - being the average number
(n) of glycosidic rings in a cellulose macromolecule - is in the order of 1200 for unbleached
soft wood kraft.

From kraft pulp various types of paper and pressboard having varying density are made. By
adding various nitrogen containing compounds the ageing characteristics of the cellulose may
be improved. With normal kraft paper the design maximum hotspot temperature inside the
top of windings is 98 oC, while with thermally upgraded paper the equivalent hotspot design
temperature is 110 oC.
CH2OH OH CH2OH
O HO O
HO
HO OH
O O HO
O
OH CH2OH OH

n -
Figure 1: Structural formula of cellulose.

1.4 AGEING OF CELLULOSE

IEC’s loading guide suggests in accordance with Montsinger, that the ageing of transformer
insulation is doubled (or halved) for every 6 oC in the range 80 -140 oC. This is a simplified
version of the Arrhenius law used in IEEE’s loading guide. It is the mechanical strength
(usually tensile strength) of the paper that is essential and a reduction down to 50 % is
considered as a value indicating the end of life. Measurement of DP value of paper has
become more frequently used than mechanical strength. It is a simpler tool for assessing
samples taken from older transformers. Figure 2 shows a typical correlation between tensile
strength and DP value. However, as it is practically impossible to take samples from the
critical hot-spot region one is left to making intelligent estimates based on ageing models.

Figure 2: Correlation between tensile strength and DP value for kraft paper.

Page 6 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

The DP value of the cellulose is reduced from its starting value (DP0) due to molecular
cellulose chains being cut. The relation between the chain scissions (η) and measured DPt is:
DP0
η= −1 (1)
DPt

The ageing of paper is often described by the following Arrhenius relation:

− EA
1 1
− = A ⋅ e RT ⋅ t (2)
DPt DP0

Here, R is the molar gas constant (8,314 J/mole/K), T the absolute temperature in Kelvin and
EA is the activation energy in kilojoule per mole. The pre-exponent value A is a constant
depending on the chemical environment.

It is now commonly accepted that cellulose ageing may be described by the following
processes:
− Hydrolysis, which is a catalytically governed process where the rate of chain scissions
depends on carboxylic acids dissociated in water. As both water and carboxylic acids are
produced during ageing of cellulose this process is auto accelerating.
− Oxidation, suggested to be catalyzed by hydroxyl radicals (HO•), which are produced by
decomposition of hydrogen peroxide, H2O2 and of organic hydroperoxides (ROOH).
− Pyrolysis which can take place without access to water and/or oxygen, or any other agent
to initiate the decomposition. At normal operating or overload temperatures (i.e. <140oC)
pyrolysis is of little relevance.

In a real transformer all these processes – hydrolysis, oxidation and pyrolysis act
simultaneously, resulting in a non-linear Arrhenius plot. This hampers the application of one
model describing the full complexity of the degradation processes. Which process will
dominate depends on the temperature and the condition (i.e. oxygen, water and acid content).
Furthermore, it is important to realize that different degradation processes in real life do not
act independently of each other. The ageing rates’ dependence on temperature (i.e. activation
energy) will vary depending on the process dominating in the specific temperature region as
suggested in Figure 3. It looks like the ageing by oxidation is less temperature dependent than
hydrolysis and while hydrolysis doubles for every 6-7 oC, oxidation needs a temperature
increase of 8-9 oC to double.

Figure 3: Arrhenius plot of ageing rates due to different ageing mechanisms. The arrow
shows the effect of increased water content on the preexponent A-value for hydrolysis. X-
axis is inverse temperature

Page 7 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

The condition of the insulation system is important for which ageing mechanism that will be
dominating. For kraft paper it seems from lab experiments that the degradation rate is close to
being proportional to the water content, while the effect of oxygen is less pronounced. The
difference in ageing rate between an almost oxygen-free environment and air-saturated oil
appears to be in the region of 2 to 3. It appears that low molecular weight water soluble acids
play a significant role in the hydrolysis, but further work is needed to shed more light on this.
In addition some thermal upgrading systems seem to have a significant effect, at least at
hydrolysis.

1.5 INTERACTION WITH OTHER AGEING PROCESSES

Cellulose is not the only material that ages: the oil will also age, predominantly from
oxidation, forming various carboxylic acids, some of which may contribute to the paper
ageing. Peroxides formed during interaction with active metals are another source of
synergetic effects. Furthermore, sludge may form from oil ageing and this sludge may, when
deposited on the paper, reduce the cooling of the winding and therefore lead to increased
ageing of the cellulose by increasing the temperature.

1.6 DIAGNOSTICS
Water
Water is produced by ageing and is also an important ageing accelerator that significantly
influences the ageing rate. In order to assess the ageing rate it is therefore important to
measure the water content of the cellulose correctly. It is now clear that the indirect scheme
using equilibrium curves for assessing moisture in paper via measured water content in oil
can give significant errors, due to changes in the water solubility for aged oils. Errors of a
factor of two are possible. It appears that this is mainly due to an increased content of low
molecular weight acids in the oil that increases the water solubility, thereby changing the
partitioning of water between oil and cellulose. Nevertheless, in spite of its shortcomings, this
indirect method still has its use. CIGRE WG A2.30 is presently reporting on moisture
dynamics of transformers.

Furanic compound analysis

Furanic compounds are a group of chemicals formed both by oxidation and hydrolysis of
cellulose. A lot of faith has been put in these as chemical markers for assessing the ageing
state of the cellulose. However it is now apparent that temperature dependent partitioning
between oil and cellulose has to be considered, as have moisture, acidity, stability etc. Also
one has to consider that production of furanic compounds will depend on the degree of
ageing - being distributed through the winding - and the amount of materials involved.
Finally, it appears that thermally upgraded cellulose does not produce furanic compounds to
the same extent that normal kraft paper does. This has all led to a more reluctant attitude
towards accepting simple relations between furanic compound concentrations and winding
ageing. CIGRE TF D1.01.13 pursues this complex issue.

Dissolved gas analysis

Water and carbon oxides are the main by-products from cellulose degradation. However,
other degradation processes produce these gases and they may come from the outside
atmosphere. Since the cellulose material is the main producer of carbon oxides these gases
can be used for a rough assessment of the cellulose ageing. This assessment is based on gas

Page 8 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

levels, production rates and on ratios. Recently CIGRE joint task force JTF D1-01/A2-11
made an extensive summary of typical absolute values and gas increase rates.

Other markers
Presently work goes on with gas chromatography and HPLC techniques to identify other oil
soluble markers suited for diagnostics and many interesting substances are identified (e.g.
carbohydrates, alcohols and acids). More work is indeed necessary on the low molecular
weight acids, also due to their participation in hydrolytic ageing.

Dielectric diagnosis
Dielectric response methods (FDS, PDC and RVM) are promoted for measuring water in
cellulose as explained by CIGRE TF D1.01.09. It appears that ageing by-products like
carboxylic acids may act somewhat like water in these measurements. However, the effect on
the dielectric response from ageing seems small compared to that of water, and dielectric
methods can therefore not be expected to offer an alternative to chemical diagnostics.

1.7 MAINTENANCE ASPECTS

The chemical degradation of the cellulose can, as explained above, be described with some
confidence by Equation 2, which can be rearranged for life assessment:
1 1
− EA

· e [ year ]
DPt DP0
Expected Life = RT
(3)
A · 24 · 365

DP0 is in the range 1000 - 1250 for a new transformer. DPt is defined by the user from his
preferred margin of safety for the transformer, but frequently a DP value of 200 is used as an
end of life criterion. Paper is then very brittle and cannot withstand folding. We see that life
is inversely proportional to the pre-exponent A value, which is governed by contamination of
the insulation system as explained above.

From a maintenance perspective the important questions are:

− How can ageing accelerating substances in the materials of the transformer be avoided in
new units?
− How can one avoid occurrence of ageing accelerating substances in the transformer
during service?
− How can ageing accelerating substances be removed from a transformer and when is such
refurbishment a profitable project?

The basis for a long transformer life is good design, and good materials which are adequately
dried. One may keep contamination low by moisture absorbers, vacuum pumps and sealing
membranes. To some extent oil maintenance will have a positive effect on the cellulose
insulation as well, but if a transformer has ended up with a significant contamination, more
efficient techniques are needed.

The possibility of removing any substance from the cellulose will depend on the extraction
efficiency in the surroundings of the cellulose (oil, air or vacuum). This will mainly be
governed by partitioning of the contaminant between cellulose and oil, the vapour pressure of
the contaminant and chemical bonding between contaminants and cellulose. The partitioning
of water is fairly well understood and techniques for removing water from the solid insulation

Page 9 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

are described extensively in Cigré Brochure No 227. However as acids of low molecular are
not treated specifically it is useful to give them some attention here. These acids will, like
water, stick to the cellulose. There seems to be a synergetic dependence between solubility of
water and low molecular weight acids in the oil. While the lowest molecular weight acids
have a boiling point close to that of water, the boiling temperature increases with the
molecular weight.

Relevant techniques for refurbishing cellulose insulation are: (1) Vapour phase techniques,
(2) Hot oil spray (which also can be combined with low frequency heating), (3) Hot air
circulation and vacuum extraction, and (4) Liquid circulation with extraction. One should be
aware that clamping pressure increases with moisture and that drying may lead to a reduction
even below the initial pressure when the transformer was new and dry.

1.8 TEST TECHNIQUES FOR MATERIAL AGEING PERFORMANCE

Standardized tests can be found in the referenced national and international standards. They
may be of direct or indirect interest for power transformers. Care should be taken that any
accelerated test replicates service conditions. For example: liquids with increased water
solubility may at high temperature dry out cellulose to an unrealistic level compared to
service conditions where in almost all cases most of the water will remain in the cellulose -
thereby reducing hydrolysis and leading to wrong conclusions.

Page 10 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

2 INTRODUCTION

2.1 BACKGROUND AND AIM OF WORK

This report discusses the ageing of cellulosic insulation materials in mineral oil impregnated
insulation systems for transformers and reactors. The focus will mainly be on critical paper at
the windings hot spot, but will also cover cellulose in general as it will age in similar ways
and produce markers that must be considered in a diagnostic evaluation.

In most countries the age of the transformer populations spans a long period with average age
in the 30 year range; the oldest ones being installed in the 1940s and 1950s. It is well known
that the transformer insulation system ages. As the transformers - when acquired - were
designed for a 30 years lifetime, one can understand that owners are concerned when to
replace them and how risk increases with age. However, experience tells that transformers
may remain in service for 50-60 years without failures, and that owners may be able to
postpone reinvestments. The question of transformer ageing and renewal is further
accentuated with the operation of power grids and stations moving towards increased loads,
higher potential overloading and larger and more frequent load changes.

For correct functioning, each transformer winding needs to be insulated turn from turn and
from coil to coil. Several solid insulating materials can be used for this purpose. In power
transformers, cellulosic insulating paper, also known as kraft paper, is by far the most widely
used material, although nowadays other synthetic insulating materials are used to insulate the
areas where the operating temperature is believed to be high (hybrid insulation), or for entire
transformers specially designed to operate at high temperatures. These materials (e.g. aramide
papers) should be considered very special cases, and their behaviour will not be considered in
this document (for more information on these materials and their applications, see document
IEC 60076-14/TS).

In a recent CIGRÉ study [1] utilities requested more information about paper/oil ageing on:
− Which diagnostic indicators are correlated with ageing?
− What are the ageing mechanisms?
− Which parameters like temperature, water, dissolved by-products act on insulation ageing?
− How to slow insulation ageing rate?
− Which maintenance practices could be recommended to extend transformers life?

Regarding ageing of cellulose insulation, the most important issue is how this ageing
influences the mechanical rigidity of a winding: the major failure scenario being winding
movement resulting from high mechanical stresses occurring during short circuit and possibly
during multiple inrush currents. It is the rigidity of the paper; its ability to withstand shear
stresses between conductors and spacers, and the changing of the winding compression; the
possibility for changes in the shape and elasticity from plastic deformation of paper and
pressboard again, which are important. We will mainly cover the first of these, but give some
information about the others.

As the transformer ages the mechanical strength (e.g. tensile strength) of the paper will be
reduced. This comes about because the fibers of the cellulose are weakened due to chemical
reactions cleaving the long chains of the cellulose molecules. We will cover these reactions in
detail, since they are important for understanding the rate of the ageing and because many of

Page 11 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

these chemical processes produces markers that have a potential for diagnoses of the state of
the solid insulation.

Ageing accelerates with temperature from what we will call normal temperatures to direct
failure conditions from e.g. bad soldering or clogged cooling channels. During failure
condition the temperature may reach several hundred degrees centigrade. We are here only
concerned about normal ageing (i.e. ageing as occurring before thermal failures).With normal
operation we understand the temperatures the transformers are designed for during
continuous nameplate load and accepted overloads. The temperatures considered here would
be below about 150oC. These conditions are described in the standards (e.g. IEC 60076-7).
The IEC loading guide describes how loss of life is influenced by time and temperature. One
element lacking here is consideration of the condition of the insulation system; the content of
water, oxygen, other ageing accelerators, oil quality etc. Most likely the same chemical
reactions do not occur during normal modes (most often being below 90oC) and failure
modes, and extrapolation from one condition to the other must be avoided.

There are a number of possible and related ageing effects that are relevant and should be
considered in a risk assessment, which are not covered in this work. We mention here but a
few of these. Several studies have confirmed that even if the cellulose system ages and the
mechanical strength is reduced, the withstand voltages of the insulation is little affected [2].
We are here not concerned about effects ageing has on withstand voltages. Over time the
transformer will age and its condition change: water content increases due to water ingress
and formation. Moist cellulose fibers in the insulation system may initiate breakdown
streamers [3] and also reduce breakdown voltages [4]. Water content in cellulose will also
reduce the PD-inception voltage [5]. If water content becomes very high locally it may result
in dielectric losses and formation of blisters in pressboard. With time cellulose fibers may
together with sludge clog oil channels and reduce cooling of the winding. Finally, there has
been a concern on copper sulfide lately, as several transformers have failed due a formation
and growth of this conductive substance within and on the paper. For the time being this is
considered to originate mainly from oil and copper. However, the presence of paper may
influence the process.

2.2 PAPER AGEING AND TRANSFORMER RISK ASSESSMENT

Reduced strength of paper from normal ageing may be a direct or indirect cause of a failure.
Future progress of cellulose ageing and the risks following this requires preparedness from
utilities and service providers. It is in this context we must see this work.

What we mainly discuss here is the ageing of paper; mainly as studied in a series of
laboratory experiments. The scatter in experimental results is significant. Furthermore, even
if the conditions in these experiments well simulate the condition in a real transformer, the
real situation is always more complex: temperature and water content varies along the
winding, there are several different qualities of paper (eg. thermally upgraded or normal Kraft
paper), and the paper is present in many layers with temperature and water content gradients.
Also cellulosic materials like pressboard and wood are used. Simple formulas may give
insight in ageing mechanisms, but are never sufficient for assessing the condition of a whole
winding. Having said that we still believe that one should base the analysis on the weakest
link consideration, where it will be the conditions around the hotspot region in the
transformer that are the most critical. Ageing will be most pronounced in this region within

Page 12 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

the top end of the winding, and electro-mechanical forces will be strong here. If the tensile
strength in this region is insufficient to withstand these forces, this will be the most likely
place for a failure to occur.

Furthermore, even if we have an end-of-life criterion for the strength of the paper or the
degree of polymerization of the cellulose, there is the always question of how strong the
stresses are. The possibility of a failure is different for transformers, which are placed in grids
where the transformer rarely experiences a short circuit or where the short-circuit currents are
lower than what the transformer is dimensioned for, than for transformers exposed to frequent
and high short circuit stresses from e.g. line to ground failures on power lines due to
lightning, or when moved from one place to another.

Finally, in asset management an engineer will not only look at failure probability, but also on
failure consequences to get the correct picture of the risk involved with a certain specific
transformer or a larger fleet of transformers.

2.3 REINVESTMENTS, MAINTENANCE AND CONDITION CONTROL

Analyses of the ageing of a transformer involves two major aspects: on one hand to decide to
take actions to extend the life of the transformer and on the other hand to decide when to
scrap a transformer. Knowledge of the general behavior of an impregnated insulation system
and its ageing is the key to understanding the possibilities for improving service conditions
for the cellulose (e.g. water content, oxygen and acid content) to slow down the ageing rate of
the cellulose. There exist guides for the oil maintenance [6,7]. This report will discuss how
effective different maintenance techniques may be for refurbishment and life extension of the
cellulose in a transformer.

Operating conditions vary throughout the world: In Europe transformers are predominantly
free breathing, while in e.g. USA and Japan they are sealed from the environment using
rubber diaphragms or nitrogen cushions. This will influence the environmental conditions for
the cellulose. Sealing is sometimes introduced as a retrofit, and should as such also be
evaluated as a maintenance or refurbishment action.

The analysis of the answers from a questionnaire concerning maintenance practices [1]
showed the main following tendencies:
• Transformers are used in production, transmission and distribution applications with an
average age about 30-40 years.
• Maintenance is important: the main trend is towards condition based maintenance (CBM)
and the needs tend to increase maintenance with age with a peak towards the middle of
transformers life. The main triggers are conditions assessment and DGA monitoring from
oil sampling.
• Utilities are trying to apply life extension of insulation by using any of the following
maintenance methods :
− Removal of water and ageing products: vacuum degassing, oil filtering and
heat/vacuum drying, oil reclaiming.
− Maintaining low oxygen content.
− Reducing the temperature by modifying the cooling.
− Providing better protection against external stresses.

Page 13 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

2.4 IEC, IEEE AND CIGRE PUBLICATIONS RELATED TO OVERALL SITUATION

IEC 60422 – Supervision and maintenance guide for mineral insulating oils in electrical
equipment
IEEE C57.106-1991 Guide for Acceptance and Maintenance of Insulating Oil in Equipment
IEC 60076-7 - Loading Guide for oil-immersed power transformers
CIGRE Brochure 248 – Guide on economics of transformer management

Page 14 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

3 MATERIALS

3.1 INTRODUCTION
In power transformers, cellulose is used in various forms as a solid insulation material.
Insulation made from pure "kraft"1 cellulose provides outstanding oil impregnation
characteristics, thus excellent performance in electrical fields, has good geometric stability in
oil, is easy to mould, wrap, bend, saw, grind, mill – depending on thickness – and is (still) by
far the most economic solid insulation for oil cooled power transformers.

Nowadays other, synthetic insulating materials are used to insulate areas where the operating
temperature is designed to be high (hybrid insulation), or for entire transformers specially
designed to operate at high temperatures. Such materials, for example aramid paper, are
considered very special cases, and their behaviour will not be discussed in this document. For
more information on these materials and their applications see IEC 60076-14.

3.2 COMPOSITION OF CELLULOSE

3.2.1 Source
Cellulose is the most important constituent of most plants. It is never found pure in nature.
The cotton fibre with a cellulose content of more than 95 % is probably the purest natural
source. Most commonly, in wood, plant stalks, leaves and the like, cellulose is associated
with other substances such as lignin and so-called hemicelluloses, both in considerable
quantities. Thus, according to the species, wood contains on a dry weight basis 40 to 55 %
cellulose, 15 to 35 % lignin, and 25 to 40 % hemicelluloses. The most common sources of
cellulose for industrial use are wood pulp and cotton lint. In the paper industry, cotton fibres
and cotton linters are only used for very specific applications. The bulk of the cellulose pulp
is made from wood.

As raw material for the manufacturing of paper pulp, either soft wood (spruce, pine, fir, etc.)
or hard woods (birch, beech, maple, eucalyptus, etc.) are being used. The advantage of soft
wood is its longer fibres. The typical fibre length of spruce is 2.5 to 4.5 mm whereas the
typical fibre length of beech is in the range of 0.7 to 1.7 mm. For paper and board with a high
mechanical strength, softwoods that grow in regions with low average temperatures
(Scandinavia, Canada, Chile) are optimum.

3.2.2 Processing/Treatment
For the manufacturing of paper and pressboard for electrical insulation, mainly unbleached
softwood kraft pulp is used. The cellulose is refined from the tree by the so-called "sulphate"
of "kraft" process.

The trees are supplied to the pulp factory as 1 to 4 m long logs. The logs are debarked and cut
into chips with a side length of 20 to 40 mm and a thickness of 5 to 10 mm, which can be
stored up to 6 months. The chips are cooked in a mixture of sodium hydroxide, sodium
carbonate, sodium sulphide and sodium sulphate. Today this cooking process is continuous.
The wood chips are impregnated either at ambient pressure and a temperature of 60 to 70 °C
1
Kraft denotates strength or force in German. It is written in small letters.

Page 15 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

or with elevated pressure at 110 °C for 10 to 20 minutes The chips are then cooked at 160 to
180 °C and 8 to 9 bar for 0.5 to 2 hours. After the cooking process, the solid and the liquid
phase are separated. The fibres are then neutralized with an acidic solution and subsequently
washed with water. For the manufacturing of electrical grade pulp, this washing process is of
utmost importance, since in this process step any ionic impurities are removed from the pulp.
Electrical grade cellulose is normally not bleached; therefore the pulp is dried right after the
washing process. The pulp is dewatered in a process that is similar to the paper making
process. The dewatering machine consists of a screen-, press- and drying part. The drying
part can either be heated rolls or a so called "flash drying" equipment, where the wet fibres
are fluffed and blown into a heated chamber where they are dried at a temperature of about
260 °C.

3.2.3 Composition of sulphate wood pulp

In the pulp manufacturing process, lignin and hemicelluloses are partially removed. The
typical composition of unbleached softwood kraft is the following:
75 - 85 % Cellulose
10 - 20 % Hemicelluloses
2-6% Lignin
< 0.5 % Inorganics

3.2.3.1 Cellulose
It is generally accepted that cellulose is a linear condensation polymer consisting of
anhydroglucose joined together by glycosidic bonds.

