UNDERSTANDING FAILURE MODES OF TRANSFORMERS

Victor Sokolov
ZTZ-Service, Ukraine

Abstract

The paper discusses factors that impact the reliability of large power transformers and their
typical failure modes primarily when in service for many years.

Introduction

A failure is usually a "tuning fork" of Life Management procedures. Failure analysis delivers key
information providing insight for determining “what happened?” and “what to do?” in terms of
managing network reliability , assessing risk, optimizing maintenance, and estimating end of life.
Ultimately, the information aids in improving design and manufacturing of equipment. Failure
modes and causes may differ markedly depending on user specifications, transformer
application, design features, and, in particular, on the susceptibility to service deterioration and
external exposure. In order to understand the cause of failure properly all factors such as design
anamolies, operating conditions and the mechanisms which reduce safety margins should be
considered. This paper attempts to examine large power transformer reliability based on ZTZ-
Service database statistics. Typical failure-modes and failure causes are discussed, using design
review as a main instrument of investigation.

Failure Statistics

Updated Failure Statistics

Many experts describe failure occurrences in terms of the “bathtub curve” where it is predicted
that transformer failures increase through time. However, available statistics have not yet
revealed a correlation between the number of failures and advancing years in service. In fact, the
statistics show peak failures occurring around 19-21 years after the transformer has been in
service 1,2 ,3.

In spite of the fact that a huge transformer population has already been in service for 25-40 or
more years there is still little information available about the units that have failed primarily due
to thermal degradation of insulation material.

The ZTZ-Service database covers failure events since 1959-60. Analyzed equipment includes
large power transformers of different applications including over 5,000 units rated 100 MVA and
above and shunt reactors in the 400-750 kV range primarily from CIS countries. Since 1994
database has been supplemented with information obtained from worldwide failure events.

In order to gain some insight into failure statistics over a wide time range including recent
periods, four large groups of transformers of similiar type installed in CIS countries have been
analyzed: autotransformers 125-200 MVA, 220/110 kV (observation period 1964-2005, 27,505
transformer-years), 125-200 MVA, 330/110 kV (1963-2005, 9,477 transformer-years; 167 MVA
500/220 kV (1965-2005, 13,749 transformer-years), and generator transformers 400 MVA, 330
kV (1969-2005, 1,600 transformer-years).

Euro TechCon 2005 43

Failure rates have been determined as the ratio of the number of failures of a given population
over a given period of time to the number of accumulated service years for all transformers in
that period of time. Hence Failure rate (%pa) = Failures /Transformers x service years
Fig.1 and 2 represent statistical distributions of the failure rates over a wide range of time.
It was found that failures diagrams are only partly predicted by the classical ‘Bathtub’ curve.
One can suggest that failure profiles in the time range until 35-40 so far consists of wear-in in
company with random failures.

0,90 2,50
0,80
0,70 2,00
0,60
1,50
0,50
0,40
1,00
0,30
0,20 0,50
0,10
0,00 0,00
0

5
40

0
.1
..1

..2

..2

..3

.3

.1

.1

..2

..2

..3

.3

-4
5-
5.

0.

5.

0.

0.
35
10

15

20

25

15

20

25
3
3

3
Figure 1
Bar graphs of failure rates of autotransformers 220/110 kV (left) and
autotransformers 330/110 kV (right)

4,00
0,60
3,50
0,50
3,00
0,40
2,50
0,30 2,00
0,20 1,50

0,10 1,00

0,00 0,50
0,00
-3

-5

10 0

15 5

20 0

25 5
30

5
0
-1

-1

-2

-2

-3
-4

3 -5 10 15 20 25 30
-
0
3

30
35
-

3 - - - - -
0 5 10 15 20 25
5

Figure 2
Bar graphs of failure rates of autotransformers 500/220 kV (left) and
GSU transformers 400 MVA, 330 kV (right)

It’s remarkable that we have not observed symptoms of increasing failure rates for GSU
transformers with time. This finding also relates to very large GSU transformers rated from 600-
1000 MVA which are installed in CIS countries. One may suggest several reasons for this
phenomenon:

 Specification of moderate ratio of generator-transformer rated powers equal

approximately to 0.8.
 Application basically OFWF cooling system for the units rated above 600 MVA
 Specification of large generator transformers without LTC and NLTC
 Application only high quality Naphthenic based inhibited oil, membrane sealed
conservator and permanent regenerative filters to absorb initial by-products

