You are on page 1of 10

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/375860407

Analysis of Major Failures of Power Transformers

Conference Paper · November 2023

CITATION READS

1 231

3 authors, including:

Stefan Tenbohlen
Universität Stuttgart
439 PUBLICATIONS 5,944 CITATIONS

SEE PROFILE

All content following this page was uploaded by Stefan Tenbohlen on 29 January 2024.

The user has requested enhancement of the downloaded file.


Analysis of Major Failures of Power Transformers
S. Tenbohlen1, Z. Hanif1, D. Martin2,
on behalf of Cigre WG A2.62
1
University of Stuttgart, Stuttgart, Germany
2
Essential Energy, Port Macquarie, Australia
E-mail: stefan.tenbohlen@ieh.uni-stuttgart.de

Abstract— Accurate information about the service experience of high voltage equipment is of significant value to both the
electric utilities and to manufacturers of such equipment. It helps the manufacturers improve their products, and provides
important inputs for the utilities when organizing maintenance and benchmarking their performance. Statistical analysis of
the past failure data can display useful features with respect to the future failure behavior. Equipment reliability data are also
required when assessing the overall reliability of an electric power system, including studies of the electric energy supply
security. Furthermore, international standards applicable to high voltage equipment are being improved on the basis of service
experience and reliability data.
This contribution addresses the analysis of transformer failures collected by Cigre WG A2.62. Based on a transformer
population with more than 425,000 unit-years and 1,159 major a failure rate of app. 0.3 % p.a. was determined. Failure location
and mode analysis is presented for different voltage classes, along with external effects. Winding related failures appear to be
the largest contributor of major failures. Bushing failures most often lead to severe consequences like explosion or fire.

Index Terms—Failure statistic, power transformer, failure rate.

I. INTRODUCTION

Power transformers are an integral part of power systems, and their reliable operation directly impacts that of the whole
network. Internationally, utilities are aligning their strategic directives to meet the necessary requirements for reliable
power system operation. Periodic review of failures statistics becomes necessary where they can be used to influence
transformer design and technology, maintenance and condition monitoring practices with changing system loading,
operation and network configuration.
CIGRE has a rich history of collecting statistical data on failure, with the earliest transformer survey being performed in
1983 [1]. Learnings from these surveys have helped manufacturers improve their products, users understand risks, and to
plan maintenance. As a statistical analysis uses historical data to forecast events, it is worth repeating this work
periodically to ascertain how reliability has changed. These days there is more use of renewables, an emphasis to
decarbonise, interest in information technology to oversee assets, and the emergence of electric vehicles. It will be
interesting to observe if any of these new technologies impact the reliability of the power transformer. Thus, it is important
to periodically revisit failure statistics. The last CIGRE international survey was published in 2015 by WG A2.37 [2].
However, it was not possible to compare failure rates dependent on age (hazard curves), because the age distribution or
the operating units was not known. Consequently, a new survey was performed which does have these failure rates. Since
2015 there was an analysis of Australian power transformer failure, where the hazard curves were calculated. A distinct
wear-out distribution was observed to occur after a random failure period [3] - [4]. Thus, for the new A2.62 survey the
methodology was repeated.
The questionnaire developed by WG A2.37 was optimized to collect utility failure statistics in a standardised way [5].
Besides information about the population under investigation, failure data was collected for various groups of
transformers in terms of the failure locations, failure causes, failure modes, actions, external effects and others parameters.
Based on the experience of WG A2.37, the decision was taken to limit the data collection to only major failures of
transformers and reactors operating at 100 kV and above. Additionally, to the previous working group analysis, retirement
data was also collected for this work. A transformer may be retired before it actually fails, and therefore not be included
in a failure list, and so the failure rate calculation may underestimate the number of transformers which have failed
functionally. Two types of retirements are distinguished in this report: retirement due to condition/age, and, retirement
due to network requirement. In the first case, the transformer is retired from service before it fails, after detection of a
condition that prevents from safe or reliable operation. A controlled outage combined with scrapping is the typical
scenario for this type of retirement, and in practice this would be the desired end of life of a transformer. For the second
type of retirement, the transformer is retired without a real concerning issue, but due to major network upgrades like
voltage level, rated power or loss requirements; the unit does not have necessary to be scrapped, as it could still be used
as a spare unit or back-up. The retirements were also used to calculate hazard curves by being included as suspensions.
A major failure was defined as any situation which required the transformer to be removed from service for a period
longer than 7 days for investigation, remedial work or replacement. The necessary repairs should have involved major
remedial work, usually requiring the transformer to be removed from its installation site and returned to the factory. A
major failure would require at least the opening of the transformer or the tap changer tank, or an exchange of the bushings.
A reliable indication that the transformer condition prevents its safe operation is considered a major failure, if remedial
work (longer than 7 days) was required for restoring it to the initial service capability. In some cases, also failures were
assigned as major, if remedial work was shorter than 7 days and extensive work with oil processing had to be done (e.g.
exchange of bushings).
II. DATA COLLECTION

