XVII International Symposium on High Voltage Engineering, Hannover, Germany, August 22-26, 2011

ASSESSMENT OF POWER TRANSFORMER RELIABILITY

1* 1 2 3 4
S. Tenbohlen , F. Vahidi , J. Gebauer , M. Krüger and P. Müller
1
University of Stuttgart, Pfaffenwaldring 47, 70569 Stuttgart, Germany
2
Maschinenfabrik Reinhausen GmbH, Falkensteinstr. 8, 93059 Regensburg, Germany
3
Omicron electronics GmbH, Im Ried 1, 6833 Klaus, Austria
4
EWZ, Tramstr. 35, 8050 Zürich, Switzerland
*Email: [email protected]

Abstract: Power transformers are one of the most important parts of power systems and
often the most valuable asset in a substation. Understanding how and when transformers
are likely to fail is a critical point in the asset management of large networks. Due to aging
and inadequate maintenance some critical failures in power transformers may occur.
Thus, the aim is to reduce the failure rate as low as possible. One way to minimize failure
probability is the analysis of old failures and their conditions in order to understand the
reasons for severe failures and to improve maintenance procedures by means of this
knowledge.
In this contribution the results of a failure data survey of 20 utilities in Germany, Swiss,
Austria and the Netherlands based on a newly developed questionnaire are presented.
The investigated transformer population covers more than 23800 unit-years and reveals a
failure rate of 0.3% for 110kV and 0.6% for 220kV and 380kV for major failures. The
hazard curve function shows considerable low failure rates for a transformer age below
30 years. Tap changer and windings are with one third each the main components
leading to major failures.

1 INTRODUCTION Detailed statistics about the failure location in the

respective equipment, failure cause or mode and
Power transformers are one of the important parts repair activities are not included. Therefore the
of power systems and often the most valuable benefit of this statistic regarding asset
asset in a substation. Understanding how and management is limited. [2, 3]
when transformers are likely to fail is a critical point
In contrast to this a questionnaire was developed
in the asset management of large networks. Due to
by the CIGRE working group A2.37 (Transformer
aging and inadequate maintenance some critical
Reliability Survey) which constitutes a helpful tool
failures in power transformers may occur. Thus,
to collect utility failure statistics in a standardized
the aim is to reduce the failure rate as low as
way [4, 5]. Beside information about the population
possible. One way to minimize failure probability is
under investigation failure data is collected for
the analysis of old failures and their conditions in
various groups of transformers in terms of failure
order to understand the reasons for severe failures
locations, failure causes, failure modes, actions,
and to improve maintenance procedures by means
external effects and others. Thus valuable
of this knowledge.
information for asset management purposes can
An international survey on failures in large power be achieved.
transformers in Service was launched in March
In this contribution the results of a failure data
1978 and was limited, for practical reasons, to the
collection in Germany, Austria, Swiss and the
countries represented in CIGRE Study Committee
Netherlands are presented based on the newly
12 (power transformers, now A2). The survey
developed questionnaire.
involved transformer and reactor units designed for
networks with a highest system voltage of not less
2 DATA ACQUISITION
than 72 kV, without any limitation on rated power,
not older than 20 years, and installed on
generation, transmission and distribution systems 2.1 Methodology
[1]. The data used in this contribution are acquired by
In Germany, official statistics are compiled that means of the reliability questionnaire form of
present the analysis of disturbances from CIGRE WG A2-37. Each utility filled a
participating utilities in the country. All disturbances questionnaire form and all the answers were
are recorded in a standardized way. The main collected in a database. In order to achieve a
objective of this survey is the systematic collection maximum security and anonymity, the failure data
of data on the availability and disturbances of the were anonymized by a code on the basis of the
electrical power supply. So, main figures are geographical location and a sequential number.
frequency, duration and extent of interruptions.
 XVII International Symposium on High Voltage Engineering, Hannover, Germany, August 22-26, 2011