Euro TechCon 2005 44

It is clear that failure profiles cannot be determined by a single function due to the presence of
several mechanisms of degradation. However, each common failure-mode can be described by a
particular function. Advanced wear in transformers during early years of service (first three
years) is associated basically with insulation and LTC failures. We then observe a rise in failure
rates in the time range of 7-15 years which is mainly due to the weak design of bushings and
mechanical movement of windings. These failure histograms closely mirror the total failures
histogram. One can observe wear in form of OLTC failures, though failure rate due to contact
deterioration can be approximated with lognormal distributions. There is no obvious evidence of
failures due to paper insulation aging. However, a clear trend of increasing insulation dielectric
failures with time was found.
70 40

60 35

50 30
25
40
20
30
15
20
10
10
5
0
0
0-3 3…5 5…10 10..15 15-20 20-25 25-30 30-35 35-40
0-3 3…5 5…10 10..15 15-20 20-25 25-30 30-35

Total failures Bushings failures

40
35 30

30 25
25 20
20
15
15
10 10

5 5
0 0
0-3 3…5 5…10 10..15 15-20 20-25 25-30 30-35 35-40
0-3 3…5 5…10 10..15 15-20 20-25 25-30 30-35

Mechanical winding failures LTC failures

16
14
12
10
8
6
4
2
0
0-3 3…5 5…10 10..15 15-20 20-25 25-30

Dielectric winding insulation failures

Figure 3

Euro TechCon 2005 45

Failure-modes of Power Transformers Versus Application

Failure analysis was performed on the basis of worldwide information available.

During the period 2000-2005, 108 major failures of large power transformers manufactured
by nine different companies were observed.

Table 1 compares power transformer failures reported from 1996-98 by Doble clients (52
failures) to ZTZ-Service. The data show that the average age of failed transformers is still
between 20-22 years. However, the percentage of failures from units “older than 25” is
becoming more significant. Note the meaningful number of “early failures” occurring in a
period less than 5 years.

Doble clients ZTZ-Service 2000-2005, %

1996-98, % GSU Transmission
Total number of failures 52-100% 45-100% 63-100%
Average age, years 22.4 21 20.5
Over 25 years % 43 44.1 32
Less than 5 years, % 7.5 2.94 9.4

Table 1
Major failures of power transformers rated 100 MVA and above

Table 2 (shown on the following page) represents failure-modes separately for

transmission autotransformers, generator transformers, and coupling
transformers. It was found that transformer application correlates with particular
failure modes. The most frequent failure modes that are inherent in specific
groups are:

Transmission transformers (autotransformers)

 Dielectric mode windings insulation (31.6%), basically HV and TW windings

involved
 Bushings insulation (38%)
 Winding movement (9.5%), basically common and tertiary windings involved

Generator transformers

 Dielectric mode windings (49%), predominantly HV windings involved insulation

 Thermal-mode failures (13%), basically attribute to the units that operate ratio of
generator-transformer rating 0.9, LV windings involved
 Leads and connections overheating (13%).
Auxiliary Power Plant Transformers (coupling transformers)

 OLTC insulation failures & contacts heating (43%)

 Leads and connections (14%)
 Winding movement (14%) , basically LV and sometimes TW involve

Failure -mode Component Transformer application

Coupling GSU Transmission
transformers
Dielectric Winding minor (turn, 28.5 37.8 14.3
coils) insulation
Major insulation - 11.2 17.3
Thermal Conductor insulation - 13.3 4.8
Mechanical Winding distortion 14.3 4.4 9.5
Magnetic Core/magnetic shields* - 4.4 4.8
circuit
Current Leads, connection 14.3 13.3 3.2
carrying
Accessories Bushing 13.3 38
OLTC** 42.8 4.4 7.9
DETC 2.1 -
Total 100 100 100
failures,%

Table 2
Failure-modes (major failures) of power transformers for different applications

Dielectric-related Failure Modes

Failure Causes

Statistics show that about 30-50% of the total number of dielectric failures have been
associated with damage to the windings.

A failure occurs when the dielectric withstand strength of the insulation space is exceeded by
operating stresses. Basically, three main causes of failure may be considered:

 Insufficient safety margin due to the underestimation of operational stresses

 Operational stresses exceeding specified levels

Euro TechCon 2005 47

  Critical deterioration of the safety margin

With time one can apparently expect a proportionally greater number of failures resulting
from deterioration of the safety margin which automatically increases the effects of initial
design margins as well as historical operational stresses. All three causes should be
considered in determining transformer integrity.