The different members of the Working Group A2.62 approached the main utilities within their regions to invite them to
participate in the data collection process by providing their anonymous transformer fleet data, failure and retirement
records. The utilities which participated in this data collection process completed the Excel questionnaire given in annex
A, and the responses were monitored by each of the leading Working Group member for each utility. Special attention
was paid to the correct data entry, like the proper time frame when the major failures occurred, the proper voltage level,
the details about occurrences, the consequences of failures and the retirement reasons. All the responses were compiled
into a general database for analysis.
In order to achieve data security and anonymity, the failure data from each source was labelled with a code based on the
geographical location and a sequence number. Information about the transformer manufacturer was not collected.
III. DATA ANALYSIS

For this reliability survey, data from 1,204 major failures and 1,916 retirements were collected from a population of
37,104 transformers with a voltage rating ≥ 100kV. These failures occurred between 1987 and 2021 contributed by 66
utilities from 27 countries. The reference period describes the period for which the participants reported failures.
• Number of major failures = 1,204
• Number of retirements= 1,916
• Number of transformers = 37,104
• Total population = 425,030 transformer years
• Number of utilities = 66
• Number of countries = 27
• Year of manufacture = 1919 - 2020
• Reference period = 3 - 34 years
Table I shows the investigated population according to voltage rating of the transformers. About 65% of total transformer
population has a lower voltage rating, between 100 ≤ U < 200 kV. The transformer population decreases substantially
with increasing voltage rating, which while unsurprising means fewer failures to analyse. Only 6 participating utilities
have transformers with voltage rating above 700kV, comprising 0.68% of the total population. The calculated failure rates
for transformer population are also given in Table 1. The failure rate is below 1 % except for the voltage class ≥ 700 kV.
TABLE 1: TRANSFORMER POPULATION DATA DEPENDENT UPON VOLTAGE CLASS.

Population Highest System Voltage [kV]


Information 100≤ U < 200 200 ≤ U < 300 300 ≤ U < 500 500 ≤ U < 700 U ≥ 700 Total

No of utilities 60 54 39 16 6 66

No of transformers 24,106 6,619 4,929 1,194 256 37,104

Transformer-years 268,887 86,881 51,506 15,561 2,459 425,294

No of failures 619 247 228 74 36 1,204


Failure rate 0.23% 0.28% 0.44% 0.47% 1.46% 0.28%
No of retirements 1,153 422 194 76 71 1,916

A. Failure Rate
The failure rates are calculated using the number of failures and transformer years with the assumption that the number
of in-operation transformers remains constant during the reference period. The full failure population (1,204 failures) is
used in this analysis. The failure rates according to the voltage class for the substation, generator step-up transformers,
and shunt reactor transformers, are given in Figure 1 respectively. The overall failure rates in all three types of
transformers are below 1%. The failure rate of GSUs with voltage above 700 kV and shunt reactor operating at voltage
between 300 kV to 500 KV is higher than 1 %. However, it is important to note that the number of failures, as well as the
transformer population of GSUs and shunt reactors, is significantly lower than substation transformers. Thus, the
calculated failure rates should be considered with caution.
Figure 1: Failure rate in transformers dependent upon voltage class and application.

