Thesis Jur Erbrink

On-load Tap Changer Diagnosis on HighVoltage Power Transformers using
Dynamic Resistance Measurements
On-load Tap Changer Diagnosis on HighVoltage Power Transformers using

Dynamic Resistance Measurements
Proefschrift
ter verkrijging van de graad van doctor

aan de Technische Universiteit Delft,
op gezag van de Rector Magnificus prof. ir. K.C.A.M. Luyben,
voorzitter van het College voor Promoties,
in het openbaar te verdedigen op maandag 28 maart 2011 om 12:30 uur
door Juriaan Jonathan ERBRINK
elektrotechnisch ingenieur
geboren te Arnhem
Dit proefschrift is goedgekeurd door de promotor:

Prof. dr. J.J. Smit
Samenstelling promotiecommissie:
Rector Magnificus,
Prof. dr. J.J. Smit,
Prof. ir. L. van der Sluis,
Prof. dr. hab. ir. E. Gulski,
Prof. dr. hab. inz. R.A. Malewski,
Prof. dr. ing. S. Tenbohlen,
Prof. dr.-ing. E. Gockenbach,
Dr. ir. B. Quak,
Prof. dr. ir. J. Biemond,
Voorzitter
Technische Universiteit Delft, promotor
Technische Universiteit Delft
Poznan University of Technology
Instytut Elektrotechniki w Warszawie
Universitt Stuttgart
University of Hannover
Seitz Instruments AG
Technische Universiteit Delft, reservelid
The investigations for this thesis were financially and technically supported by
Seitz Instruments AG in Switzerland and by Alliander in the Netherlands.
ISBN: 978-94-6169-042-5
Summary
On-load Tap Changer Diagnosis on High-Voltage Power
Transformers using Dynamic Resistance Measurements
It is a modern tendency to perform condition-based rather than time-based
maintenance. In this way, utilities attempt to reduce maintenance costs, extend the
service-life of equipment and prevent possible catastrophic failures. Conditionbased tap changer maintenance requires diagnostic measurements to identify the
condition of the on-load tap changer (OLTC) in order to determine when and
what maintenance is necessary. In addition, diagnostic measurements can be used
for the pre-failure detection of defects not seen during maintenance and to assess
the condition of parts that are not easily accessible for visual inspection. Although
many service-aged OLTCs are in service and although literature shows that it is
responsible for a major part of transformer failures, it is not common to assess the
condition before returning the power transformer to operation after maintenance.
Maintenance errors, contact degradation and mechanical defects can thus remain
undetected.
This thesis elaborates on the condition diagnosis of the electrical contacts of intank high-speed resistor tap changers using dynamic resistance measurement
(DRM). While originally used to diagnose circuit breakers, it is also suitable for
measuring the resistance while the tap changer is in operation. This study describes
how to perform DRM on tap changers and how to interpret the results.
Following the introduction, an overview of tap changer technology,
degradation and failure is given in Chapter 2. Several contact resistance models
from literature are discussed and the applicability to DRM is clarified.
Subsequently, Chapter 3 discusses the possible methods available to diagnose the
OLTCs technical condition. The connection between dissolved gas analysis
(DGA) and contact resistance is studied in more depth to determine whether there
is a correlation in the methods. A statistical analysis of DGA and DRM results of a
population of transformers was undertaken but no relationship was found for the
dataset selected.
Chapter 4 shows that contact diagnosis can be applied to different types of tap
changers in service and that DRM can detect a large variety of defects and
degradation mechanisms using a low measurement current. Its results provide
more information about the tap changer condition than static resistance
measurement and important information about the significance and location of
several aging mechanisms can be extracted from the results. The dynamic
vii
measurements are discussed in detail, including the effect of the measurement

current and circuit resistance.
Because each diagnostic method is sensitive to certain defects, a set of methods
is selected and implemented (Chapter 5), creating a diagnostic system. The building
blocks of the diagnostic system are discussed and hardware demands were
determined for each block. The chapter discusses which results should be clearly
presented after a measurement, so that a power transformer can be re-energised
without the risk of failure.
The aging mechanisms revealed through DRM are addressed in Chapter 6. The
different degradation mechanisms examined are: film formation on the contact
surface, coking at high-temperature spots caused by the load current flow, and
pitting and contact erosion due to arcing. DRM results from approximately 450
OLTCs were examined during this project and measurement data from a number
of these tap changers was compared with findings based on visual inspections. A
substantial number within this population were shown to have degraded contacts.
Chapter 6 proves that dynamic resistance measurement is an effective tool for
assessing the technical condition of tap changers, proving to be highly sensitive to
dominant defects, in particular contacts affected by the long-term aging effect on
the change-over selector or the tap selector, as well as defects that compromise the
main function of the OLTC (switching between taps without interrupting the load
current). Typical DRM plots which reveal these various defects are presented with
an explanation of how these defects can be recognised in different types of tap
changers, in particular in relation to transformer subcomponents that are not
accessible for visual inspection during regular maintenance because they are
enclosed in the transformer tank.
Laboratory experiments (Chapter 7) and several test cases were then used to
verify the aging mechanisms to which DRM is sensitive. In addition to a DC
current of 1 A, other measurement current levels were used during the experiments
(Appendix C). The effect of artificially degraded contacts on measurement results
was studied using a test model. The study shows that the surface film on contacts
is measurable. Contact resistance was shown to increase significantly when
contacts move over a surface film, which can conceal more significant contact
degradation. Arcing contact wear influences the contact timing, thereby changing
the shape and amplitude of the resistance graphs. Interruptions of the current can
appear in the graphs when excessively worn contacts are tested.
This thesis also demonstrates how DRM results can be represented as
condition indices using an interpretation flowchart that categorises them using
boundary values (Chapter 8). Complex diagnostic information is converted into
simple condition indices. This process is dependent on the type of tap changer
being measured. Irregularities can be attributed to the arcing switch, the tap
selector or the change-over selector, and to the drive system. On the basis of the
DRM results, quantities which can be used to reveal the most common defects
were determined. Basically six situations are distinguished: Contacts in good
condition, contacts with an increased contact resistance, tap changers with open
contacts, transition time deviations, worn contacts or phase synchronisation
problems. An example of a flowchart to be used for decision support was
developed and its practical applicability to a large population of OLTCs was
viii
verified. The deviation of the measurement current with respect to the expected
value was used and limits were determined for decision-making. The technical
condition can thereby be determined, allowing the selection of a set of critical tap
changers that will benefit from failure-reduction efforts.
ix
Table of contents
SUMMARY..................................................................................................................... VII
TABLE OF CONTENTS ................................................................................................. XI
1 INTRODUCTION ......................................................................................................... 1
1.1 On-load tap changers in power transformers ............................................... 1
1.2 Failure aspects of on-load tap changers ......................................................... 3
1.3 Tap changer diagnostics.................................................................................... 5
1.4 Objectives of this thesis .................................................................................... 5
1.5 Approach............................................................................................................. 6
1.6 Outline of this thesis ......................................................................................... 7
2 OLTC TECHNOLOGY AND DEGRADATION ........................................................ 9
2.1 OLTC technology overview............................................................................. 9
2.1.1 Selector switch type tap changer ........................................................... 10
2.1.2 Diverter switch type tap changer........................................................... 11
2.2 OLTC degradation mechanisms.................................................................... 12
2.3 Contact degradation ........................................................................................ 12
2.3.1 Clean contacts .......................................................................................... 13
2.3.2 Contacts with a surface film................................................................... 17
2.3.3 Contacts with high local temperatures ................................................. 23
2.3.4 Contact failure due to overheating........................................................ 24
2.3.5 Corrected model of tap changer resistance ......................................... 27
2.4 Conclusions....................................................................................................... 30
3 ON-LOAD TAP CHANGER DIAGNOSIS ................................................................. 31
3.1 Oil and insulation diagnosis ........................................................................... 32
3.1.1 Dissolved gas analysis ............................................................................. 32
3.1.2 Study of the agreement between DGA and DRM results ................ 34
3.2 Mechanical diagnosis ....................................................................................... 37
3.2.1 Acoustic and vibration signature........................................................... 37
3.2.2 Motor power measurement.................................................................... 38
3.2.3 Position measurement............................................................................. 39
3.3 Contact status diagnosis.................................................................................. 39
3.3.1 Temperature difference measurement.................................................. 40
3.3.2 Static resistance measurement ............................................................... 41
3.3.3 Dynamic resistance measurement (DRM)........................................... 43
xi
3.4 Discussion ......................................................................................................... 43