Visual damage is seen predominantly on HV windings and Tap Windings due to inherent
comparatively elevated dielectric stresses. Typically, the following failure modes are
involved:

 Breakdown of the space “ bushing shield –turret or tank wall”

 Creeping discharges across insulation between phases or space “winding-core”
 Overlapping the winding (HV) from line coils to neutral ones
 Breakdown between tap leads or from lead to earth
 Short-circuit between adjacent coils or group of coils
 Short –circuit between turns

In most cases the main reason for failure is associated with critical deterioration of the safety
margin. Failures have occurred with under-rated voltages or in combination with transients.
Degradation occurred mainly because of these critical factors:

 Entrance of free water through poor sealing of bushing lead or explosion vent;
 Conductive particles from the outside: metal particles from worn out pump bearings
or carbon form LTC diverter switch compartment
 Conductive particles from the within: conductive by-products from oil oxidation,
carbon from the site of localized oil overheating, and formation and sediment of
copper sulphide
 Contamination with “natural” particles, namely cellulose fibres in combination with
water

Failure Mechanisms Involving Insulation

Both experiments and experiences have shown that the dielectric strength of transformer
insulation is determined by the dielectric strength of oil. The following failure mechanisms
have been typically involved: breakdown of large oil gap, breakdown of oil duct between
coils, surface discharge, creeping discharge, and occurrence of destructive PD in oil layers
between conductors (turns).

Breakdown of Large Oil Gaps. Large oil gaps, particularly those that are not divided by
barriers are very sensitive to oil contamination from particles as well as to the distortion of
the electrical field on electrodes surfaces.

Dielectrophoretic forces attract particles to HV electrodes from a certain distance depending

on the electric field distribution at this electrode, on the particle shape and moisture

Euro TechCon 2005 48

concentration. The stirring of oil (e.g. from starting cooling pumps resulting in particle
migration) or the lowering of the oil temperature causing an increase in the relative saturation
and accordingly an increase of particle conductivity could result in breakdown even under
operating voltage. Poor electrode quality enhances the likelihood of failure.

Several failures have occurred due to the poor performance of 500 kV bushing shields;
namely the hidden defect metallic mesh with sharp edges. Figure 1 shows partial discharge
activity under the effect of switching surge and power arc to the turret .

Figure 1
Hidden defect on metallic mesh of resin covered shield from a 500 kV bushing

Breakdown of Oil Duct Due to Free Water. Poor or deteriorated seals of lead bushings
as well as the poor sealing of explosion vents allow the entry of rain water into transformers
which result in a sudden breakdown of the oil duct between coils of HV or TW windings .

Figure 2
Short-circuit between coils due to introducing free water
Left: damage of 400 kV winding due to penetration of water through poor bushing top
sealing
Right: Short circuit between coils of regulating winding disposed under explosion vent

Special studies of the impact moisture on winding insulation 4 shown that coil type windings
are mostly susceptible to moisture contamination. The presence of moisture within the oil
duct can reduce strength to an operating voltage of 6-10 kV, which is typical operating

Euro TechCon 2005 49

voltages between coils for HV and Tap windings. On the other hand wet paper results in
reduction of the dielectric withstand strength by only 20%.

Fig. 3
Impact of moisture on dielectric withstand strength of winding insulation
1. Disc winding with radial cooling duct (sensitive construction)
1a-Dry insulation and oil; 1b-Wet (not dried) insulation (strength reduction by 20%);
1c-high moisture in oil and on surface (strength reduction by 5 times)
2. Disc winding without paper insulation, high moisture in oil
3-Helical type winding without cooling ducts, high moisture in oil and on surface.

Surface discharge. The occurrence of surface discharge is basically associated with

transients. Two failure mechanisms are proposed: 1) Oil breakdown progressing into
insulation destruction and 2) Surface discharge as self-firing phenomenon.
The magnitude of the electric field tangential component that can result in PD and forcing oil
out of the pressboard could be the criterion for the dielectric strength across the insulation
surface. Using non-aged, dry and clean insulation it has shown that surface discharge can
occur under an electric field stress of 6.5-12.5 kV/mm on condition if the ratio of average and
maximum field intensity in the oil gap is 0.4-0.5 or less (e.g. sharp electrode). Apparently, the
contamination of surfaces with conductive particles reduces the value of critical field
intensity.