In this survey failure data of transmission, 2.3 Failure data

distribution and generator-step-up transformers
The collected failure data account for major failures
with operating voltage of 110kV, 220kV and 380kV
only. To clarify the meaning of failure, some terms
were analysed.
are given according to the definitions within CIGRE
WG A2.37 as follows:
2.2 Investigated Population
The questionnaire consists of three major sheets. Failure
The first sheet presents general information about Any unscheduled situation which requires the
the population of the operating transformers for the equipment to be removed from service for
indicated reference period. Also, voltage ratio, investigation, remedial work or replacement is a
rated power, typical loading and some other failure. Failure can be divided into minor and major
specifications for different applications of failures both with forced and scheduled outages.
transformer units are available in this section. The
other sheets of questionnaire submit failure data, Major failure
essential definitions and failure analysis. Any situation which requires the equipment to be
removed from service for a period longer than 7
For each utilities a reliability questionnaire form days for investigation, remedial work or
exists which contains mentioned data. To be exact replacement is a major failure. Where repairs are
it is required to summarize all these forms into one required, these involve major remedial work,
single unified form. usually requiring the transformer to be removed
The resulted unified form shows the number of from its plinth and returned to the factory. A major
applications based on their voltage throughout failure would require at least the opening of the
Germany, Austria, Swiss and Netherlands. tank, including the tap changer tank or an
exchange of bushings. Also a reliable indication
The characteristics of transformers from these that the condition of the transformer prevents a
countries are in terms of age and condition of safe operation should be counted as a major
population so similar that they have been classified failure if remedial work (longer than 7 days) is
in same category. needed for restoring original service capability (e.g.
detection of strong PDs).
The analysis takes 112 failures within the last 11
years into account. All the failures are analyzed Minor failure
relating to a total population of more than 23800 A minor failure requires remedial work that lasts
unit-years. The investigation results are presented shorter than 7 days.
in term of external effects of failures, failure
location, failure mode, failure cause and the action After receiving the completed forms from utilities
that was taken after failure. The following table the data was checked and adjusted as follows:
shows the population information investigated in
this contribution: • To have a uniformed survey, just the
failures after year 2000 are counted.
Table 1: Investigated population data of the
transformers dependent on system voltage and • 20 utilities submitted data with different
application reference periods. The smallest reference
POPULATION HIGHEST SYSTEM period was 5 years.
INFORMATION VOLTAGE
Application 110kV 220kV 380kV • There were some data with nominal
Substation - Distribution 1292 0 0 voltages less than 69 kV, in this case the
failures were not counted.
Substation -
73 581 478
Transmission
Power Station - • Bushings failures are assigned to major
66 127 73 failures although the repair time was partly
Generator Step-Up
given as less than 1 week.
Because of limited failure data for generator step-
up units, the following failure data analysis was not According to these preconditions, the investigation
performed dependent on the application of the contains 112 major failures within 20 utilities from
power transformer. Having a larger database it is Germany, Austria, Swiss and the Netherlands.
planned to do a more specific analysis also taking
the application and voltage class into account [4].
 XVII International Symposium on High Voltage Engineering, Hannover, Germany, August 22-26, 2011

3 DATA ANALYSIS has to be considered that the investigated

population and consequently also the number of
3.1 Failure rate failures of generator step-up units were quiet low.
Therefore the failure rate of 1.61% for 380 kV
Failure rate
G.S.U. is statistically not proven.
To determine the failure rate, the following formula
is used [1]:
i
Table 4: Failure rate analysis of generator step-up
∑n i
(1)
transformers as a function of voltage class (Years
from 2000 to 2010)
λ= i
1
⋅ 100%
∑N
1
i
Voltage level 110 kV 220 kV 380 kV

Where: Number of
2 4 12
failures
ni = Number of transformers that failed in the i th
Transformer
726 1355 744
year years
N i = Number of transformers in service during the Failure rate (%) 0.28% 0.30% 1.61%
th
i year
For the calculation of failure rates a constant 3.2 Hazard curve
transformer population was assumed for the
investigated time period. There are some useful reliability functions to
The calculated failure rates are given in table 2 analyze the failure situation dependent on
dependent on the voltage level. These results transformers age. Calculating the failure rate for
agree fairly well with the failure rates given in [2], ever smaller intervals of time, results in the hazard
function. It shows the momentary probability of a
where the failure rate is 0.31 % or the 110 kV units
failure dependent on the transformer age. In order
and 0.64% for the 220 kV and 380 kV units. The
to calculate the hazard rate the age distribution of
increase of failure rate with increasing voltage is
all transformers is required. Due to simplicity
obvious.
reasons this age distribution is unfortunately not
Table 2: Failure rate analysis of power included into the used questionnaire. Therefore the
transformers as a function of voltage class (Years transformer age distribution of the investigated
from 2000 to 2010) population is calculated by using the age
Voltage distribution of a utility participating in this survey as
110 kV 220 kV 380 kV
level a reference for the 220kV and 380 kV voltage
Number of classes. Assuming that the transformer fleets of
36 44 32
failures the other utilities have a similar age distribution in
Transformer these voltage classes – which is very probable -,
11474 7111 5226
years the full transformer age distribution can be
Failure rate calculated by scaling up the data of the reference
0.31% 0.62% 0.61%
(%) utility. The achieved cumulative distribution of
operational years of the investigated population in
the voltage classes 220KV and 380 kV is shown in
In [1] a general failure rate, irrespective of the
figure 1.
voltage class and function of the units was given of
the order of two percent. Reason for this deviation
Transformer
could be the different investigated population and years
taking into account also failures with downtimes of
1200
less than one week.
1000
Table 3: Failure rate analysis of power
transformers (transmission and distribution) as a 800

function of voltage class (Years from 2000 to 2010)