B. Age Distribution
Fifty-eight of the utilities supplied the age distribution of their transformer fleet. The populations for Europe and North
America are shown below in Figure 2 normalised to fleet size, to emphasise features in the age distribution. The relative
weighting of the fleet age distribution will impact the accuracy of the failure rate calculation. For instance, a fleet being
young will create more uncertainty when attempting to extrapolate a failure rate for aged transformers compared to an
older fleet. Europe has a large proportion of new units, but also ones around fifty years. North America has a high
proportion of older units.
This data was analysed to investigate relationships between age, voltage class and probability of failure. Both
nonparametric and Weibull distribution methods were used. The Weibull distribution is fitted over the data, and then
compared with the nonparametric model to assure a close fit. In this case, the nonparametric model was dividing the
number of failures for each year of age by the service-years observed for that age (the hazard rate), which is shown in
Figure 3. The data was combined and checked using an Excel spreadsheet, before being processed using Python scripts.

Europe North America

0.03

0.025
Fraction of fleet

0.02

0.015

0.01

0.005

0
1 6 11 16 21 26 31 36 41 46 51 56 61 66
Age (years
Figure 2: Age distribution Europe and North America.

C. Hazard Curve
For the European and North American populations, their respective hazard curves for failures are shown in Figure 3,
where the transformer was subsequently scrapped. These hazard curves are for all voltage levels 100 kV and above.
Failure Weibull Upper CI Failure Weibull

1.0% 1.0%

0.8% 0.8%

0.6% 0.6%

0.4% 0.4%

0.2% 0.2%

0.0% 0.0%
0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70
Age (years) Age (years)

Figure 3: Hazard curve of failures for European (left) and North American (right) transformers

Curve fitting software was used to fit the Weibull distributions. The Weibull shape parameter was 3.0 for the European
data and 2.8 for North America. A shape parameter higher than 1 indicates a wear-out with age pattern in the data. The
95% confidence interval was plotted for the shape and scale parameters, shown in the European data. However, there was
an insufficient number of failures from North America for a reasonable confidence interval to be determined using the
model, and so the 95% confidence interval is not indicated for this set of transformers. While the European data has a few
high points for ages greater than 60 years, its Weibull distribution is similar to the North American data. For instance, at
70 years the hazard function calculated using the Weibull distribution is around 0.2% for both datasets.
Substation transformers are normally not so heavily loaded which could make ageing more pronounced. Failures due to
ageing can be masked by random or external failure reasons and early replacement of transformers. Utilities often use
replacement strategies, in which preferably old transformers are replaced, which do not anymore fulfill the operational
requirements (e.g. losses, noises, regulating range). Thus, potential failures of replaced units do not contribute to the
hazard curve.

D. Failure Location
The failure data of the full transformer population, with 1,204 major failures, were analyzed as a function of failure
location, external effects, failure mode, failure cause, detection mode, and action taken after failures.
The failure data is classified according to primary location (component) of major failures. The failures for which location
was not described were excluded from this analysis. Thus, the number of failures reduces to 848. In Table 2 the failure
location analysis is shown according to voltage class for the full population. It can be appreciated that winding, tap changer
and bushing related failures were the major contributors, followed by lead-exit related failures.
Figure 4 shows the failure locations for all major failures. In entire population the failures are predominantly winding
(36.7%), bushings (24.9%), and tap changer (18.8%) related. The contribution of bushing related failures increased with
increasing voltage class. The contribution of tap changer related failures appeared to decrease with increasing voltage
level, this may be attributed to the fact that at higher voltage level, less transformers are equipped with tap changers. The
contribution of tap changer related failures slightly decreased by 3 % in comparison with the statistics from WG A2.37
[2]. On the other hand, the bushing related failures increased by 10% in comparison with WG A2.37. Lead exit (4.7%),
core (2.7%), and insulation (3.5%) are minor contributor to major failures. For 4.2% of the failures the location was given
as unknown.
In substation and GSU types, winding related failures are major contributors. However, substation transformers have
higher participation of winding (38.1%), bushings (24.3%), and tap changer (20.3%) failures than GSUs with 29%, 14.5%,
and 14.5% respectively. GSUs have a higher contribution from core and magnetic circuit failures (11.3%) which is a
minor contributor (1.5%) in substation transformers. On the other hand, in shunt reactors, bushing related failures are
main contributors with 48%, followed by 24% in windings and 10.8% in core & magnetic circuits. The higher contribution
of bushing and core failures in shunt reactors can be attributed to the fact that shunt reactors are severely exposed to
vibrations, which can lead to bushing and core failures. Lead exit, tapping leads, and insulation (phase to phase, winding
to winding, and winding to ground) are minor contributor to major failures in all types of transformers.
TABLE 2: FAILURE LOCATION ANALYSIS DEPENDENT ON VOLTAGE CLASS