One particular “aging problem” is the accumulation of conductive and polar particles in oil
deposited on surfaces. Insulation surface contamination has been observed in the form of
adsorbed oil-aging products with cellulose or deposit of conducting particles and insoluble
aging products in areas of high electrical stresses. Contamination results in the distortion of
electrical fields and the reduction of surface discharge voltages. Studies on winding
insulation show that the deposit of sludge and a high contamination level can reduce the
dielectric withstand strength of impregnated insulation system under the effect of switching
impulses by 18-24% 5.

Euro TechCon 2005 50

The failures occur when there is a breakdown between coils and the HV winding under the
effect of switching surges and lighting impulses. In several cases the operation of SF6 circuit-
breakers triggered these failures.

Figure 4

Damage of insulation contaminated with conductive particles

Left: Failure of 330 kV winding contaminated with oil by-products under lighting surge
Right: Surface discharge across the barrier under 500 kV winding. Surface contaminated with
metal particles-products of wear out of oil pump bearing

Creeping discharge. This is, likely, the most dangerous failure mode that typically
results in catastrophic failures under normal operating conditions. The phenomenon occurs in
the composite oil-barrier insulation and progresses in several steps:

1) Partial breakdown of oil gap.

2) Surface discharge in oil across a barrier (an appearance of black carbonized marks on the
barrier).
3) Microscopic sparking within the pressboard, resulting in traces of carbon in the
pressboard.
The presence of some excessive moisture stimulates vapor bubble formation and the
degradation of material. The creeping process can continue for minutes to months or even

Euro TechCon 2005 51

years, until the treeing conductive path causes shunting of an essential part of the transformer
insulation resulting in a powerful arc.
.
Three critical factors are recommended in evaluating the likelihood of creeping discharge
occurrence:
1) Specific insulation design configuration (e.g. presence of creeping path across pressboard
between electrodes; winding disk-to-disk transposition touch to adjacent barrier; touch of
barrier to bushing or grounded details). Voltage class prone to insulation damage is 220 kV
and above.

2) High enough dielectric stresses: magnitude of tangential component of electric field stress
(1,0 kV/mm ).

3) Presence of source of initial critical ionization of high energy causing carbonized marks
on barriers: gas (air) bubbles (pumps cavitations, residual air after refilling with oil, and
intense local oil heating) penetration of free water, metal particles contamination, and static
electrification.

Figure 5
Creeping discharge progressing across the barrier of insulation between
phases of 330 kV autotransformer

Destructive PD Occurrence Between Turns. The process of turn-to-turn failures starts

from the occurrence of PD over 400 pC within oil layer between conductors with graduated
increasing PD intensity up to the paper destroying level of 100,000 pC and above. Tests on
winding models have shown that the total duration of failure progression is in the range of 5-
40 hours. The destructive stage could last up to 10 Hours . The period of time elapsing before
failure is too short to expect any preventive detection by means of DGA. Presumably only PD
on-line monitoring could detect a faulty state.

PD incipient dielectric stress for turn insulation under AC voltage is typically very high (over
20-30 kV/mm). In order to cause PD a combination of factors would be required: substantial
voltage between conductors, significant deterioration of dielectric properties and additional
strength reduction factors (e.g. reduction of PD incipient voltage at high temperature). It is
expected that stressed interleaving disc windings would largely be impacted.

Euro TechCon 2005 52

Recently, a number of transformer and shunt reactor failures occurred as turn-to-turn short-
circuits during normal operation due to dramatic contamination of insulation with copper
sulphide 6-10. It originated basically from non- inhibited oil containing presumably non-
corrosive sulphur components such as natural inhibitors. The impact of temperature electrical
stresses and time resulted in the transformation of non-corrosive sulfur to corrosive
molecules. Further reactions produced copper sulphide and deposit on paper.

Similar failures occurred also with HVDC transformers particularly with valve windings,
which are subjected to frequent and intensive transients, and DC fields that promote the
deposition of conductive particles on surfaces.
Tests show that copper sulphide sediment can result in an increasing dielectric loss factor up
to 38% at 100C 6, reduction of contaminated paper breakdown voltage from 20 kV/mm to
less that 1 kV/mm 9, and reduction of the PD initiation voltage between conductors from 20-
30 kV to 2 kV or less (5% probability) 10.
The damaged area of failed winding was typically confined within several coils.
The appearance of windings contaminated with copper sulphide and those contaminated with
oil sludge particles is very similar (fig. 6). In order to identify sulphur contamination
separation of the paper layer from conductor down to the copper would be required.

