Why Transformers Fail
WHY TRANSFORMERS FAIL
Hongzhi Ding, Richard Heywood
John Lapworth and Simon Ryder
(email@example.com) (firstname.lastname@example.org )
Doble PowerTest Ltd.
5 Weyvern Park, Peasmarsh, Guildford, Surrey, GU3 1NA, United Kingdom
Knowledge from the tear down investigation of faults and failures in power transformers is of
vital importance in understanding the results from the dissolved gas analysis (DGA) and
electrical condition assessment measurements and preventing further incidents. This technical
article discusses with examples the common failure modes observed in the scrapped power
transformers. The review will also outline what we can do that is effective in preventing
power transformer failures, with examples, too, showing how the developing failures could
be saved through continually Transformer Asset Health Review by effective DGA analysis
combining with effective condition assessment.
While assisting in the investigation of unexpected transformer failures is an important aspect
of the work, there are many examples of transformer component defects and faults that were
detected well before an unexpected failure could occur, i.e. during routine dissolved gas
analysis (DGA), electrical condition assessment and maintenance operations. Many in-
service power transformers are now required to operate beyond their original design life,
mainly as a consequence of missing-match between the large number of ageing transformers
and the limited resources available to source replacements for them; and also because these
ageing transformers are still in reasonably good working condition although their condition
and ability to carry peak loads are usually unknown. As part of the transformer asset health
review and life extension program, over the years Doble PowerTest have records of detailed
forensic teardown inspection of more than hundred large power transformers. This involves
witnessing the process of scrapping and making a thorough inspection of each component to
assess its condition. This teardown of power transformers has enabled the condition
assessment of components that would not normally be addressed during routine maintenance
because of their inaccessibility. Knowledge of the causes of transformer in-service failures,
together with assessments made during strip downs of transformers removed due to high risk
exposure, have given significant insight into modes of deterioration/failure in particular
design groups. This has been translated firstly into a diagnostic strategy for assessing the
condition of power transformers nearing the end of their life and then integrated into asset
health and asset risk reviews and finally utilized in aged transformer replacement planning.
Doble PowerTest’s experiences thus far reveal most transformer failures are not due to old
age, but localised damage or ageing due to some limitations in design and manufacture,
application and maintenance [1-4]. Sometimes a power transformer does fail without any
warning notice. In most cases, however, the symptoms of developing fault and failure can be
detected, prevented or eliminated.
Transformer Design and Construction
As electrical devices that transfer energy from one electrical circuit to another by
electromagnetic coupling without moving parts, power transformers are normally regarded as
highly reliable assets because they are designed and constructed by time-proven technology
and materials. It is generally believed that the transformer designed and built at the turn of the
century was already a mature product as the essential features of the device remain
unchanged to date, although the transformer continues to evolve [5, 6].
The principles that govern the function of all electrical transformers are the same regardless
of size or application . The typical power transformer is submerged in mineral oil for
insulation and cooling and is sealed in an airtight metallic tank. Low- and high- voltage
power lines lead to and from the coils through bushings. Inside the transformer tank, core and
coils are packed close together to minimise electrical losses and material costs. The mineral
oil coolant circulates by convection through external radiators. Figure 1 shows three winding
assembly on core viewed from the HV side and after the tank being removed.
Three winding assemblies on core, HV side view
The essential parameters that characterise the ideal transformer depend, to a large extent, on
the properties of the core. The properties that are critically important in transformer core
materials are permeability, saturation, resistivity and hysteresis loss. It is generally believed
that it is in the core that the most significant advances in power transformer design and
construction have been made .
The performance of power transformers depends on dielectric insulation and cooling systems.
These two systems are intimately related, because it is the amount of heat both the core and
winding conductors generate that determines the permanence and durability of the insulation,
and the dielectric insulation system itself is designed to service to carry off some of the heat.
It is vital that the insulation utilised in a power transformer must be able to separate the
different circuits; isolate the winding core and outer case from the circuits; provide
mechanical support for the electrical coils and withstand the mechanical forces imposed by
power system surges and short circuits. Generally Kraft paper has been utilised for winding
conductor insulation, high density pressboard for inter-winding and inter-phase insulation,
and crêpe paper for lead insulation. The critical properties that determine the functional life
of dielectric oil/paper insulation are chemical purity, thermal stability, mechanical and
What Causes a Power Transformer to Fail?
