2017 Electrical Insulation Conference (EIC), Baltimore, MD, USA, 11 - 14 June 2017

A Multi-Dimensional Diagnostic Fingerprint for

Power Transformers
Ricardo M. Rampersad, Member, IEEE, Arvind Singh, Member, IEEE, Craig J. Ramlal, Member, IEEE, Sean Rocke
Member, IEEE and Clarissa Arneaud

Abstract—Power Transformers are the most critical assets in II. T ESTING M ETHODS
a power system after generators. They are costly, have long lead
times for acquisition and their failure compromises the security To determine the state of a power transformer, monitoring
of the system. As a result of this, a number of testing methods techniques are conducted. The results are analyzed somewhat
have been developed over time to appraise their condition with independently of each other and compared to acceptable values
a view to extending their lives as well as predicting their end of set by governing standards such as, the American National
life as early as possible. Typically these tests are administered at Standard (ANSI) and the National Electrical Testing Associa-
different times depending on the maintenance scheme adopted tion (NETA).
by the utility and analysed individually. This paper describes the This paper explores the added discrimination that can be
development of a multi-dimensional fingerprint for transformer achieved by using a multi-dimensional diagnostic fingerprint
health estimation. Test results from a number of testing methods
as the basis for transformer monitoring. The fingerprint uses
are analysed and a binary code system developed to more
precisely pinpoint the transformer state. the same test data that is available through standard tests but
integrates the test results through a common framework. The
Keywords—power transformers, transformer testing. main difference between this and the extant method is that the
extant one views the tests as a hierarchy and perform them in
isolation with positives in some tests being triggers for other
I. I NTRODUCTION tests to be performed at later times.
The following is a list of the current transformer tests that
Power transformers are complex devices which are re- were considered for this study along with brief descriptions:
sponsible for transmitting power between circuits of different 1) Insulation Resistance/ Polarization Index: The insula-
voltage levels. They comprise electrical, mechanical, chemical tion resistance and polarization index test, gives an
and thermal systems which all interact in a closed space. indication of the level of contamination and mois-
Generally, transformers are considered to have truly failed ture existing on the surface of the solid insulation in
when their insulation systems have failed (or can no longer transformers [1]. Unacceptable values are indicative of
sustain a moderate to severe fault current). However, their possible fissures, pinholes or cracks in the insulation,
failure is usually due to ageing which can be accelerated by as these give contaminants a place to gather [2].
a lack of proper maintenance, abnormal system conditions 2) Turns Ratio: The transformers turns ratio test is used
or growing load which increases the operating temperature to detect failure events in the transformer windings. Its
of the device. These factors all contribute to a reduction in most important function is to ensure that the nameplate
the usable life of the transformer. As the average age of the turns ratio and the actual turns ratio has a deviation
worlds transformer fleet increases, the need for more robust equal to or less than 0.5% [3]. Unacceptable turns
and continuous monitoring correspondingly increases. Power ratio test results indicate an issue with that transformer
transformer failure can cause: windings.
• Substantial costs for the repair or replacement of the 3) Excitation Current: Used for the detection of incipient
transformer, repair or replacement of neighboring dam- winding deformation. High excitation currents are in-
aged equipment and care for injured persons. dicative of partial short-circuits between one or more
• Loss of revenue or penalties arising from contractual turns in the transformer windings [2].
arrangements. 4) Winding Resistance: Used for the detection of high
resistance joints, fractures, corrosion or losses in the
• A negative reputation for the utility as extended un-
cross-sectional area of conductors.
planned outages lead to dissatisfied customers.
5) Dissipation Factor: The dissipation factor test gives a
• Additional operating stress on other in-service equip- measure of the total health of the entire insulation
ment as they will now be required to transport the lost system of the transformer. This test is very sensitive
power for extended periods of time if a new transformer and can be used to detect faults in the bushings of
has to be procured. the transformers, which could have been missed by an
Practically, this means that more detailed knowledge of insulation resistance test. It is also a confirmatory test
a power transformer’s state is becoming more important in for fault events detected during the insulation resistance
proper asset management schemes. and polarization index test.