CH2OH OH CH2OH
O HO O
HO
HO OH
O O HO
O
OH CH2OH OH
n-2

Figure 4: Structural formula of cellulose.

The degree of polymerization - being the average number (n) of glycosidic rings in a
cellulose macromolecule - may be as high as 14.000 in native cellulose but the purification
procedures, as described above, usually reduce it to the order of 1200 (unbleached soft wood
kraft).

When the cellulose molecule is fully extended it takes the form of a flat ribbon with highly
hydrophilic hydroxyl groups protruding laterally and capable of forming both inter- and intra-
molecular hydrogen bonds. The surface of the ribbon consists mainly of hydrogen atoms
linked directly to carbon and is therefore hydrophobic. These two features of the molecular
structure of cellulose are responsible for its supramolecular structure and this in turn
determines many of its chemical and physical properties.

Page 16 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

3.2.3.2 Hemicelluloses
Hemicelluloses substances are a non-uniform group of complex polysaccharides that are not
cellulose. Their degree of polymerization is substantially lower than that of cellulose and is
between 50 to 250. Hemicelluloses form the actual matrix in which the cellulose fibres are
embedded. The hemicelluloses present in the cellulose fibres also play a significant part in
the mechanical strength, i.e. in the fibre-to-fibre bonding, because their hydroxyl groups form
the hydrogen bridges necessary for chemical bonding.

3.2.3.3 Lignin
Lignin is a complex high-polymeric natural substance that is difficult to degrade. It forms
together with the cellulose the main component of wood.
Lignin
OH
OH

HO OMe OMe
OMe O
HO
HO O
O
Lignin O O O
MeO OMe
HO OMe MeO

O O
O O

OH HO
HO OH
HO OMe MeO OH
OH HO

Figure 5: Structural formula of lignin.

3.3 PRODUCTS FOR ELECTRICAL INSULATION

3.3.1 Manufacturing
Wet sheet

Paper

Single ply

Multiply

Pressboard

Mouldeable

Calendered

Hot press dried

3.3.1.1 Wet sheet

The initial production process of the paper and pressboard production is identical. The
manufacturer supplies the unbleached sulphate wood pulp in a high-purity electrical quality2.
The cellulose is dissolved in filtered water and temporarily stored in vats. Refining which
brushes up the fibres, thus strongly increasing its surface, follows this. At the same time, the

2
UKPE = unbleached kraft pulp electrical grade

Page 17 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

refining creates very fine ramification, which permits sheets with a very high mechanical
strength to be produced.

The treated fibres are further diluted with water and then subjected to intensive cleaning. This
cleaning removes not only any metallic or mineral particles present but also any insufficiently
opened-up fibre bundles.

The cleaned cellulose-water mixture is now supplied to the sieve of the machine. This filters
out the fibres and forms a paper ply. Due to the nature of the production process, the fibres
are predominantly orientated parallel to the running direction of the sieve. This causes a
difference in properties of paper and board in machine (MD) and cross machine direction
(CMD).

3.3.1.2 Paper

Single ply
The paper ply is fed into the pressing and drying section of the paper machine. Finally the
finished product is rolled. In separate processes, the paper is, depending on the requirements,
calendered, creped and / or slit.

Multiply
For the manufacturing of multi-ply papers, paper machines with more than one former are
used. The wet sheets of the different formers are couched together to one layer, which is then
fed to the dewatering section of the paper machine.

3.3.1.3 Pressboard
An endless felt carries the paper web from the sieve and transports it to the automatic forming
roll. The individual plies are rolled wet on the forming roll to obtain the required wet
thickness and are couched without a binder. The wet sheet is separated from the forming roll
by cutting.

Mouldable
The wet sheet is fed into the press and compressed. The sheet is then dried without pressure
in a continuous oven. Because of its good deformability, this material is used as the raw
material for manufacturing of moulded parts.

Calendered
The wet sheet is fed into the press and cold pressed. The sheet is then dried without pressure
in a continuous oven. The dried sheets are additionally compressed on a (heated) calender.

Hot press dried

The wet sheet is dried and compressed at the same time in a heated press under high pressure
and at high temperature. The drying parameters are set so that the sheets are dried during the
process, providing a very hard and rigid board.

Page 18 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

3.3.2 Special treatments

3.3.2.1 Thermally upgrading

The purpose of "thermally upgrading" or "thermally uprating" is to reduce the rate at which
the paper decomposes over the lifetime of the transformer. Ageing effects are reduced either
by partial neutralization of water forming agents (melamine, dicyandiamide, polyacrylamide)
or by inhibiting the formation of water through the use of stabilizing agents
(cyanoethylation). The latter method requires modification of the cellulose in the pulp stage
and is thus a complicated process and is not used anymore. All processes result in an
increased content of nitrogen in the solid insulation, but the treatment is by no means
standardised, and the concentration of additives may vary.

H2N N NH2
C C NH
R CH2 CH R H2C
N
N N C C CONH2
C H2N NH CH C N

NH2

Melamine Dicyandiamide Polyacrylamide Acrylnitrile

Figure 6: Agents for thermally upgrading of paper.

A significant advantage of thermally upgraded papers is their better resistance to loss of

physical strength in operation. Often, the expression “55 C-rise paper” is used for standard
(plain, non-upgraded) paper, whereas “65 C-rise paper” stands for thermally upgraded paper.
The numbers refer to the average oil rise temperature, indicating that the designed hot-spot
temperature with upgraded paper is higher than with untreated paper. This allows an
increased continuous load rating of the transformer.

A wide range of thermal upgrading systems is available3, and a significant difference between
them is the amount of nitrogen used. Because the thermal upgrading chemicals are in most
cases not present in kraft pulp, the degree of chemical modification is often determined by
testing the amount of nitrogen present in the treated paper. Typical values for nitrogen
content of thermally upgraded papers are between 0.5 and 4 percent.

The thermal characteristics of kraft and thermally stabilized kraft paper can be reflected in
different thermal classes (105 °C respectively 120 °C), corresponding to the temperature for
50 % reduction of tensile strength after 20.000 hours (IEC 60216-1 and IEC 60216-2).

3.3.2.2 Crepe papers

Insulating crepe papers are made from electrical grade kraft papers. They consist of 100 %
pure sulphate wood pulp, and are available in standard or thermally upgraded quality.

Insulating crepe papers are used mainly as wide tapes or cut narrow rolls for insulating parts

3
A paper is considered „thermally upgraded“ if it meets the life criteria as defined in ANSI/IEEE C57.100; 50%
retention in tensile strength after 65'000 h in a sealed tube at 110°C [Definition draft by ANSI standard].

Page 19 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

of the high voltage electrical apparatus. The degree of creping matches to individual
requirements, permitting manual or machine wrapping and a close joint between layers. The
pronounced mechanical extension characteristics of creped papers enable insulation of rings,
bended tubes and conductor leads.

Special mechanically strong crepe papers with high tensile energy absorption are used as a
final wrapping around continuously transposed conductor (CTC) bundles.

3.3.2.3 Diamond dot printed papers

Diamond dot printed papers (abbrev. DDP or DPP) are used to form a bonding and insulation
system of excellent mechanical stability. In most cases, the resin is applied in dots forming
squares, arranged in a diamond pattern and printed to either both sides or only one side of the
paper surface. Through oven heating, the thin layer of B-stage epoxy resin4 melts and cures,
providing a bond that is resistant to the solvent effect of transformer oil. The epoxy resin has
a superior adhesion to copper and aluminium and is widely used for distribution transformers
with the result of solid winding blocks. Occasionally, it is used as a final wrap on CTC
conductors of power transformers. These papers are available in standard or thermally
upgraded quality.

3.3.3 Products and application (summary)

Table 1: Main components of power transformer insulation.

Solid insulation Mass Temp. Composition Material Description
(%) (°C) (IEC Standard)
Paper (conductor 20 - 25 80 – 110 Sulphate wood pulp1 60554-3-1: 1.1-1.4
wrapping, soft paper
bushing)
Radial spacers (in the 10 - 15 80 - 100 Sulphate wood pulp 60641-3-1: B.3.1
winding)
Cylinders and strips 15 - 20 60 - 80 Sulphate wood pulp 60641-3-1: B.3.1
(between windings) Laminated pressboard2 60763-3-1: LB.3.1.1/2
Corrugated board and 10 - 15 60 - 80 Sulphate wood pulp 61628
cylinder wraps 60641-3-1: B.2.1/3.1
Angle rings, caps, 5 - 10 60 - 80 Sulphate wood pulp 60641-3-1: B.4.1/5.1
snouts, lead exit
insulation
Washer ring plates and 5 - 10 60 - 80 Sulphate wood pulp 60641-3-1: B.3.1
spacing blocks Laminated pressboard 60763-3-1: LB.3.1.1/2
Clamping rings and 25 - 30 60 - 80 Laminated sulphate wood pulp 60763-3-1: LB.3.1.1/2
plates Laminated wood low density3 61061-3-1: C1R / C2R
Laminated wood high density 61061-3-1: C3R / C4R
Crepe paper (lead <5 80 - 110 Sulphate wood pulp 60554-3-3: 3.1 - 3.2
insulation, screen
rings, RIP-bushings)
Presspaper4 Sulphate wood pulp 60641-3-2: P.2.1/4.1/6.1/7.1
Varnish Shellac (natural resin)5 Shellac
Mass Fraction of cellulosic material in solid insulation system of oil cooled power transformers.
Temp. Refers to typical operational temperature under full load.
1. Alternatively materials made from cotton fibres or a mixture of cotton fibres and sulphate wood pulp are being used.
2. Commonly used adhesives: Casein (water based), synthetic resins (e.g. polyester).
3. Commonly used adhesives: phenolic resins, the reaction product of formaldehyde and phenols.
4. Mainly distribution transformers.
5. In the early years of power transformer manufacturing, Shellac had been applied as a varnish on transformer windings. It is basically a
thermoplastic material with a melting point, which increases on prolonged heating, attaining finally a tough and rubbery state.

4
This epoxy is solid, but will melt and later cure during heating.

Page 20 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

In Figure 7 and Figure 8, insulation components of a core-type transformer are depicted.

Clamping plate Clamping plate

Tube Snout

Cylinder Spacer block

Angle ring Cap

Moulded lead Stress ring

exit
Paper wrap around
copper wire

Cylinder
Figure 7: Cross-sectional view of a 400 kV transformer end insulation (220 kV-side).

Angle ring Crepe paper wrapping

of stress ring
Edge protection ring
Spacer (radial spacer)
Cylinder
Paper (wrapping)
Strip (axial strip, rib,
stick)

Figure 8: Cross-sectional view of a 400 kV transformer end insulation (detail).

Page 21 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

3.4 AGEING EFFECT ON MATERIAL PROPERTIES

3.4.1 Degree of polymerization

The degree of polymerization is measured in accordance with IEC 60450. Since not only
cellulose but also hemicelluloses and lignin are dissolved, the DP is a sum parameter. It is a
measure for the average molecular weight of the cellulose molecules and so the most
important parameter to assess the ageing status of the cellulose insulation of electrical
equipment.
With other methods, such as size exclusion chromatography (SEC) not only the average
degree of polymerisation, but the molecular weight distribution can be determined.

3.4.2 Tensile strength

The tensile strength of cellulose insulation can be influenced by the choice of raw materials
and the manufacturing process. The mechanical strength of insulating paper used for
conductor insulation as well as for pressboard used as structural components, plays a vital
role for the functioning of electrical equipment. The tensile strength of cellulose insulation
materials has a direct correlation to the degree of polymerisation.

3.4.3 Burst strength

The burst strength, which is in fact a biaxial tensile test, shows a similar correlation with the
degree of polymerisation as the tensile strength. This correlation gives generally lower values
of retained mechanical strength for a particular value of DP than does tensile strength [8].

3.4.4 Elongation
The elongation of the materials is also directly linked to the degree of polymerisation. When
the paper reaches its end of life, the elongation becomes very small and thus the paper brittle.

3.4.5 Folding endurance

As the elongation, the folding endurance is also a measure for the brittleness of the material.
Its decrease follows the degree of polymerisation similar as the elongation.

3.4.6 Interaction with water

3.4.6.1 Water absorption in air

Saturation (steady state)

Cellulose is due to its chemical nature highly hygroscopic. On the molecular level, cellulose
can theoretically absorb 28 % water when all the hydroxyl groups form a hydroxyl bond with
one water molecule. In practice the following saturation levels have been found:

Page 22 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

18
16

Water content in paper (%)

14
12
10 20°C
8 40°C

6
4
2 60°C
80°C
0
0 20 40 60 80 100
Relative humidity of air (%)

Figure 9: Water saturation of paper in air [after 9].

Rate of water absorption (transients)

The rate of water absorption is governed by the following material parameters:
- Thickness of the board / paper
- Density of the board / paper
- Temperature
- Oil impregnation (without / with oil)
- Environment (air / oil flow)

Influence of the thickness Influence of the density

Pressboard B.3.1 at 23°C and 50 % RH Pressboard 3 mm at 23°C and 50 % RH
8 8
Water absorption (%)
Water absorption (%)

0.85 g/cm³
6 1 mm 6
3 mm 1.25 g/cm³
4 5 mm 4
8 mm
2 2

0 0
0 7 14 21 28 0 7 14 21 28
Time (days) Time (days)

Influence of the oil Influence of the temperature

Pressboard B.3.1, 3 mm at 50 % RH Pressboard B.3.1, 3 mm at 23°C and 50 % RH
8 8
Water absorption (%)
Water absorption (%)

30 °C
6 6
not oil impregnated 10 °C
4 4

2 2
oil impregnated
0 0
0 7 14 21 28 0 7 14 21 28
Time (days) Time (days)

Figure 10: Rate of water absorption of previously dried pressboard samples, exposed to air
(weight increase) [Weidmann experimental investigation].

Page 23 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

3.4.6.2 Dimensional changes due to water absorption

When cellulosic insulation picks up water it swells. The major dimensional change concerns
the thickness of the material. In machine and in cross machine direction, the change is about
one order of magnitude lower.
8
Dimensional change (%)

Weight
Water absorption (%)

6
Thickness
4

2
Length, machine and cross machine

0
0 7 14 21 28
Time (days)
Figure 11: Water absorption (weight) and dimensional change of predried, non-impregnated
precompressed board, 3 mm exposed to air at 23 °C and 50 % relative humidity
[Weidmann experimental investigation].

3.4.6.3 Moisture distribution in oil filled systems

Insulating oils, such as transformer oil, have a low affinity for water. However, the solubility
increases markedly with temperature. In a transformer in service, most of the water stays
within the solid insulation. With rising temperatures a small amount of the water moves from
the paper to the oil. With the help of the graph below, and assuming the same temperature for
oil and solid insulation, it is possible to determine the average amount of water in the
cellulose insulation by measuring the water content of the oil [10]. In order to get proper
results, it is essential to know the exact temperature of the oil at the moment when the sample
is taken. This moisture distribution is only valid for steady-state conditions, which may take
significant time to be established. The moisture distribution of aged paper - oil systems will
deviate from this as discussed later in 6.1.1.

20 °C 30 °C
12 40 °C
Water content in paper (%)

10

8
50 °C
6
60 °C
4 70 °C
80 °C
2

0
0 10 20 30 40 50 60 70 80
Water content in oil (ppm)
Figure 12: Moisture equilibrium of paper in oil [after 11].

Page 24 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

3.5 IEC PUBLICATIONS RELATED TO CELLULOSE MATERIALS

IEC 60216-1 (2001) Electrical insulating materials - Properties of thermal endurance - Part 1:
Ageing procedures and evaluation of test results
IEC 60216 -2 (2005) Electrical insulating materials - Thermal endurance properties - Part 2:
Determination of thermal endurance properties of electrical insulating materials -
Choice of test criteria
IEC 60554 Specification for cellulosic papers for electrical purposes
IEC 60641 Specification for pressboard and presspaper for electrical purposes
IEC 60763 Specification for laminated pressboard
IEC 61061 Non-impregnated densified laminated wood for electrical purposes
IEC 61628 Corrugated pressboard and presspaper for electrical purposes
IEC 61629 Aramid pressboard for electrical purposes

Page 25 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

4 AGEING KINETICS – INFLUENCE FROM MATERIALS AND CONDITION

4.1 BACKGROUND; DESIGN AND SERVICE CONDITIONS

The loss of mechanical strength and the ageing of the oil-impregnated cellulosic insulation is
one of the important factors that limit the life expectancy of a transformer. It is commonly
acknowledged that the ageing is governed by temperature. A lot of research in the last years
shows that water and oxygen influence the ageing of solid and liquid insulation significantly.

In many countries the transformer populations have reached an average age around 30 years.
Transformers, using normal kraft paper in the insulation, are normally assumed to operate for
decades at nominal load under a specified ambient temperature with a design maximum
hotspot temperature inside the top of windings of e.g. 98 oC [12]. With thermally upgraded
paper the equivalent hotspot design temperature is 110oC [13]. The ageing of the transformer
population accentuates questions concerning life expectancy and good service and
maintenance schemes.

Usually the transformers are operated at lower loads than they are specified for. This gives a
potential margin in the life expectancy. However, often the condition of the insulation in the
transformers is not as good as in a new modern transformer; water, oxygen and contaminants
are present and will accelerate ageing.

Ageing depends on temperature; according to Montsinger’s rule - established in the 1930s -

life is halved for every 6-10 degrees temperature increase depending on the material [14].
Later, major studies performed by French and British research groups showed that ageing
was accelerated by presence of water and oxygen. IEC’s loading guide [IEC 60076-7] does
recognize the effects of contamination of the insulation system, but gives no information
about its importance. CIGRÉ Task Force D1.01.10 was established to report on ageing of oil-
impregnated cellulose, and in this context also maintenance and possibilities for diagnostics.
New knowledge has surfaced and needs to be assessed and reported to the transformer
manufacturers and users. This Task Force forms one part of the work of CIGRÉ Working
Group A2.24 that looks at thermal performance of transformers in a wider perspective, also
considering thermal models and designs of transformers and questions concerning operations
of transformers in the grid.

4.2 TRANSFORMER INSULATION AGEING

IEC’s loading guide [12] suggests in accordance to Montsinger that the life of a transformer
can be described according to the equation:
Life duration = e-p*θ (3)
where p is a constant (a value 6 is suggested in the range 80-140oC) and θ the temperature in
degrees Celsius5. This is a simplified version of Arrhenius law used in IEEE’s loading guide
[13] stating that per unit life (PUL) can be expressed as:
PUL = A·eB/(θH+273) (4)
where A and B are constants, θH is the hottest-spot temperature6, which at design hotspot
gives PUL equal to one. As no end-of-life criterion for a transformer is available presently,

5
Here θ for unit life is set to 98oC, and life halves/doubles for every 6oC up/down.
6
Note that in this case the reference temperature is 110oC according to US practice of using upgraded paper.

Page 26 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

both guides advocate an approach where instead ageing rates are considered.

In the IEC guide this reads:

Rate of ageing = constant×epθ (5)
When this is taken further in a relativistic analysis, one then looks at ageing rate at a given
temperature compared to that of the design hot spot temperature:
V = ageing rate at θH/Ageing rate at 98oC =2 (θH-98)/6 (6)
For a transformer – in the context of thermal ageing – it is the mechanical strength of the
paper that matters. The ageing of paper results in a decreased mechanical strength and
reduces the ability of the transformer to withstand short circuit stress. Early work showed
how mechanical strength of paper can be measured in several ways as indicated in Figure
13a. Tensile strength, elongation and folding strength all decay with time, and faster at higher
temperatures. Plotting the three different parameters against time in semi-logarithmic paper
(Figure 13b) shows the same rate of change for all three properties. Some countries also use
bursting strength as a reference, tested with a paper spanned over a tube with an internal
rubber bladder that is blown up until the paper bursts. Bursting strength is reported to have a
more linear correlation with DP [8,15] as seen in Figure 14.

a)
)) b)

Figure 13: a):The effect of temperature on three representative mechanical properties of

vacuum dried and oil impregnated insulating paper aged under oil in sealed
containers having nitrogen gas in contact with the surface of the oil [16].
b) Comparison of the three mechanical properties, showing equal ageing rate for
all three properties. The same is valid at 120oC and 135oC.

Because it is impossible to perform meaningful mechanical tests on paper samples

unwrapped from conductors in a transformer one is stuck with the possibility of measuring
the degree of polymerization (DP) of such samples [18,19,20]. The DP value is an average
value of chain lengths of the cellulose molecules given as a number of glucose rings in a
cellulose chain. It is measured through measurement of the viscosity of a paper solution [17].
For new paper the DP-value is about 1200, and after a new transformer has undergone high
temperature drying a DP-value around 1000 is representative. With other methods, such as
size exclusion chromatography (SEC) not only the average degree of polymerisation, but the
molecular weight distribution can be determined as shown in Figure 15 [18, 19].

Page 27 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

Figure 14: (a) Relative decrease of mechanical parameters with decrease of DP of cellulose
paper. Redrawn after [15]. (b) Correlation between mechanical properties and
remaining DP of upgraded creped kraft paper. Redrawn after [8].

Figure 15: Molecular size distribution of new and aged kraft paper. The measurement is done
by gel permeation chromatography (GPC) and light absorbance. X-axis is elution
volume [ml]. For new paper, the right peak is the cellulose, while the left peak
probably shows hemicelluloses. [Investigation by Dr David Hill, University of
Queensland].

During ageing both mechanical properties and DP-value will be reduced with time as shown
in Figure 16. The figure also shows how the tensile strength depends on the polymerisation of
the paper. Often a DP value of 200 is considered as the end-of-life for the cellulose
insulation. Figure 16c shows that at this DP level the tensile strength is about 30 % of initial.
Interfiber forces are also reduced when DP is getting down in this range; the ratio between
long-span to short-span tensile strength falls [20]. Here long span and short span refers to the
distance between the jaws of the tensiometer; long span involving a longer paper strip with
more bonds between fibres than for the short span where the jaws are very. Down to DP-

Page 28 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

values of 600-700 the papers mechanical strength will stay as for new.