Figure 6

Appearance of winding with conductive deposit

Left: Discoloration due to copper sulphide deposit
Right: Discoloration due to oil sludge sediment. Winding construction involved the
arrangement of radial spacers to provide directed oil flow through the coils.

Thermal Failures

Analysis has shown the following failure causes for thermal failures:

Euro TechCon 2005 53

  Overheating of tap leads located between regulating coils of HV windings connected
to the no-load tap changer.
 Overheating of the coils of windings blocked with insulating collars preventing oil
flow and proper cooling.
 Underestimation of winding temperature, especially of LV windings in large
generator transformers with OFAF cooling system including insufficient cooling oil
through the windings.
 The main reason of failures was not normal aging but design deficiency.

Figure 7
Failure of 700 MVA generator transformer after 25 years due to overheating the
two top coils of LV winding (CTC wire) resulting in short circuit between
parallels and then between turns

There has been a rather common opinion e.g. 11 declaring:

“If the respective degree of polymerization of the insulating paper falls down into
interval DP = 400 … 300, the operation of the transformer must be ended definitively”.
Experience has shown that aging profile of large transformer is typically greatly nonuniform.
There have been numerous transformers particularly large generator transformers where DP
levels of some top winding components could be expected to be less than 300-250.
Accepting the aforementioned statement would necessarily require the removal of a
significant number of transformers, many of which exhibit in some locations DP numbers of
less than 250 but that operate quite satisfactory.

The question becomes: Maybe mechanical weaknesses of the conductor insulation are not so
dangerous as it was traditionally suggested? This question makes sense. Conductor insulation
is subjected basically to compressive stress. The reduction of the DP below 200-250 would
be likely not so critical for continuous disc windings and particularly for layer windings.
The exception would be for CTC (continuously transposed conductor) wire, which could be
subjected to very high compressive stresses. It’s remarkable that for the last few years there
have been a number of transformer failures associated with short-circuits insulation between
CTC wire strands because of overheating and critical decomposition of insulation.

Euro TechCon 2005 54

More sensitive to aging deterioration could be also winding construction allowing conductors
titling and bending under short-circuit stresses, and having sensitive spots affected by
mechanical stresses e.g. unsuccessful transpositions.

Figure 8
Construction of transposition allowing damage of insulating
under the effect of axial and twisting stresses

Damage to Leads and Connections

Over 13% of failures of highly loaded generator transformers are associated with overheating
leads and connections.
Basically three failure modes have been observed: Overheating the insulation of winding exit
leads; Overheating soldered connections; Overheating bolted connection to bushings

Overheated Lead Insulation

There have been several cases associated with the overheating of winding leads that
contained the same wire as the windings, which is typical when winding a transformer from
CTC wire

Euro TechCon 2005 55

 Figure 9
Overheating and burning out leads insulation in 700 MVA GSU transformer

Left: Burning out internal layers of HV lead

Middle: Overheating and short-circuit between parallels of LV lead exit
Right: Overheating of HV winding lead termination

Design review and relevant calculations have shown that the performance of leads with
winding wire without increasing cross-sections can be a subject of special concern especially
when a thick lead insulation is used. One should emphasize that design review is likely the
only effective tool to identify the problem. Considering a limited amount of overheated
insulation DGA and Furans analysis show clear symptoms of fault only at the stage when
short-circuit between strands and insulating burning occur.

Mechanical Failures

About 10% of transformers fail due to movement of winding under the effect of short-circuit
stress. Most failures (70% ) occur after 28-42 years of service and others in mid-age
(14-16 years).

Up to 80% of failures occur due to radial buckling of the common windings of

autotransformers and LV windings of step-down transformers. One generator transformer
failed under effect of short-circuit on the LV side. Tilting deformation and significantly
loosed winding clamping was revealed on the LV winding of step-down transformers that

Euro TechCon 2005 56

experienced three phase short-circuits with limited current but for long duration (during 1;530
sec). Tilting of conductors on HV windings was found also in GSU transformers as a result of
frequent short-circuit events on the HV side. Design review of the failed transformer revealed
that the 300 kV winding was performed as helical type and the safety margin to axial stresses
was only 0.84.