978-1-5090-3967-8/17/$31.00 ©2017 IEEE 108

 6) Oil Screening: From NETA an entire host of tests,
in accordance with the American Society for Testing
and Materials (ASTM), must be performed on the
transformers oil. These tests can give a measure of the
general health of the transformer. These test include:
• Interfacial Tension (ASTM D 971)
• Acid Number (ASTM D 974)
• Oil Moisture (ASTM D 1533)
• Power Factor (ASTM D 924)
• Dielectric Breakdown Voltage (ASTM D 877 / D
1816)
• Oil Colour (ASTM D 1500) Fig. 1. Subsystems of Power Transformers [7] [8]
7) Furfural Analysis: Furfural analysis could be used to
confirm the results of the oil test and to possibly indicate formers. Failure causes are defined as the sequence of causes
if the cellulose paper is the source of the unacceptable that initiate a process that leads to a failure mode over a certain
oil test results. time [7]. Fail events are the effects as a result of the failure
8) Dissolved Gas Analysis (DGA): The DGA measures causes such as insulation failure or winding deformation. And
the combustibles gases dissolved in the transformer the failure mode of the transformer speaks about how the
oil due to normal or abnormal operating conditions. transformer actually failed, for example, short circuit between
If the transformer is operating above nameplate values a phase winding and core. According to [2] there are three
or under faulted conditions, different gases would be failure modes of transformers, there are; Electrically Inducted
produced in the oil due to the breakdown of electrical Factors, Mechanically Inducted Factors and Thermally Induced
insulating materials and related components inside the Factors as shown in Figure 2.
transformer [4]. In accordance with the ASTM D 3612
and the IEEE C57.104 [5] different gases and their
levels could be detected and used to determine what
type of faults are occurring. If unacceptable gases are
detected by the DGA, to confirm or properly interpret
results Patil and Chaudhari [6] suggested that the SFRA
test could be done to pin point the exact fault location.
9) Sweep Frequency Response Analysis (SFRA): SFRA
is used to confirm winding deformation, movements
in the transformer core and coil assembly along with
other internal faults.These issues can arise when the
transformer is subjected to high fault currents. The test
involves measurement of the impedance of transformer
windings over a wide range of frequencies. The results
are then compared to previous results from the same
transformer or a similar transformer.
10) Leakage Reactance: Used to detect winding movement Fig. 2. Fault Tree Examples
which usually occurs due to heavy fault current or me-
chanical damage during transportation or installation. It is also possible to have more than one Failure Causes and
Measurements are usually performed on one phase at a Failure Events, producing the same Failure Mode as shown in
time and are compared to the short-circuit impedance Figure 3 [8] that depicts the fault tree for the transformer’s
nameplate or factory test values. Changes of more than winding failure. Fault trees can be developed for the all the
3 % are considered significant. other components of the transformer.

A. Transformer Failure Mode Analysis B. Traditional Testing Procedure

Analysis of the different failure modes of the power trans- In the typical testing procedure shown in figure 4, diagnostic
former requires the device to be broken down into subsystems. testing methods are looked at in isolation. Positives on partic-
These subsystems as shown in Figure 1 consists of the trans- ular tests, such as the turns ratio test may then trigger further
formers active parts, insulation system and accessories, with testing using other methods. At each stage however, the tests
each having its own set of components [7]. The development results are not directly integrated with those of other tests.
of both current and proposed testing methods were be built
around the failure events of each individual component. C. Integrated diagnostics
Using fault trees allows for the quick analysis of Failure The interpretation of several tests, especially electrical ones,
Causes, Failure Events and Failure Modes of the power trans- have coarse levels of discrimination. They therefore do not

109
 quality index rating (0 to 5) and a descriptor of the possible
cause of fault. The quality index is determined by the number
of faults in a transformer; the higher the index, the more
critically ill the transformer’s state is deemed.
The ”possible faults” have been based on experience with
data gathered from field test results and inspections. In the
future, the procedure can be refined as the current resolution
of test results are low, simply deemed as either ’1’ or ’0’based
on whether the results were questionable or acceptable.
For example, the results from a moisture in oil (ASTM D
1533) is deemed unacceptable and tagged with a ’1’. This ’1’
simply indicates that the moisture content in the oil is out of
the acceptable limits but it does not identify the issue with the
transformer. Therefore, if the resolution is increased, the source
of the moisture could be attributed to the aging of cellulose.
Similarly, frequency ranges in FRA are not independently
treated with. These refinements would add to the robustness
Fig. 3. Fault Trees for Transformer Windings [8]
of the technique.

III. C ONCLUSION
The simple multi-dimensional fingerprint proposed in this
paper shows how results from various tests can be integrated
into a simple framework which allows maintenance personnel
to better understand the state of the transformer through unique
codes. The method is still rudimentary though and more work
needs to be done in categorizing complex test results such
as DGA and FRA which do not lend themselves to simple
thresholds for pass or fail.