Figure 16: Development of DP and tensile strength of paper for ageing kraft paper (3% water
at 110oC), and correlation between the two parameters from multiple experiments
[21].

4.3 PAPER DEGRADATION; THE GENERAL KINETIC MODEL

As ageing proceeds the molecular weight and DP of the cellulose is reduced due to molecular
cellulose chains being cut. The relation between the chain scissions (η) and measured DP is:
η = DP0/DPt – 1 (7)
DP0 is the initial degree of polymerisation and DPt is value after an ageing period t. Most
analyses of degradation have been based on the work of Kuhn and co-workers in 1930 [22],
which was extended by Ekenstam in 1936 to relate rates of degradation to DP [23]. Ekenstam
considered random, first order chain scission and showed a direct relationship of reciprocal
DP with time and that this relation combined with the Arrhenius equation to include
temperature dependence can be mathematically expressed as:
−E
1 1
− = A ⋅ e RT ⋅ t
DPt DP0
(8)
−E
1 ⎛ DP0 ⎞
⎜⎜ − 1⎟⎟ = A ⋅ e R T ⋅ t
DP0 ⎝ DPt ⎠
Here, R is molar gas constant (8,314 J/mole/K), T the absolute temperature in Kelvin ( equal
to θH + 273) and E is the activation energy in kilojoule per mole. A is a constant depending
on the chemical environment.

So from several reasons it is convenient to focus on changes in the DP value or the rate at
which chain scissions occur. The equation above can be explained and supported by the
following procedure and considerations: If we instead of plotting DP vs. time as shown in
Figure 17a, plots 1/DP as shown in Figure 17b we get fairly straight lines, showing that the
rate of change (k) is quite linear over time up to a certain value of 1/DP. This is in accordance
with a model saying that Δη/Δt = k, which is a first order reaction rate model. Plotting the
natural logarithm of k vs. 1/T gives straight lines as would be the case for a thermally
activated process described in an Arrhenius plot. In equation 6 it is the value of E describes
the slope of the curve in Figure 17c; the higher this value is the more temperature dependent
will the reaction rate be (steeper curve). Table 2 shows how the activation energy is related to

Page 29 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

the temperature increase giving a 50% life reduction: Chemists prefer to relate the energy to
joule per mol, while physicists use eV per molecule. In principle the A-values determine the
intercept of the curves in 14c with a virtual Y-axis from 0; the higher the value the higher the
location of the curve above the abscissa and the faster the ageing.

Figure 17: Ageing of kraft paper with a high initial water content versus time for four
different temperatures. a: DP-value versus time, b: 1/DP versus time, c: Reaction
rate versus inverse absolute temperature [21]. (Wet paper is real data, dry paper
is somewhat idealised; i.e. assuming an activation energy equal to that of wet
paper.)
Knowing the end-of-life (EOL) criterion, we can reorganize equation 6 to express life
expectancy as a function of temperature T, and the parameters E and A:
1 1
− E

* e [ year ]
DPt DP0
Expected Life = RT
( 9)
A * 24 * 365
Which is equal to what was is suggested in the standards, except from the EOL-criterion
being based on DP value instead of mechanical rigidity of the paper as Montsinger did.

Table 2: Correlation between activation energy and temperature rise for halving of life. This
is only valid in a limited temperature range around maximum operating temperature.
Activation
kJ/mol 70 90 110 130 150
energy
o
ΔT (50%) C 11,7 9,1 7,4 6,2 5,4

4.4 STRUCTURE OF CELLULOSE AND THE REACTION RATE

The structure of cellulose is also said to influence the rate of degradation. In the fibers and
micro-fibers the cellulose is found in semi-crystalline regions where these regions of high
crystallinity are separated by regions of low crystallinity or amorphous structure. The
amorphous regions are said to degrade quicker than the more crystalline regions (Erofeev et
al., 1982) and even in the amorphous regions there are said to be areas that are more reactive
than others. The existence of weak links within the cellulose chain have been proposed to
explain initial rapid degradation rates before a steady rate was achieved
[24, 25, 26, 27, 28, 29]. A weak link where the cellulose chain folded (every 30 units) was
proposed by Chang [24], which coincided with the fibril lengths of 30 nm. Other reports have

Page 30 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

suggested a weak link every 500 units [26]. The degradation rate of crystalline regions of
cellulose is lower due to the higher degree of order within the cellulose chains reducing
accessibility of ageing agents and the increase in strength induced by crystallinity. During
ageing one can see that the DP reduction levels off at a value of about 100. At this level, only
microcrystalline regions of the cellulose remain7.

Looking closer to Figure 17b one sees that the curve describing the reaction rate at 130 oC is
not straight. The steepness (or reaction rate) falls off with time. This is explained by a
reduction with time of amorphous regions or locations in the cellulose structure where chain
scissions may occur. In a recent study [30] it is proposed to extend Equation 6 to:

1 − 1 = k10 ⎡1 − e −k2t ⎤ ( 10)

DPt DP k 2 ⎢⎣
0
⎥
⎦

Where k10 is the initial rate at which bonds break and k2 is the rate at which k10 changes. k10
and k2 will vary with temperature in the same manner using the Arrhenius equation.
Equation 8 takes into account that the rate of chain scissions changes with time as would be
expected for an inhomogeneous material such as kraft paper, with a mixture of amorphous
and crystalline regions.

In addition to the cellulose also hemicellulose and lignin will age. The lower “hump” in
Figure 15, is associated to the hemicellulose. One can see that also these molecules are
reduced in size with ageing.

4.5 DEGRADATION PROCESSES

The degradation of paper insulation within transformers has been studied for many years and
several theories postulated on the mechanism and the kinetics involved. It is commonly
accepted that the main factors influencing the degradation paper insulation are temperature,
water, oxygen, acids and contaminants.

There are many reports in the literature that have tried to quantify the degradation rates with
respect to water, oxygen and temperature. These reports have been reviewed for transformer
operating conditions by Emsley [31, 32, 33]. Other factors such as electric fields and oil
degradation products are considered to be of secondary importance according to Fabre and
Pichon [11].

The temperature at which a transformer operates will affect the rate of degradation. The rate
of most chemical reaction rates doubles every 2-25°C, depending on the activation energy.
Reported activation energies for cellulose aged in oil vary from 85 kJ/mol
[34, 35] (degradation in vacuum) to 120 kJ/mol [36] (hydrolytic degradation). Emsley [31]
calculated an average activation energy of 111 ± 6 kJ/mol from results from the main
significant studies available at that time, using the assumption that the rate of degradation
approximates to zero order and that all processes have the same activation energy. However,
biotechnologists state that normally hydrolyses have an activation energy in the 115-130
kJ/mol range and oxidation via free radical influence give activation energies in the 50-65
kJ/mol range [37]. It has been reported that the rate increases significantly above 140°C [38],
due to a change in the activation energy or the pre-exponential factor of the Arrhenius
7
Actually, microcrystalline celluloses used as filler in medical tablets is produced by hydrolysis of cellulose.

Page 31 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

equation. For ageing in a water and oxygen free environment, where only pyrolysis takes
place activation energies are found to increase; Fung doing experiments under vacuum at 200
to 280oC experienced activation energy of about 150 kJ/mole [39] and even higher values are
suggested. Potential oxidation of the paper greatly effects the pre-exponential factor [31]; i.e.
any residual oxygen in a so called oxygen free environment will effect the Arrhenius
parameters significantly. Zou et al. [40] associated an increased rate above 140°C with
pyrolysis (thermal degradation) of cellulose and assumed that below this, the dominant
mechanism is hydrolysis.

To summarize: One may say that we have three main processes of degradation:
− Hydrolysis
− Oxidation
− Pyrolysis

In a real transformer all these processes – hydrolysis, oxidation and pyrolysis act
simultaneously, resulting in a non-linear Arrhenius plot - which hampers the application of
one single activation energy - describing the full complexity of the degradation processes.
Which process will dominate depends on the temperature and the condition. Probably also
synergetic effects takes place between the different reactions; e.g. oxidation may activate
hydrolysis. However, we will for illustrative purposes assume independent processes. The
total degradation then being the sum of degradation from each process becomes:
⎛ − EOxi − E Hyd − E Pyr
⎞
⎜
η Tot = AOxi e R⋅T
+ AHyd e R⋅T
+ APyr e R⋅T ⎟ ⋅ t (11)
⎜ ⎟
⎝ ⎠
where oxi, hyd and pyr are used as subscripts to identify the activation energy E and the
environment factor A for the singular processes mentioned above. Somewhat simplified one
can say the degradation rates from these reactions will depend on activation energy and the
environmental for each process at the given temperature. The ageing rates dependence on
temperature will vary depending on which process that dominates in the specific region as
suggested in Figure 18.
ln (reaction rate)

Heat
H2O O2

1/T

Figure 18: Sketch of ageing rates due to different ageing mechanisms. The arrow shows the
effect of increased water content increasing the A-factor for hydrolysis.

Page 32 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

4.5.1 Hydrolysis
Presence of water will increase the rate of degradation. At the beginning of a transformer’s
life, the kraft insulation contains less than 0.5 % water, and the oil is also dried. The water
content levels within the transformer may increase up to 5 % during its lifetime (Fallou,
1970). Fallou (1970) showed that the rate of degradation of the paper at initial value of 4%
water content was 20 times greater than that at 0.5 % water content. So, in principle as the
transformer ages the rate at which the insulation deteriorates is expected to increase.
Recently, Lundgaard et al [41] suggests that the hydrolysis of cellulose is a catalytic process
where the reaction rate depends on dissociated acids or rather H+-ions that can get into the
amorphous zones of the cellulose. This mean that the A-factor will depend both on the
existence of organic acids - as earlier also suggested by Ivanov [42] – and water, that is able
to dissociate these acids. Low molecular weight watersoluble acids, that are formed by the
paper ageing and to some degree also by the oil ageing, are more efficient than the larger
hydrophobic acids, which mainly stems from the oil ageing. The fact that acid catalyzed
hydrolysis generates organic acids and at the same time is governed by their presence makes
the process auto-acceleratory.

4.5.2 Oxidation
Oxygen will also accelerate the degradation reaction. If the oxygen level in the oil is held
below 2000 ppm the rate of degradation of the full insulation system is reported to be five
times lower than that of a free breathing transformer [43]. Transformer oil can reach about
30 000 ppm when the oil if fully saturated, but in reality most free breathing transformers in
service (warm) only contain 20 000 ppm as a result of dissolution of gas from the air. Cox’
results suggest that by extracting the oxygen to below 300 ppm, using for example semi-
permeable membranes the effect of oxygen can be reduced to a sixteenth of that of normal
operating conditions [44]. These results contrast with experimental evidence that ageing of
paper in oil with access to oxygen is only about 2-3 times higher than ageing under vacuum
[11, 41]. Oxidation promotes accumulation of additional ageing accelerators as e.g. acids.

It is suggested [41] that the oxidative depolymerisation is catalyzed by hydroxy-radicals

(HO•), which are produced by decomposition of hydrogen peroxide, H2O2 and of organic
hydroperoxides (ROOH). Hydrogen peroxide can for example be formed from oxygen and
water by reactions catalyzed by transition metal cations (such as Cu+/Cu2+ or Fe2+/Fe3+).
Hydroxy-radicals are formed from H2O2 or ROOH in a reaction catalyzed by traces of Fe3+ or
other active metals, together with small amounts of autooxidizable compounds such as
phenols, aromatic amines or thiols. It is here also suggested that the oxidation is reduced in an
acidic environment, which would reduce the importance of these reactions with time.

4.5.3 Pyrolysis
By pyrolysis we mean a process that can take place without access to water and/or oxygen, or
any other agent to initiate the decomposition. At normal operating or overload temperatures
(i.e. <140oC) such processes are considered to be of little relevance. At high temperatures,
which may occur at defects such as poor soldering or magnetic induced local failure currents,
pyrolysis may well occur. Generation of CO and CO2 may follow. However this is outside
the scope of this report.

4.5.4 Influence from oil

Liquids for transformers are mainly mineral oils, with different composition and refining,
containing different oil molecules (aromates, naphtenes and paraphines), sulphur compounds,

Page 33 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

inhibitors, passivators, etc. Also synthetic liquids like vegetable and mineral oil based esters,
silicone oils etc are used. In principle ageing products from these liquids may influence the
paper ageing.

Hydrolysis may be influenced by low molecular weight acids from oils. Oxidation of the oil
produces both low molecular weight acids and larger acids. Oil testing according to IEC [45]
includes measurement of both high and low molecular weight acids in the aged oil and in a
water trap respectively, so in principle data on this is available. However, it seems that a
significant fraction, or even most, of the low molecular weight acids found are paper
degradation products.

In their book, Lipstein [46] et al. describe oil ageing in general and also the influence on
paper ageing. Their focus was mainly on transformer oils from Russian crude oils. They
suggest that in a process of autocatalytic oxidation of hydrocarbons polyatomic
hydroperoxides are formed, which as explained above through the reaction with copper
would form radicals that would initiate oxidation of paper. They observed that paper ageing
in synthetic askarels was slower than in mineral oils, and attributed this to less production of
aggressive compounds in these liquids.

It has been suggested that paper will age slower in certain esters due to a drying effect of the
liquid due to its higher (some 100 times) water solubility. This is believed to be a highly
unlikely effect, as anyway the solubility of water in cellulose will be much higher than in the
ester, and the water concentration in the paper in a real transformer - having a weight ratio
oil/cellulose of 10-20 - would not be that much influenced.

However, even if we know that there is a correlation between oil and paper ageing, much of
this is still not properly investigated. Here one should not forget that the additional
complication of the environment with metals in the real transformer.

4.6 AGEING OF CELLULOSIC INSULATION IN TRANSFORMERS

4.6.1 Kraft paper

For kraft paper the general picture is that ageing rates doubles for every 6-8 degrees Celsius.
Based on equation 7, a start DP=1000 and end-of-life DP=200, paper would last about 20
years at a temperature of 98oC using an A-value of 1·108, which has been found representative
for test conditions where access to oxygen and water is hindered. Free access to air and
oxygen would reduce expected life with a factor of about two. How this reduction depends on
the content of oxygen in the oil is not clear. Life expectation is approximately inversely
related to water content and at around 2% the life would be reduced by a factor of around 10.
One knows that there is a strong synergy between water and organic acids, but details on how
to assess this synergetic effect are not yet known.

In his thorough review of paper ageing Emsley found an activation energy of 111 kJ/mol,
with 95% confidence bounds of 105-117 kJ/mol, in average for all the experiments reviewed.
The same value (114kJ/mol) Lundgaard found by mapping a large group ageing experiments
together. However, he found that when differentiating between experiments where oxygen
was added and those where water was added he got respectively 96 kJ/mol and 125 kJ/ mol,
supporting the idea that the activation energy is different for the oxidation and hydrolysis.

Page 34 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

Based on choosing an activation energy of 111 kJ/mol for ageing of cellulose under optimum
conditions an A value of 1 to 2×108 seems reasonable.
Figure 19 shows how water will influence the A-values based on the findings of Emsley from
[31] (with 95% confidence limits) and data from [41]. Note that it is the region of water
contents up to 2% which is of most practical interest, since at equilibrium conditions higher
water contents than this in the hottest regions of the windings are very unlikely.

Figure 19: Ageing factor ”A” for initial water content in cellulose in laboratory tests. The unit
of A is h-1.

1200
Formic

1000 Acetic
Levulinic
Naphtenic
800
Stearic
DP-value

No acid
600

400

200

0
0 200 400 600 800
Time [hours]
Figure 20: Ageing of kraft paper at 130oC with 2,5 % water content with different acids in
oil. The oil was prepared to a neutralisation value of 0,4 mg KOH/g oil before
impregnating the paper with it. Formic, acetic and levulinic acid were to a large
extent absorbed by the paper, while naphtenic and stearic were not absorbed [47].
The direct correlation of ageing rate to water content can only be considered a preliminary
step as acids do play a role together with the water. It is known that low molecular acids
together with water are the real driver of the hydrolysis, potentially being the most significant
process [42, 47]. One would expect that a relation like A ∝ KH2O⋅NLA , where KH2O is the
concentration of water in cellulose and NLA is the neutralisation value of low molecular acids
in the oil or paper, to be a potentially better parameter for assessing the importance of
hydrolysis. Figure 20 indeed supports the focusing on the low molecular acids. One must be

Page 35 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

aware that in this study the partitioning coefficient of the various acids resulted in very
different acid concentrations in the paper. This is in line with experiences from service aged
transformers, where the concentration of water soluble acids can be 100 times higher in the
paper than in the oil. It is clear that the high molecular weight acids not being water
dissolvable and mainly originating from oxidation of oil – which dominates the neutralisation
value of the oil – do not play any significant role for the cellulose ageing. However, much
work remains before a more practical A-value for the cellulose ageing can be determined
from oil diagnostics. The neutralisation value (total acid number) – as used today- is not good
for assessing the content of low molecular weight acids in the paper.

A hypothesis emerges – at least for open breathing transformers – that the ageing starts with
oxidation in an alkaline or neutral environment. With time, this process is limited as the
environment becomes more acidic and the auto-accelerated acid catalyzed hydrolysis takes
over.

4.6.2 Thermally Upgraded Paper

Thermally upgrading of paper may reduce the ageing rate by a factor between 1,5 and 3. The
ageing acceleration from the presence of water is seen to be strongly reduced for the
upgraded paper, while ageing with access to oxygen seems to be less influenced. For
upgraded paper there is still some uncertainty about the activation energies, and which
process dominates ageing. The reduced sensitivity for hydrolysis is in line with theories
presented by Bauer et al. [48], and are supported by two larger investigations performed by
Shroff [49] and Lundgaard [41, 50]. In these investigations paper from two different
manufacturers was used.

Lundgaard et al. have shown that upgraded paper produces more acids than regular kraft
paper. This effect has to be verified in further experiments. Lundgaard found that the
activation energy for the ageing of upgraded paper was lower than for kraft. Most likely the
ageing of the paper he investigated is mainly caused by oxidation, due to the alkaline
environment provided by the additives that would favor oxidation and sequester the acid
catalyzed hydrolysis.
a) -8 b) -8
Kraft dry Kraft Dry
Insuldur dry Insuldur Dry
ln (Reaction rate)

-12 -12
ln (Reaction rate)

Kraft wet Kraft + Oxygen

Insuldur wet Insuldur + Oxygen

-16 -16

-20 -20

-24 -24
0.0025 0.0026 0.0027 0.0028 0.0029 0.0025 0.0026 0.0027 0.0028 0.0029
1/T 1/T
Figure 21: Arrhenius plots from ageing experiment on kraft and Insuldur paper: a) influence
of 3% water and b) influence of oxygen [50]. Values are taken from the initial
slope in the 1/DP plot.

Page 36 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

4.6.3 Ageing of wood

In today’s power transformers, substantial quantities of laminated wood (commonly called
"plywood") are often used, replacing laminated pressboard products mainly for economic
reasons.

Laminated pressboard is produced by laminating individual sheets of high-density pressboard

together with either a synthetic (polyester resin) adhesive or casein glue [51]. In contrast,
laminated wood is produced by joining pieces of wood veneers (typically beech) together
with a thermosetting synthetic adhesive (typically phenolic) under pressure and heat. There
are many grades of laminated wood products available for the transformer industry. These
depend on the density of the wood and the directionality of the beech veneers [52]. Due to the
following parameters, these two materials behave differently:
− Lignin content
− Microscopic structure of the fibers
− Macroscopic structure of the sheets
− Type and amount of adhesive

Despite the composition of these wood products being quite different from kraft presspaper
and pressboard, there has been done very little ageing research with laminated wood in the
past. Yet, a recent study has discovered interesting test results [53]. Two different types of
laminated wood (low and high density) from two different suppliers were tested in parallel
with laminated pressboard at temperatures from 105 to 135 °C, and an ageing time of up to
16 months. Regular testing of the condition of the liquid and solid insulation included
flexural strength, water and acid contents.

To sum up:
Laminated pressboard showed a significantly better ageing performance than laminated
wood:
− Laminated wood can produce aggressive, corrosive acids – explaining problems with
vapor phase systems (corrosion of pipes and valves), as reported repeatedly by the
transformer industry. Moreover, acids accelerate the ageing process of all cellulosic
insulation.
− Solid ageing by-products from laminated wood can clog cooling ducts, in the worst case
causing premature paper ageing due to elevated temperatures.
− The mechanical strength of laminated wood deteriorates more rapidly than the strength of
laminated pressboard.
− The pronounced water production by laminated wood ageing reduces the electric strength
of the oil, and accelerates the ageing process of all cellulosic insulation.
− High gas production of wood at elevated temperatures can cause dangerous bubbling, e.g.
during temporary overload condition, which in turn can lead to partial discharge (PD)
inception. In consequence, this effect limits the thermal overload performance.
− Despite the fact that the laminated woods test samples were all in accordance with the
corresponding IEC specification [52], the materials showed significant differences in their
ageing behavior.
− Another not to be neglected fact is that due to the nature of laminated wood fabrication
(veneer patch work), the inclusion of voids is inevitable. In the case that such voids are
not impregnable with oil because they are completely surrounded by adhesive, PD-
inception at low electric fields is the consequence - with hazard of catastrophic
breakdown early or after years in operation.