Design review using modern methods has shown that in most cases dynamic stability is not
sufficient to stand specified stresses. We used the method, which was developed by Dr.
Lazarev (Zaporozhye). The method allows pinpointing not only the likely damaged winding
but also the form of loss stability (Fig. 10). In most cases when wire from annealed copper
with conductor yield strength of less than 100 MPA, a radial form of loss stability could be
anticipated.

Figure 10 Forms of loss radial stability

Left : Half-shifted form of loss stability. No radial support, insufficient compressive force
Right: Shifted form of loss stability. No radial support, sufficient compression force

For example, a step-down 80/33 kV transformer failed due to the dramatic distortion of LV
windings (Fig 11). It was found that the transformer, which was manufactured in 1974 has
very low radial stability (Tabl. 3). Taking into account a long service life and inevitably
loosening clamps, a half-shifted form of deformation is expected.

Table 3
Winding, tap position Safety margin
Radial Axial
RW, max 6.4 6.4
LV, max 0.54 0.94
HV, max - 3.63

Euro TechCon 2005 57

 Figure 11
Radial buckling of LV winding after 3 phases short-circuit on LV side
Half-shifted form of deformation revealed.

Failures Associated with Magnetic Circuits

There have been a few cases of major failures associated with faults in magnetic circuit
systems, however, a number of cases occurred which caused intensive gas generation and
unwanted scheduled outages. Probable defects can be grouped under two general headings:
1) Defects associated with main magnetic flux, and 2) Those associated with stray flux.

Defects Associated with Main Magnetic Flux form loops from circulating currents
linked with main flux. In fact this group makes up about 20% of magnetic circuit failures and
it results in the dissipation of high energy and intensive gas generation with the activation of
Buchholz relay. The cases observed were basically attributed to loosening winding press
bolts and short-circuit to metallic press rings or to core yoke.

Defects Associated with Stray Magnetic Flux present the main cause of localized oil
overheating and gas generation, and DGA concern. They can be classified into two groups:

1) Overheating under effect of eddy current induced by intensive stray flux

2) Overheating and (or) sparking in a loop for circulating current, linked with stray flux.
In the first group typical defects are overheating of the core frame due to absence or improper
disposition of magnetic shields on the frame (Fig.12 left), overheating of pressing of the
pressure bolt that situated just under the core yoke (Fig.12 right), overheating a part of the
tank wall due to improper shielding.

Euro TechCon 2005 58

 Figure 12

Local overheating due to eddy current induced by stray flux

Left: overheating the bottom frame and adjusted insulation in 730 MVA
generator transformer due to improper disposition of magnetic shields
Right: Overheating the pressing jack in 417 MVA generator transformer

Loop currents depend on the electromotive force induced by the magnetic flux F , resistance
of the members that form a loop Z cir , and contact resistance Rtr

E cir
I cir ; (1)
zcir Rtr
Fm
Ecir 2f  (2)
2
Loop resistance is of an order Zcir 10-3 Ohm and inducing electromotive force even of 1V
Results in current up to 1000 Amps.

Two mechanisms of overheating of members that form circulating current loops have been
observed:

1) Loose contact in circulating loop provided with construction (Fig. 13)

2) Shorting between core members forming the loop:

 Shorting magnetic shunts to core and the tank
 Shorting bottom frame to tank (Fig. 14)
 Shorting top frame to tank

The latter forms loops of large dimensions, allowing induced voltage up to 10 V and resulting
in heating and arcing.

Euro TechCon 2005 59

 Figure 13
Overheating in the loosened areas between members forming
circulating current loops in 700 MVA transformer

Figure 14
Traces of overheating in location of contacts
Core frames with tank bottom

Euro TechCon 2005 60

Bushing Failures

HV bushing remains one of the weakest transformer components responsible sometimes for
more than 30% of transformer failures. Recently it was reported10 that sixty-three failures of
bushings on large power transformer have occurred since 1995 from one manufacturer. It is
remarkable that the age of failed bushings was only between 2 and 15 years. Fifteen failures
were accompanied with an explosion and likely destruction of the transformers. Forty-nine
bushings were removed from service due to signs of PD gases. Twelve failures were
associated with overheating paper of the core after just 6-8 years of service.

A recent failure survey in Australia and New Zealand 3 shows that bushing explosions are the
main reason for oil fires. Survey data associated with fires events from 2002-04
include eleven transformer failures and ten fires were caused by oil-paper bushing and cable
box failures. It was found that Risk Transformers causing oil fire = 0.09 % or
~ 1 / 1000 Transformer years.