3.5 Conclusions....................................................................................................... 48
4 DYNAMIC RESISTANCE OF ON-LOAD TAP CHANGERS...................................... 49
4.1 Measurement setup.......................................................................................... 50
4.1.1 Indirect measurement of the OLTC..................................................... 50
4.1.2 Direct measurement of the OLTC ....................................................... 51
4.2 Measurement current ...................................................................................... 55
4.2.1 Winding configuration ............................................................................ 55
4.2.2 Amplitude of the measurement current............................................... 56
4.2.3 Sensitivity .................................................................................................. 61
4.3 Circuit resistance .............................................................................................. 62
4.4 Secondary short circuit.................................................................................... 66
4.5 Determination of the contact resistance ...................................................... 69
4.6 Conclusions....................................................................................................... 69
5 POWER TRANSFORMER DIAGNOSTIC SYSTEM .................................................... 71
5.1 International standards for power transformer testing.............................. 73
5.2 Transformer diagnostic system...................................................................... 75
5.3 Measurement procedure ................................................................................. 77
5.3.1 Dynamic resistance of the OLTC ......................................................... 77
5.4 OLTC drive system condition ....................................................................... 81
5.5 Hardware and data processing....................................................................... 83
5.6 Conclusions....................................................................................................... 84
6 EVALUATION OF TYPICAL DRM DIAGNOSTIC DATA ....................................... 85
6.1 Regular DRM plots.......................................................................................... 86
6.2 Irregular DRM plots........................................................................................ 93
6.3 Irregularities in the change-over selector ..................................................... 95
6.3.1 Long-term aging in selector switch type OLTCs ............................... 95
6.3.2 Current interruptions due to the change-over selector ...................104
6.4 Irregularities in the arcing switch ................................................................106
6.4.1 Long-term aging in the arcing switch .................................................107
6.4.2 Contact wear of the arcing switch.......................................................109
6.4.3 Timing difference between phases......................................................110
6.4.4 Maintenance errors ................................................................................110
6.4.5 Current interruptions due to the arcing switch.................................112
6.4.6 OLTC transition times..........................................................................113
6.5 Defects in both parts of the OLTC ............................................................115
6.6 Conclusions.....................................................................................................116
7 LABORATORY VERIFICATION OF SELECTED FAILURE MECHANISMS ..........119
7.1 Measurability of the surface film using DRM ...........................................119
7.1.1 Correlation between DRM and contact degradation .......................119
7.1.2 Effect of the surface film on DRM ....................................................123
7.1.3 Accelerated degradation tests...............................................................124
7.2 Arcing contact wear.......................................................................................128
xii
7.2.1 Arcing contact dimensions...................................................................128

7.2.2 Arcing contact timing............................................................................131
7.3 Conclusions.....................................................................................................134
8 DRM KNOWLEDGE RULES FOR DECISION SUPPORT ......................................137
8.1 Condition assessment....................................................................................137
8.2 DRM knowledge rules...................................................................................138
8.3 Quantities reflecting the OLTC condition.................................................141
8.4 Condition indexing ........................................................................................143
8.5 Application example ......................................................................................146
8.5.1 Open contacts ........................................................................................147
8.5.2 Contact resistance ..................................................................................148
8.5.3 Flowchart verification ...........................................................................152
8.6 Conclusions.....................................................................................................155
9 CONCLUSIONS & RECOMMENDATIONS.............................................................157
9.1 Conclusions.....................................................................................................157
9.2 Suggestions for future work .........................................................................160
REFERENCES ..............................................................................................................163
APPENDIX A - OLTC TECHNOLOGY, DEGRADATION AND PROTECTION ....173
APPENDIX B - INDIRECT MEASUREMENT OF THE OLTC
DYNAMIC RESISTANCE .............................................................................................189
APPENDIX C - EXAMPLES OF STATIC RESISTANCE MEASUREMENTS
WITH INCREASING CURRENT ON SERVICE-AGED OLTC CONTACTS..............195
APPENDIX D - DESIGN CONSIDERATIONS OF A TRANSFORMER
DIAGNOSTIC SYSTEM ................................................................................................203
LISTS ............................................................................................................................217
ACKNOWLEDGEMENTS ...........................................................................................225
SAMENVATTING ........................................................................................................227
CURRICULUM VITAE .................................................................................................231
xiii
CHAPTER 1 - INTRODUCTION
1
Introduction
1.1 On-load tap changers in power transformers
Large amounts of electrical power are transported and distributed by the electricity
grid. In addition, long distances have to be bridged between the generation and
consumption of electrical power and high voltages are used to reduce the power
losses during this transport. Different voltage levels are used in the grid and these
voltages are linked using power transformers.
A power transformer basically has two functions:
To link different voltage levels in the high-voltage power grid in such a

way that electrical power can be exchanged.
To keep the voltage at an acceptable level when the load changes.
Figure 1.1 Example of a 50/10 kV (14 MVA) power transformer with on-load tap changer.
The second function, voltage regulation, is
accomplished by adjusting the transformation
ratio of the power transformer. For that
purpose, the transformer winding is equipped
with tapped windings that can be selected by an
on-load tap changer. The active and reactive
power flow can be controlled by the tap changer.
On-load tap changers are also used for phase
regulation in phase shifter transformers. In
general, it can be said that on-load tap changers
are used to select tapped windings along the
main transformer windings. The function of an
on-load tap changer can therefore be described
as:
To select another transformer tapped

winding without interrupting the load
current
Figure 1.2 Example of an in-tank

selector switch type on-load tap
changer; 1) selector switch, 2)
coarse tap-selector.
An on-load tap changer is driven by a complex