Page 37 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

4.7 IEC, IEEE AND CIGRE PUBLICATIONS RELATED TO AGEING KINETICS

IEC 60076-7 (2005) Guide for loading of oil-immersed transformers
IEEE Std C57.91-1995 IEEE Guide for loading Mineral-oil-immersed Transformers
IEC 60450 Measurement of the average viscometric degree of polymerization of
new and aged electrical papers
IEC 61125 Unused hydrocarbon based insulating liquids - Test methods for
evaluating the oxidation stability
IEC 61061 Non-impregnated densified laminated wood for electrical purposes

Page 38 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

5 DEGRADATION PRODUCTS IN IMPREGNATED INSULATION SYSTEMS

5.1 INTRODUCTION
The ageing of paper and board leads to the formation of many different products. All the
main constituents of paper, i.e. cellulose, hemicellulose and lignine, will degrade. Many
different degradation products will be formed. Some are fairly stable, others may degrade
further. Some of the most important degradation products will be briefly discussed in this
section, from different points of view.

In theory, the ultimate degradation products of cellulose are water and carbon oxides.
Obviously, in real life no transformer can be operated up until the point where all paper
components are totally degraded. To be strict, in a real transformer at the end of technical life
the main degradation product is in fact slightly modified cellulose. However, this fills the
same function as the unaged material, and what concerns us are mainly what could be
described as by-products. A degradation product is discussed in this section if it meets one
ore more of the following criteria:

− Its presence may constitute a threat to the safe operation of the transformer
− It affects the ageing rate of the insulating system
− It can be used as an indicator of ageing, or has some other diagnostic value.

Another aspect that deserves to be stressed is that the oil and the solid insulation will age
together. Sometimes a certain product can be formed from either of them. Furthermore, any
degradation product, regardless of its origin, paper or oil, will be found in both (although the
distribution may vary much from case to case). Also, the degradation products of one may
affect the degradation of the other. For these reasons it is not possible, nor meaningful, to
keep at all times a strict distinction between paper degradation products and oil degradation
products.

5.2 DEGRADATION PRODUCTS FROM CELLULOSIC INSULATION MATERIALS

5.2.1 Water
Water is a dominant degradation product of paper and board. It is formed by dehydration
reactions following hydrolysis (which itself actually consumes water), but is also an end
product in the oxidation of oil as well as that of paper. The amount of water formed may
constitute several % by weight of the total mass of solid insulation. Figure 22 shows the
amount of water formed in laboratory ageing experiments, as a function of the number of
chain scissions of the cellulose chain.

Most of the water that is formed is retained in the solid insulation. Typically less than 1% of
the total amount of water is dissolved in the oil. Some examples of equilibrium curves are
shown in Figure 23. The distribution of water between paper and oil is discussed in more
detail in chapters 5 and 7.

Substantial amounts of water may also be the result of ingress from the outside. This,
together with others aspects of water in transformers, has been discussed in detail in the
report of CIGRE WG 12.18 [54].

Page 39 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

3,5 Moser Dahinden

Fabre Pichon
3
Fallou
2,5
Water [%]
2

1,5

0,5

0
0,0 2,0 4,0 6,0 8,0
Chain scissions

Figure 22: Water production during degradation of oil impregnated cellulose insulation. The
curves represent excerpts from several experimental studies [2,11,35].

Figure 23: Some experimental curves for water in oil (Nytro 10X) vs. water in paper
(Munksjø) for different temperatures. Data from study carried out by Spicar,
1972 on unaged paper and oil. These curves do not necessarily apply to other
combinations of oil and paper.

5.2.2 Acids
The hydrolysis of paper produces acids. The mechanisms are fairly well understood from the
study of model compounds [55, 56]. The initial hydrolysis reaction causes scission of the
cellulose chain. Dehydration reactions follow, where 5-hydroxymethyl-2-furfuraldehyde is a

Page 40 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

major product. This substance readily decomposes into levulinic acid and formic acid. It
should be noted that these acids also undergo further reactions. Levulinic acid may form an
acidic polymeric compound (known as “caramel”) while formic acid may decompose into
carbon monoxide and water.

Oxidation also produces acids. There may be free acids formed (e.g. small carboxylic acids),
as well as acidic groups attached to the cellulose molecular chain. Some of the latter are
present in the cellulose already in new paper and board.

Obviously the oxidation of paper (as well as oil) is associated with the presence of oxygen.
The higher the oxygen content the higher the rate of oxidation. In the absence of oxygen the
oxidation rate will be insignificant. Reducing the O2 content will reduce the oxidation rate but
not necessarily in direct proportion to the change of O2 content. However, as already pointed
out, acids can be formed by other routes than oxidation, especially hydrolysis.

Many of these acids will have a low molecular weight thus being volatile and having a low
boiling point. They will also have a high polarity, meaning that they will be hydrophilic,
tending to dissolve well in paper.

5.2.3 Furanic compounds

“Furanic compounds” refers to a whole family of compounds, all of which could be
described as furane derivatives. The most abundant is 2-furfural (2FAL), but 2-acetylfuran
(2ACF), 5-methyl-2-furfural (5MEF), 5-hydroxymethyl-2-furfural (5HMF), and 2-
furfurylalcohol (2FOL) have also been found in oil and paper. Their determination in
insulating oil is described in IEC 61198.

5HMF and 2FAL are formed by dehydration reactions following hydrolysis of the cellulose
and hemicellulose [55, 56]. But perhaps more important, all the furanic compounds
mentioned above are also formed by oxidative pyrolysis [57].

The presence of furanic compounds is not generally considered to influence the ageing of oil
or paper significantly. However, the measurement of furanic compound content, especially
2FAL, has found some use in transformer diagnostics. There is believed to be some
correlation between the degree of polymerizaton of paper and 2FAL (or total furanic
compound) content of the oil. This is discussed in more detail in the chapter on diagnostics.

5.2.4 Carbon oxides

CO and CO2 are ultimate degradation products of all the constituents of paper and board. In
lesser amounts also CO and CO2 are oxidation products of the oil. The oils content of these
gases is always measured in Dissolved Gas Analysis, and the production rates and relative
amounts of CO2 and CO are used in the interpretation of the results. Very high CO2 contents
from paper degradation may influence the acidity of the oil, and may thus complicate the
assessment of oil condition.

Page 41 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

5.3 IMPORTANT DEGRADATION PRODUCTS OF OIL

5.3.1 Acids and peroxides

Although this report deals primarily with the degradation of paper, acids that originate from
the oxidation of oils deserve to be mentioned. There are two main reasons for this. The first
one is that the acids formed from paper and from oil are partially the same compounds. The
other reason is that acidic oxidation products from the oil may influence the degradation rate
of paper. Part of the acidity of aged paper is due to absorption of acids that are the product of
oil degradation. Other polar oxidation products such as alcohols, aldehydes and ketones may
also be partially absorbed.

A multitude of different acids can be formed when oil is oxidized. Mineral insulating oil
contains many thousands of different molecules that are susceptible to oxidation, and (at least
in theory) the number of different acids that can be formed is even larger. They will include
both carboxylic acids and phenols, and range from the simplest possible representatives from
both groups to very large and complex molecules. Likewise they will represent a very wide
range of acid strength. However, it has been demonstrated that the strongest and most polar
acids, such as low molecular weight carboxylic acids, are those that will have the greatest
influence on paper degradation [47].

Some observations from oxidation stability testing of oil may be interest in this context. The
most widely used test method for mineral insulating oil is IEC 61125, method C. In this test
air is bubbled through the oil under test. Any volatile acids formed are carried with the air
flow and trapped in an absorption vessel containing water. The amount of acids in the water,
as well as that remaining in the oil, is determined. The proportions may vary widely, but
usually the amounts of soluble acids (left in the oil) and volatile acids (trapped in the water)
are of the same magnitude. Thus, a substantial proportion of the acids produced by oil
oxidation can be assumed to affect the ageing of cellulosic insulation.

While acids are formed in a relatively late stage in the oxidation process, and significant
amounts of sludge even later, peroxides are among the first oxidation products to appear.
They are very reactive and a key intermediate in the oxidation process of paper as well as oil.

Both acids and peroxides will help mobilize metals (in particular copper) which can act as
catalysts in the oxidation process. Metals can be dissolved as salts (soaps) with acids or
peroxides, but there are surely a wide range of different complexing agents present in mineral
oil that assist in the solubilization of metals.

In inhibited oils, i.e. oil that contains a synthetic antioxidant such as DBPC
(ditertbutylparacresol, sometimes called BHT - butylated hydroxytoluene), the production of
acids and peroxides is very low as long as there is sufficient inhibitor left in the oil.

5.3.2 Sludge
Sludge consists of degradation products that have low solubility in the oil. Sludge may be
present in the form of suspended particles, or as deposits. Most of the sludge is due to
oxidation of the oil, but degradation products from other materials (including paper) most
likely contribute to the total amount. The low solubility of sludge is due to the high polarity,
and possibly to some degree to a high molecular weight, since some degradation products
may polymerize.

Page 42 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

There are analysis methods for the determination of sludge in transformer oil, i.e. in
IEC 60422. Note that the amount of “sludge” determined by such a method may include not
only sludge already present as a solid, but also sludge that is precipitated when the oil is
diluted with a nonpolar solvent (such as heptane). This could be seen as a kind of “potential”
sludge. A rise in oil acidity can also be used as an early warning of sludge formation. For
evaluation of the hazardous influence of sediment and sludge, as well as for estimating the
effectiveness of the maintenance measures, e.g. reclaiming, it is favourable to determine the
type of sludge: chloroform soluble or chloroform insoluble. The presence of chloroform
insoluble sludge (which is formed by further oxidation of chloroform soluble sludge and
points to progressive oxidation processes) can be estimated by washing the glass filter with
the sediment with chloroform (see procedure described in IEC 60422). In case of particles
still present on the glass filter, chloroform insoluble sludge is present.

Large sludge deposits may be a hindrance to oil flow, and thus prevent proper cooling of all
parts of the windings. Sludge is usually strongly acidic and contains peroxides, and has been
shown to promote paper degradation [46]. Since sludge is mainly oxidation products of oil,
preventing the oil from becoming acidic also prevents the formation of sludge. Thus, any
measures taken to prolong life of the oil will also be measures against sludge formation.

Sludge can be at least partly removed by on-site treatment. One example is the use of special
regenerative oils, that have higher solubility for sludge than normal transformer oil. If a
transformer is temporarily filled with such oil, some of the sludge will be dissolved. Oil
reclaiming has the same effect. However, especially in the latter case, the efficiency of sludge
removal will depend very much on oil temperature during the treatment. The higher the
temperature, the more sludge will be dissolved. Sludge present as suspended particles and
“potential” sludge, i.e. not yet precipitated sludge, will of course be removed whatever the
temperature.

5.4 OTHER DEGRADATION PRODUCTS

5.4.1 Hydrocarbons
Even though hydrocarbons in the oil, as measured in the Dissolved Gas Analysis, are usually
attributed to oil degradation, some of the total amounts surely originate from paper
degradation. However, this is normally not taken into account when interpreting DGA results.

5.4.2 Hydrogen
The development of hydrogen, as well as further fault gases like methane seems to be
promoted by the presence of insulating paper in comparison to oil alone [58].

5.4.3 Miscellaneous
Cellulose fibres, although perhaps not a degradation product in a strict sense, may affect the
safe operation of a transformer. Especially at an advanced stage of ageing, when there are
large amounts of particles in combination with a high water content of the solid insulation
and the oil and the dielectric breakdown voltage of the oil may be significantly lowered.
However, this condition is easily corrected, at least temporarily, by reconditioning of the oil
[6].

Page 43 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

Finally there is a multitude of oil-soluble paper degradation products, which may find use in
transformer diagnostics. Just like furanic compounds are beginning to be used today as
ageing indicators, there may be many other compounds, whose concentrations in the oil can
be measured, and that may give useful information about the condition of the solid insulation.
Some of these are discussed in the chapter on diagnostics.

5.5 IEC, AND CIGRE PUBLICATIONS RELATED TO DEGRADATION PRODUCTS

IN IMPREGNATED INSULATION SYSTEMS
IEC 60422 Mineral insulating oils in electrical equipment - Supervision and maintenance
guidance
IEC 60296 Fluids for electrotechnical applications - Unused mineral insulating oils for
transformers and switchgear
IEC 62021 Insulating liquids - Determination of acidity - Part 1: Automatic potentiometric
titration
IEC 61198 Mineral insulating oils - Methods for the determination of 2-furfural and related
compounds
IEC 60599 Mineral oil-impregnated electrical equipment in service - Guide to the
interpretation of dissolved and free gases analysis
IEC 60666 Detection and determination of specified anti-oxidant additives in insulating oils
CIGRE Working Group 12.18 "Life Management", Brochure N° 227, 2003

Page 44 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

6 DIAGNOSIS

6.1 ESTABLISHED CHEMICAL DIAGNOSIS FROM OIL SAMPLES

6.1.1 Water
Water in the insulation influences the life of a transformer in many ways: Accelerating
ageing, increasing losses, reducing insulation strength and introducing the risk of bubble
formation during overload. To avoid premature ageing of the cellulose insulation, the
equipment’s water content must be kept to a minimum when it is being commissioned.
However, changes in the water content during cellulose degradation are inevitable and
substantial (resulting from the cellulose degradation process). A large number of studies have
been conducted on the various phenomena involved in the equilibrium of water between oil
and paper based on temperature. They involve the complex interpretation of the water
measurement in the oil to evaluate the humidity of cellulose insulation [31,59]. The
knowledge of sampling temperature and of oil state (new or aged) is important for the right
evaluation of the humidity in the solid insulation.

6.1.1.1 Dependence between humidity in oil and humidity in the solid insulation.
Theoretical curves – sometimes known as Oommen curves – have been drawn, describing
equilibrium conditions between new oil and solid insulation at different temperatures [10].
Application of these curves presuppose sufficient time to reach equilibrium of water between
solid and liquid insulation. They are theoretically derived from two experimental curves,
describing the water content concentration in paper versus relative water content in oil being
in balance with that of air on one hand (Figure 24a) and water content concentration in oil
versus relative water content of air – again in balance with the relative water content of air on
the other hand (Figure 24b). These curves are in line with earlier direct measurement on
unaged materials as shown by Fabre and Pichon [11] and Spicar (Figure 23). The temperature
dependence is evident and well known. Their applicability for other liquids having different
water solubility is not well studied, but it is known that water solubility depends among other
factors on the aromatic constituents of an oil, so most likely the curves are less general than
one likes to assume.

Equilibrium phenomena are dependent on temperature, thickness and humidity of the

insulation material. The diffusion takes a long time at low temperatures and low water
content contents. Practical experience shows that an equilibrium state requires more than
three weeks of a steady mode of operation [60]. In fact, the diffusion time is even longer,
since thick structures (e. g. laminated wood) require long times to reach equilibrium.

Page 45 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

Figure 24: a) Dependence between absolute water content in paper and relative humidity [9].
b) Dependence between absolute water content in new oil and relative humidity
according to [7].

Figure 25: Oommen curves for water content distribution between new oil and paper.

However, using these curves for aged insulation systems can give large errors: Aged oils
show a significant increase in the water absorbing capacity and the curve Figure 24b is
strongly shifted [60, 61].

Based on that profile, the familiar equilibrium curves (Figure 26) for new oil (continuous
lines) were supplemented by curves for some analysed, aged oil (dashed line). The oil was

Page 46 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

strongly aged (acidity 0,22 mg KOH/g oil and dissipation factor 3,6·10-3). The error by
disregarding changes in oil’s water solubility due to ageing is significant: equilibrium curves
for a new oil indicate water content in the solid insulation between 3 and 5% (red dots),
which would indicate a wet transformer, while, by considering the equilibrium curves of aged
oil, the resulting values in the solid insulation were only 1.5 – 2.2% (black circles), which
were later confirmed by direct measurements on the paper.

Figure 26: Water content equilibrium curves for paper in new (red dots) and aged oil (black
circles). After [60].

Additionally, it is important to know the transformer temperatures during oil sampling to get
a reasonably good assessment of water content in the cellulose insulation.

6.1.2 DGA
Over the last 30 years, the analysis of dissolved gases in oil (DGA) has been recommended as
a good practice for detecting and characterizing incipient faults in the paper-oil insulation;
see Table 3. It has been shown that the amount of dissolved carbon monoxide and carbon
dioxide in oil could be correlated with the degree of polymerization and the tensile strength
of the paper [62]. In 1990, Goto et al, [63] suggested a method to predict the residual life of a
transformer based on the total CO and CO2 generation. However, it should be noticed that
carbon monoxide and carbon dioxide are also – normally in lesser quantities - produced by
the thermal degradation of the oil, layers of paint, varnish and phenolic resins, which are used
as transformer components. Carbon dioxide could also result from atmospheric
contamination especially when an open, free-breathing configuration is used. Experiences
have shown [64,65] that the ratios between CO2 and CO must remain moderate (lower than
10). The IEC standard IEC 60599 [66] suggests CO2/CO ratios above a value of 3 for normal
ageing. Otherwise a fault involving excessive paper degradation can be suspected.

Recently the use of ANN (Artificial Neural Network) has been tested to model the age
estimation of the paper insulation in power transformers with these by-products [67]. These
two diagnostic gases can also be combined with furanic compounds analysis in oil to relate
cellulosic insulation degradation state [68].

Page 47 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

6.1.3 Furanic compounds

Several mechanisms for the thermal degradation of cellulose have been proposed, depending
on whether the predominant step is pyrolysis or hydrolysis of the glucose units [33]. In the
first case, pyrolysis of glucose, the reaction proceeds via levoglucosan (1,6-anhydro-β-D-
glucopyranose) a dehydrated sugar which, after a series of bond rearrangements and further
losses of water and formaldehyde (or hydrogen plus carbon monoxide) yields furfuraldehyde
(2-furfural) among other products. The hydrolysis of glucose units proceeds through the
formation of an epoxide or an enol followed by internal re-arrangements and further
dehydration and elimination of formaldehyde to yield furfuraldehyde and other products.
Figure 27 shows a proposed mechanism.

Figure 27: Cellulose degradation mechanism.

It is noteworthy that water and carbon oxides are the main by-products of the thermal
degradation of cellulose. Furanic compounds and carbonyl derivatives are the second major
degradation products, whereas alcohols, acids, and aromatic and aliphatic hydrocarbons are
minor products [69].

6.1.3.1 Origin
In 1984, Burton et al. [70] suggested that the analysis of furanic compounds (2-furfural, 5-
methyl-2-furfural, 2-furfurylalcohol, 2-acetylfuran and 5-hydroxymethyl-2-furfural)
dissolved in oil could be used to characterize the thermal decomposition of the cellulose
insulating paper at temperatures as low as 100°C.

Since the first work conducted by Burton, a large number of papers dealing with the analysis
of furanic compounds in oil (especially 2-furfural) have been published [71,72]. Techniques
that use liquid chromatography analysis [73] are most often used. Less complex and costly
than gas or liquid chromatography coupled with mass spectrometry, these methods allow
furanic compounds to be detected at thresholds close to the ppb range.

It has been proposed that furanic compounds may primarily originate from the degradation of
hemicellulose rather than cellulose. According to several authors [59,74,75], this could
decrease their potential for use given that hemicellulose is the most fragile part of the paper
and that it is not really representative of insulating paper (it accounts for 10 - 20 % of the
latter’s composition). However, the degradation mechanisms proposed at temperatures
between 100° and 200°C show that 2-furfural (2FAL) should also be generated through the 5-
hydroxymethyl-2-furfural (5HMF) resulting from the cellulose. Table 3 gives the overview of
the studies.

Page 48 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

Table 3: Species already used for paper diagnosis (Oil analysis).

Molecules Appearance Method of References
detection
PAPER

Real transformers Burton et al. (1984) [70]

In the laboratory at 20,80 and 110° C Liquid Unsworth et al. (1990) [71]
Furanic
chromatography
compounds In the laboratory 110-140°C Shroff et al. (1985) [49]
Lab tests and real transformers
A.De Pablo et al.(1997) [73]
( Cigre TF 15.01.03)
Yoshida et al. (1987) [62]
CO and CO2 Gas
In real transformers Goto et al. (1990) [63]
Chromatography
Dominelli (1995) [65]

6.1.3.2 Correlation with the degree of polymerization

The correlation between the paper’s loss of mechanical properties, which are intrinsically
linked to a decrease in its degree of polymerization and changes in some furanic compounds,
mainly 2-furfural, was verified (logarithmic relation established by Shroff and Stannett [49]
in 1986).

In 1996, Hill, Darveniza and Saha [76] attempted to establish a model for generating furanic
compounds based exclusively on the splitting of glycosidic chains which is directly related to
a decrease in the degree of polymerization. In 1999, based on research conducted by CIGRE
Working Group 15.01.03 [73], De Pablo [77] developed a new degradation model. Later
validated by Serena [78], the model suggests that each time three cellulose chains are broken,
one 2-furfural (2FAL) molecule is generated. Correlation between DP and the furanic
compounds evolution obtained in laboratory ageing tests is shown on Figure 28.

The establishment of these relationships is important for interpreting the furanic compounds
but it is clear that the resulting diagnosis will depend on our knowledge of the amount of
paper involved. The non-negligible effect of the water and the acidity of the oil on the
formation kinetics of these compounds did not allow the authors to establish a foolproof
model. The kinetics could also probably vary if the equipment involved operates in a nitrogen
atmosphere or is insulated with thermally upgraded paper [79]. Shkolnik et al [80] also noted
this variation and proposed a new model in 1999 exclusively for those papers.

It is now more clear that a general dependence between the 2-FAL and an average degree of
polymerization of the cellulosic insulation is not likely to exist, but will be dependent on the
operating conditions.

Page 49 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

Figure 28 : Furanes in oil vs. DP. Collected laboratory data [73].

6.1.3.3 Critical thresholds

Although the analysis of furanic compounds combined with DGA analysis presents a certain
level of reliability, it is still difficult to use their threshold values or ratios for diagnostic
purposes due to a lack of knowledge of how these compounds are formed and evolve over
time.