R EFERENCES
[1] IEEE, “Guide for diagnostic field testing of fluid-filled powertransform-
ers, regulators,and reactors,” IEEE Std C57.152, pp. 1–121, 2013.
Fig. 4. Typical Testing Procedure [2] Bartley and H. William, “Analysis of transformer failures,” International
Association of Engineering Insurers 36th Annual Conference, pp. 1–5,
2003.
offer as much insight into the evolving state of the transformer [3] “Acceptance testing specifications for electrical power distribution equip-
ment and systems,” InterNational Electrical Testing Association, 2011.
as maintenance personnel may like. As transformer fleets age,
[4] L. Hamrick, “Dissolved gas analysis for transformers,” International
having greater insight into their continuously changing state Engineering Testing Association (NETA) World, 2011.
especially near the end of life will become advantageous. [5] IEEE, “Guide for the interpretation of gases generated in oil-immersed
The multi-dimensional transformer fingerprint, integrates the transformers,” IEEE Std C57.104, pp. 1–36, 2008.
results from all the monitoring techniques to give a more [6] S. S. Patil and S. E. Chaudhari, “An Attempt to Investigate the
nuanced understanding of state of the power transformer. Transformer Failure by using DGA and SFRA Analysis,” Properties
With continuous online monitoring techniques being constantly and Applications of Dielectric Materials (ICPADM), 2012 IEEE 10th
developed, it is hoped that this sort of analysis will provide International Conference, pp. 1–4, 2012.
maintenance personnel with an added level of insight into the [7] M. Akbari, P. Khazaee, Sabetghadam, and P. Karimifard, “Failure
Modes and Effects Analysis (FMEA) for Power Transformers,” Power
state of the transformer. Transmission Distribution Research Center Niroo Research Institute
Results from each diagnostic test are tagged as either being (NRI) Tehran, Iran. 28th international power system conference, iran,
unacceptable(1) or acceptable(0) . This results in a unique November2013, 2013.
binary code for different transformer states as depicted in Table [8] A. Franzn and S. Karlsson, “Failure modes and effects analysis of
1. transformers,” Electrical Engineering, 2007.
Each binary code is unique and is associated with a single