600
Voltage level 110 kV 220 kV 380 kV
400
Number of
34 40 20
failures 200
Transformer
10748 5756 4482
years 0
1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55
Failure rate (%) 0.31% 0.69% 0.45% Age

Figure 1: Cumulative distribution of operational

In tables 3 and 4 the failure rates are shown years of investigated transformer population
dependent on the application of the transformer. It (220kV and 380 kV)
 XVII International Symposium on High Voltage Engineering, Hannover, Germany, August 22-26, 2011

The hazard function is computed using the (component) in the transformer where the failure
information from figure 1 and the following formula: was initiated (Fig. 3).

f (T ) Other
H (T ) = ⋅100% 4.5%
Winding
N (T ) Tap
32.1%
Changer
33.9%
H (T ) : Failure hazard rate in percentage
f (T ) : Number of failures at age interval T
N (T ) : Number of transformers in operation and
surviving at age interval T

In figure 2 the hazard function is shown as a dotted

line. In order to obtain a better interpretation of the Lead Exit
Cooling 8.9%
results an averaged curve is provided using a five unit Core and
year moving average method. 0.9% magnetic Electrical
circuit Screen
Bushings
7.1% 0.9%
11.6%
Hazard Rate
5,0%

Figure 3: Failure location analysis based on 112

4,0% major failures between 2000 and 2010

3,0% Like in former surveys major failures are originating

from several transformer components. Tap
2,0% changer and windings are with one third each the
main reasons for major failures. Bushings, lead
1,0% exits and core are listed with a minor percentage
as a reason for major failures. This result agrees
0,0% fairly well with the statistics from 1983, if only
0 5 10 15 20 25 30 35 40 45 50
Transformer Age failures with downtimes longer than one day are
regarded [1].
Figure 2: Failure hazard rate (dotted line) and five
period moving average of failure hazard as a 3.4 Failure mode analysis
function of transformer age (220kV and 380 kV)
Information about the failure mode was also
collected in the questionnaire which describes the
Below 30 years the failure rate is around 0.2% and
nature of the failure illustrating what actually
therefore considerable low. Especially at the
happened when the failure occurred. The
beginning of the operational life no period of
definitions of the failure modes are according to [6].
particularly high failure rate could be observed that
Dielectric failure means PD, tracking, flashover.
could possibly indicate design or manufacturing
Electrical failure means open circuit, short circuit,
problems. At an age of 30 years the failure rate
poor joint, poor contact, ground deterioration,
increases to an order of 0.7%. After an age of 40
floating potential.
years the hazard curve is increasing strongly to
There is no single prominent failure mode. The
levels of far beyond 1%, which can be associated
categories of dielectric and electrical are with 27%
with end of life wear-out failures. It has to be
each the most dominant (Fig. 4).
regarded that for transformer ages above 40 years
the operational experience is low. Therefore the
Dielectric Electrical
calculation of hazard rate is statistically inaccurate, 27.7% 27.7%
which can be seen by the peaks in the dotted
curve in figure 2.

The hazard curve for generator step-up units only

is not shown. The small investigated population
and the low number of failures would give a result
which is statistically not valid. But there are
indications that the increase of failure rate starts
some year earlier. Unknown
3.6% Thermal
Physical 15.2%
3.3 Failure location analysis Mechanical chemistry
17.0% 8.9%
The failure data of the full population were
analysed as a function of the primary location Figure 4: Failure mode analysis based on 112
major failures
 XVII International Symposium on High Voltage Engineering, Hannover, Germany, August 22-26, 2011

3.5 Failure cause analysis 3.7 Action analysis

The circumstances during design, manufacture or The actions taken after a major failure are depicted
operation that led to the failure are analysed. in figure 7. It can be explicitly seen that except the
Because it is quiet difficult to determine the root scrapped portion of transformers (35.7%), 24.2%
cause of a failure, 25% are unknown causes. of the failed ones were repaired onsite and 39.3%
Among the different failure causes aging of were repaired in a workshop.
transformer is with a contribution of 17.9% the Repair in Scrapping
most mentioned one. Astonishingly design and workshop 35.7%
39.3%
manufacturing are mentioned quiet often as a
failure cause. This cannot be proven by the quiet
low failure rate during the first 30 years of
operation. Lightning and overvoltage are almost
negligible as failure cause. An explanation is that
almost all transformers in the investigated
population are protected by surge arresters.