It is generally believed that failure occurs when a transformer component or structure is no
longer able to withstand the stresses imposed on it during operation.
During the course of its life, the power transformer as a whole has been suffering the impact
of thermal, mechanical, chemical, electrical and electromagnetic stresses during normal and
transient loading conditions. The condition of the transformer deteriorates gradually right
from the start, resulting in
Reduction in dielectric strength (i.e., the ability to withstand lightening and switching
Reduction in mechanical strength (i.e., the ability to withstand any through faults);
Reduction in thermal integrity of the current carrying circuit (i.e., the ability to
Reduction in electromagnetic integrity (i.e., the ability to transfer electromagnetic
energy at specified conditions including over-excitation and overloading).
A failure ultimately occurs when the withstand strength of the transformer with respect to one
of the above key properties is exceeded by operating stresses.
A useful way of thinking about failure of a power transformer could be illustrated in Figure 2,
as proposed by CIGRE WG 12.18 [7, 8]. In its early life of service, the power transformer has
a sufficient spare safety margin between the various types of transient service stress and
capability. Here “strength” and “stress” are used generically to cover any kind of incident-
thermal, mechanical or electrical events. However, after a period of general ageing this may
not be the case. At some point in the deterioration process, probably long before the useful
life is run out, one or more parts of the transformer may well have changed just enough or
even failed such that the transformer no longer performs as required, e. g. even if a transient,
such as an overvoltage or close-up short circuit has been successfully withstood failure could
occur at the next transient.
Mechanisms of failure that are involved in a large transformer are often complex. Typical
transformer functional failure mechanisms are summarised in Table 1, as per the CIGRE
WG12.18. Note this is a functional failure model only for transformer core and coil assembly,
not including on load tap changers (OLTC) and bushings.
It is also important to distinguish the fault and the failure. A fault is mainly attributed to
permanent and irreversible change in transformer condition. The risk of a failure occurrence
depends not only on the stage of the fault developing but also the transformer functional
component involved. The failure could be repairable on site, depending on the type of fault as
well as the severity of the failure.
Transformer functional failure model [7,8]
System, Component Possible Defect Fault and Failure Mode
Abnormal oil ageing
Abnormal paper ageing
Flashover due to:
Excessive paper ageing
Destructive partial discharges
Localised surface tracking
Loosing winding clamping
Failure of solid insulation due to:
Failure of leads support
Winding displacement (radial,
Loosing core clamping
Short-circuit (open circuit) in
Excessive gassing due to:
Short-circuited turns in winding
Short-circuit due to:
Increasing AgeNew Old
Reducing Strength with
time and after incidents
Increasing AgeNew Old
Reducing Strength with
time and after incidents
A conceptual failure model proposed by CIGRE WG 12.18 [7,8]
From our records and case histories data, failures of power transformers are commonly
associated with localised stress concentrations (faults), which can occur for several reasons
Design and manufacture weakness, e.g. poor design of conductor sizing and
transpositions, poor joints, poor stress shield and shunts, poor design of clamping,
inadequate local cooling, high leakage flux, poor workmanship, etc.;
The microstructure of the material utilised may be defective right from the start, e.g.
containing micro-voids, micro-cracks etc.;
Corrosive attack of the material, e.g. sulphur corrosion on paper and conductor can also
generate a local stress concentration.
Weakness in transformer design, construction and materials could be covered by low loading.
However, increasing loading and extended periods of in-service usage will uncover these
Common Failure Modes
Failure modes of large power transformers are not always straightforward. But purely from
an assumption of the failure experienced in a large power transformer, most transformer
failures can be classified into either one or a combination of more than one of the following
Breakdown of insulation as a whole, due to severe solid insulation ageing;
Breakdown of insulation by part, due to premature ageing by localised high temperature
Mechanical failure of windings.
Common among many of the transformer failure modes is a shorted turn. The shorted turn
was developed as a result of breakdown of the solid insulation which causes winding
temperature shoot-up. The breakdown of solid insulation could be due to the natural wear of
insulation or repeated overloading or cooling system deficiency, which often result in severe
ageing of winding insulation. This type of failure (shorted turns without any prior warning or
obvious system cause) is a typical ‘end of life’ failure mode. If the transformer runs
abnormally hot and/or develops less than its normal out voltage, one can safely assume the
possibility of shorted turns.