110
 TABLE I. M ULTI - DIMENSIONAL F INGERPRINT

Insulation Resistance

Single Quality Index

Winding Resistance
Leakage Reactance
Fault Combination

Moisture Content

Core Excitation
Liquid Screen

Insulation PF

Turns Ratio
Bushing PF
Liquid PF

SFRA
Furan
DGA

Recommendations for improvement and

Condition Assessement Possible Fault maintenance of reliability

Normal Condi- Continue to monitor transformer health,

1 0 0 0 0 0 0 0 0 0 0 0 0 0 tion Continue 0 none perform tests again at their specified
Service times
Possible slack connection at Monitor condition (compare with previous and
2 0 0 0 0 0 0 0 0 0 0 0 1 0 Investigate Results 1
bushing or tap changer future values) (Retest in 6 months)
Repeat DGA in 6 months and compare results to
3 1 0 0 0 0 0 0 0 0 0 0 0 0 Investigate Results 1 Possible internal fault - gas level above normal determine rate of gas generation (for any fault for
DGA)
Retest in 6 months, perform additional tests, fur-
4 0 0 0 0 0 0 1 0 0 1 0 0 0 Investigate Results 1 Possible geometric change in the core ther deterioration may require detanking to inves-
tigate (consider residual magnetism)
Retest in 6 months, perform additional tests, fur-
5 0 0 0 0 0 0 1 0 1 0 0 0 0 Investigate Results 1 Possible geometric change in winding ther deterioration may require detanking to inves-
tigate
Retest in 6 months, perform additional tests, fur-
6 0 0 0 0 0 0 1 0 1 0 0 1 0 Investigate Results 1 Possible geometric change in winding with loose connections ther deterioration may require detanking to inves-
tigate
Retest in 6 months, perform additional tests, fur-
7 0 0 0 0 0 0 1 0 1 0 1 0 0 Investigate Results 1 Possible geometric change with damaged turns ther deterioration may require detanking to inves-
tigate
Cause for
Significant probability of major geometric change with shorted turns Detank to investigate further, check connections
8 0 0 0 0 0 0 1 0 0 1 1 1 0 Concern 2
and slack connection for tightening
Investigate
Cause for
Detank to investigate further, check connections
9 0 0 0 0 0 0 1 0 1 0 1 1 0 Concern 2 Significant probability of major geometric change in winding
for tightening
Investigate
Cause for
Check all flanges for leaks, change silica gel and
10 0 0 1 1 1 0 0 0 0 0 0 1 0 Concern 2 Recent ingest of moisture into oil and a potential slack connection
consider reconditioning the oil
Investigate
Perform dehydration to remove moisture, Do a
Cause for
Possibly wet insulation since bushing PF , insulation PF and Insulation partial discharge test, Consider hot collar and C2
11 1 0 0 0 0 1 0 1 0 0 0 0 1 Concern 2
resistance flagged and DGA (partial Discharge) tests to determine if to replace bushing, Perform
Investigate
IR survey
Cause for PD test, Perform dehydration to remove moisture,
12 1 1 0 0 0 1 0 0 0 0 0 0 1 Concern 2 Significant probability of insulation degradation since Furans, insula- monitor the DGA results, consider a retest in 3
Investigate tion PF and insulation resistance months( for all in this category)
Cause for Investigate cooling and loading conditions, per-
13 1 1 0 0 1 0 0 0 0 0 0 0 1 Concern 2 Significant probability of overheating in insulation due to DGA and form IR Survey , Perform DGA in 3months,
Investigate Furan. This can result in lost of life Perform dehydration to remove moisture
Remove From
Significant probability of damaged windings including shorted turns Detank and investigate(Possible badly damaged
14 0 0 0 0 0 0 1 0 1 1 1 1 0 Service and In- 3
and loose connections winding)
vestigate
Remove From Perform dehydration to remove moisture, consider
SFRA (Deformation within windings), Bushing, Insulation PF, Defec-
15 0 0 1 0 1 1 1 1 1 0 0 0 0 Service and In- 3 replacing bushing, consider detanking to check for
tive Bushing with wet insulation and Deformed winding/s
vestigate deformation)
Remove From
Wet insulation (Insulation PF and Insulation Resistance) with Geomet- Perform dehydration to remove moisture, detank
16 0 0 0 0 0 1 1 0 0 0 1 1 1 Service and In- 3
ric change and shorted turns in winding. and investigate (may require replacing coil)
vestigate
Remove From
Wet insulation (Insulation PF and Insulation Resistance) with Geomet- Perform dehydration to remove moisture, detank
17 0 0 0 0 0 1 1 0 1 0 1 1 1 Service and In- 3
ric change and shorted turns in winding. and investigate (may require replacing coil)
vestigate
Remove From DGA (Arcing), wet insulation, insulation degradation or loss of life, Remove from service, Perform dehydration to
18 1 1 0 1 0 1 0 0 0 0 0 1 1 Service and In- 3 arcing from a slack connection in oil (bushing connection to lead is remove moisture, remove oil to check connections
vestigate slack) and perform IR survey on bushing)
Remove From
19 1 1 1 0 1 1 0 0 0 0 0 0 1 Service and In- 3 Highly Degraded insulation Perform dehydration to remove moisture, retest
vestigate DGA in 1 month
Severe Degra- Remove from service and detank for investiga-
20 0 0 0 0 0 1 1 0 1 1 1 1 1 dation in Ex- 4 High Probability of Significantly Damaged windings tion(possible damaged winding), Retest DGA in
pected Life one week)
Severe Degra- Perform dehydration to remove moisture and con-
21 0 1 1 0 1 1 1 0 0 1 0 0 1 dation in Ex- 4 Significantly Deteriorated insulation with suspected geometric change sider detanking to investigate, Retest DGA in one
pected Life week
Severe Degra-
Perform dehydration to remove moisture, perform
22 1 1 1 1 1 1 0 0 0 0 0 0 1 dation in Ex- 4 Severely wet and degraded insulation and high PD activity
PD test, Retest DGA in one week
pected Life
Remove from Remove from service, Detank and investigate
23 0 0 0 0 0 1 1 1 1 1 1 1 1 Service Failure 5 Severely damaged windings
windings, or consider replacing transformer
Imminent
Remove from Remove from service, Detank and investigate
24 0 0 0 1 0 1 1 0 1 1 1 1 1 Service Failure 5 Severely damaged windings, Acidic Oil windings, Perform Reclamation, or consider re-
Imminent placing transformer
Remove from
25 1 1 1 1 1 1 1 1 1 1 1 1 1 Service Failure 5 Failed transformer Replace transformer
Imminent

111