drive mechanism for which timing is an
important issue. The OLTC can easily switch 100,000 times during its lifetime,
depending on the function and location of the transformer. The operation of the
on-load tap changer is usually performed by an autonomous control system, but an
on-load tap changer can also be operated from a control room or manually at the
transformer. Different types of tap changers are in service worldwide. These types
can be grouped according to:
The physical location of the tap changer: tap changers that are installed
inside the transformer (in-tank type) or tap changers that are bolted onto
the transformer (compartment type).
The electrical location, resulting in types that are installed on the highvoltage side of the transformer and tap changers that are installed on the
low-voltage side.
The transition impedance that is used: reactor type or high-speed resistortype tap changers.
The number of contacts available for load switching: diverter or selector
switch type tap changers.
The winding configuration in which the tap changer is incorporated: wye
or delta connected.
The switching cycle of the transition contacts: flag-cycle or pennant-cycle
operation (symmetrical and asymmetrical).
The capabilities of switching load: on-load tap changers (OLTCs) or deenergised tap changers (DETCs).
The contact material of the change-over selector, for example copper or
silver-plated contacts.
Diversity can also be seen in the switching technology of OLTCs. Tap changers
having their arcing contacts submerged in insulating oil are already in use for many
decades. This type of on-load tap changer dominates the market and, due to the
long lifetime of power transformers, degradation of this oldest design is of concern
nowadays.
New developments, like vacuum switching or solid state (thyristor) technology,
are being developed and implemented with the purpose to prolong the
maintenance interval. Solid state tap changers, mostly found in transformers with a
low power rating, are expensive, have high no-load losses and have difficulties with
handling high short circuit currents. Vacuum tap changers (see Figure 1.3) transfer
the load current using vacuum bottle switches and can handle high power ratings,
but are also relatively new and therefore not yet widely implemented. The
application of vacuum switching technology reduces the need for maintenance and
diagnosis.
Besides these newly installed
technologies,
manufacturers
continuously improve the design of
oil type OLTCs. For example,
experience with tap changer
degradation
led
to
design
improvements using new contact
materials, higher contact pressure
or wiping contacts. OLTCs that use
oil as arc quenching medium are
still the most frequently installed
type due to their cheap and
improved design. Therefore, this
type will still be in service for at
least 50 years and it is expected that
condition assessment of OLTCs
1.3 Example of an on-load tap changer
will still be beneficial in the future, Figure
using vacuum switching technology (one phase
although the future population will of the switch is shown); 1) Vacuum switch bottle,
face different problems.
2) transition resistor, 3) connections to tapped
windings.
1.2 Failure aspects of on-load tap changers

Tap changers, like transformers, have a long lifetime and a high level of reliability.
However, the average age of the total transformer population is also high [1].
Therefore, degraded transformers remain in service and failures occur regularly [2].
The on-load tap changers are responsible for the major part of the unplanned
transformer unavailability, as follows from power transformer failure studies [312]. This makes the OLTC the most vulnerable part of a power transformer [13].
Its failure can result in unavailability of the power transformer, damage to the tap
changer and surrounding transformer and even the total loss of the power
transformer. A catastrophic failure of a power transformer not only involves
3
replacement costs, but safety, environmental issues and, in case a spare transformer
is not directly available, additional costs for not delivered power and penalty costs
should also be considered.
In general, it is the quality of the paper winding insulation that mainly
determines the transformers lifetime; repairing relatively small parts like
degenerated contacts is still considered worthwhile as long as the transformer had
yet not reached the end of its life. For example, if the degree of polymerisation of
the cellulose insulation indicates that the paper still has 50% of its life left but the
tap changer is in a critical condition, the tap changer can be given an extensive
overhaul (costing for example 70,000-150,000) to prolong the transformers life.
A new transformer (costing for example between 500,000 for a 40 MVA unit and
2,500,000 for a 500 MVA unit) will not be needed for at least another 10-15 years.
Diagnostic measurements (costing relatively little and requiring relatively few manhours, for example 2 maintenance engineers working for 8 hours each for a
resistance measurement) can be used to determine when such maintenance work
can be cost-effective.
Figure 1.4 shows two examples of such failure studies: 41% of the transformer
failures were related to the on-load tap changer according to an international study
[4] and 56% according to a Dutch study [6].
Other
12%
Core
3%
Other
10%
Windings
10%
Tap-changer
41%
Windings
19%
Bushing
12%
Bushing
6%
Tap-changer
56%
Leakage
18%
Leakage
13%
Figure 1.4 Failure distribution of an international population [4] and a population of Dutch
50-150 kV power transformers [6].
Dutch failure studies come with higher OLTC failure percentages compared to
other studies and those percentages are lower in the US than in Europe [14]. These
differences can be explained by the difference of types that are in service. The
Dutch on-load tap changer population contains a considerable number of old tap
changers, the design of which makes them more prone to degradation mechanisms
than modern designs by tap changer manufacturers which incorporate design
improvements. Old tap changers are clearly also more service-aged than the
modern types of tap changers. Therefore the current population of OLTCs could
benefit from tap changer diagnosis. In addition, service-aged tap changers provide
interesting case studies that can be used to show the effect of aging on diagnostic
parameters.
1.3 Tap changer diagnostics

Maintenance is normally recommended by the supplier after a fixed time interval
or a fixed number of operations. A modern tendency is to perform conditionbased rather than time-based maintenance. In this way, utilities try to reduce
maintenance costs, extend the service-life of equipment and prevent possible
catastrophic failures. Condition-based maintenance requires diagnostic
measurements: one needs to know the condition of the tap changer to determine
when and what maintenance is necessary.
In addition, diagnostic measurements on OLTCs can be used for the pre-failure
detection of defects not seen during maintenance and to assess the condition of
parts that are not easily accessible. Although many service-aged OLTCs are in
service and the literature shows that it is responsible for the majority of
transformer failures, it is not yet common to assess the condition before returning
the power transformer to operation after maintenance. Maintenance errors, contact
degradation and mechanical defects can remain undiscovered. A number of
diagnostic tests can be performed and utilities combine different tests to determine
its need for maintenance. A few utilities apply dynamic resistance measurement
(DRM) to OLTCs during regular maintenance. Preliminary results showed that it is
sensitive to major degradation mechanisms in on-load tap changers.
However, few scientific studies of the dynamic resistance of OLTCs have been
performed: most research has focused on the contact degradation process. This
thesis investigates the suitability of dynamic resistance measurements as a simple,
direct means of detecting defects and degradation mechanisms in OLTCs and
hence of checking the main functions of the tap changer. In particular, underlying
knowledge about how to perform DRM on tap changers and how to interpret the
results obtained in this way requires more scientific research.
Diagnostic measurements, including DRM, result in a large amount of raw
measurement data. The structured interpretation of these results can be helpful in
drawing conclusions about the tap changer condition. Interpretative support that
relates the DRM results to the actual condition of the tap changer is important.
1.4 Objectives of this thesis

This thesis elaborates on the condition diagnosis of high-speed resistor-type tap
changers for the pre-failure detection of frequently occurring defects. The aim is to
establish the sensitivity of dynamic resistance measurements and to provide
interpretation support.
The specific research objectives of this thesis are to:
1. Assess the prioritisation of tap changer degradation mechanisms and defects,
thereby including the application of existing contact resistance models to tap
changers.
5
2. Derive a diagnostic methodology which can be used to detect and localise the
dominant electrical and mechanical degradation mechanisms of an on-load tap
changer without opening the transformer tank.
3. Perform a sensitivity verification of dynamic resistance measurements for prefailure phenomena in OLTCs.
4. Determine the measurement parameters that are most suitable for detecting
these defects, such as the presence of a contact film or pyrolytic carbon,
contact wear and pitting.
5. Propose a systematic method of interpreting the dynamic resistance curves of
OLTCs, using concise evidence-based rules, which can be used to support
strategic decisions about the maintenance or overhaul.
1.5 Approach
Firstly, after introducing OLTC technology, an overview of dominant defects and
degradation mechanisms is set up. Several contact resistance models from literature
are discussed and their applicability to DRM is explained. Based on the dominant
failure modes, existing diagnostic methods will be evaluated. The sensitivity,
suitability for the early detection of aging, their limitations and their mutual
relationship will be discussed.
Secondly, dynamic resistance measurements will be discussed in more detail, in
particular the important topics that should be considered when performing
dynamic resistance measurements, such as the effect of the measurement current
amplitude, a short circuit on the secondary side of the transformer and the
reproducibility of the measurements.
Thirdly, the sensitivity of dynamic resistance measurement to defective and
degraded tap changer contacts is verified using field measurements, visual
inspection results, laboratory experiments and transformer failure data. The
sensitivity of DRM to the dominant degradation mechanisms is verified using
artificial defects and accelerated degradation. In total, 20 laboratory experiments
are performed, each consisting of 3-16 measurements. These experiments use a
test model that was developed for this project.
Several methods are selected to assess the dominant failure modes of an OLTC,
and an off-line solution for condition diagnosis is proposed. Up until now, about
700 post-maintenance measurements have been performed with this system, of
which 95% have been measured with a prototype of the proposed system. The
results of these field measurements are used to demonstrate the defects and
degradation mechanisms that can be found using DRM and to discuss their
significance, in order to determine a suitable corrective action. In addition, the
additional value of performing dynamic resistance measurements after tap changer
maintenance is investigated.
Concise rules are formulated based on the knowledge gained from the experience
of field and laboratory measurements. These knowledge rules can be arranged in
such a way to create a decision-support flowchart. In addition, an example will be
worked through.
1.6 Outline of this thesis