Electrical breakdown is a common failure mode for power transformers, too. The electrical
breakdown could be developed by a number of reasons such as ageing of insulation,
excessive moisture content, deformed windings etc. Moisture reduces the dielectric strength
of insulation and can promote the occurrence of surface creeping discharges on the
pressboard barriers and lead to a flashover. Deformed windings indicate not only a high level
of force that may have broken or abraded the winding conductor insulation, but also a
reduction in electrical clearance. This mechanical failure of windings will then manifest itself
as an electrical breakdown which actually causes failure of transformer.
Poor design and overheating are very much interrelated and make for high failure modes. In
the bottom end, lack of cooling causes either general or localised high temperature
overheating, resulting in rapid insulation deterioration and damage progression. Breakdown
of insulation between the core and main tank may lead to circulating currents in the
core/frame/tank and result in local overheating. Circulating current in the tank can produce
hotspots in the tank and across gasket joints, resulting in partial discharges emanation from
the ground potential surfaces of the tank and parts mounted on the tank. Note local
overheating in current carrying circuit, if not extremely severe, often will not itself cause
direct failure of the transformer, but will reduce the mechanical strength of the insulation so
that when the transformer is subjected to a system fault close to the terminals, it will then fail
. This is similarly true for winding movement.
Poor design and loose clamping are very much interrelated and make for high failure modes,
too. The most well-known design problem with loose clamping is arcing/sparking fault at the
loose clamping bolts, which compromises the mechanical strength of the transformer and
makes diagnosis of dielectric faults using DGA difficult. The arcing/sparking discharges also
lead to deterioration of the oil and the production of fine carbon, which compromises the
dielectric integrity of the transformer.
Three Case Studies on Transformer Failures
Case 1: Transformer Failure Due to Shorted Turns
In April 2009, a 30 year old 750MVA 400/275/13kV autotransformer tripped on Buchholz.
Analysis of a subsequent DGA sample from the Buchholz clearly indicated a fault. Electrical
protection had shown unusual waveforms on the middle phase immediately before the trip.
Condition assessment tests after the trip are shown in Table 2 (turns ratios) and Table 3
(winding resistances). Measured ratio for the middle phase differs from expected value by
three times more than allowed (0.5%), indicating lost turns, and lower than expected value
indicates the fault is in series winding. Winding resistance measurements confirm fault in the
middle series winding, which was unlikely to be unlikely to be economically repairable.
During the scrapping, after the wraps of the middle phase series winding were removed, the
shorted turns in the 2nd
discs of series winding was found and these seemed to be
particularly severe. Figure 3 shows a picture of failure by shorted turns. There was extensive
loss of conductor and conductor insulation in the upper part of the series winding, which is
unlikely to be economically repairable.
Turns ratios measurement on a 750MVA autotransformer after Buchholz trip
Measured RatiosExpected ratio Applied HV-N voltage, kV
A phase B phase C phase
0.3 1.435 1.452
1.455 12.0 1.456 1.455
Notes: Measurements made at 0.3 and 12 kV, using Doble M4000 Insulation Analyser and Doble TTR
Winding resistances on a 750MVA autotransformer after Buchholz trip
Winding A phase B phase C phase
Series (400 to 275 kV) 0.1783 0.3296 0.1778
Common (275 kV to neutral) 0.5236 0.5222 0.5236
Notes: Measurements made at 5A using Tinsley 5896 Transformer Microhmeter.
According to WTI’s, transformer was at 15°C.
Failure of transformer by shorted turns
Comparison in colour of conductors in A/red phase series winding top disc: from left to right
is outermost strand, middle strand and innermost strand
The worst degree of polymerisation (DP) measurement obtained was 142/146 (average 144)
from the middle strand of top disc of the middle phase series winding. The next worst result
was 151/161 (average 156) from the middle strand of top disc of A/red phase series winding.
The DP results on paper samples showed that apparently the insulation condition of the series
winding had reached the end of its life. The DP analysis on paper samples also showed that
the winding hotspot was located in the middle strand of the upper part of series winding.
Figure 4 shows a visual comparison of conductors taken from A/red phase series winding top
disc, from left to right is the outermost strand, middle strand and innermost strand,
respectively. Note the severe discoloration of the middle strand conductor which implies not
only the location of series winding hotspot, but also the inadequate cooling design in the
The learning point from this case study is that the short turn was developed as a result of
severe winding conductor insulation ageing which was partly a function of the age of the
transformer and the loading to which it had been subjected. The thermal design of the series
winding, however, led to localised overheating of certain areas, including the point of failure.