Highest System Voltage [kV]


Failure Location 100≤ U < 200 200 ≤ U <300 300 ≤ U < 500 500 ≤ U < 700 U ≥ 700 Total
% % % % % %
Tapping Leads 4 1.1 1 0.5 0 0 3 4.5 0 0.0 8 0.9
Bushings 60 16.3 45 23.5 69 34.7 23 34.8 14 43.8 211 24.8
CT 0 0 1 0.5 1 0.5 0 0 0 0 2 0.2
Phase to phase
2 0.5 2 1.0 1 0.5 0 0 0 0 5 0.6
Insulation
Electrical screen 2 0.5 0 0 3 1.5 0 0 1 3.1 6 0.7
Windings 165 44.8 53 27.7 61 31.9 23 34.8 9 28.1 311 36.6
Winding to
ground 6 1.6 3 1.5 3 1.5 0 0 1 3.1 13 1.5
insulation
Tank,
Conservator, 9 2.4 4 2.0 6 3.0 0 0 2 6.3 21 2.5
Piping
Unknown 19 5.2 6 3.0 10 5.2 0 0 1 3.1 36 4.2
Winding to
Winding 6 1.6 3 1.5 1 0.5 1 1.5 0 0 11 1.3
Insulation
Lead exit 16 4.3 8 4.0 12 6.0 4 6.1 0 0 40 4.7
Tap changer 73 19.8 60 30.2 16 8.3 10 15.2 0 0 159 18.6

Core and
5 1.4 5 2.5 7 3.6 2 3.0 4 12.5 23 2.7
magnetic circuit
Internal surge
1 0.3 0 0 1 0.5 0 0 0 0 2 0.2
arrester
Total 368 191 191 66 32 848

Figure 4: Failure location analysis based on 848 major failures

E. Failure Mode Analysis


The failure mode describes the nature of the failure illustrating what happened when the failure occurred. The failure
mode analysis of transformers according to voltage class is shown in Table 3. Electrical (35.9%) and dielectric (30.7%)
mode failures were the most prominent, followed by mechanical (18.8%) type failures. Dielectric failure mode means
partial discharge, tracking and flashover. Electrical failure mode means open circuit, short circuit, poor joint, poor contact,
ground deterioration, floating potential. Electric mode failures were the major contributor in the 100kV to 200kV, 200kV
to 300kV, and the 500kV to 700kV voltage classes, with the highest contribution occurring in 500kV to 700 kV. Whereas
dielectric mode failures were most prominent in transformers with voltage rating between 300kV to 500kV. Compared
to the WG A2.37 statistic the percentage of electrical mode failures has increased by almost 20%.
TABLE 3: FAILURE MODE ANALYSIS DEPENDING ON THE VOLTAGE CLASS

Highest System Voltage [kV]


Failure mode 100≤ U < 200 200 ≤ U < 300 300 ≤ U < 500 500 ≤ U < 700 U ≥ 700 Total
% % % % % %
Dielectric 92 24.4 46 29.5 70 44.6 24 36.4 10 31.2 242 30.7
Mechanical 88 23.3 29 18.6 22 14 4 6.1 5 15.6 148 18.8
Electrical 136 36.1 56 35.9 44 28 37 56.1 10 31.2 283 35.9
Unknown 29 7.7 12 7.7 7 4.4 0 0 4 12.5 52 6.6
Thermal 10 2.6 8 5.1 12 7.6 1 1.5 1 3.1 32 4.1
Physical 13 3.4 5 3.2 2 1.3 0 0 2 6.2 22 2.8
Chemistry
LTC & Failed 9 2.4 0 0 0 0 0 0 0 0 9 1.1
testing
Total 377 156 157 66 32 788