Furanic compounds evolve according to a relative rate which is always about the same, i.e.
mainly 2-furfural (2FAL) (always the main compound), 5-hydroxymethyl-2-furfural (5HMF)
and 5-methyl-2-furfural-2 (5MEF), followed by 2-acetylfuran (2ACF). As for 2-
furfurylalcohol (2FOL), only some authors were able to note major changes in this compound
which could be caused by degradations at higher temperatures [64,81], but this was not
systematically noted in thermal degradation studies. Based on a CIGRE Task Force 15.01.03
study [73] where the furanic compounds absorbed on the paper had also been analyzed,
5HMF was the compound primarily found. However, it appeared to be retained to a far
greater extent than 2FAL on the paper, thus making it less accessible in oil.

In general, most of the authors agree that there is a single critical threshold for equipment of
about a few hundred ppb in 2-furfural. According to a US consultant company, for a
representative population of U.S. transformers, 90% have less than 100 ppb of 2-furfural. In
Europe, the same conclusion was reached in a statistical survey done on more than 5,000
transformers [73].

It should be noted that when European statistics are compared against North American
figures, the levels of furanic compounds that are generated are greater in Europe. This fact
can be related to the predominant use of thermally upgraded paper in North America since
the beginning of the 80’s. Paper can be thermally upgraded by impregnating it with
dicyandiamide (dicy) or by other processes such as the Insuldur method. In this respect, it has
been suggested [64, 82] that these molecules, which are added to the paper to increase its
thermal resistance, could lead to a degradation of the furanic compounds by-products over

Page 50 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

time. Another possible hypothesis is that these compounds are not formed as easily in
upgraded paper.

Finally, one should not forget that in some transformers one uses mixes of upgraded and non-
upgraded paper, and in almost all transformers the pressboards are non-upgraded even if all
paper is upgraded.

Even if some critical thresholds based on laboratory experiments have been proposed [83],
the utilisation of those classifications are still very difficult in the field. A CIGRE statistic on
transformers in service is about to be carried out. This will be the principal aim of the CIGRE
TF D1.01.13 “Furans for diagnostic” which began its activities in June 2005.

6.1.3.4 Partition coefficients between oil and paper

Furanic compounds present in the impregnated insulation system are primarily found in the
paper. Allan [84] proposes some partition coefficients for a few furanic compounds which
would, at least in the case of 2-furfural, not be dependent on the temperature (between 25°
and 90°C). For example the author suggests a ratio of 2FAL concentration in oil to the
concentration in paper of about 0,2 at 25oC. This ratio allows one to suppose that even after
the oil is regenerated (which would eliminate its furanic compounds), the balance of the
furanic compounds in the paper-oil complex would once reestablish the content in the oil,
thus restoring the information concerning the condition of the cellulose insulation. The
coefficients found for the other compounds at 25oC (i.e. for 5HMF, 2FOL, 2ACF and 5MEF)
are 0, 0.03, 0.61 and 0.49, respectively. These coefficients are more dependent on the
temperature.

Myers and Sans [85] suggest that the partition of the furanic compounds between the oil and
paper is affected by the oil’s rate of water saturation. According to these authors, the greater
the amount of water dissolved in oil, the lesser the concentration levels of furanic
compounds. Griffin [86] observed roughly the same phenomenon, and believes that the
division of furanic compounds in the transformers is fairly complex even if it is not
dependent on temperature. He also noted that at 200°C, a damp paper retains most of the
furanic compounds it generates, which is not the case for paper heated in the same way but
previously dried.

However, the furanic compounds would appear to migrate quickly in the oil; at 80°C it would
appear that equilibrium is attained in less than 48 hours.

Pahlavanpour et al [87,88] also observed that the furanic compounds are more absorbed in
paper than in oil. They also noted that the correlation between these specific compounds
detected in oil and the DP value of the insulating paper is not easy to establish since they are
depending of a lot of factors such as temperature, type of oil, type of paper, transformer
design, and so on.

To refine the ageing models using the furanic compound measurements, it will be crucial to
find the partitioning of furanic compound between paper and oil as well as the parameters,
which could influence them. Some recent works [89,90] showed that the ratios paper/oil of
furanic compounds are:
- Greatly influenced by the humidity of the paper (which is related to the water content in
oil).

Page 51 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

- Less influenced by temperature unless it is in conjunction with an acidic oil (0,16 mg

KOH/ g oil).
- Different for each furanic compound (particularly between 5HMF and 2FAL).
- Also influenced by the type of paper.

As an example, the

Figure 29 shows the 2-furfural partitioning between paper and oil in different conditions.

40

35

30
in new oil with dry paper
in acidic oil with dry paper
Partition (paper/oil)

25
in new oil with 5% paper humidity
20
in acidic oil with 5% paper humidity

15

10

0
20 30 40 50 60 70 80 90 100

Temperature (° C)

Figure 29: 2-Furfural partition between paper and oil versus temperature.

6.1.3.5 Stability
There is still no consensus concerning stability of the compounds above 110°C. For instance,
Griffin [91] believes that the presence of water in the transformer could degrade 5HMF.
Allan [84], for his part, suggests that the furanic compounds are stable at operating
temperatures in the absence of any oxygen but after 8 weeks at 110°C when oxygen is
present, he observed a non-negligible loss of 2-furfural (20 to 40%) and a substantial loss of
the corresponding alcohol (40 to 100%). Lutke et al. [92] observed the same kind of
instability in presence of oxygen, even at operating temperature as low as 60oC.

In 2000, Emsley et al. [93] made roughly the same observations: with the exception of 2FOL,
the furanic compounds are stable up to about 140°C. However, in his opinion the lack of
stability of 2FOL would be an asset since its presence could only be attributed to a very
recent fault. The team headed by S.D. Myers [85] even suggests that the four furanic
compounds normally found at very small concentration levels than 2-furfural appear to be
transient species and would not be found during a transformer’s normal ageing process.

As part of another study, Emsley [31] suggests that in the case of transformers undergoing
normal ageing, i.e. without any thermal faults, the presence of furanic compounds could
mainly be attributed to hydrolytic degradation. This leads one to assume that at low
temperatures their development could be attributed to a greater extent to the acidity or water
content in the oil rather than the thermal degradation of the cellulose insulation. Scheirs

Page 52 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

[59,75] made the same observations when he noted furanic compound levels that were much
higher where water content tends to accumulate than elsewhere in transformers.

6.1.3.6 Establishing ratios between the generated compounds

Numerous authors have attempted to establish ratios between the various furanic compounds
in order to determine the default temperature involved and to help determine the equipment’s
lifespan. In 1988, Burton [94] attempted to determine the fault temperature between 140° and
350°C by using a 2FAL/5MEF ratio. In 1990, a CIGRE working group [95] also proposed a
2FAL/5HMF ratio which would allow the fault temperature to be determined between 60°
and 130°C. In 1992, Grant [96] proposed a 2FAL/2FOL ratio, while in Japan, Nanba et al.
[97] established a relation between 2FAL and 5HMF that appeared to be highly linear. In
2001, a study by Soares et al. [98] showed that the 5HMF/2FAL ratio doubles between 80°
and 120°C.

6.1.3.7 Limits of furanic compounds

Furanic compounds certainly give information about the paper degradation. But this is
possibly all that can be stated. Years of experience show us that their evolution is dependant
on test conditions or on a transformers operating condition. Recently, the effects of water,
oxygen and acidity upon ageing and type of paper have been studied [41, 47,50,92,58].

Moreover, work done by Mulej et al. [99] allowed to understand why many workers find
much lower concentration of 2- furfural in transformer oil samples as compared to results
from laboratory ageing test for the same DP. According to them, type and quality of oil
influence not only the speed of ageing and production of furanic compounds but also their
quantity, stability and solubility in oil.

Lutke et al [92] even concluded that due to these limitations, it will not be possible to predict
the remaining life of a transformer solely based on the content of furanic compounds.

6.2 POTENTIAL NEW CHEMICAL DIAGNOSIS FROM OIL SAMPLES

Several papers were published in the 1960s [100,101] on the use of gas chromatography for
measuring compounds resulting from the thermal degradation of cellulose. These
experiments allowed a large number of compounds to be identified, including levoglucosan,
formaldehyde, acetaldehyde, acetone and acetic acid.

In the early 1990s, Nick Dominelli [65,72] began a wide study on the possibility of detecting
new compounds, both volatile and oil-soluble, resulting from the degradation of insulating
paper. Table 4 lists the compounds, which have been observed thus far in laboratories or from
in-service transformers.

6.2.1 Sugars
Many studies on paper decomposition showed that levoglucosan (1,6-anhydro-ß-D-
glucopyranose) plays a key role in the thermal degradation of paper. Emsley [33] showed that
this compound could be a key intermediate product in the degradation of cellulose in power
transformers, which could lead directly to the formation of furanic compounds.

Page 53 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

Scheirs et al. [59,75] anticipated and observed the creation of monosaccharides and
polysaccharides during these processes. They also proposed some mechanisms to explain the
formation of 2-furfural. According to these authors, 2-furfural results mainly from the
decomposition of hemicellulose and, to a lesser extent, cellulose decomposition via the
formation of levoglucosan. Unfortunately, this anhydrosugar is not very soluble in oil.
However, in 1998, Saiid Husain [102] was able to determine its presence in in-service
transformers. The technique used involved performing an aqueous extraction followed by
liquid chromatography analysis coupled with a refractive index detector. Soares et al. [98]
also noted the presence of levoglucosan, but this time at temperatures over 250°C.

Lessard et al. [103,104] followed the thermal degradation of insulating paper in a laboratory
ageing test by using High-Performance Ion Chromatography (HPIC) for measuring specific
carbohydrates such as levoglucosan and cellobiose extracted from paper. Changes in these
sugars could be linked to the loss of DP.

Vergne [74] also noted the presence of a sugar, arabinose, in the oil of an actual transformer.
He gave up on the analysis since he was not able to optimize the oil extraction (aqueous
extraction). The analysis of aqueous extractions of paper obtained from models aged at
100°C and 140°C allowed the detection of four sugars, i.e. arabinose, galactose, glucose and
xylose. It is interesting to note that the only compounds, which Vergne was able to observe at
100°C were HMF and arabinose.

The possibility of following potentially oil-soluble carbohydrates, alcohol carbohydrates or

organic acids that could be formed in the cellulose degradation process has not been ruled
out. The major challenge is to find an appropriate and efficient extraction technique to detect
them even in small concentrations in the oil before the furanic compounds are formed.

6.2.2 Organic acids

Koreh et al. [105] used liquid chromatography coupled with mass spectrometry to determine
the presence of 3-furoic acid methyl ester. An interesting fact is that it seems to develop at
concentration levels on par with 2-furfural. The 2- and 3-furoic acids where also identified on
a few occasions [65,102].

Vergne [74] observed the development of various organic acids in oil. These consist of
levulinic, formic, acetic, lactic, carbonic, glycolic and furoic acids. According to the author,
the presence of the three first acids would appear to be mainly caused by the oxidation of the
oil. According to Lundgaard et al. they could also be formed by reactions following
hydrolysis [41,55,56].

6.2.3 Volatile compounds

Soares et al. [98] were able to observe in the laboratory a substantial list of volatile
compounds, with the main ones being acetone, hydroxypropanone, butenal, butadione and
butanone.

Acetone is another product which was observed by a few authors (Dominelli [65], Awata
[106] and Keiichi [107]) both in the laboratory and in a transformer. This small volatile

Page 54 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

molecule, which is formed in negligible quantities in oil, is highly soluble in oil. In addition,
Awata et al. were able to observe the molecule’s appearance in oil well before that of 2-
furfural and, what is reassuring, they were able to note a good correlation between its
development over time and the paper’s DP. Some authors already perceived that the molecule
had excellent potential. Keiichi et al. developed a semi-conductor sensor capable of
continuously monitoring changes in acetone.

Dominelli [65] was also able to observe the presence of various molecules of methyl sulfides
where dimethyl sulphide was the major product. These molecules could originate from the
residual lignin of the insulating paper sulphonated during the pulp preparation process. In this
respect, these volatile and very specific molecules could prove to be significant witnesses to
the cellulose degradation process.

The thesis work of Vergne [74] also helped determine the presence of methyl ethyl ketone
and phenol.

6.2.4 Furanic analogous compounds

According to Dominelli [72], Wodley [101], Grant [96] and Lessard [108], the thermal
degradation products of insulating papers and pressboards are not limited to the commonly
studied furanic compounds. In the past few years, unknown peaks most likely resulting from
species issued from paper decomposition were observed by several authors [81,109,95] in
chromatograms obtained by High-Performance Liquid Chromatography (HPLC) using a
reverse-phase column. The UV spectra and polarity of the compounds found indicate that
they are of the furan family. However, additional work is still required to identify these peaks
and to link their origin and evolution to the paper degradation.

6.2.5 Phenolic resin degradation

Phenolic resins are present in laminated wood as well as in resin impregnated paper. In 1991,
Dominelli [109] observed the presence of phenolic compounds (phenol, m-cresol, o-cresol
and dimethyl phenol 2,3 and 2,6 and 2,3,5, trimethyl phenol) in power transformers and
linked their presence to the phenolic resins. Vergne [74] identified the same family of
molecules in French transformers. These compounds could be very useful for the overall
characterization of a transformer’s insulation.

Page 55 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

Table 4: Potential new markers in oil samples.

Molecules Appearance Method of detection References

FROM PAPER AGEING

Levoglucosane In the laboratory at T DSC (TVA) coupled with gas Husain (1998) [102]
>250°C chromatography/mass
spectrometry
Transformers Aqueous extraction, liquid Soares et al.(1995) [98]
chromatography/ refractive index
detector
Organic acids Ageing model at 140°C Ion chromatography Vergne (1992) [74]
and transformers

Sugars Ageing model at 140oC Ion chromatography Vergne (1992) [74]

Ageing model at 150°C Ion chromatography Lessard et al. (2003)

[103,104, 108]
Acetone In the laboratory at 120°C SN02 sensor Keiichi et al.(1994) [107]
In the laboratory at T DSC coupled with gas chrom- Soares et al.(1995) [98]
>250°C atography/mass spectrometry

Japanese transformers Not available Awata et al.(1997) [106]

In the laboratory at 150°C Gas chromatography Dominelli (1993) [72]

Dimethyl sulfur Laboratory at 150°C and Gas chromatography Dominelli (1993 ) [72]
160°C (pressboard and
thermostabilized paper)

Furanic analogous Transformers Liquid chromatography Burton et al. (1984) [70]

polar compounds Grant(1992) [81]
Lessard (2003) [108]
Methyl ethyl ketone Transformers Gas chromatography coupled Vergnes (1992)[74]
with mass spectrometry

3-furoic acid methyl Transformers Liquid chromatography coupled Koreh et al. (1998) [105]
ester with mass spectrometry

FROM AGEING OF PHENOLIC RESINS

Phenolic compounds Transformers Methanol extraction, HPLC/UV Dominelli(1991) [109]
analysis at 215 nm
Gas chromatography coupled Vergne (1992) [74]
with mass spectrometry

Page 56 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

6.3 MEASUREMENT OF THE DEGREE OF POLYMERISATION OF CELLULOSIC

INSULATION

6.3.1 Context
It is possible to characterize the state of degradation of the insulating paper used in power
transformers by measuring its degree of polymerisation (DP). This method is used to
determine the average number of glucose units chemically linked in a cellulose molecule.
During cellulose degradation, the degree of polymerization decreases from about 1200 units
for new paper to as low as 150 units for paper that is highly deteriorated, brittle and friable.

The loss of mechanical strength of the paper insulation in transformers is directly caused by
breaks in chains in the cellulose.

6.3.2 Sampling

6.3.2.1 Introduction
The quality and precision of the assessment related to this test primarily depends on the
quality of the sampling. The DP is a localized value (only a few mg of paper are used) and
can vary substantially (up to twice its value) depending on the location of the sample in the
apparatus. A variation in the DP can even be observed for different layers of the sampled
paper.

It is crucial to have very good knowledge of the design of the equipment involved.

6.3.2.2 Basic rules

In general, the sample taking should follow a few basic rules:
- At the winding subject to the highest thermal stress: to know the ageing state of an
apparatus, the sampling can be limited to areas subject to the most potential damage.
- At the maximum local temperature: this will allow the paper that is presumably the
most degraded on the winding to be characterized. For most designs, the temperature
will be higher at the top of the winding than at the bottom.
- At the minimum voltage: this location must be considered if the winding is to be
restored after the sample is taken (to avoid making it prone to dielectric problems).
- Reference sample: It is very important to compare the results obtained with a realistic
reference. A paper sample obtained from the area that is subject to the least stress in
the equipment under study will allow for a more realistic comparison than with new
paper that has never been subjected to any ageing. Whenever possible, samples should
be taken from a conductor that is accessible e. g. from the tertiary winding.
- Amount of paper required: To facilitate laboratory work (e.g. sample preparation,
replicates), ideally 1 gram of paper is required; for instance, this corresponds to about
three pieces of impregnated conductor paper roughly 20 cm in length.
- Scope of application: Since this test is intrusive (the sample is destroyed), it cannot be
performed on all the transformers received at the shop. In addition, unless the
equipment is being de-tanked, the windings cannot be accessed. However, samples
may be taken from insulated high voltage leads accessible from inspection hatches at

Page 57 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

the top of the transformer tank as an indication of the degree of polymerisation in a

hot region of the transformer [ 110].
- The samples must be taken carefully; it is possible that part of the winding paper must
be repaired after the sampling so that the windings are not affected.

6.3.2.3 Inspections or repairs of equipment

In case of inspections or repairs of equipment that will eventually go back in service,
adequate samples are easier to obtain. A visual inspection of the windings will help in this
process. Sample-taking will be limited to a minimum number of samples and should be
strategic, i.e. it should be representative of the general condition of the equipment or winding
involved. In fact, if the inspection shows that the cellulose insulation has obviously not
considerably aged, the test should not be done. However, such a test will be very useful in
specific cases where the residual lifespan of a winding is potentially at risk; should a winding
be kept or replaced?

6.3.2.4 Insulating paper life monitoring

Planting of several paper samples could be done in new transformers to monitor the residual
life of the insulating paper [111]. A first sample should be sent for analysis after the factory
testing as an initial DP benchmark. This first DP, taken after the dry out, should be around
1000 or higher. These specific samples should be mounted and accessible from the top of the
equipment (it will only need partial oil-drain, but no de-tanking). From such paper samples
also the water content and the level of contamination (acids, furanic compounds, etc.) the
paper will have in mutual exchange with the oil may be found. Experience has shown
uncertain correlation with winding ageing due to different temperatures, put possibly the
method may be used if one “corrects” for temperature differences using the earlier described
formulas.

6.3.2.5 Micro-sampling of paper

The possibility of sampling a very small piece of paper (not much more than a piece of a few
cm²) from transformers in-service has also been experienced with satisfaction by a utility.
This sampling which should be carried out with an accurate technique (the insulation
integrity shall not be affected), combined with a furanic compounds and a dissolved gas
analyses (DGA), should lead to a better insulation life assessment [110].

6.3.2.6 Post mortem inspections

When transformers to be scrapped, it takes little additional effort to gather samples of the
solid insulation for the determination of its ageing condition. For a comprehensive ageing
replica it is necessary to take samples systematically throughout the transformer.
- compare real ageing with laboratory accelerated ageing experiments.
- compare predicted ageing (temperature profile!) with real ageing of different solid
insulation components.
- verify the cooling performance of the transformer active part.
- detect of unintended, hot spot areas, design flaws.
- identify the location of the true hot spot(s).

Page 58 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

- count back of initial DP values of different solid insulation components (when the
transformer was put in service).
- assessment of the condition of similar transformers still in operation.

6.3.3 Test procedure

The determination of the degree of polymerisation shall be carried out in accordance with
IEC 60450. For a well-performed DP measurement repeatability would typically be 2 % and
reproducibility will be about 10 % [112].

The viscosity of a solution of macromolecules is dependent on the molecular weight at a

given (low) concentration. This phenomenon is used for the determination of the DP of
cellulosic insulating materials.

A small amount of deoiled and fluffed paper or board (50 to 100 mg) is dissolved in
Cupriethylenediamine (Cuen). The viscosity of the paper solution is determined using a
capillary viscosimeter. From this result the degree of polymerization is calculated using
experimentally established equations and constants that can be found in the relevant standard.

6.3.4 Results from post mortem analysis

In the years 2003 to 2005 post mortem inspections were made on three power transformers
taken out of service [113]. On the example of one of these units the interesting and partly
unexpected findings shall be illustrated:

- Voltage: 380 / 220 kV

- Phases: 1
- Rating: 600/3 MVA
- Cooling: OFAF
- In service: 1962 - 2005

The single-phase unit consisted of two wound main limps and two side limps. The tap
winding was in a separate tank. HV and LV winding were disk type, tertiary winding was
helical type. All the solid insulation was made of pressboard, hot press dried for cylinders,
spacers and washers, laminated for strips and spacing blocks and low density for moulded
parts such as angle rings and caps. All the material surfaces were in very good condition and
did not show traces of ageing. It was possible to take as many as 300 samples, 169 of which
have been analyzed. In Table 5, the measured DP values are summarized.