Case 2: Transformer Failure Due to a Flashover
In the middle of 2006, a 42 year old 30MVA 132/11kV station transformer failed releasing
oil to the ground from the busting discs. It was believed the transformer might have been
subjected to through fault before its failure.
Failure of transformer by a flashover in the main tank
Close-up of the middle phase winding bottom end-blocks
from HV side (left) and LV side (right)
During the strip down it was found that the failure actually involved one severe
arcing/sparking fault in the main tank, which was located between the bare copper strip
connected to the middle phase LV winding line end and the middle phase top steel clamping
platform in the LV side, where the arcing seemed to be particularly severe so that both the
bare copper strip and the corner of the steel clamping platform had been damaged. Figure 5
shows a picture of failure by flashover in the main tank.
Further inspection of the core and windings during scrapping found direct evidence of
mechanical deformation of all windings from three phases particularly in middle phase.
Figure 6 shows the severe displacement of the middle phase winding bottom end-blocks.
Note that the missing end-blocks on the LV side had been found on the tank floor. It was
therefore believed that all windings from three phases particularly the middle phase windings
had been subjected to very significant circumferential forces and had significantly twisted
and relaxed as a result. It was further thought that the relaxed winding clamping had caused
the downward movement of the middle phase LV winding line end, which reduced electrical
clearance between the bare copper strip and the steel clamping platform corner and
eventually caused a flashover in main tank.
The learning point from this case study is that the flashover was developed as a result of
reduced electrical clearance which was due to winding mechanical deformation caused by
short circuits, and poor design of having no physical support to the middle phase LV winding
line lead that connected to the bus-bar.
Case 3: Transformer Failure Due to Axial Collapse of Winding
In late 2004 the decision was made to scrap a 50 year old 120MVA 275/132/11kV
autotransformer which had suffered a serious tap changer fault.
The fault was first noted during planned maintenance, and it appeared that the middle/B
phase tap selector became misaligned by one tap compared with the A and C phase tap
selectors. After the transformer was returned to service, the voltage control scheme
eventually sent the transformer to the end of the tap range. At the end position the B phase
diverter was required to switch the entire tap winding, rather than one tap step as it was
designed to. This resulted in serious damage to the B phase tap changer and large currents
flowing in B phase of the transformer. Fault investigation tests were made on the transformer
and results of additional winding capacitance and power factor measurements are listed in
Table 4. The results from B phase clearly indicated a serious problem. The large reduction in
capacitance between the series and common and tap windings seemed to indicate axial
collapse of the tap winding.
During the strip down it was found that the transformer failed due to axial collapse of the B
phase tap winding, following a fault in the B phase tap changer. This would have been
impractical to repair.
Figure 7 shows a picture of failure by axial collapse of the tap winding. Note there had been
no serious design defects or unusual design features found during the scrapping. The
transformer seemed to have no faults other than with the B phase tap winding. The degree of
polymerisation analysis on paper samples were in the age 450-750, which indicates little
ageing and considerable useful life remaining.
Fault investigation on a 120MVA autotransformer
Collapsed B phase tap winding
What Can We Do That Is Effective in Preventing Transformer Failures in Our
Why transformers fail is easy to understand. However, getting more transformer engineers to
do their part in preventing failures is the hard part. So, what can we do that is effective in
preventing transformer failures in our substations?
The simple answer is that a power transformer must be replaced when it no longer meets the
requirement of system reliability and before it fails . In order to be able to replace the
transformer before it fails, it is considered necessary to have a transformer asset health review
methodology to analysis and prevent in-service failure [1-3]. This involves using information
from a wide range of sources, including oil tests, on-line and off-line condition assessment
tests and visual inspections. However, knowledge of transformer designs and of their
strengths and weaknesses is essential to understanding the other information. Given the age
of many of the transformers, such information is now in many cases only obtainable through
witnessing the scrapping of transformers.
The following three case examples illustrate how developing failures could be managed and
even saved by effective DGA analysis combining with effective condition assessment tests.
Case 4: Developing Failure Due to Loose Clamping and Leakage Flux
In early 2009 a 43 year old 240MVA 275/132/13kV autotransformer was taken out of the
service as per the planned replacement. This transformer had been suffering from the known
loose clamping for many years, and the strip down inspection of a sister transformer one year
before it was removed from the system had provided valuable information about the likely
condition of this transformer believed to be significantly in risk of failure.