Figure 5: Failure mode analysis for 788 failures

F. Failure Cause
Determining the root cause of failure can be a difficult and extensive task. Failure can often be attributed to different
failure causes. 16 % of the failure causes are described as unknown. Furthermore, 23 % are described as other reason,
although the causes suggested in the questionnaire actually covers all possible causes in the opinion of the WG members.
This means that a total of 39% of the causes cannot be traced back to a definite cause. This number is increased by 5% in
comparison with WG A2.37. The results presented in this section should therefore be treated and interpreted with caution.
Table 4 represents the failure cause analysis according to the voltage class of the transformers. External short circuits,
and material are major causes for transformer failure. Electrical short circuit related failures are highest in transformers
with voltage class 100 – 200 kV, and these causes exhibits decreasing trend across the voltage classes. On the other hand,
the percentage of material related failures is highest in transformer greater than 700 kV, and contribution of these causes
appeared to decrease with decreasing voltage class.
TABLE 4: FAILURE CAUSE ANALYSIS DEPENDING ON THE VOLTAGE CLASS.

Highest System Voltage [kV]


Failure cause
100≤ U < 200 200 ≤ U < 300 300 ≤ U < 500 500 ≤ U < 700 U ≥ 700 Total
% % % % % %
Overload 4 1.0 4 2.4 1 0.58 0 0 1 3.1 10 1.2
Ageing 27 7.3 12 7.8 20 11.6 12 20 4 12.5 75 9.5
External short circuit 76 20.7 10 6.5 11 6.4 1 1.6 0 0 98 12.5
Unknown 61 16.6 20 13.0 28 16.3 10 16.6 6 18.7 125 15.9
Loss of clamping 8 2.2 3 2.0 1 0.58 3 5 0 0 15 1.9
pressure
Vandalism 3 0.8 1 0.6 0 0 0 0 0 0 4 0.5
Other reasons 53 14.4 57 37.2 49 28.6 10 16.6 11 34.3 180 22.9
Material 31 8.4 11 7.2 26 15.2 12 20 7 21.8 87 11.1
Overheating 11 3.0 3 1.9 3 1.7 2 3.3 0 0 19 2.4
Over voltage 20 5.5 6 3.9 7 4.1 4 6.6 0 0 37 4.7
/Lightening
Oil Quality 22 6 2 1.2 2 1.1 2 3.3 0 0 28 3.5
External Pollution 5 1.3 1 0.6 3 1.7 0 0 0 0 9 1.1
Collateral damage 3 0.8 1 0.6 3 1.7 0 0 0 0 7 0.9
Transport/Site 7 1.9 5 3.2 7 4.1 0 0 0 0 19 2.4
Installation
Corrosive Sulphur 9 2.4 3 2.0 1 0.58 1 1.6 0 0 14 1.7
Improper maintenance 9 2.4 12 7.8 5 2.9 1 1.6 3 9.3 30 3.8
or repair
Loss of cooling 2 0.5 1 0.6 1 0.58 0 0 0 0 4 0.5
Design/ Manufacturing 7 1.9 0 0 3 1.7 2 3.3 0 0 12 1.5
Repetitive through 9 2.4 1 0.6 0 0 0 0 0 0 10 1.2
faults
Total 367 153 171 60 32 783

Figure 6: Failure cause analysis for 783 failures.


G. External Effects
The impacts of transformer failure can impose significant economic, financial and environmental implications. Table 5
represents the external effects due to failures with reference to the voltage class. Figure 7 shows the external effects of
728 major failures. Most of the major failures do not have any external effects (61.7%). 14% of failures lead to fire while
for an equal amount (14.3%) of major failures, external effects were related to explosions or bursts without fire. 9.3% of
failures resulted in oil leakage only.
TABLE 5: EXTERNAL EFFECTS OF FAILURES DEPENDING ON THE VOLTAGE CLASS

Highest System Voltage [kV]