Improper
Clamping Material maintenance Lightning Unknown
pressure 8.9% 2.7% 0.9% Onsite Onsite Onsite
1.8% 0.9%
Overvoltage Overheating External Repair > 1 Repair > 1 Repair < 1
Manufacturing 0.9% 1.8% short-circuit month week week
Installation
10.7% 8.9% 4.5% 16.1% 3.6%
0.9% Abnormal
Deterioration
0.9%
External Figure 7: The analysis of the actions taken after
Pollution
1.8% 112 transformers failure
Design Aging
14.3% 17.9%
Figure 8 and 9 show the failure location of
Unknown scrapped and repaired transformers respectively.
25.0%

Winding
60.0%
Loss of cooling Collateral
0.9% Damage
0.9%
Other reasons
0.9%

Figure 5: Failure cause analysis based on 112

major failures

3.6 External effects analysis

In figure 6 the various external effects which are
Other Lead Exit
caused due to the transformer failures are shown. 5.0% Tap 7.5%
Core and Bushings
Most of the major failures do not result in external Changer magnetic 2.5%
effects (88.4%). Some other external effects which 15.0% circuit
10.0%
are detected in some cases are “Fire” with 6.3%
and “Explosion or Burst” with 2.7%.
Figure 8: Failure location analysis of 40 scrapped
transformers
None
88.4% Due to their impact winding failures lead normally
to a situation where the failed transformer is
scrapped. On the other hand tap changer and
bushing failures are normally repaired.

Others Explosion Leakages

0.9% Fire or Burst 1.8%
6.3% 2.7%

Figure 6: External effects of 112 transformers

major failures
 XVII International Symposium on High Voltage Engineering, Hannover, Germany, August 22-26, 2011

Other Winding 5 ACKNOWLEDGMENTS

4.2% 16.7%

Tap The authors appreciate the fruitful discussions

Changer within CIGRE Working Group A2.37 “Transformer
44.4%
Reliability Survey” and thank the transformer
specialists of the utilities involved in the data
collection for their valuable support.

6 REFERENCES

Lead Exit [1] A. Bossi, et al., “An International Survey on

Cooling 9.7%
unit Failures in Large Power Transformers in
1.4% Core and Electrical
magnetic Screen Service” – Final report of CIGRE Working
Bushings
circuit
16.7%
1.4% Group 12.05, Electra, No.88, pp. 22 – 48,1983.
5.6%

[2] VDN-Störungs- und Verfügbarkeitsstatistik ,

Figure 9: Failure location analysis of 72 repaired Verband der Netzbetreiber VDN, Berlin,
transformers www.vde.com/fnn, Berichtsjahr 2004.

4 CONCLUSION [3] J. Jagers, S. Tenbohlen, “Evaluation of

Transformer Reliability Data Based on National
A questionnaire was developed by the CIGRE and Utility Statistics”, 16th International
working group A2.37 (Transformer Reliability Symposium on High Voltage Engineering,
Survey) by which utility failure statistics in a Cape Town, South Africa, August 24-28, 2009,
standardized way can be collected. In contrast to Proceedings of the ISH, paper No. F-3, page
several public available statistics the results of this 1488-1493
questionnaire deliver valuable information which
can be used for asset management of power [4] Terms of reference of CIGRE WG A2.37
transformer fleet. Thus transformer failure data can “Transformer Reliability Survey”, Jan. 2008,url:
be analysed and interpreted for various types of http://www.cigre-a2.org/Site/What/pa_od.asp.,
transformers in terms of failure locations, failure last accessed: May 01, 2011
causes, failure modes, actions, external effects
and failure rates in transformers. [5] Questionnaire of CIGRE WG A2.37
The presented results of the performed failure data “Transformer Reliability Survey”, May 2011,url:
survey are based on a population of 2690 http://www.uni-stuttgart.de/ieh/wga237.html.,
transformers with more than 23800 unit-years and last accessed: June 01, 2011
112 major failures in Germany, Swiss, Austria and
the Netherlands. They show a failure rate of 0.3% [6] “Life Management Techniques for Power
for 110kV and 0.6% for 220kV and above. The Transformers”, CIGRÉ Brochure No. 227, June
hazard curve function shows considerable low 2003.
failure rates for a transformer age below 30 years.
Tap changer and windings are with one third each
the main reasons for major failures. Winding
failures lead normally to scrapping of the
transformer whereas tap changer and bushing
failures are mostly repaired.