During the scrapping it was found that approximately one third of the clamping bolts showed
signs of having been loose in the past. Certain clamping bolt bosses showed signs either of
spark erosion or of hammering (elongated slots). Overall the winding clamping was in a very
poor condition and looks much worse than what was seen from the scrapped sister
transformer a year before. The loose clamping had resulted in severe arcing/sparking
discharges developing at a large number of the clamping bolts/bosses, producing fine carbon
contamination everywhere particularly on the top frame surfaces. The loose clamping had
also resulted in relaxed coil assembly leading to the development of partial discharges and
fine carbon contaminations produced inside the windings. Figure 8 close-up shows the severe
loosing clamping fault. Note how one of the missing clamping bolts had become embedded
in the insulation above the tertiary winding, as shown clearly in the picture on the right hand
side. Here it was electrically shielded by steel clamping ring. The same picture also shows a
bent clamping bolt.
During scrapping the possibly burned electrostatic shields were also noted and those seemed
to be extremely severe. Figure 9 shows a picture of the copper foil having become severely
overheated by the leakage flux, resulting in damage to the bottom end clamping platforms as
well as to the adjacent insulation. This was not particularly apparent from dissolved gas
In conclusion, the findings of the severe loose clamping plus the burned electrostatic shields
provided conclusive evidence to confirm that this transformer had reached the end of its life
and certainly was not capable of continuing service.
Loose clamping faults from LV side (left) and A phase end (right)
The burned electrostatic shields: general view (left) and close up (right)
Case 5: Developing Failure Due to Localised Overheating
In early 2009 the decision was made to scrap a 1996-made 240MVA 400/132kV (no tertiary)
autotransformer, believed to be significantly in risk of failure from localised high temperature
overheating in current carrying circuit.
This transformer had been suffering from a thermal fault in the main tank before it was
removed from service. Fitting a frame earth resistor had not stopped the development of the
fault. It was believed, therefore, that the thermal fault had not been caused by a circulating
current in the core/frame/tank.
The dissolved gas levels in the main tank had been typical of the larger transformer
population until a year before the transformer was removed from the system. There was then
a rapid rise in the ethylene level, accompanied by rises in the hydrogen, methane and ethane
levels. The last sample before the transformer was removed from service contained 324 ppm
of ethylene, 302 ppm of methane, 144 ppm of hydrogen and 123 ppm of ethane. The
dissolved gas signature clearly indicates a serious thermal fault in the main tank which
developed through 2008. The rate of deterioration seems to have increased during the year.
The carbon monoxide level had been less than 500 ppm for much of the service years but the
ratio of carbon dioxide to carbon monoxide varied between 2 and 45. These both seemed to
suggest little to moderate solid insulation ageing only. However, the relative proportions of
gases suggested a localised high temperature overheating fault involving solid insulation
(relatively high hydrogen and methane, low acetylene, ethylene/ethane ratio < 4).
Based on winding resistance measurements, it was suspected that there was likely a bad joint
in the C phase LV current carrying circuit, but internal investigation inspected all joints and
connections around the C phase LV terminal and there was no clear indication of any
problem. It was finally concluded that the fault must be inside the C phase common winding.
During scrapping, after the C phase common winding was pushed out, it was found that all
joints were shown to be healthy and there was no clear indication of any problem. Figure 10
shows a picture of a developing failure point within the common winding due to localised
overheating. The localised high temperature thermal fault had caused extensive loss of
conductors and insulations but had not led to a short-circuited turn developing yet.
Developing failure point within common winding due to local overheating
The learning point from this case study is that this developing failure does not seem to have
been caused or exacerbated by the design of the transformer, although the root cause of the
thermal fault was not actually known. It could, however, be caused by any one of the
following reasons: microscopic conductor damage from new; weak joint in conductor; slack
damping/fretting which resulted in the loss of insulation; and a system transient.
Case 6: Developing Failure in a Transformer Saved by DGA Analysis
This is the case of a 750MVA 400/275/13kV autotransformer built in 1967 and currently still
in service. Over the last few years this transformer has developed severe thermal fault twice
but all saved by effective DGA analysis.