External effects 100≤ U < 200 200 ≤ U < 300 300 ≤ U < 500 500 ≤ U < 700 U ≥ 700 Total
% % % % % %
Explosion with fire 37 10.1 12 9.1 33 23 16 26.6 6 18.7 104 14.2
Leakage only 24 6.5 17 12.9 20 13.9 4 6.6 3 9.3 68 9.3
None 270 74.2 82 62.6 66 46.8 23 38.3 8 25 449 61.6
Explosion w/o fire 30 8.2 20 15.2 20 13.9 17 28.3 15 46.8 102 13.9
Unknown 3 0.8 0 0 2 1.4 0 0 0 0 5 0.6
Total 364 131 141 60 32 728

Figure 7: External effects analysis for 728 failures

IV. CONCLUSION AND RECOMMENDATION

The working group has collected 1204 major failures which occurred in the period 1987 and 2021, within a total
population of 425,030 transformer years, contributed by 66 utilities from 27 countries. This paper presents an excerpt of
the findings of Cigre WG A2.62. The full results will be published in a brochure in 2024.

The overall failure rates of shunt reactors, GSU, substation and the combined group of transformers were below 1%. The
failure rate of GSUs with voltage above 700 kV and shunt reactor operating at voltage between 300 kV to 500 kV is
higher than 1 %. However, it is important to note that the number of failures, as well as the transformer population of
GSUs and shunt reactors, is significantly lower than substation transformers. The hazard curves were also calculated for
the European and North American populations, and for all voltage levels 100 kV and above. For these populations a low
hazard rate was observed below age of 20 years. From the available data, a particular age of transformer cannot be derived
where the probability of failure is high, but statistics show a slight failure rate increase with age.
In entire population, the failures are predominantly winding, bushings, and tap changer related. Windings related failure
proved to be largest contributor of major failures, irrespective of the transformer voltage class. Substation transformers
had higher contributions of winding failures than GSU transformers and shunt reactors. Shunt reactors on the other hand
had higher contributions of bushing related failures than GSU and substation transformers.
Electrical and dielectric modes failures were the most prominent irrespective of the voltage class of transformers.
Substation and GSU transformers had higher contributions of electrical type failures. While shunt reactors had higher
contributions of dielectric type failures.
In root cause analysis study, external short circuits appeared to be the major contributor. However, the root cause of a
large portion of the population was unknown and, in some cases, different failure causes were present. Thus, results of
root cause analysis should be evaluated with caution.
It is recommended to develop proper guidelines for collecting and compiling data in a standardized way. Accordingly, a
detailed questionnaire should be circulated among worldwide utilities along with the guidelines of the recording of the
failure data after failure cause analysis.

V. ACKNOWLEDGEMENT
The authors appreciate the discussions within CIGRE Working Group A2.62 “Analysis of AC Transformer Reliability”
and thank the transformer specialists of the utilities involved in the data collection for their valuable support.

VI. REFERENCES

[1] A. Bossi, J. Dind, J. Frisson, U. Khoudiakov, H. Light and e. al, "An international survey on failures in large power transformers in service,"
Electra, no. 88, pp. 21-48, 1983.

[2] S. Tenbohlen, J. Jagers, G. Bastos, B. Desai, B. Diggin and e.a., "Transformers Reliability Survey," Cigre Technical Brochure 642, Paris, 2015.

[3] D. Martin, J. Marks and T. Saha, "Survey of Australian Power Transformer Failure Modes and Retirements," IEEE Electrical Insulation
Magazine, vol. 5, no. 33, 2017.

[4] D. Martin, J. Marks, T. Saha, O. Krause, G. Russell and A. Alibegovic-Memisevic, "On the Development of Power Transformer Failure Models:
an Australian Case Study," in IEEE PES General Meeting, USA, 2017.

[5] CIGRE WG A2.62, “Questionnaire "Transformer Reliability Survey",” 2021. [Online]. Available: https://www.ieh.uni-
stuttgart.de/aktuelles/news/CIGRE-working-group-A2.62/. [Accessed 1 September 2023].

[6] D. Martin, J. Marks, T. Saha, O. Krause and N. Mahmoudi, "Investigation into Modelling Power Transformer Failure and Retirement Statistics,"
IEEE Transactions on Power Delivery, vol. 4, no. 33, August 2018.

View publication stats

You might also like