Following this introduction:
Chapter 2 introduces the reader to the subject by explaining the operation

of some common types of on-load tap changers. An overview of tap
changer degradation mechanisms is discussed and a model for tap changer
contact degradation is derived.
Chapter 3 describes the possible options for on-load tap changer

diagnosis to assess the condition of the tap changer, including the
mechanical drive mechanism and contact degradation.
Chapter 4 provides more detail on dynamic resistance measurements (for

example the measurement setup, circuit resistance, measurement current
amplitude and short circuit on the secondary transformer side).
Chapter 5 presents a measurement device for transformer diagnosis based

on dynamic resistance measurements, OLTC drive motor power
measurements, position measurements on the drive shaft of the on-load
tap changer, transformer turn ratio measurements and DC winding
resistance measurements.
Chapter 6 discusses the possible irregularities of the dynamic resistance

measurement data.
Chapter 7 deals with the measurability of a surface film that is formed on

the tap changer contacts in the early stage of the long-term aging effect.
Experiments with worn arcing contacts are also discussed.
Chapter 8 presents the main issues related to the assessment of the

condition of on-load tap changers and explains the advantages of
condition indexing. Decision-support for dynamic resistance
measurements is provided, based on the knowledge rules derived in
previous chapters. An example of the application is given.
Chapter 9 presents conclusions and recommendations for future research.
CHAPTER 2 OLTC TECHNOLOGY AND DEGRADATION
2
OLTC Technology and
Degradation
The primary function of an on-load tap changer is to select another tap without
interrupting the load current. This can be accomplished in many ways, resulting in
a considerable diversity of tap changer designs [15-16]. This chapter will first
provide an overview of OLTC technology. The terminology used throughout this
thesis follows that given in [17].
2.1 OLTC technology overview

All on-load tap changer designs must transfer the load current and are therefore
equipped with an arcing switch. Two different arcing switch principles are in use,
namely a diverter switch and a selector switch. The main difference between these
designs is that a diverter switch type OLTC uses a tap selector to pre-select taps
without switching current, in combination with a diverter switch to switch the load
from the selected to the pre-selected tap. A selector switch type OLTC combines
the selection of fine tap windings with the switching of the load current. To
expand the regulating range of the arcing switch, the design can be extended by a
change-over selector. The change-over selector can be implemented as a reversing
change-over selector, a coarse change-over selector or a combination of both.
Another difference in design principles is the transition impedance that is used
to control the circulating current that exists when two taps are selected during the
transfer of the load current from one tap to another. A reactor type OLTC, which
uses a reactor as transition impedance, is mainly used in the US. The reactor type
OLTC normally consists of a diverter switch designed as a compartment type
OLTC and can be found on the low-voltage side of the transformer. Most OLTCs
in Europe are in-tank high-speed OLTCs which use a transition resistor during the
transfer of the load current. These tap changers are normally located in the highvoltage transformer windings. Finally, tap changers using vacuum switches have
9

been developed and installed in several hundred transformers. Operational
experience of the vacuum type OLTC has been gathered for a few years and
indicates satisfactory performance.
2.1.1 Selector switch type tap changer
Selector switch type OLTCs combine the switching of the load current with the
selection of fine winding taps. Selector switches can be used in a stand-alone
manner but their voltage regulating range can also be extended with a (multiple)
coarse change-over selector. When all of the fine tap windings are selected by the
selector switch, a coarse tap winding can then be inserted before the selector
switch can continue.
An example of a selector switch type OLTC is shown in Figure 2.1, with its
selector switch on top of the coarse change-over selector. The selector switch is
accessible at the top of the transformer, while the coarse change-over selector is
mounted deeper within the transformer tank, underneath the selector switch. As a
result, the selector switch can be checked during regular maintenance by removing
the rotor insert, the dirty oil can be replaced, the selector switch inspected and its
arcing contacts renewed when necessary.
The selector switch and the coarse change-over selector both consist of a stator
on which the static contacts are mounted. These contacts are connected to the taps
on the transformer windings, shown on the left of Figure 2.1. The rotor is located
inside the stator and rotated by the drive system. The rotor makes a connection
between the stator contacts. When switching the load current the selector switch
will cause arcs; therefore, its insulation oil is separated from the main transformer
tank and the arcing contacts are made of tungsten. The entire selector switch
movement will take about 30-200 ms.
Figure 2.1 Selector switch type OLTC with coarse change-over selector (left). Right: picture
of the stator contacts of a selector switch when the rotor and insulation oil are removed.
10

The selector switch changes the voltage ratio with one voltage step at a time. The
coarse change-over selector depicted in this figure makes a voltage change of 8
steps when switching from coarse tap A8 to A9 (Figure 2.1). The tap changer as
depicted in Figure 2.1 has now selected the main winding on tap A8. The current is
brought to the selector switch stator by the rotor of the coarse change-over
selector (using contact A in Figure 2.1). The selector switch has selected three fine
tap windings using tap A3. The current leaves the tap changer using a slip ring.
Appendix A describes the switching cycle of a selector switch type OLTC in more
detail.
The transition resistors are necessary to prevent the short circuit of adjacent
fine tapped windings and to provide a non-interrupted current path for the load
current during the switching operation. The resulting circulating current is
determined by the step voltage and the resistance of the selector switch.
2.1.2 Diverter switch type tap changer
A diverter switch type OLTC combines a
diverter switch and a tap selector. The tap
selector has two sets of contacts available for
tap selection. One of the two contacts is
selected by the diverter switch and is under
load. The other contact selects the next tap
(pre-selected tap) without switching current.
The diverter switch then switches from the
selected to the pre-selected tap. Diverter
switch type OLTCs use transition resistors to
limit the circulating current in the diverter
switch. Figure 2.2 shows a schematic
overview of a tap selector and a diverter
switch. The tap selector can be combined
with a change-over selector to expand the
range of the tap changer, see appendix A. By
switching the reversing change-over selector
one can add or subtract the selected Figure 2.2 Diverter switch with
resistors (top) and a tap
transformer windings, thus doubling its range. transition
selector (bottom).
A coarse change-over selector can add a
section of the regulating winding to the main
winding, in this way changing the voltage of all tap selector contacts.
In contrast to the selector switch type tap changer, most diverter switch type
OLTCs have no rotor inside the arcing switch. A stationary insert is used instead.
The diverter switch can easily be maintained by removing the diverter switch
insert. The dirty oil can be replaced, the diverter switch inspected and its arcing
contacts replaced.
11
2.2 OLTC degradation mechanisms