In late 2005, the transformer was removed from service because of rapidly increasing
dissolved gas results which indicated a bare metal fault inside the main tank (high ethylene as
the dominant gas). The following electrical tests, including winding resistance measurements,
pointed to a winding joint problem associated with the tertiary winding, most likely involving
connections to the tertiary bushings. An internal inspection revealed faulty joints in the
internal connection between one of the main tank tertiary bus-bars and the left hand tertiary
terminal (3C2) in the tertiary loading box at the A phase end of the tank. This was originally
a multi-part single aluminium bar, whereas the 3B2 and 3A2 leads were double copper bus-
bars. The fault appeared to be due to a poorly bolted connection in the cranked part of the
connection where it left the tertiary loading box to rise up towards the top of the main tank to
connect to the tertiary bus-bars. As part of the repair a second parallel copper bus-bar was
added to the 3C2 lead.
Unfortunately, after the transformer was returned to service and tertiary loading (by a shunt
reactor) resumed, further gassing was observed. Analysis of tertiary winding resistance
measurements made after the 2005 repair suggested another high resistance joint problem
with the 3A2 connection. During a planned outage in 2008 these resistance measurements
were repeated and confirmed. After the oil was drained a visual inspection took place and a
large carbon deposit was found at the base of 3A2 bushing on the joint between the flexible
and the bushing.
Developing failure point due to local overheating: bus-bar joint in tertiary connections (left)
and bushing joint (right)
Figure 11 shows a developing failure point in the main tank due to local overheating. Note
the left picture shows an overheated bus-bar joint in tertiary connections and the right picture
shows overheated bushing joint.
Tertiary winding connections in the studied 750MVA autotransformer
Tertiary winding resistance measurements before and after repair
Measured resistance, mΩ
21.6ºC, 41% RH
17.4ºC, 56% RH
16ºC, 89% RH
(1) TA to 3A2 11.023 8.211 7.802
(2) TA to 3B2 15.732 15.874 15.836
(3) 3A2 to 3B2 10.783 7.624 7.745
(4) 3B2 to 3C2 8.085 8.032 8.010
(5) 3C2 to 3A2 18.650 15.354 15.590
(6) TC to 3B2 8.124 8.112 8.046
(7) TC to 3C2 0.4281 0.4273 0.3722
Notes: Measurements made with Tinsley resistance meter
After the repair the winding resistances were measured again, and these confirmed that there
were no further tertiary resistance anomalies. Note that in Figure 12, the tertiary winding
connections in this transformer are somewhat unusual in that all three corners of the tertiary
are brought out to the A phase end of the transformer for tertiary loading, while the original
arrangement of bringing out one corner (TA and TC leads) for closing and earthing externally
is retained out at the C phase end. Table 5 summarises the tertiary winding resistance
measurements before and after repair.
The learning point from this case study is that developing failures due to bad joints in main
tanks of transformers could be saved just by effective DGA analysis combining with effective
condition assessment tests.
Fault and failure investigations on power transformer components have an important role in
improving reliability and managing the risk of transformer failure. The identification of the
primary cause of failure and the subsequent analysis enable recommendations for corrective
action to be made that hopefully will prevent similar failures from occurring in the future.
Most unexpected power transformer failures happen because of maintenance oversights and
over loadings. Couple your understanding of how power transformer components are
supposed to function with a careful look at tell-tale damage, and you can prevent recurrences.
When design error and/or weaknesses developing over time are uncovered, enhanced
monitoring/investigation on sister units built by same manufacturer will help in preventing
future failures and therefore aid in managing the risk of unexpected failure.
 R. Heywood, J. Lapworth, L. Hall, and Z. Richardson, “Transformer lifetime
performance: Managing the risks”, 3rd IEE International Conference on Reliability of
Transmission and Distribution Networks, London; February 2005.
 R. Heywood and A. Wilson, “Managing reliability risks-Ongoing use of ageing system
power transformers”, Doble Israel Conference 2007.
 A. Wilson, R. Heywood and Z. Richardson, “The life time of power transformers”,
Insucon 2006, 24-26 May 2006, Birmingham, UK.
 H. Ding and S. Ryder, “When to replace aged transformers? Experiences from forensic
tear downs and research”, Euro TechCon 2008. Liverpool, 18-20 November 2008.
 M. J. Heathcote, J & P Transformer Book, 13th
edition, Elsevier 2007.
 J. W. Coltman, “The transformer”, IEEE Industry Applications Magazine, pp. 8-12,
 CIGRÉ Working Group 12.18, “Guide for life management techniques for power
transformers”, CIGRÉ Brochure No. 227, 20 January 2003.
 CIGRE WG 12.18 “Life management of transformers, draft interim report”, July 1999.