Page 59 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

Table 5: Measured DP values from the insulation of a 600 MVA transformer that has been in
service for more than 40 years.
Insulation Bottom Middle Top
End LV Angle ring 882-1089
HV Angle ring 482-973 488-912
Spacer 620-753
Inter LV/Tertiary Cylinder 713-917 724-1000 676-824
winding Strip 886-1169 795-1083 791-1063
HV/LV Cylinder 841-1082 818-1083 613-975
Strip 724-1182 659-1067 765-935
HV/HV next Cylinder 853 811-820 758
limp Strip 811 735-995 731
Winding Tertiary Paper 655-742 559-599
Spacer 761 640-728
LV Paper 634-651 624-758 580-608
Spacer 643-793 608-792 585-690
HV Paper 676-798 645-743 493-556
Spacer 751-776 640-739 582-722
Lead exit 220 kV Paper 891
380 kV Paper 813-853
Stress ring LV Paper 938 835
HV Paper 683 619

The locations and the values of the measured DPs are depicted schematically in Figure 30.
The horizontal, radial DP profiles at three levels across the windings are depicted in Figure
31.

The cellulosic insulation of this transformer was in very good condition. Even the most aged
parts showed a DP of 500. The highest values were found for the end insulation underneath
the LV winding and the cylinders of the main gap. In summary the following statements can
be made for this transformer:

- The insulation of the copper conductor is more aged than the insulation between the
windings.
- The spacers in the windings are more aged than the adjacent conductor paper.
- Pressboard cylinders are less aged than the strips in physical contact.
- Strips and cylinders next to the windings are significantly less aged than in the middle
of the Hi-Lo insulation.
- Hot spot at top of the high voltage winding.

Page 60 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

Tertiary LV HV
#

#
#
#
#
#
#
#
#
#
#
#
#
#

> 900
801-900
701-800
601-700
501-600
401-500
Core

301-400
# # # # # # # # # # # # # # # # # # # # # # # # # 9 8 7 6 5 4 3 2 1
201-300
< 200

Windings

# #
# #
# #
# #
# #
# #
# #
# #
# #
# #
# #
# #
# #
#
# #
#

Figure 30: DP distribution, schematic, sectional view of transformer, between limbs.

1200

1000
Degree of Polymerization

800
bottom
600 middle
top
400
Tertiary LV HV
200

Figure 31: Radial DP profiles of transformer, between limbs.

Page 61 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

6.4 DIELECTRIC DIAGNOSIS OF SOLID INSULATION

Dielectric measurements are standard methods for assessing the condition of a transformer or
of the oil. Insulation power factor (IPF) or the equivalent measurement of tan δ is performed
at ac power frequency. If these values are too high it is an indication that the insulation
system is contaminated [114]. The polarization index (PI) is the ratio of the insulation
resistance measure after 10 minutes to the dielectric resistance measured after 1 minute using
a dc voltage; this is used for the same purpose. A PI below 2 indicates insulation
deterioration.

Lately, the so-called dielectric response methods have got a considerable attention as a
possible diagnostic method [115,116,117,118]. Simply put, this is measurement of the
complex impedance or admittance of the transformer insulation in e.g. the main gap between
the high voltage and low voltage winding over a wide (and low) frequency range (e.g. 10-3 Hz
to 103 Hz). Tan δ (or IPF) at power frequency is then only one point at these curves. The
dielectric response can be measured in several ways [119]:
ƒ Measurement of polarization and depolarization currents after application of stepped DC
voltages, called polarization and depolarization current measurement (PDC)
ƒ Measurement of currents and phase sag using a swept frequency sinus voltage, from which
capacitance and tan δ can be calculated. This is called frequency domain spectroscopy
(FDS).
ƒ Measuring the voltage increase on an ungrounded transformer after it has been polarized
for a certain time; the so-called return voltage measurement (RVM).

One can - with certain restrictions – by calculation transpose results from one to the other
measuring method (e.g. Fourier transformation). The insulation is considered as a capacitance
with oil, paper and pressboard in series and parallel. All methods depict how the full
windings behave; e.g. results are influenced by the average condition and it is therefore
impossible to distinguish behaviour from the upper and lower part of the winding. The
measurements are influence both by the oil and the solid, and interfacial polarization
(Maxwell-Wagner effects) is important. Gäfvert has described this in detail, together with the
importance of the geometry and design of the winding [120]. Cigré TF 15.01.09 showed in
their work how the response of a transformer was influenced by the conductivity of the oil,
and that this had confused the conclusions from the interpretation rules as given for the
RVM-method [119].

It is possible to separate the influence of the oil and the solid in a response measurement. The
oil has a frequency independent permittivity and a conductivity that depends on composition,
contamination and temperature. The conductivity can either be measured directly (IEC 61620
- Determination of the dielectric dissipation factor by measurement of the conductance and
capacitance) or read out from response measurements at either short times in the time domain
measurement or high frequencies in the frequency domain measurement. The response of the
solid is found for the longer relaxation times in a time domain measurement and for the lower
frequencies in the frequency domain measurements. Jonscher describes response
measurement and analyses for dielectric materials generally and in detail also on granular and
fibrous materials in his two books [121, 122]. From response measurements at different
temperatures one can design a response master curve for a pressboard [115]. This master-
curve varies with water content, and promising techniques for revealing the water content of

Page 62 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

cellulose from response measurement are proposed [120, 119].

One hopes to establish similar techniques for revealing the ageing condition of paper and
pressboard as those proposed for its humidity. However, the techniques here are not even in
their infancy. Oil impregnated pressboard is in many ways similar to the transformer
windings; it consist of multiple interfaces between cellulose fibres with oil in the open spaces
between. The permittivity and conductivity will be influenced by acids, and polar substances
formed by the ageing of the cellulose. How and if the impact of the ageing substances bound
to the paper fibres and dissolved in oil are separable from the influence of water only is an
open question. Here one should not forget that there exist different qualities of paper and
board. For example should the conductivity of aqueous solutions from cellulose vary, then
the inherent response may vary between paper or board types.

Recent results from laboratory ageing of kraft paper indicate that the effect of ageing on the
frequency domain dielectric response is much smaller than the effects of water content [123].

Page 63 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

7 LIFE MANAGEMENT OF TRANSFORMERS

7.1 MAIN QUESTIONS REGARDING CELLULOSE INSULATION

We have thus far treated the ageing of cellulose insulation in transformers and the
possibilities of monitoring the condition by various diagnostic methods. The next question
follows quite naturally: how can the degradation be avoided or reduced? Of course reducing
temperatures by good design and cooling, and use of good materials to get a better
performance is important. This however, is a question relating to specifications and designs
of the full transformer and to a large degree outside our scope. We are here mainly concerned
with the performance of cellulose insulation itself in the chemical environment of the
transformer. Given this as a restriction, we can limit ourselves with the following questions:
• How can ageing accelerating substances in the materials of the transformer be
avoided?
• How can one avoid occurrence of ageing accelerating substances in the transformer
during service?
• How can ageing accelerating substances be removed from a transformer, and when in
the transformers life cycle is that a profitable project?

Mainly three factors; design and materials, contaminants in the insulation system and
operational conditions, will determine the ageing of a transformer. For the solid insulation –
paper and pressboard - it means a combination of mechanical and dielectric performance,
which are interlinked and synergetic. The mechanical performance of cellulosic insulation is
given in terms of mechanical strength (e.g. tensile index) or degree of polymerisation, which
are strongly influenced by ageing.

Paper ages more quickly under the influence of water and low molecular acids increasing
hydrolysis, and from oxygen, free radicals and metal ions formed under reactions with free
radicals (Fe++, etc) contributing to oxidation. Consequently, these substances should be
extracted or their formation should be prevented in order to reduce ageing.

7.2 REMAINING LIFE ESTIMATION

If the present condition of the insulation system and the operational conditions of the
transformer is known this allows for a rough estimation of life expectation. Such estimation
should be based on the hotspot temperature of the transformer, known from design criteria
and load conditions. The critical parameter is believed to be the mechanical strength of the
paper. Once the paper cannot withstand the shear stresses in the winding during a high
through current transient (e.g. short circuit), the conductors will move. If breakdown does not
immediately occur, a change in the geometry will increase future stresses and increase
possibilities for further development of the defect. Short circuit stresses occur during
lightning strokes, breakdown of nearby equipment, from unintentional energizing with earth
apparatus connected, and from connecting generators in phase opposition. Frequent large
inrush currents may also play a role. The magnitude and frequency of short circuit stresses
must be considered when analyzing the need for scrapping a transformer. A transformer in a
cable network does for example experience fewer short circuits than a transformer connected
to overhead lines with frequent lightning strokes.

Page 64 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

Two schemes for assessing ageing are possible:

• Prediction of reduced mechanical strength using equations for strength development
[12].
• Testing or prediction - based on algorithms for ageing kinetics - of DP-value in paper
and correlating that to tensile strength.

In any case it is the responsibility of the transformer owner to choose a criterion for end of
life. Often a DP value of 200 is recommended, a value that according to the figure shown
below will indicate a 30% remaining tensile strength.

According to IEC loading guide, that to a large degree is based on Montsingers work [14],
ageing rate doubles (or life is halved) by every 6-8 oC temperature increase. This essentially
describes an Arrhenius relation where T is the absolute temperature. In the loading guide this
is simplified and expressed as a relation with temperature valid within 80-140oC

Rate of ageing = constant×epθ

where p describes how sensitive the ageing is to temperature, which in the IEC standard is
expected to be a doubling for every 6oC, and θ is the hotspot temperature of the transformer
in degrees Celsius. The Constant describes the influence of the environment on the change in
functionality (e.g. tensile strength) and is usually ignored.

The base of all this is a definition of a material test where the criterion is that paper aged
under a certain condition (e.g. THot = 98oC) should not loose more than a certain percentage
of its property within the period. This scheme is used for design and for material testing
before manufacturing. The disadvantage for service aged transformers is that it is difficult to
take samples that are suited for checking tensile strength from a service aged transformer.

Figure 32 shows how tensile strength of kraft paper is reduced with degradation of the
cellulose polymer chain.

Figure 32: Correlation between tensile index and DP-value for non-upgraded kraft paper.
Based on results from [41].

Page 65 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

The chemical degradation of the cellulose can – as explained above - with some confidence
be described by Ekenstams and Emsleys formulae (see chapter 3.3):
1 1
− E

* e [ year ]
DPt DP0
Expected Life = RT

A * 24 * 365

DP0 is in the range 1000 – 1250 and DPt is defined by the user from his preferred margin of
safety for the transformer. E is the activation energy, R the molar gas constant, T the hotspot
temperature and A an environmental coefficient as explained above.

One can in principle assess ageing progress by sampling as one may get valid samples from
transformers after opening. However, a major uncertainty lies in how to link ageing at an
accessible location to that of the critical hotspot region. One may with some confidence
assume similar E and A values in the full winding, or adjust the A value for expected
deviations in water content. For this one would need temperature models for the transformers.
Again one should realize that expected low DP of insulating paper will seldom be the only
reason for taking a transformer out of service.

7.3 NEW TRANSFORMERS

7.3.1 Quality of materials.

The most materials used in the transformer manufacturing are “traditional” materials, used
for decades. The low content of conductive contaminants in the insulating material is very
important and specified in numerous IEC standards. The factory drying (e.g. down to 0,3 -
0,5%) has the goal of removing water and low molecular acids from paper, pressboard,
laminated wood, thus providing a high dielectric strength.

The insulating oil should – as detailed in chapter 5 - be stable against oxidation as well as
against corrosion. Oxidation of oil produces peroxides and water soluble acids which as polar
substances are selectively bound to cellulose, thus contributing to its degradation. Cellulose
can be oxidized directly: through air dissolved in oil or through oxidation products in the oil –
free radicals and peroxides. Also bare metal surfaces will contribute to formation of free
radicals, and take part in ageing. Therefore metal passivators may be used to prevent this
effect. The oil degradation products are of polar nature and may further deteriorate into
carboxylic acids. Some information about the types of acids formed is available from
standardized oxidation tests of oils, where high molecular weight acids are measured in the
oil and low molecular weight acids are measured in the water trap on the exhaust of the
ageing cell. Furthermore the oils tendency to form sludge may influence the cellulose through
reduction of cooling and also chemically.

Choosing either kraft or thermally upgraded paper will of course influence ageing rates. As
earlier explained, upgraded paper is less sensitive to acid catalyzed hydrolysis, but about
equally susceptible to oxidation as kraft paper is.

7.3.2 Transformer design

The most important choice is probably the operating temperature of the insulation system,
which normally is decided from a criterion of normal life expectancy under nominal

Page 66 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

conditions and hot spot temperatures of either 98 oC for normal paper and 110 oC for
upgraded paper. In principle this is based on a criterion for a reduction in tensile strength of
not more than 50 % over the transformer life. Using upgraded paper with a 98 oC hotspot
temperature would give an increased life expectancy. This criterion is probably far too
conservative as many transformers operate well with far lower strength of the paper as
discussed in the IEEE loading guide [13]. These considerations are valid for design and are
based on ageing tests of paper under dry and oxygen free condition. In service, conditions can
be far worse and an accelerated ageing must be expected with increased water content,
oxygen and degradation by-products content.

Next to design and materials, the choice of protection from ingress of ageing accelerating
substances like water and oxygen will determine the progress of ageing of the transformer. In
the US and many other countries most of the transformers are delivered with a sealing system
based either on a nitrogen cushion or a rubber bellow in the expansion tank. This will prevent
ingress of water and oxygen, but also keep any water and easily evaporable acids inside the
transformer.

In Europe, many transformers are open breathing, with silica gel water absorbers at the
breathing inlet of the expansion tank. To have well dimensioned absorbers is vital. With open
breathing systems equipped with such absorbers one may expect that some of the water
formed in the insulation will be absorbed by the silica gel. However, experience shows that
some European transformers may become quite wet (3-4%) so this effect cannot be very
significant. Possibly water content loss from the insulation will depend on the operating
temperature. Furthermore, the maintenance of the dehumidifiers is essential, especially at
transformers working with a variable load, thus actively “breathing”.

Strongly oxidized oil can also represent a source of humidity, especially at higher operating
temperatures.

An open breathing transformer will operate with an insulation in balance with the gasses in
the atmosphere and an oxygen concentration up to 30 000 ppm in the oil is normal. The
correlation between ageing rate and oxygen concentration has not been extensively
investigated, but Lampe’s study [124] suggests that one needs to reduce the oxygen content
from around 30 000 ppm down to below 2000 ppm to significantly reduce the effect of
oxidation of a transformer, contrary to oxidation being linearly dependent on the oxygen
concentration as often assumed.

7.3.3 Condition control

Here we describe some common on-line methods for extracting water, oxygen and acids out
of the transformer.

Partial degassing
Small vacuum pumps for continuous use in an oil circulation circuit are commercially
available. This technique mainly aims at a deoxygenation of the transformer oil and cellulose.
Which efficiency is needed to arrive at a certain concentration of e.g. oxygen depends on the
size of the transformer, and whether or not it is open to surrounding air. Additionally they
will remove some water. For humidity control, such pumps may be debatable; they are not
very efficient (efficiency again depends on the temperature of the oil and the water vapour

Page 67 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

pressure in the degassing unit), but may over time help keeping generated water content away
from cellulose. There is also a possibility that volatile acids may be extracted from the oil.
However, continuous degassing, may interfere with DGA analysis, reduce its sensitivity or
change the basis for its interpretation.

Cold traps
To keep water content down, cold traps connected to the conservator tank are being used.

Molecular sieves or cellulosic absorbents

Traps in the oil circulation path can be used to extract water content selectively [125,126].
These can either be molecular sieves or tanks with pressboard kept at a reduced temperature.
Molecular sieves can absorb water irreversibly. Since they have a potential catalytic activity,
it will be important to investigate the compatibility with oil, particularly at elevated
temperatures, as well as their influence on DGA. In case of cellulosic absorbers, the influence
on the diagnostic parameters, e.g. furanic compound must be considered. Also, their capacity
depends on the weight of absorber material, and regular replacement is necessary.

7.4 TREATMENT OF SERVICE AGED TRANSFORMERS

7.4.1 General
Here we will consider any action that singularly, periodically or continuously is introduced to
slow down the ageing processes of the cellulose insulation in a transformer and extend its
service life. The actions that will have an impact on the ageing can be split into three
categories:

ƒ Reduction in service temperatures that will reduce ageing according to the influence
temperatures have on the kinetic processes as discussed above. This will influence
ageing of both oil and cellulose.
ƒ Oil maintenance [6] can be grouped under periodic actions like degassing,
reconditioning (removal of particles and water content), reclaiming (removal of
substances from ageing of oil and cellulose) by chemical absorption and finally
replacement of the oil. These processes may also improve the condition of the
cellulose.
ƒ Cellulose maintenance by reducing ageing accelerating contaminants from the
cellulose. Normally this involves factory or on-site drying processes like vapour phase
drying, hot-oil spray, winding heating, vacuum treatment promoting evaporation and
transport of e.g. water and low molecular weight acids.

The efficiency and benefit of a condition management or maintenance technique will depend
on the extent of ageing of the transformer and the possibility for this technique to improve the
condition of the cellulose insulation and thereby retard the ageing.

For a new transformer in good condition one can retard ageing by keeping oxygen away from
the oil and paper by e.g. rubber bellows and degassers to avoid oxidation or by keeping the
water concentration in the paper low by dehumidifiers as described above. If the ageing rate
is kept low, the need for future actions like oil reclaiming or drying of windings will occur
later.

Page 68 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

1000

800
Maintenance

600
DP-value

400
Life extension

200

0 20 40 60
Years in service
Figure 33: Simulation of maintenance action of cellulosic insulation and its effect on
retarding ageing.

If the transformer has a significant water and acid content, drying and oil reclaiming may
retard ageing of the solid insulation. The benefit of such actions presupposes that the state of
ageing is not too advanced to get a significant life extension as is sketched in Figure 33.
Looking at the mechanical strength vs. DP value of kraft paper, shown in Figure 32, where
the mechanical strength of the cellulose is almost unchanged down to a DP-value of 500-600,
the transformer conditions can be roughly divided into three groups:
• “New” units having a DP of around 900, water content around 0,5 %, and negligible
content of water soluble, low molecular weight acids.
• Units with DP’s in the 900 to 350 range, having a moderate content of water content
and acids.
• Aged units with a DP lower than 350, usually with high acidity and water content.

The cost-benefit of any maintenance action will depend both on the efficiency of the process
and the extent of ageing and contamination of the transformer.

7.4.2 Transport (diffusion) of contaminants

Most maintenance and condition control techniques address either how to avoid oxygen,
water and active acids being formed and/or getting into the cellulose insulation, or how to
remove this contamination from the cellulose insulation once there. To assess the efficiency
of certain equipment for maintenance one has to consider vapour pressures of the
contaminants (water or acids) and diffusion of these out through paper, pressboard and
cellulose in relation to the performance of the equipment and the duration of application. The
concentrations are not homogeneous; water concentration is highest in low temperature
regions; here the diffusion is lowest and it is most difficult to remove it.

When the transformer insulation is heated, the water (and probably also most other polar
ageing by-products) moves from the paper to the surrounding oil/air, because water solubility

Page 69 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

of oil increases as shown in Figure 23. This is the basis for procedures for drying of the paper
insulation. Next, the water transport must be made as efficient as possible; both for the
diffusion of water from the inner layers of the insulation to its surface, and for the transport of
the water from the surface of the cellulose to the drying apparatus. Diffusion processes
increase with temperature. Therefore, high temperatures are advantageous to increase
efficiency of any maintenance process on the cellulose insulation, as shown in Figure 34.
Diffusion is described by Fick’s law invoking a material dependent diffusion constant.
However, in the case of transformers one should note that the basic form of Fick´s law may
not fully describe the diffusion process. The diffusion coefficient D depends on the water
content concentration [127].

Figure 34: Sketch of time to get close to equilibrium for pressboard with type B.3.1 (EN
60641-3-1) thickness 1 mm, showing the higher the humidity in paper and/or the
higher the temperature is, the quicker is the establishment of an equilibrium
between insulation and oil [128].

7.4.3 Removal of contaminants from cellulose

The possibility for removing any substance from the cellulose will depend on the extraction
efficiency in the surroundings of the cellulose (oil, air or vacuum). This will mainly be
governed by three factors.
• The partition coefficient between paper and e.g. oil for gasses, water and low
molecular acids.
• The vapour pressure of substances like water and acids at the process temperature.
• Chemical bonding between contaminants and cellulose.

The partition coefficient of water is fairly well understood (see 6.1.1.1), and techniques for
removing water from the solid insulation are described extensively in Cigré Brochure No 227

Page 70 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

[54]. However as low molecular acids are not treated specifically it is useful to give them
some attention here. These acids will behave similarly to water and mainly stick to the
cellulose [129]. There seems to be a synergetic dependence between solubility of water and
low molecular weight acids in the oil.

While the lowest molecular weight acids have a boiling point close to that of water, the
boiling temperature increases with the weight as indicated in Table 6. In reality it is the
vapour pressure at the actual temperature that is relevant. A low boiling point indicates a high
vapour pressure. So heating the transformer will help by increasing the vapour pressure of the
water and acids, but efficiency may be low for some substances. There also is a possibility
that the acids will bind to the cellulose and that this will bind water even firmer to the
cellulose.

Table 6: Molecular weight, boiling point and vapour pressure of some organic acids.
Acid Type Mr (g/mol) Tboil (oC) at Vapour pressure
1 bar P [mm Hg] Ref temp [oC]
Formic 46 101 44,8 20
Acetic 60 118 11,4 20
Levulinic 116 243 1 102
Naphtenic 285 233-375 1,1*10-7 - 1,4*10-5 20

7.5 EFFECT OF OIL TREATMENT ON CELLULOSE AGEING

7.5.1 Role and characteristics of the oil

Good transformer oil should provide electric insulation and cooling of the windings, and be
chemically stable. Important characteristics of the oil are the thermal oxidation stability and
corrosivity. During ageing oil should produce little sludge and low molecular acids. The
stability will depend on its chemical composition. Practice varies regarding using additives
like oxidation inhibitors or not. Oils where inhibitors are used may be refined differently to
uninhibited oils: less of the natural inhibitors (mostly polyaromatics and sulphur compounds)
in the oil itself are kept. The inhibiting substances - acting as radical scavengers - are
consumed by oxidation, delaying the oxidation of the oil itself. Once these substances are
used the oxidation of the oil itself starts and the content of aldehydes, ketones and carboxylic
acids rapidly increases. It is advisable to reinhibit the oil once the content of inhibitor has
fallen to a low level (e.g. 30 % of new value) to avoid the oil becoming acidic and increasing
the need for extensive reclaiming.