The high failure rate of OLTCs [14] emphasises the need to understand tap
changer degradation. Figure 2.3 provides an overview of the different degradation
mechanisms (electrical, mechanical, thermal and chemical) that are discussed in
appendix A, the overview is based on literature, observations during overhaul and
post-mortem investigation. The arrows indicate the location at which the
degradation mechanisms act and some properties of the subcomponents (as
discussed above) are also summarised. Contact degradation is described in this
chapter, more information about other degradation mechanisms can be found in
appendix A.
Figure 2.3 Overview of the different degradation mechanisms. The arrows indicate the
location at which the mechanisms act. Important features of the OLTC parts are listed.
2.3 Contact degradation

An OLTC has sets of contacts that switch different currents at different recovery
voltages. For example, the main contacts of the arcing switch are designed to
transfer the load current to the transition contacts. The arcing contacts of the
arcing switch are designed to break the load current and the circulating current. In
contrast, the contacts of the tap selector and the change-over selector are not
designed to switch current. Therefore these sets of contacts wear differently.
Change-over selector contacts (including tap selector contacts) do not wear as
fast as arcing switch contacts that wear due to the switching of load currents.
These contacts will not switch significant currents, but can show pitting of the
12

contacts and the development of pyrolytic carbon [18-19]. This contact
degradation is not due to the arcs caused by switching the current but by a long
term overheating process.
Usually, the contacts of the change-over selector are infrequently used and can
be motionless for long periods. This activates the second degradation mechanism
of change-over selector contacts: a long-term aging effect on contacts under oil,
henceforth referred to as the long-term effect. The long-term effect is the most
common degradation mechanism on the change-over selector in the Netherlands
and causes most change-over selector failures.
Three basic stages of the long-term effect can be distinguished and are shown
in Figure 2.4. This section deals with these degradation stages. Formulae for
contact resistance found in literature will be explained. Finally, in Section 2.3.5, the
formulae will be completed using experimentally derived correction factors, which
correct for OLTC operation and for the measurement current amplitude.
Figure 2.4 The long-term effect starts with the formation of a surface film on the contacts.
The increased resistance can cause coking. Thermal runaway can finally cause failure of the
OLTC due to open contacts.
2.3.1 Clean contacts

A clean contact involves a metal-to-metal connection between two contacts. The
resistance of this contact material can be described as:
RB
X'
A
2-1
where RB is the bulk resistance of the material in , the specific resistivity of the
contact material in m, X the contact thickness in meter and A the contact crosssectional area in square meter. Since the contact
material is deformable, contact between the two
contact
surfaces is made over a small contact area, not at a
D/2
area A
single point.
D
C
X
For example, let us consider the cross-sectional
contact area between an ideal pair of radial contacts
(two contacting hemispheres, see Figure 2.5) of
diameter D. Due to the contact force, a contact area A
is created a distance D/2 - C from the centre of each
hemisphere, which may be written:
A ( )2 X 2
2
[20]
2-2
over the range X = 0 to D/2-C, so that two contacting

hemispheres have a bulk material resistance of:
Figure
2.5
Two
contacting hemispheres
create a contact area A a
distance D/2 - C from the
spheres centre.
13

D 2 C
Rc 2
D X
2
D 2
2
2
( ) X dX
ln
D
D
2 X
1
D 2 C
[20]
2-3
where C is the part of the radius of each hemisphere involved in the contact. For
example, a 200 amp tap changer with 22 N contact force will have D = 8 mm and
C = 4 m, so that Rc = 10.5 for copper contacts at 20 C [20]. This theoretical
resistance would exist if there was no contaminant film between the contact
surfaces and perfect contact was made over the entire contact area. Practical
experience of copper contacts indicates an as-built resistance closer to 50 or 60
, suggesting that true contact is not achieved uniformly over the whole surface
[20].
The small contact area constricts the current from flowing uniformly: the
current lines are bunched near the micro contact. This gives rise to an additional
resistance, the constriction resistance (often called the contact resistance).
Assuming an infinite volume of bulk material above and below the circular contact
spot, we can calculate the constriction resistance to be:
2-4
[21]
2a
where a is the radius of the circular contact spot in meter and assuming that a is
much smaller than the apparent contact area.
In practice, electrical contact will be made at more than one spot due to the
roughness of the material. It follows that parallel metal-to-metal micro contacts
must be assumed to be present in OLTCs. In addition to the total cross-sectional
area of the contact spots (Rparallel), there will then be an additional resistance due to
interaction between the spots (Rinteraction). Ref. [22] describes this effect on the
assumption of n circular spots on clean contact surfaces, where the contact
resistance of a circular clean contact is given by:
n
Rc R parallel Rinteraction
2 ai
i 1
2 ai
i 1
n n 1
n 2 i 1,i j j 1 dij
i 1,i j j 1
n
ai a j
dij
( ai )
2-5
i 1
which can be approximated as:
Rc
[22]
2-6
where ai and aj are the radii of contact spot i and j in meter, and dij is the distance
between the centres of spot i and j in meter. The approximation of Equation 2-6
assumes that there is no correlation between the size of the contact spot and its
position, so only the mean radius is considered in the second term (ai = aj = a).
Hence, the number of contact spots and the distance between them determine the
ratio of the parallel to the interaction resistance; see Figure 2.6. When the number
14

of contact spots is low, the contact resistance is mainly determined by the parallel
resistance of the spots, while with a large number of spots the resistance is mainly
determined by the interaction between the spots.
The voltage drop over this cluster of clean circular spots may be written [22]:
Ii
(
2 ai
Ij
j i
2-7
ij
where U is the voltage over the contact interface in Volts and Ii and Ij are the
currents through contact spot i and j in Amperes.
Ref. [21] describes the effect of parallel contact spots by the equation:
1
1
)
2na 2
2-8
Resistance (arbitrary units)
Rc (
Figure 2.6 The number of micro-spot contacts determines the ratio of the parallel resistance
of all micro-spots to the resistance due to the interaction between the contact spots [22].
where n is the number of contact spots with radius a

uniformly distributed in a circular area of radius (see
Figure 2.7).
A higher contact pressure lowers the resistance of
such a clean circular contact spot by increasing the
contact area due to plastic deformation of the metal.
OLTC contacts are therefore designed to ensure that the
contact pressure is high and the surface film is wiped off Figure 2.7 n contacts
with radius a are
each time the contacts are operated.
distributed over an
By way of example, Figure 2.8 shows the effect of apparent contact area .
temperature and contact force on the contact area of a
cluster of circular contact spots of equal radius (see [23]). The numerical simulation
performed by [23] employs a finite element model to calculate the material
deformation due to mechanical and thermal stress near an aluminium contact
15

boundary (similar to that shown in Figure 2.5). It may be seen from Figure 2.8 that
higher contact forces and contact temperatures yield a larger contact area. The
best-fit relation between the temperature T in degree Celsius and the contact area
Ac in square metre for contacts with a high contact load (2.3 kN and 4 kN) can be
found from Figure 2.8 [23] to be the exponential function:
Ac 5.16 106 7.27 107 e1.0510
T / T0
2-9
with T0 = 1C.
Figure 2.8 Example of a numerical simulation performed by [23], in which the contact area
increases with higher temperature and contact force.
Ref. [23] gives the relationship between the resistivity and the contact temperature
as:
0
T0
Te C (T T )
0
[23]
2-10
where 0 T is the bulk resistance, 0 and T0 are reference values, is the bulk
T0
contact resistivity factor (related to the applied load and surface condition), T is the
C (T T0 )
is the softening effect factor reflecting the increase in
temperature, e
contact area at high temperatures (as a function of the softening point of the
contact material, which is about 200 C for pure copper), and is the contact area
correction factor. The change in contact resistance with temperature according to
Equation 2-10 (adapted for degree Celsius, for = 1000 and = 1) is plotted in
Figure 2.9. It will be seen that the resistance initially increases with temperature,
due to the bulk resistivity effect [23]. When the temperature rises above the
softening point, the exponential term of Equation 2-10 causes the resistance to fall
as the contact area increases due to the softening of the material [23].
16