Failure Modes. Experience has uncovered the following failure modes:

 Internal discharges leading to internal gas and pressure build up and ultimately an
electrical breakdown between the central conducting tube and the bushing flange,
which could be caused by the paper not being properly impregnated with oil. Design
review of some core construction revealed overstressing of some condenser layers and
possible mechanical sliding (displacement) across the central tube
 Mechanical failure of the central support tube allowing loss of oil within the bushing.
 High temperature of the central tube and adjusted paper, during overloading.
 Deterioration of copper grounding layer in contact with aluminum foil.
 Vacuum formation in oil-gas separation system due to underestimation of volume of
nitrogen cushion, followed with water enter

CIGRE WG A 25 “Bushings Reliability” was set up in 2004 with the main aim to improve
bushing reliability or at least to prevent the decrease of bushing performance (trend due to
economic pressure), the long term impact of which can be catastrophic for transformer
reliability.

Effect of Conductive Residue on Porcelain. Special attention should be made towards

failure modes associated with the degradation of the dielectric withstands strength of oil and
across the core and porcelain surfaces that progresses in flashover along the surface. These
phenomena are typically originated from critical aging the oil, formation of semi-conductive
residue on the lower porcelain;

Discharges across the inner part of the transformer end porcelain are an outcome of a
typical aging-mode phenomena in the bushing. The failure process is initiated and developing
within the oil channel between the core and lower porcelain. Another option is formation
conductive residue on the external porcelain surface by means of attracting conductive by-
products from transformer oil.

Euro TechCon 2005 61

 Figure 15
Formation of semi-conductive residue on lower porcelain
Left: Internal staining with aged oil by-products containing metal colloids
Right: Deposit of by-products of external surface attracting by electrical field from
aged oil in 400 kV generator transformer. Concentration residue on side facing tank wall

Electric field intensity in the oil channel and across the surfaces of core-end components and
inner porcelain is established both by the bushing insulation construction and by disposition
of the bushing end relative to the grounded parts and the winding.

Impact of Transformer on Bushing State. The transformer in many instances

significantly affects the oil temperature within the bushing. Hot transformer oil is one of the
main sources of the bushing heating. Another two sources are dielectric losses in the core and
resistance losses in the central conductor. Heat radiated from the tank top cover is a source of
elevating temperature of the cooling medium (air around the bushing). Current density
through the central conductor and actual transformer/bushing current ratio including
permissible transformer overloading determines hot spot temperature within the bushing
affecting paper temperature.

The transformer distorts the electrical field within and around the bushing. Strengthening the
electrical field within the bushing, specifically, in the oil between the core and lower
porcelain due to the approach of conductive layers to the grounded components and
transformer winding should be considered. Accordingly, the contamination of transformer
oil with conductive particles may result in those particles being attracted by the bushing’s
electrical field and depositing on the surface (porcelain) and dramatically deteriorating the
dielectric strength

Euro TechCon 2005 62

 Figure 16
Electrical field of the bushing porcelain surface
The electrical field intensity across the porcelain surface facing the tank
wall may be as much as six times greater than that on the opposite side

Failure of OLTC’s

The failure rate of power transformers associated with OLTC problems varies in the range of
5-20 %. Failure analysis incorporating design review and considering both OLTC and the
transformer itself highlights some factors that impact on reliability issues:

 In many instances mechanical and dielectric performance of the transformer is

determined by the state of tap winding and leads.

 Choice of OLTC with a low ratio of maximum rated through fault current for LTCand
the maximum current for the transformer. According to IEC 60542 this ratio shall be
at least 120%, and the temperature rise of LTC contacts above the oil shall be not
more than 20C. Experience has shown that for rarely moved contacts the temperature
rise shall be less that 15C.

 Underestimation of the impulse transfer function allowing in some cases voltage

values between steps or with respect to ground above the test voltage of the OLTC.

 Many failures initiated with short-circuit between steps in diverter or selector

switches were accompanied with distortion of the tap winding. In fact, it is rare that
transformers are designed to withstand a short-circuit between LTC taps.
Accordingly, a comparatively minor failure in OLTC (e.g. burning out resistance) has
resulted in major transformer failures and long-term unit non-availability.