Acids do influence the paper ageing. It is not well known what types of acids are formed
from the oil, but at least some of them are easily water soluble. Today oil acidity is measured
as a sum of all acids and expressed as the total acid number. Information about the ratio
between the easily evaporable low molecular weight acids and the less evaporable high
molecular weight acids can be found from the oxidation stability tests [130]. However, in a
transformer we expect that it is the paper ageing that is the main source of low molecular
weight acids.

Page 71 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

Figure 35: Progress of ageing rate (1/DP); effect of air and inhibitor at 85°C [58]. Experiment
on pressboard/authentic oil with weight ratio of 1:10.
It is not known whether oxidation inhibitors added to the oil may influence the paper ageing,
although it is reasonable to expect that they should have at least some effect on the oxidation
rate, as found in an experiment by Kachler et al. shown in Figure 35. However, Shroff in his
laboratory study found no support for such conclusions [49]. These effects need to be further
investigated before any general advice can be given.

7.5.2 Oil reconditioning and reclamation

Oil reclaiming is a major restoration done mainly to increase the life of the oil in a
transformer. The oil is passed through sorbent (e.g. Fullers earth) columns where acids and
polar compounds from the deterioration of the oil and paper are efficiently removed. After
reclaiming the oil needs to be inhibited – also in the case of previously uninhibited oil.
Usually, reclaiming is combined with a reconditioning of the oil using filters to remove
impurities and vacuum columns to degas and dry the oil. The oil has primarily been
considered in this context as described in the IEC oil maintenance guide [6].

Maintenance of the oil will not extend the service life of the cellulose unless ageing
accelerators like water, oxygen and acids are removed from the cellulose. If the oil is
circulated several times through the columns, a good, long-term stability in the oil’s acid
number is reported [131] and also a reduction in water content in cellulose is found.

The following considerations indicate to what extent water content reductions may be
expected. Table 7 shows some examples of drying times at some water contents and
temperature levels for a 300 kV transformer of 80 MVA with 27 tons of oil and 5 tons of
cellulose.

Page 72 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

Table 7: Estimated relative efficiency of on-site drying of a transformer done by oil

circulating through a drying column. Capacity of drying equipment: 5000 litres/hour
with full removal of water at each pass [21].
Relative Relative
Weight Weight
Water Weight Water Water Water Water time to time to
water in water in
content water in Concentration Concentration removed removed reduce reduce
oil at oil at
in cellulose in oil at 20oC in oil at 70oC per hour at per hour at humidity humidity
20oC 70oC
cellulose [kg] [mg/Litre] [mg/Litre] 20oC [kg] 70oC [kg] by 1% at by 1% at
[kg] [kg]
20oC 70oC
1% 50 0,027 0,270 0,9 9 0.0045 0,045 67 7
2% 100 0,081 0,810 2,7 27 0,0135 0,135 22 2
3% 150 0,162 1,89 5,4 63 0,027 0,315 11 1
In this model it is not considered that the oil in the transformer will be a blend of conditioned and unconditioned oil, and that the diffusion of
water through the cellulose and convection of oil from paper to process equipment may slow the process. Furthermore isothermal conditions
are assumed. Time is relative to time at 3% water in cellulose at 70oC

The example has relevance for the methods offered for on-site oil reconditioning or
reclamation. The drying of the cellulose is a possible bonus effect from processes mainly
focused on degassing, reconditioning and reclamation the oil itself. For all these methods the
cellulose and pressboard are dried via the oil transported through the processing apparatus.
The ability of the methods to get water and ageing by-products out of the winding will
depend on, and increase with, the temperature of the insulation system in the transformer
during the processing. To remove water will - as stated before - require time: temperature
dependence of solubility and diffusion for water is basically known. It is easier to remove one
percent water from wet insulation than from dry insulation because then the water
concentration in the oil is higher and more water is brought to the process equipment.
Preferably reclaiming should be done on transformers under load to keep a high temperature
and thereby increasing diffusion in cellulose and keep the concentration of ageing by-
products in the oil high.

Also, the higher the temperature, the more efficient is the removal of sludge. A prolonged
treatment at high temperature may remove significant amounts of sludge. Removal of sludge
will improve cooling and reduce ageing of the solid insulation Special regenerative oils, that
have better solubility for sludge than normal insulating oil, have been developed. .

Oil processing will also remove acids from the oil. The efficiency on the cellulose will
depend on the partitioning of the acids between paper and oil. One must be aware that it is
easier to remove the high molecular weight acids than the low molecular weight acids, but
that the normal measurement of acidity does not discriminate between the two.

Experiences from Sweden show that a reclaiming process reduced the water content in the
cellulose “somewhat” from the original 3% to a level that was still above 2% [132].

The effect of oil reclaiming on both water and acidity of the solid insulation has been
investigated on a 47 kV, 6 MVA transformer manufactured in 1947 prior to scrapping it
[133]. The transformer got a new low voltage winding in 1969. The transformer contained
500 kg of cellulose and 10 000 litres of oil. Samples were taken of oil and of solid insulation
from 1 mm pressboard barriers, and intake and star point connections. (No samples were
taken from the winding as it was found to be varnished.) The transformer was processed with
a reconditioning/reclaiming unit where the Fuller’s earth was regenerated each night. The
winding was heated to 65oC using a low frequency heating system. A total number of 10
passes (40 hours) of the full oil volume with Fullers earth and 16 passes (64 hours) with only

Page 73 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

vacuum was applied.

The condition of the oil was improved from a neutralisation value of 0,15 mg KOH/g to 0,01
mg KOH/g. For the solid insulation a reduction of 30-40% in both water content and acidity
could be seen as shown in Table 8. The acidity was measured using water extraction.

Table 8: Example of water content and neutralisation value in cellulosic insulation before and
after oil reclaiming.
Neutralisation value [mg KOH/g] Water content [%]
Before After Reduction[%] Before After Reduction[%]
LV outtake lead 2.42 1.23 49 2.0 0.9 55
LV starpoint 1.76 1.00 43 2.4 1.2 50
1 mm press-
2.08 1.45 30 1.6 1.1 31
board barrier

One can conclude that even if reclaiming is not intended for restoring the paper insulation,
the methods has an undocumented potential of also improving the conditions of the cellulose
by removing some water and acids. However, the measured total acidity of the oil does not
reflect the conditions of the paper where water soluble low molecular weight acids dominate.

In principle also, continuous degassing equipment can remove water content and acids. The
efficiency will of course depend on the capacity of the pumps and the efficiency (pressure) of
the degassing unit.

7.6 METHODS FOR CELLULOSE RECONDITIONING

By reconditioning the cellulose of a transformer one normally refers to singular actions using
techniques able to remove the main portion of water from the cellulose. To get proper drying,
one will usually take the transformer out of service and remove the oil. In general, draining
the oil away and applying vacuum makes water transport easier than in oil, particularly when
the drying can be done at elevated temperatures. In this case it will be the diffusion of the
water through the paper that limits the drying. Up to now this has mainly been done in a
factory, but now such services are being offered for on-site use.

Different methods for drying can be used. Some are based on techniques for drying of un-
impregnated windings. For unimpregnated units one can apply temperatures in the 120oC
range. For units from service one can use similar techniques after the oil is drained, or one
can remove water and contaminants via circulating oil. The main methods are:
Vapour phase; A hydrocarbon liquid (kerosene) is evaporated whereafter it condenses on the
insulation thereby heating it and building an azeotropic8 mixture with the water in the
cellulose. A series of pressure lowering allows the efficient extraction of water content. The
water and kerosene vapours are extracted by vacuum pumps and separated.
Hot oil spray; Hot oil is sprayed onto the windings to heat the cellulose and evaporate the
water, which thereafter is removed by vacuum.
Low frequency heating (LFH) is used to heat the windings and evaporate water, which is
thereafter removed by vacuum. It can be combined with hot oil spray because it heats parts
where the hot oil does not contact.

8
Mixture between water and kerosene with steady concentration ratio

Page 74 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

Hot air – vacuum; Here both hot air blowing to heat and evaporate water, and vacuum to
remove it, are used.
Liquid circulation as described in 6.5.2 can also be used to remove contaminants. One can
also exchange the oil with another liquid that better dissolves water and contaminants
(e.g.esters, oils with higher aromatic content) [134].

The methods above are listed according to their ability to quickly remove water from all parts
– the vapour phase being the most efficient. The hot vapour used in the vapour phase
techniques will condense on all cold surfaces. For hot-oil spray the oil will hit the outer
insulation surfaces and be less efficient for “screened” parts [135]. One has to apply vacuum
for a sufficiently long period to extract the water from the wood, the paper and the
pressboard. In some processes the winding is heated by applying low frequency current
during the drying process. This will definitively improve drying of the winding. How much it
will help on ribs and barriers has to be verified. One should be careful not to apply too high
temperatures when impregnated surfaces are in contact with air to avoid oxidation of the oil.

Figure 36 gives an example of the results from a low frequency heated vacuum drying of a 32
MVA transformer. Temperature was 100 – 110oC, vacuum in the transformer tank was
2 mbar, and the process was applied during 250 hours. When using these methods one has to
have storage tanks for the oil. Thermal insulation of the transformer tank will improve the
heating efficiency. Details about the processes can be found in [136].
LV-Winding Main Insulation HV-Winding Tap-Winding
100 100 100 100

75 75 75 75
Height in %

- Before drying
50 50 50 50 - After drying

25 25 25 25

0 0 0 0
0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3
CH2O in % CH2O in % CH2O in % CH2O in %

Figure 36: Water distribution in a winding arrangement before and after LFH-drying [136].

It is an open question how efficient these methods are for removing acids from the paper. We
would expect that some of the low molecular carboxylic acids having a low boiling point
would evaporate at the elevated temperature that is necessary for drying. This is in line with
the observed high acidity of the extracted water like shown in Figure 37.

Page 75 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

Figure 37: Extraction of low molecular organic acids during the LFH-drying [136].

Increase of water content in a transformer winding will result in swelling of the cellulose,
increased clamping pressure and plastic deformation of the cellulose. After drying, the paper
will shrink. Due to the plastic deformation that has occurred, a net reduction in clamping
pressure can be expected with slackness in the winding as a possible consequence. For a
winding, where the clamping pressure is essential to avoid displacement during short circuit
stresses this has to be considered [137]. Transformer design and locally expected mechanical
stresses should be considered when assessing the consequences of a possible winding
slackness.

7.7 IEC AND CIGRE PUBLICATIONS RELATED TO LIFE MANAGEMENT OF

TRANSFORMERS
Cigré brochure 227 “Life management techniques for power transformers”, Cigre WG
A2.18 Chairman V.Sokolov,
IEC 60422 (2005) Mineral insulating oils in electrical equipment - Supervision and
maintenance guidance

Page 76 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

8 PROCEDURES FOR VERIFYING AGEING PERFORMANCE OF

CELLULOSIC INSULATION

8.1 PROOF TESTING

Standardized tests can be found in the following national and international standards. They
may be of direct or indirect interest for power transformers. Care should be taken that any
accelerated test replicates service conditions. For example liquids with increased water
solubility may at high temperature dry out cellulose to an unrealistic level –compared to
service conditions where in almost all cases most of the water will remain in the cellulose -
thereby reducing hydrolysis:

8.1.1 IEC 60216-1 to 60216-6 "Electrical insulating materials - Properties of thermal

endurance"
This standard specifies the general ageing conditions and procedures to be used for deriving
thermal endurance characteristics and gives guidance in using the detailed instructions and
guidelines in the other parts of the standard. Simplified procedures are also given, with the
conditions under which the procedures can be used. Although originally developed for use
with electrical insulating materials and simple combinations of such materials, the procedures
are considered to be of more general applicability and are widely used in the assessment of
materials not intended for use as electrical insulation. Applying this standard, it is assumed
that a practically linear relationship exists between the logarithm of the time required to cause
the predetermined property change and the reciprocal of the corresponding absolute
temperature (Arrhenius relationship). For the valid application of the standard, no transition,
in particular no first-order transition should occur in the temperature range under study.

8.1.2 IEEE C57.100-1999 "Standard test procedure for thermal evaluation of liquid-
immersed distribution and power transformers"
A test procedure is established to provide a uniform method for investigating the effect of
operating temperature on the life expectancy of liquid-immersed transformers. The test
procedures are intended to provide data for the selection of a limiting hottest-spot
temperature for rating purposes, provide data which may serve as the basis for a guide for
loading, and permit the comparative evaluation of a proposed insulation system with
reference to a system that has proven to be acceptable in service.

8.1.3 TAPPI T 573pm-03 "Accelerated temperature ageing of printing and writing paper
by dry oven exposure apparatus"
This test method describes a laboratory procedure for accelerating the ageing of printing and
writing paper in sealed glass tubes through exposure to elevated temperature in an oven.

The test method applies to all types of printing and writing paper whether it is plain base
paper, has internal additives, is coated, is printed or contains any variants of printing and
writing paper found in normal usage. The method specifies the sample preparation and
conditions of exposure required to obtain information on the stability of paper to ageing with
regard to changes in mechanical strength properties brought about by exposure of such paper

Page 77 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

to elevated temperature. The method provides the means to compare the stability of different
papers on a relative basis, but does not attempt to project the exact life expectancy for a given
paper. Life expectancy depends in large part on the limits of acceptability beyond which
various paper properties are no longer useful, as defined by end-users. For a given paper,
those limits will be different for different end-users. This method does not provide a means to
measure or to calculate such life expectancy.

8.2 LABORATORY SCALE EXPERIMENTS

Laboratory experiments are typically conducted in sealed glass vials of only few hundred
millilitres. The amount of aged paper is accordingly small. For the ageing, the vials are
placed in an oven with forced air circulation. The following parameters can be varied in such
experiments:
- Temperature
- Time
- Ratio cellulosic insulation - oil
- Air / oxygen / nitrogen blanket
- Water content of oil / cellulose
- Acid number of oil
- Addition of potential catalysts (copper / steel)
- Oil type (non inhibited / inhibited; new / preaged)
- Paper / board type (Thickness, density, upgraded)
- Addition of molecular sieve or other drying agents

8.3 SIMULATION OF REAL TRANSFORMERS E.G. WEIDMANN TEST

APPARATUS

8.3.1 Principle
The concept underlying the design of the ageing vessels and the test procedure was to create
conditions similar to those encountered in transformers. The mass and surfaces in the thermal
model experiments are similar to those in a 200 MVA transformer [2]. The long term test
vessels are constructed as shown in Figure 38. A vessel consists of 9 independent ageing
chambers, which are heated or cooled jointly.

1: Ageing vessel with nine independent test chambers

2: Test chamber (volume: 20 dm³; wall area: 60 dm² made of sand
blasted steel sheet)
4 3: Expansion glass tube
3 4: Silicagel conservator
5: Flow baffle
5 6: Heater
2 7: Sensor for temperature control
14
11 8: Temperature measurement
12 9: Cooling coils
13 10: Outlet valve
8 11: Material samples (specimens)
9
12: Copper sheet; surface: 3 dm²
13: Core steel; surface: 6 dm²
14: Thermal insulation
1 6 7 10

Figure 38: Ageing vessel with open expansion system.

Page 78 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

8.3.2 Expansion systems

Open expansion system: Some of the ageing vessels are operated using the concept of “open
expansion”. In this system, the oil in the expansion vessel is continuously in contact with dry
air via a silica gel dryer and an oil filled bubbler.

Closed expansion system: In this case the expansion vessel contains a hermetic seal in the
form of metal bellows, which completely isolates the oil/solid insulation from the
atmosphere.

8.3.3 Temperature control

This system can be run with constant temperature as well as with temperature cycles. In case
of temperature cycles the heating is switched off for e. g. 2 hours per day. During the cooling
phase, the temperature in the chamber drops by 40 to 60 K depending on the nominal
temperature.

8.4 SLOW THERMAL RAMPING

Isothermal testing of polymer ageing requires long-term exposure if realistic results are to be
obtained, which means a timeconsuming and hence expensive exercise, which may have to
be repeated many times over many different conditions. Rates of ageing can be accelerated
by increasing the temperature of tests or by using a more aggressive environment, but then
results must be extrapolated to working conditions, assuming that the same mechanisms still
apply. Slow thermal ramp technique, similar to thermo-gravimetric analysis, allows the
qualitative and, in some cases, quantitative assessment of ageing to be made in a relatively
short time: 15 to 20 days, compared to up to 2 years for isothermal tests. The degree of
polymerisation is used as a key indicator of degradation. [138,139].

Page 79 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

REFERENCES

[1] CIGRE SC A2 2003 maintenance Questionnaire, Merida

[2] H.P.Moser, V.Dahinden: ”Transformerboard ”, Weidmann, Switzerland, 1999
[3] O.Lesaint, A.Saker, P.Gournay, R.Tabazeon, J,Aubin, M.Mailhot.“ Streamer
propagation and breakdown under ac voltage in very large oil gaps.“, IEEE Trans
DIE, Vol.5, No.3 June 1998, pp 351- 359
[4] L.Lundgaard, K.Herstad, M.U.Anker, J.Sletbak: "Flashover Along Solid Surfaces
Parallel to the Electric Field in Liquid Insulation at 50 Hz", CIGRE 1986, Paris, paper
15-08.
[5] G.Berg, L.E.Lundgaard, L.Machazek: “Discharge Signatures from a Wedge
Geometry in Transformer Insulation of Paper”, ISH , Delft, 2003.
[6] IEC 60422 (2005) – Supervision and maintenance guide for mineral insulating oils in
electrical equipment)
[7] IEEE C57.106-2002 “Guide for Acceptance and Maintenance of Insulating Oil in
Equipment”
[8] T.V. Oommen and L.N. Arnold, "Cellulose insulation materials evaluated by degree
of polymerization measurements," Proceedings of the 15th Electrical Electronics
Conference, Chicago, Oct. 1981, pp. 257-261.
[9] R. Jeffries: “The sorption of water by cellulose and eight other textile polymers”,
Journal of the Textile Institute Transactions, Vol 51, No.9,1960, pp 339-374.
[10] Y.Du, M. Zahn, B. C. Lesieutre, A. V. Mamishev, Water content Equilibrium in
Transformer paper-Oil System, IEEE Electrical Insulation Magazine, 1999, vol. 15,
11-19.
[11] J.Fabre, A.Pichon : “ Deteriorating processes and products of paper in oil. Application
to transformers”, CIGRÉ paper 137, (Paris), 1960.
[12] IEC 60076-7 Loading guide for oil-immersed power transformers
[13] IEEE Std C57.91-1995 IEEE Guide for loading Mineral-oil-immersed Transformers
[14] V.W. Montsinger: “Loading transformers by temperature”, Trans. AIEE, April 1930,
pp776-192
[15] W.G. Lawson, M.A.Simmons, P.S.Gale: “Thermal aging of cellulose paper
insulation”, IEEE Trans. El.Ins., Vol. EI-12, 1977, pp 61-66.
[16] F.M.Clark: ”Insulating Materials for Design and Engineering practice”, John Wiley
and Sons, New York, 1962,p 287.
[17] IEC 60450 – Measurement of the average viscometric degree of polymerization of
new and aged electrical papers
[18] A. M. Emsley, M. Ali and R. J. Heywood; “A size exclusion chromatography study of
cellulose degradation”, Polymer, Volume 41, Issue 24, Pages 8461-8747 (November
2000), Pages 8513-8521
[19] D. J. T. Hill, T. T. Le, M. Darveniza and T. Saha: “A study of degradation of
cellulosic insulation materials in a power transformer, part 1. Molecular weight study
of cellulose insulation paper”, Polymer Degradation and Stability, Volume 48, Issue
1, Pages 1-198 (1995), Pages 79-87.
[20] A.M. Emsley, R.J. Heywood, M. Ali, X. Xiao: “Degradation of cellulosic insulation
in power transformers .4. Effects of ageing on the tensile strength of paper”, IEE
Proceedings - Science, Measurement and Technology, Volume: 147, Issue: 6.,
Page(s): 285-290