8
=(2.99x10-5
+ 1.68x10-6 T)exp (-1.17x10-2 T)
0
0
100
200
300
400
500
600
700
Temperature (C)
Figure 2.9 Relationship between the contact temperature and resistivity. The resistivity first
rises with temperature due to the bulk resistivity effect, and then falls again as the contact
material softens [23].
2.3.2 Contacts with a surface film

The resistance of a contact will gradually increase during its lifetime. There is
empirical evidence [20][24-26] that this long-term effect starts with surface
oxidation and the formation of organic polymers [19]; these effects reduce the
conductivity. Anorganic layers consisting of tenorite (CuO), cuprite (Cu2O) and
sulfides (CuS and Cu2S) form on copper contacts [19]. Ref. [26] found that the
oxide and sulfide layers always formed in the same sequence; a thin sulphide layer
followed by the curpite later, then the tenorite, see Figure 2.10.
At higher temperatures an organic film is also formed [26]. It follows from [25]
that the organic film is composed of organic compounds in the transformer oil and
consists of polyacrylates and polyfurans [19], giving a red-orange color to the
contact [26]. This layer bonds to the oxide layer formed on the contacts to give a
stable film of low conductivity.
The formation of this surface film is referred to as the early stage of the longterm aging effect.
Figure 2.10 1600x magnification of a service-aged tap changer contact surface. The contact
was mounted in resin to make this cross-sectional picture. A thin copper sulphide layer forms
on the copper contact, followed by a cuprite and a tenorite layer. [26]
17

Experiments described in the literature [18][27] show that the surface film grows
thicker as the contacts age, and that the growth rate of this layer is strongly
dependent on the surface temperature of the contacts. Ref. [18] gives an
expression for this process as a function of time, based on the endurance testing of
OLTC contacts:
2-11
s k1 * 0 k 2 * t k 3
where s is the thickness of the film layer in , 0 is the surface temperature in C at
which the surface film grows on the contacts and t is the duration of the aging test
in hours. k1, k2 and k3 are coefficients that depend on the contact material and oil
composition [18]. For copper contacts immersed in Shell Diala insulating oil, this
expression becomes:
2-12
s k0 * 03.862 * t 0.3559 [27]
-6
with k0 = 1.88310 with an appropriate unit to balance the equation.
Figure 2.11 shows the growth of this surface film as a function of time at
different temperatures for copper and brass contacts. It may be seen from this
figure that the film grows faster on copper than on brass, which means that copper
contacts are more liable to this long-term aging effect. OLTC manufacturers
therefore select the contact material carefully in order to slow the development of
the resistive contact film.
Figure 2.11 Regression lines of the surface film growth on copper and brass contacts at
different surface temperatures in Shell Diala D insulating oil [24].
The contact resistance and the power losses at the contact increase as the surface
film grows. Figure 2.12 gives an example of how the contact resistance increases as
a function of thickness when an oxidation film develops on a copper contact in air.
The contact first shows a low contact resistance when the film is relatively thin, but
above film thickness of 1000 the resistance increases sharply leading to contact
failure [28]. This is due to changes in the conduction mechanism: tunnelling effects
are found in thin films (thinner than 50 100 ), while the bulk resistivity of the
film dominates in thick films [28]. Thick surface films of tens of micrometers have
been found on tap changer contacts [26].
18
Figure 2.12 Resistance of an oxidation film on

copper contacts in air, as a function of film
thickness [28].
The contact resistance can now be written:

Rc Rclean Rtunnel RFILM
2-13
where Rclean is the so called constriction resistance of the metal-to-metal contact,
Rtunnel is the tunnelling resistance for thin films (creating a quasimetallic contact)
and RFILM the resistance of the thick film. Due to the thick film on degraded
OLTC contacts, RFILM can be high compared to Rtunnel and Rclean, as in the example
of Figure 2.12, so Rtunnel and Rclean can be neglected during DRM on degraded
OLTC contacts [29].
Ref. [30] describes the contact resistance for one circular contact spot of
contact radius a and film resistivity as:
Rc Rclean R film
2a a 2
2-14
where the film surface resistivity is calculated by multiplying the film resistivity f
by the film thickness f.
For multiple parallel contact spots with a resistive film, the resistance according
to Equation 2-5 becomes:
Rc
1
2
i
a
2
ai 2i
d
i j
ai2 a 2j
ij
( ai 2i )( a j 2 j )
ai2
(
)2
ai 2i
[30]
2-15
If the contact force increases, the number of contact spots and the contact area
increase. This causes both the film resistance and the constriction resistance to fall,
as shown in Figure 2.13. When the contact pressure is released, the material
19

remains deformed so the radius of the contact spots remains large. As a result, the
contact resistance will only increase slightly when the contact pressure is
released [31]. The bulk resistance is independent of the contact force [31].
Figure 2.13 Increasing the contact force has no effect on the bulk resistance, but the film
resistance and constriction resistance both fall because the number of contact spots and the
contact area increase [31].
The resistive behaviour of the film can be simplified by assuming that the film is
conducting when the thickness <c and insulating when >c [32], which
simplifies the behaviour of the film shown in Figure 2.12. The growth of the layer
can thus be described as a process that reduces the effective contact area. If we
assume that the effective contact radius r is reduced to ra, the contact area of one
circular spot can be described as a function of time by the expression:
r
1 K
a
Dt
Dt
1 K 2
2
a
a
2-16
[32]
where the diffusion coefficient of the oxidation layer D (which describes the speed
at which the layer grows into the contact area) can be written:
D D0 e
[32]
2-17
where D0 is the frequency factor, B the temperature-dependent diffusion

parameter and the homologous temperature (the temperature in relation to the
melting point Tm):
T
Tm
2-18
Ref. [32] determined contact lifetime of copper contacts using a coefficient K:
K 2.2 4 ln
c
m
2-19
The contact resistance of this model can then be written:

20
Rc ( t )
where
R0
2r(t )
1 K
R0
Dt
Dt
1 K 2
2
a
a
[32]
2-20
2-21
2a
It follows from this model [32] that Rc when
t tm
a2
K D
2-22
[32]
This is the life of the contact under the assumptions used for Equation 2-20. It
corresponds to the time at which the thickness of the film is >c everywhere, so
that the contact can be considered to have failed. This contact life depends on a
and is thus related to the initial contact resistance by (1/R0)2. It follows that a large
initial contact radius a results in a long contact life.
The simplified model can be extended by considering parallel circular contact spots
and a film on the contact surface. The average spot radius r(t) and the number of
spots n(t) can be used to derive the following expression for the resistance:
2-23
[32]
2n(t ) r( t )
Equation 2-16 then becomes:
r(t ) a 1 K
Dt
Dt
1
K
am 2
am 2
[32]
2-24
where a is the initial value of the mean radius and am is the initial value of the
maximum spot radius. The number of contact spots falls as the film thickness
grows. At time t, all contact spots with initial radius a<ac disappear, where:
2-25
ac K Dt [32]
so that the number of contact spots can be written:
xc
n(t ) n0 (1 f ( x ) dx)
[32]
2-26
where x=a/am (the mean divided by the maximum radius), xc=ac/am and f(x)
describes the contact spot radius distribution. The rise in contact resistance
calculated by [24] and experimentally determined by [33] resembles the calculated
contact resistance of Equation 2-24 (evaluated by [32]) when f(x) is a beta
distribution with = 1 and = 1, for which we can use the expression:
f ( x ) 6 x(1 x) [32]
2-27
This distribution describes an electrical contact with large contact spots in the
middle and smaller ones on the outside. The contact resistance can then be written:
21
R(t )
R0
1 K
Dt
am 2
1 2 K
Dt
am 2
Dt
1 K 2
am
[32]
2-28
In reality, contacts fail before Rc (Equation 2-22), at the point where the
resistance increases by a factor k relative to its initial value [32]. The life of a
multipoint contact with a beta distribution of the contact radii can then be
approximated as:
1
tl 0.56tm (1 ) 2
k
2-29
[32]
where tm is calculated from the maximum radius am. Equation 2-28 can be
corrected for this effect by reducing the maximum life (when k=) by a factor
0.56 (i.e. tt/0.56):
R0
1 K
D 1.79 t
am 2
1 2 K
D 1.79 t
D 1.79 t 2-30
1 K
2
am
am 2
For example, the data from Table 2.1 can be used to calculate R(t). Figure 2.14a
shows this curve for 0t<tm. It can be seen that the resistance goes to infinity as
the contact approaches the end of its life. However, contact failure can already be
expected earlier as described by Equation 2-29. In practice, the contact resistance
will not go to infinity because the thin film will be disrupted due to the applied
voltage, the contact will heat and high hotspot temperatures may arise. It is
therefore assumed here that the calculation of R(t) is valid up to k=10 (i.e. up to
10*R0). This part of the calculated R(t) curve is displayed in Figure 2.14b. A model
that can be used for OLTC contacts at a more advanced stage of contact
degradation will be discussed in the next section.
Table 2.1 Overview of simulation parameters used to plot R(t) of Figure 2.14.
Parameter
Bulk temperature
Hotspot temperature
Melting point of copper
Resistivity of copper
Average contact spot radius
Maximum contact spot radius
Number of micro contact spots
Critical film thickness ratio
Diffusion coefficient
Diffusion parameter
Coefficient used by [32]
Initial contact resistance
Critical contact resistance increment
22
TB
T
Tm
a
am
n
c/m
D0
B
K
R0
k
Value
50 C = 323 K
TB+1 = 324 K
1356 K
1.7241*10-8 m
10 m
30 m
45
0.04
1*10-5 m2/s
7
15
10
10
Reference
assumption
[32]
[24][32]
assumption
[24]
[32]
[32]
[32]
[32]
assumption
[32]
Figure 2.14 Simulation of resistance-time curve according to Equation 2-28 using the
parameters from Table 2.1. Left: the variation of contact resistance up to the end of contact
life. Right: close-up view of resistance variation up to R(t)=10*R0.
At early stages of the long-term aging effect, the film can be wiped off by
switching the OLTC through its cycle of operation. The motion of the contacts
during this cycle partially breaks down the surface contamination [20][34-35], thus
delaying the long-term aging effect.
2.3.3 Contacts with high local temperatures
Section 2.3.2 dealt with a mathematical model that described the early stage of
contact degradation. Another model is needed to describe the advanced
degradation stage, where long-term aging accelerates due to the temperature
dependence of the surface film growth rate. The contact resistance can also fall
during the long-term aging, as the surface film breaks down due to discharges [20]
or contact wiping. Small discharges can restore a better current path by disrupting
the surface film, but the contact resistance can increase by several orders of
magnitude before the contact improves again [20]. These discharges become worse
as degradation proceeds [20], and the oil dissociates due to these discharges and
the high contact temperatures. It is these effects together that constitute advanced
long-term aging.
Breakdown due to the applied measurement voltage can also indicate the stage
reached by the surface film. Under normal conditions, the film has a breakdown
voltage of 0.2-0.6 V, which increases to 1-4 V in the defective condition [36]. The
accompanying electric field over thin films results in high field strengths.
The decomposition of transformer insulation oil leads to the deposition of
carbon between the contacts (Figure 2.4B, [20][37]). This formation of pyrolytic
carbon is called coking [19]. Oil cracking occurs at local temperatures above 300
C [20]. Catastrophic conditions can thus be produced when the contact
temperature rises above 300 C [38], but even a rise in oil temperature near the
contact to above 100-105 C can be enough to cause serious defects [36]. Precise
measurement of the hotspot temperature is a complicated procedure, but the
23

voltage drop over the contact can be used to calculate the temperature rise from
the expression:
2-31
V 2 4 L(TC 2 TB 2 ) [20]
where L is the Lorenz constant, TC the temperature increment over the bulk
temperature in degree Kelvin and TB the bulk temperature of the contact in degree
Kelvin.
The highest local temperature at the contact interface can also be calculated
based on the I2R losses in the contact area:
I 2eff 2
( )
T TB
8eff S
2-32
[9]
where eff is the effective electrical resistivity of the contact layer, eff the thermal
conductivity of the contact layer, the film thickness and S the surface area. Using
this relation, [39] derived the following expression for the film thickness on
OLTC contacts in oil:
Ea
R
d K a t REaT K a t
e
dt
TB
I 2eff 2
( )
8 eff S
2-33
where K is the coke deposition rate in cm/day, a the amount of coke precursor
(g/cm3) generated in the oil every day due to OLTC operation, the density of the
coke, Ea the activation energy of the deposition process in Joule/mol and R the
gas constant in Joule/(molKelvin). Experiments conducted to determine the
parameters [39] led to the values Ka/ = 0.13 cm/day, eff = 0.0082 cm and
Ea/R=4730 K. Furthermore, [39] used eff = 0.4 W/mK, I=900 A, TB=300 K and
S=2 mm2. These parameters can be used to calculate the life of OLTCs, although
the high load current assumed by [39] is not common for the tap changers
considered in this thesis; the calculations performed by [39] may thus be
considered to correspond to an accelerated endurance test at heavy load. Equation
2-33 was evaluated numerically by [39] using these parameters, see Figure 2.15 and
Figure 2.16. Film growth and power losses increase gradually, which causes a rising
super temperature, until the contacts get overheated and failure occurs.
2.3.4 Contact failure due to overheating
In addition to coking, the contact material wears off locally and pitted spots
become visible on the contacts. The contacts are now irreversibly damaged, and
cannot be repaired by switching the OLTC through all its tap positions. An
overhaul is needed to undo the pitting of the contacts. Because of the infrequent
movement of the change-over selector, coking can occur on all stator contacts,
rotor contacts and other movable parts that carry the load current.
24
Dissipated power (W)
Figure 2.15 Film growth on OLTC contacts loaded at 900 A, as described by Equation 2-33,
which results in a rising super temperature and finally in contact failure [39].
Figure 2.16 Power loss on OLTC contacts at a heavy load of 900 A due to the film described
by Equation 2-33. The power losses increase gradually until contact failure occurs [39].
The final stage of the long-term aging effect is thermal runaway. The coking
accelerates due to increased contact resistance and the thermal resistance
characteristics of the carbon [9]. Coking between moving parts of the contacts and
around the springs that provide the contact pressure can cause the contacts to lose
pressure. Finally, a considerable amount of contact material disappears and
excessive arcing activates the tap changer protection mechanisms. Figure 2.15
shows that film growth and super temperature increase gradually until thermal
runaway occurs. Figure 2.16 shows how the corresponding power dissipation
increases when the system exceeds the design limitations and the cooling capacity
of the OLTC, leading to failure. Examples of failed contacts due to overheating are
shown in Appendix A.
Figure 2.17 shows the effect of the current on the contact life (characterised by
the moment at which the resistance increases to k*R0) of a copper-copper contact
25