Euro TechCon 2005 63

Experience highlights the necessity to pay more attention to the diverter switch reliability.
In particular, the following factors should be considered:

 Aging deterioration of the oil due to the effect of a high resistors temperature
 By-products sediment on insulating surfaces affect on degradation of dielectric
strength
 Mixture of carbon, water and polymerized by-products are difficult to filter out
properly
 Temperature rise of shunt contacts can sometimes exceed temperature rise of selector
 Reversing contacts are a weak spot that requires special attention
 Contact overheating can result if flashover between the phases followed with
explosion and fire.

Figure 17

Dielectric-mode failure in OLTC diverter switch

Left: Flashover between the contacts cause by cooking and oil contamination
Middle: Deformation of Tap winding under short-circuit stresses
Right: Insulation contamination and PD activity in diverter switch

Euro TechCon 2005 64

References
1. Stig Nilsson,Stan Lindgren “Review of Generator Step-up Transformer Failure data”,
EPRI Substation Equipment Diagnostics Conference V!, Feb 16-18,1998.

2. William H. Bartley “Analysis of transformer failures-a twenty year trend”

Proceedings of the 2000 International Conference of Doble Clients - Sec 8-5.

3.A Petersen, P L Austin “Impact of recent transformer failures and fires. Australian and
New Zealand Experiences” 2005 CIGRE A2 (Transformers) Colloquium, Moscow.

4. Ryzhenko, V. and Sokolov, V., "Effect of Moisture on Dielectric Withstand Strength of

Winding Insulation in Power Transformers," Electrical Stations [Electric Power Plants],
1981, No. 9 [in Russian].

5. L.Lungaard, D.Linhjell, J.Sund, G.Jorendal “Influence simultaneous AC stresses on

Impulse Breakdown in oil-paper insulation system” Eight International Symposium on High
Voltage Engineering , 1993.

6. T. V. Oommen ,C. C. Claiborne R. S. Girgis Wayne Ball“Sulfur Corrosion Tests and

Corrosion Effects in Transformers”TechCon NA Proceedings, 2003.

7. Corrosive Sulphur its origin detection and prevention” Siemens presentation at Cigre SC-
A2 Colloquium “Transformer Reliability and Transients”,20-24 June, 2005, Moscow, Russia.

8. Areva Presentation at ABINEE, Workshop on Corrosive Sulphur in Oil, June 2-3 2005,
Sao Paulo, Brasil.

9. Toshiba Presentation at ABINEE, Workshop on Corrosive Sulphur in Oil, June 2-3 2005,
Sao Paulo, Brasil.

10 Quality of oil makes the difference. ABB discovers the solution to transformer
breakdowns
ABB Review 3/2004.

11. J.Needly, G. Newesely “Evaluation of the extent of ageing of paper in oil immersed
power transformers”, CIGRE paper D1-302, CIGRE session 2004

12. Reiner Krump Discussion on Marshall F. Turley paper “Recent Failure Experience with
HSP Bushings”, Proceedings of the International Conference of Doble Clients, 2004, Paper
BIIT-3-A

13. V.Sokolov“Detection and Identification Typical Defects and Failure Modes in HV

Bushings”,
Colloquium SIGRE A2 in Merida, Mexico, 2003

14. V.Sokolov “Detection and Identification Typical Defects and Failure Modes in LTC”,
Colloquium SIGRE A2 in Merida, Mexico, 2003

Euro TechCon 2005 65

Biography

Victor V. Sokolov received his degree in electrical engineering from the Kharkov
Polytechnical University in Ukraine in 1962. In 1964 he completed a postgraduate program at
the National Polytechnic Institute in Moscow with a major in Physics of dielectrics. His PhD,
received in 1982 from Kiev Polytechnic University, is in the area of EHV transformer
diagnostics.

He started his professional career at the Transformer Research Center in Zaporozhye. Until
1990 Dr. Sokolov worked in the Installation and Maintenance Department at the
Zaporozhtransformer Corporation in the area of reliability.

Since 1990 he is a technical director of the Scientific and Engineering Center ZTZ-Service in
Zaporozhye, Ukraine.

Dr. Sokolov is a member of CIGRE (SC A2, Transformers). He was convener of the WG A2
18 “Transformer Life Management”, and Special Reporter on preferential subject “On-Site
Operation” at the CIGRE Transformers Session in 2004.
He has published over 100 technical papers including 13 papers for TechCon’s Conferences

Euro TechCon 2005 66