Page 80 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

[21] L.Lundgaard et al., Ageing and Restoration of Transformer Windings”, SINTEF

Energy Research Report TR A5540, December 2001
[22] W.Kuhn, K.Freudenberg, W.Durr, F.Bolz, G.Steinbrunn: Ber. 63, pp1510, 1930
[23] A.Ekenstam: “The behaviour of cellulose in mineral acid solutions: Kinetic Study of
the decomposition of cellulose in acid solutions”, Berichte der deutchen chemischen
Gesellschaft, Vol 69, Issue 3,1936, pp 553-559
[24] M.M. Y. Chang: “Crystallite structure of cellulose”, Journal of Polymer Science:
Polymer Chemistry Edition, Vol 12, No.7, 1974, pp 1349-1374
[25] R.St. John Manley: “Growth and morphology of single crystals of cellulose
triacetate”, Journal of Polymer Science Part A: General Papers, Vol. 1, No 6, 1963, pp
1875-1892.
[26] G.V.Schultz, (1948) J. Polym. Sci. C, 2, 331-340
[27] M. Marx-Figini, M. Coun-Matus : “On the kinetics of hydrolytic degradation of
native cellulose”, Die Makromolekulare Chemie, Vol. 182, No. 12, 1981, pp 3603-
3616
[28] B.V.Erofeev, A.M.Shishko, S.M.Volkovich, L.G.Pesnyakevich, D.V.Matskevich,:
“Reactivity of amorphous regions of cellulose during acid hydrolysis”, Vestsi Akad.
Navuk BSSR, Ser. Khim. Navuk, Inst. Fiz.-Org. Khim., Minsk, USSR 1982, 1, pp 12-
16.
[29] G. Testa, A. Sardella, E. Rossi, C. Bozzi, A. Seves: “ The kinetics of cellulose fiber
degradation and correlation with some tensile properties” Acta Polymerica, Vol.45,
No. 1, 1994, pp 47-49
[30] A.M. Emsley , R. J. Heywood, M. Ali And C.M. Eley: “On the kinetics of
degradation of cellulose”, Cellulose, Volume 4, Number 1 / March, 1997, pp 1-5
[31] A.M.Emsley,, G.C.Stevens: “Review of chemical indicators of degradation of
cellulosic electrical paper insulation in oil-filled transformers”, IEE Proc. Sci. Meas.
Technol., Vol. 141, No. 5, September 1994, pp. 324-334.
[32] A.M. Emsley, G.C. Stevens, “Kinetics and mechanism of the low temperature
degradation of cellulose”, Cellulose, vol. No.1, p.26-56, april 1994
[33] A.M. Emsley, “The kinetics and mechanism of degradation of cellulosic insulation in
power transformers”, Polymers Degradation and Stability, Vol 44, pp. 343-349, 1994
[34] B.Bouvier, : “Noveaux critéres pour caractériser la degradation thermique d’une
isolation à base de papier”, Rev.Gen.Elec. 1970,79 (6), pp 489-496.
[35] B.Fallou: “Synthèse des travaux effectués au L.V.I.E sur le complexe papier-huile.”,
Revue Générale de l’Electricité, 79, 1970, pp 645-661
[36] W.Skubala: “Investigation of ageing of transformer insulation with and without the
effects of an electric field”, Prz. Papierniezy, 1974, 188 (5), pp 4058
[37] O.Smidsrød; Personal communication.
[38] H.Yoshida, Y.Ishioka, T.Suzuki, T.Yanari, T.Teranishi,: “Degradation of insulating
materials of transformers”, IEEE Transactions on Electrical Insulation, Volume EI-22,
Issue 6 , 1987, pp 795-800.
[39] D.P.C Fung: “Kinetics and mechanisms of thermal degradation of cellulose in vacuo”,
Tappi 52, 1969, pp 319
[40] X.Zou, N.Gurnagul, T.Uesaka, J.Bouchard, J.: “Accelerated aging of papers of pure
cellulose: mechanism of cellulose degradation and paper embrittlement” Polymer
Degradation and Stability, v 43, n 3, 1994, pp 393-402
[41] L.Lundgaard, W.Hansen, D.Linhjell & T.Painter, (2002) Ageing of Oil Impregnated

Page 81 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

paper in power transformers”, IEEE Trans Pow.Del., Vol 19, No.1, 2004, pp 230-238.
[42] K.I.Ivanov, E.S.Panfilova, T.N.Kullkovskaya, V. P.Zhakhovskaya, V. K.Savinova,
M.G. Seminova: “Influence of the Products of Oxidation of Mineral Oils on Ageing
of Paper Insulation in Transformers”, Zh.. Prikl. Khim. (Leningrad), 1974, pp2705-
2711.
[43] W.Lampe, E.Spicar: “The oxygen-free transformer, Reduced ageing by continuous
degassing”, Cigre Paper 12.05, Cigre 1976.
[44] B.M. Cox, (1989) National Power Report, RD/L/3381/R88
[45] EC 61125 “Unused hydrocarbon based insulating liquids - Test methods for
evaluating the oxidation stability”
[46] R.A. Lipshtein and M.I. Shaknovich, Transformer Oil, (2nd edition), translated from
Russian, Israel Program for Scientific Translations, Jerusalem, 1970.
[47] L.E. Lundgaard, W. Hansen, S. Ingebrigtsen, D. Linhjell, M. Dahlund, “Aging of
Kraft paper by acid catalyzed hydrolysis », IEEE Dielectric Liquids, ICDL 2005, 26
June-1 July 2005, pp (:381 - 384 )
[48] K.Bauer, W.Molitor: Stabilisierte Cellulose”, Elektrotechnischen Zeitschrift, Ausg. A,
Band 89, heft 18, , 1968, pp 433 – 437.
[49] D.H.Shroff, A.W.Stannett: A Review of paper ageing in Power transformers. IEE
Proc. Vol.132, Pt C, No.6, Nov 1985, pp312-319.
[50] L.E.Lundgaard, W.Hansen, D.Linhjell: “Ageing of Kraft and thermally upgraded oil
impregnated paper “, Nord-IS 05, Trondheim, 2005, pp 45-49.
[51] IEC 60763 Specification for laminated pressboard
[52] IEC 61061 Non-impregnated densified laminated wood for electrical purposes
[53] HP.Gasser, C.Krause, T. Prevost, "The aging characteristics of laminated pressboard
and laminated wood in oil cooled power transformers", ISEI Conference, Toronto, pp.
403-407, 2006
[54] CIGRE Working Group 12.18 (V.Sokolov): “Life Management techniques for power
transformers”, Brochure N° 227, 2003
[55] T. Painter, Acta Chem. Scand., vol. 27,pp 2463-2484, 1973
[56] T. Painter, Acta Chem. Scand., vol. 27,pp 3839-3860, 1973.
[57] F. Shafizadeh, ”The chemistry of pyrolysis and combustion”, in R. Rowell (Ed.) The
chemistry of solid wood, Chapter 13, American Chemical Society, 1984.
[58] I. Höhlein and J. Kachler, “ Aging of cellulose at transformer Service Temperatures.
Part1: Influence of type of oil and Air on the Degre of Polymerization of Pressboard,
Dissolved Gases, and Furanic Compounds in Oil”, IEEE Elect. Insul. Mag., vol.21,
no.2, (March / April.), pp 15-24, 2005
[59] J. Scheirs, G. Camino, M. Avidano and W. Tumiatti, “Origin of furanic compounds in
thermal degradation of cellulosic insulating paper”, Journal of Applied Polymer
Science, Vol. 69, 2541-2547, 1998.
[60] I. Höhlein, U. Thieß, Determining the Water Content in Transformers, Possibilities
and Application Limits of Water content Distribution Curves, VGB PowerTech
12/2004.
[61] D:Linhjell, W.Hansen, S.Ingebrigtsen, L.E.Lundgaard: “Water content absorption in
service aged transformer oils.”, Nord-IS 2003
[62] H. Yoshida et al., “Degradation of insulating materials of transformers”, IEEE Trans.
on Electrical Insulation, Vol. EI-22. No. 6, pp. 795-800, 1987.
[63] K. Goto et al, “Mesure de la température des enroulements des transformateurs de

Page 82 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

puissance et diagnostic du vieillissement par détection du CO2 et CO », CIGRE,

12.102, 1990.
[64] P.J. Griffin et al., “Paper degradation by-products under incipient-fault conditions”,
61th Annual International Conference of Doble Clients, Sec. I-OD, 1994.
[65] N. Dominelli, “The analysis of furanic and non-furanic compounds as a transformer
diagnostic technique, preliminary paper”, Doble Conference, 1995.
[66] Mineral oil-impregnated electrical equipment in service- Guide to the interpretation of
dissolved and free gases analysis , IEC Standard 60599, 1999.
[67] J.S. Navamany, P.S. Ghosh, “Age estimation of cellulose Paper Insulation in power
Transformers using ANN”, Electrical Insulation Conference and Electrical
Manufacturing & Coil Winding Technology Conference, 2003. Proceedings, 23-25
Sept. 2003, pp277 - 281
[68] R. Clavreul, Michel Bellesort, France, CIGRE A2/Preferential Subject 2 /Question 9,
2004
[69] S.Soares, G.Camino, S.Levchik; “Comparative study of the thermal decomposition of
pure cellulose and pulp paper”, Polym. Deg. Stab., 49 (1995) p275
[70] P.J. Burton et al., “Développements récents au CEGB pour l’amélioration de la
prévision et de la surveillance des performances des transformateurs », CIGRE Paper,
No. 12-09, 1984.
[71] J. Unsworth and F. Mitchell, “Degradation of electrical insulating paper monitored
with high-performance liquid chromatography”, IEEE Trans. on Electrical Insulation,
Vol. 25, No. 4, pp. 737-46, 1990
[72] N.Dominelli et al., “Recent developments in paper degradation products analysis”,
Minutes of the 60th Annual International Conference of Doble Clients, Sect,
10.41,1993
[73] A. De Pablo and B. Pahlavanpour, “Furanic compounds analysis : a tool for predictive
maintenance of oil filled electrical equipments”, Electra, CIGRE, Groupe 15.01.03,
No 175, 1997
[74] J. Vergne, « Étude du vieillissement physico-chimique du complexe papier-huile dans
les transformateurs électriques », Thèse doctorale, Institut National Polytechnique de
Toulouse, 1992
[75] J. Scheirs, G. Camino, M. Avidano and W. Tumiatti, “Study of the mechanism of
thermal degradation of cellulosic paper insulation in electrical transformers”, Die
Angewandte Makromolekulare Chemie, 259, 19-24, 1998
[76] D. Hill, T. Le, M. Darvenia , T. Saha, “A study of degradation of cellulosic insulation
in a power transformer”, Polym. Degrad. Stab. ,48, p. 79-87, 1995
[77] A. de Pablo, “Furfural and ageing: How are they related”, IEE, 5, 1999
[78] E. Serena, “Degradation of transformers insulation”, CIGRE WG 15.01.05, Turin,
1997
[79] R.M. Morais, M. Carballeira, J.C. Noualhaguet, Furfural analysis for assesing
degradation of thermally upgraded papers in transformer insulation, IEEE
Transactions on Dielectrics and Electrical Insulation, Vol 6 No.2, April 1999
[80] A. B. Shkolnik, K. Bilgin, “Creating a preliminary model for estimating degree of
polymerization of thermally upgraded paper based on furan concentrations in
transformer oil”, 66th Annual International Conference of Doble Clients, Sec. 5-8,
1999
[81] D.H. Grant, “Furfuraldehyde-in-oil: How to be sure your transformer has not been

Page 83 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

weakened by heat”, Minutes of the 57th Annual international Conference of Doble

clients, Insulating Fluids, 1990
[82] K. Bilgin, A.B. Shkolnik, J. Kelly, (S. D. Myers ), “The effect of dicyandiamide in
insulation on 2-furfuraldehyde concentration in oil”, H2B, Techcon, p.207-216, 99.
[83] A.M. Emsley, R.J. Heywood, M. Ali, X. Xiao, “Degradation of Cellulosic Insulation
in Power Transformer”; Part 4, IEE, Prosci. Meas. Technol. ,147, 285, 2000
[84] D. Allan, “Recent advances in the analysis and interpretation of aged insulation from
operating power transformers”, Proceedings of the 5th International Conference on
Properties and Applications of Dielectric Materials , Korea, p. 202-205, 1997.
[85] D.P. Myers and J.R. Sans, “Field study of furan formation in transformer fluids as an
indicator of damage to paper insulation”, Minutes of the 59th Annual international
Conference of Doble clients, sec. 10-6, 1992.
[86] P.J. Griffin, L. R. Lewand, “A practical guide for evaluating the condition of
cellulosic insulation in transformers,” Minutes of the 62th Annual International
Conference of Doble Clients, Sect, 5-61,1995.
[87] B. Pahlavanpour et al, Experimental Investigation in to the thermal ageing of kraft
paper and mineral insulating oil, IEEE, ISEI, pp 341-345 , 2002
[88] B. Pahlavanpour,, Dr Ekuld, M.A. Martins, Insulating paper aging and furfural
formation, EIC, 2003.
[89] M.C. Lessard, Results to be published
[90] I. Höhlein and J. Kachler, “ Aging of cellulose at transformer Service Temperatures.
Part 2. Influence of Water content and Temperature on Degree of Polymerization and
Formation of Furanic Compounds in Free-Breathing Systems”, IEEE Elect. Insul.
Mag., vol.21, no.5, (Sept./Oct.), pp 20-23, 2005.
[91] P.J. Griffin et al., “Measurement of Cellulosic Insulation Degradation”, Annual
International Conference of Doble Clients, Sec. 10-3, 1993
[92] H. Lutke, I. Höhlein, A.J. Kachler, “Transformer ageing research on furanic
compounds dissolved in insulating oil”, Cigre 15-302,2002
[93] A.M. Emsley, X. Xiao R.J. Heywood, M. Ali, , “Degradation of Cellulosic Insulation
in Power Transformer; Part 2, Formation of furan products in insulating oil”, IEE,
Prosci. Meas. Technol. ,147, 3,p.110-114, 2000
[94] P. J. Burton et al, “Applications of liquid chromatography to the analysis of electrical
insulating material”, CIGRE, Section 15-08, 1988.
[95] M. Carballeira, “HPLC contribution to transformer survey during service or heat run
tests”, Electra, CIGRE, no 133, p. 44-50, December 1990
[96] D. Grant, “A study of furanic compounds generated in transformers during heat run”,
Minutes of the 59th Annual International Conference of Doble Clients, Insulating
Fluids Sec. 10-4, 1992
[97] S. Naban et al, “Adsorption phenomena of furfural in insulation paper (fundamental
test for life diagnosis of transformer)”, T. IEE Japan,112-A, No.2, 199p.139-144.,
1992.
[98] S. Soares, N. Ricardo, F. Heatly and E. Rodriges, “Low temperature thermal
degradation of cellulosic insulating paper in air and transformer oil”, Poly. Int. 50,
303-308, 2001.
[99] M. Mulej, A. Varl, M.Koncan-Gradnik, « Up-to-date experience on furans for
transformer diagnostics », XIIIth International Symposium on High Voltage
Engineering, pp. 1-6, Netherlands 2003

Page 84 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

[100] G. Seymour and A.R. Pierce, “Gas chromatographic analysis of products from
controlled application of heat to paper and levoglucosan”, Anal. Chem., Vol. 37, No.
4, pp. 525-527, 1965.
[101] F. Wodley, “Pyrolysis products of untreated flame retardant-treated α-cellulose and
levoglucosan”, J. of Applied Polymer Science, Vol. 15, pp. 835-851, 1971
[102] S. Husain, “Monitoring Performance and predicting the service life of power
transformers by HPLC”,. Vol 30, N0 1, pp N-20-26, 1998.
[103] M-C. Lessard, L. Van Nifterik, M. Massé, J.F. Penneau, and R. Grob,
“Physico-chernical characterization of the thermal aging of insulating paper in power
transformers”, Proceedings of the 1996 IEEE international symposium on electrical
insulation, pp.533-537, Montreal, June 1996.
[104] M-C. Lessard, L. Van Nifterik, M. Massé, J.F. Penneau and R. Grob, “Thermal aging
study of insulating papers used in power transformers”, CEIDP, pp.855-859, San
Francisco, October 1996.
[105] O. Koréh et al, “Determination of Furanic Compounds in Insulating Oil by High
Performance Liquid Chromatography/Mass Spectrometry Using Atmospheric
Pressure Chemical Ionization”, Rapid Communications in Mass Spectrometry, 12,
1515-1519 1998.
[106] Awata et al, “Diagnosis by acetone for determination of breathing transformers
containing an absorbent in insulating oil”, TIEE Japon, Vol. 117-B No 5, 1997.
[107] Keiichi et al, Toshiba Corporation, “A sensing system for degradation diagnosis of oil
field transformers”, IEEE, ISEI, p.29, 1994.
[108] M.C. Lessard. M. Massé, « Prediction of remaining life of the paper insulation by the
analysis of new oil-soluble compounds in power transformers », IEEE, CEIDP, Paper
2C-2, Albuquerque, October 2003.
[109] N. Dominelli, The analysis of transformer oil for degradation products from
overheated solid insulation, Minutes of the 58th Annual international Conference of
Doble Clients, sec. 6-21, 1991
[110] D.M.Allan: “Practical life-assessment for aged transformer insulation”, IEE Proc.-A,
Vol, 140, No. 5, Sept. 1993, pp 404-408.
[111] Queensland Electricity Commission’s (Australia) experience reported by D.M.Allan
[112] Cigre TF D1.01.13 (M.-C.Lessard) to be published
[113] HP. Gasser., B.Heinrich, C.Krause,: “Condition Assessment of the Cellulosic
Insulation from Power Transformers Taken out of Service”, ISH, Bejing, China, 2005
[114] P.Gill: “Electrical Power Equipment Maintenance and Testing”, chapter 3.6, book,
Marcel Dekker, New York, 1998, ISBN 0-8247-9907-0
[115] U.Gäfvert, B.Nettelblad: “Measurement techniques for dielectric response
characterization at low frequencies”, In proc. Nordic Insulation Symposium, Lyngby,
Denmark, Paper 7.1, 1990.
[116] B.Nettelblad: “Effect of water content on the dielectric properties of cellulose”. In
proc. Nordic Insulation Symposium, Västerås, Sweden, paper 8.9, 1992.
[117] A. Bognar, L.Kalocsai, G.Csepes, E.Nemeth, J.Scmidt: ”Diagnostic Tests of high
voltage oil-paper insulating systems (in particular transformer insulation) using dc
dielectrometrics”, in proc. 1990 CIGRE Conference, Paris, France paper 15/33-08,
1990.
[118] F.I.Mopsik: “The transformation of Time-Domain Relaxation data Into the frequency
Domain”, IEEE Trans El.Ins., Vol. EI.20, No.6, Dec, 1985, pp 957-964.

Page 85 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

[119] Cigre TF D1.01.09 (conv. S.Gubanski): “Dielectric response methods for Power
Transformers”, Paris 2004, Bochure no 254
[120] U.Gäfvert, G.Frimpong, J.Fuhr: “Modelling of dielectric measurements in power
transformers”, in proc. 1998 CIGRÉ Conference, Paris, France, paper 15-103, 1998.
[121] A.K.Jonscher: ”Dielectric relaxation in Solids”, London, UK: Chelsea Dielectric
Press, pp. 36-52, 1983.
[122] A.K.Jonscher:” Universal Relaxation Law”, London, UK, Chelsea Dielectric Press,
pp 177-187, 1996.
[123] D.Linhjell, L.Lundgaard, U.Gäfvert,: “Dielectric response of mineral oil impregnated
cellulose and the impact of ageing”, Paper accepted for publication in IEEE
Transactions on Dielectrics and Electrical Insulation
[124] W.Lampe, E.Spicar, K.Carrander: ”Continuous purification and supervision of
transformer insulation systems in service”, IEEE PES Winter meeting, 1978, Paper A
78 111-7.
[125] I.Fofana, V.Vasserberg, H.Borsi, E.Gockenbach: “Drying of transformers using
zeolite”, IEEE El.Ins.Mag., Vol.9, No.1, 2002, pp 97-103.
[126] V.Vasserberg, H.Borsi, E.Gockenbach: “A new method for drying the paper
insulation of power transformers during service”, IEEE El.Ins Symp., 200, pp 251-
254.
[127] W.Lampe: ”Betrag zur Berechnungen der notwendigen Trocknungszeit von
Großtransformatoren“, Archiv für Elektrotechnik, 53. band, Heft2, 1969, pp 121-132
[128] I. Tousignant: “Assessing water content in insulating paper from water content in oil“,
Eurodoble Conference, Krakow, 2004
[129] S.Ingebrigtsen, M.Dahlund, W.Hansen, D.Linhjell, L.E.Lundgaard: ”Solubility of
carboxylic acids in paper (Kraft)-oil insulation systems”, IEEE CEIDP 2004, Boulder,
Colorado, USA, pp 253-257.
[130] IEC 61125, “Unused hydrocarbon based insulating liquids - Test methods for
evaluating the oxidation stability”
[131] Ø.Berg, K.Herdlevær, M.Dahlund, K.Renstrøm, A.Danielsen, U.Thiess: ”Experiences
from on-site transformer oil reclaiming”, CIGRÉ Main session 2002, Paper 12-103,
Paris.
[132] M.Svenson, C.Bengtson, M.Tapper: “Isolationsunderhåll av krafttransformatorer”
Elforsk Rapport 01:18, Stockholm 2001.
[133] SINTEF Energy Research Memo AN 04.14.02 “Oil reclamation of 6 MVA
transformer “, Trondheim 2004
[134] V.Wasserberg, H.Borsi, E.Gockenbach. „Drying of liquid immersed solid insulation
using a hygroscopic insulating liquid“, ISH 2001, Paper no 4-72, Bangalore, India,
[135] S.D.Foss, L.Savio: Mathematical and experimental analysis of the field drying of
power transformer insulation: “IEEE Trans Pow. Del., Vol. 8, No.4, October 1993,
pp1820-1828.
[136] T.Leibfried, U.Thiess, I.Höhlein, S.Truant, B.Heirich, H-P.Gasser, B.Breitenbauch.
T.Lainck, J.Leissner: “Profil von Wassergehalt und polymerisationsgrad in der
feststoffisolation von leistungstransformatoren”. ETG-Tagung, Doc 03.01 March
2004 Köln.
[137] Ch. Krause, W.Goetz: “The change of clamping pressure in transformer windings due
to variation of the moisture content – tests with pressboard spacer stacks”, Cigre SC
12 Symposium, Budapest 1999

Page 86 of 87
 Cigré Brochure 323 – Ageing of cellulose in mineral-oil insulated transformers

[138] G Camino, S Levchik, J Schiers, W Tumiatti & M Avidano, Analytical & Applied
Pyrolysis, (1996), Proc, Conf. 12th Int. Symp., Venice, October 14-18
[139] AM. Emsley, R.J. Heywood, G. C. Stevens, "Ultra-accelerated thermal ageing tests:
slow thermal ramp methods in predicting life expectancy", Int. Conf on Ageing
Studies and Lifetime Extension of Materials, Oxford, July 1999

Page 87 of 87