in oil, as determined by numerical simulation [24]. It may be seen that a low
current gives a longer life. Coking is likely to occur when the resistance is rising
rapidly [24]. Figure 2.18 shows the simulated relationship between oil temperature
and contact life; these curves have the same general form as those of Figure 2.17.
Figure 2.17 Simulated relationship between contact resistance and contact life for a coppercopper contact in oil at various load current [24]. A higher load gives a shorter contact life.
According to [24], the long-term aging effect is accelerated by:
26
Infrequent movement of the contacts

o The surface film is not wiped off the contacts when the changeover selector is not operated.
High temperatures
o The formation of the surface film is highly temperaturedependent.
High load current
o Power losses at the contact interface increase exponentially with
the load current and lead to correspondingly greater heating of
the contacts.
Low contact pressure
o Coking and pitting of the contacts are more likely to lead to
contact jamming when the contact pressure is low. The layer of
contamination on the contact surface, which can lead to contact
malfunction or failure, is also more easily disrupted at high
contact pressure.
Copper or brass contacts
o The growth rate of the surface film is higher on copper and brass
contacts than on silver contacts [24]. Pitting and coking are
reduced by silver-coating of the contact surface; however, older
OLTC designs may use bare copper contacts.
Figure 2.18 Simulated relationship between contact resistance and contact life for a coppercopper contact in oil at various load current [24]. A higher oil temperature gives a shorter
contact life.
2.3.5 Corrected model of tap changer resistance

We describe the long-term variation of OLTC resistance R(t) by Equation 2-30 for
slightly degraded contacts and the film thickness by Equation 2-33 for severely
degraded contacts, on the basis of the following assumptions:
1. The early stage of contact degradation can be modelled by parallel circular
metal-to-metal micro-contacts, as described in Section 2.3.2 The film is
assumed to be conducting when its thickness <c and insulating when
>c. The effective contact radius r is therefore reduced to ra
2. The radii of the contact spots are distributed according to a beta
distribution
3. A constant contact force is applied
4. A constant current is applied
5. The on-load tap changer is not operated during its life
6. The contact film remains undisrupted
As indicated above, the model for R(t) is based on the assumptions that the OLTC
is not operated during its life and that the contact film is not damaged. These
assumptions are unrealistic, but can be corrected for with the aid of experimentally
fitted parameters. Switching the tap changer will slow down the development of
the contact film and reduce contact resistance. Moreover, differences in the
measured resistance occur at different measurement current amplitudes due to
damage to the film caused by the measurement current; this also needs to be
corrected for. These corrections are illustrated in Figure 2.19. The corrected value
of R(t) can be represented by the expression:
R(t ) R( t ) CDRM COLTC
2-34
27

where COLTC represents the wiping effect due to operation of the OLTC, and CDRM
the correction for differences in the measurement current during the dynamic
resistance measurement.
Figure 2.19 The calculated resistance R(t) can be corrected for wiping action of the contact
film caused by OLTC operation (COLTC) and for the damage to the film caused by the
measurement current during the resistance measurement (CDRM).
The correction factor COLTC is a number between 0 and 1, where 0 represents a

wiping action that totally cleans the contacts and 1 represents the complete
absence of wiping action. The actual value of COLTC depends on the contact design,
contact pressure and the mechanical resistance of the film to the wiping action.
Figure 2.20 presents typical experimental data that were used as a basis for the
determination of the value of COLTC. A set of degraded OLTCs, all of the same
type, was selected and the number of operations during a year was plotted against
the contact resistance (estimated on the basis of the assumption that each dynamic
resistance measurement was performed with 1 A and a constant voltage of 2 V,
thus neglecting the change in resistance during the measurement). When multiple
measurements were performed on a given OLTC, the results were averaged. A
trend line was determined from the data points of Figure 2.20 using the least
square method, resulting in the weak relationship:
R R0 (2144 0.32 n y )
2-35
where R is the resistance in m, R0 is the reference resistance of 1m and ny is the

average number of operations during a year (500 < ny < 4500). It follows that the
correction factor COLTC can be written:
COLTC
R
1 1.49 104 n y
Rny0
2-36
The data set shown in Figure 2.20 gives a (Pearson) correlation of -0.52, which
indicates that there is a weak relation between the number of operations and the
28

contact resistance. We refrain from using this relation for predicting the resistance,
due to the low correlation involved.
3000
Contact resistance (mOhm)
2500
2000
1500
1000
500
0
0
500
1000
1500
2000
2500
3000
3500
4000
4500
OLTC operations per year
Figure 2.20 Contact resistance as measured by DRM plotted against the number of
operations per year for a set of degraded OLTCs, all of the same type.
CDRM is the ratio of the measured and calculated resistance. The difference
between the two is due to damage to the film caused by the measurement current.
This correction factor thus describes the effect of the measuring procedure on the
results, not related to the degradation produced when the OLTC is in service.
CDRM depends on the measurement current; and is introduced to give a measure of
the degradation process that is independent of the measurement parameters. It is
assumed that the circuit resistance and inductance are kept constant during the
measurement, so that the measurement voltage varies with the current i. Hence,
CDRM is a function of i:
CDRM
R2,5
2-37
Ri
where R2,5 is the resistance measured at 2.5 A. Experimental results show that
CDRM can be written in terms of two exponential functions:
CDRM
R2,5
Ri
2,5
( i ioffset )
k0 kdyn1 e
i1
kdyn 2 e
( i ioffset )
i 2
2-38
where i is the amplitude of the measurement current, i1 and i2 the rate at which
the resistance decreases when the current is increased, and Ioffset, k0, kdyn1 and kdyn2
are constants used to fit the expression to the experimental results. Figure 2.21
shows the result of a set of resistance measurements performed on OLTCs
suffering from advanced long-term aging, using currents from 2.5 to 100 A. The
trend line giving the best fit with these results is represented by the expression:
29
230
CDRM k
10 220 e
( i 19)
22
30000 e
( i 19)
4
2-39
where i is the amplitude of the current in the range 2.5 A I 100 A and k is a
constant that is used to make the units of the equation equal. Since the scatter of
the measured data is acceptably low and the (Spearman) correlation is high (-0.95),
we conclude that Equation 2-39 can be used to describe the contact resistance of
degraded OLTC contacts at different measurement current levels.
400
Resistance (mOhm)
300
Measurement data
200
Measured Trend
100
0
0
20
40
60
80
100
Current (A)
Figure 2.21 The black curve is the trend line determined on the basis of a set of resistance
measurements using currents in the range 2.5 A i 100 A.
2.4 Conclusions
This chapter gave an introductory explanation of the operation of common types
of on-load tap changers, namely selector and diverter switch type OLTCs,
including a coarse tap-selector, reversing switch and tap selector. It discussed the
way in which degradation mechanisms can lead to failure of the tap changer. In
this context, failure was defined as the situation in which the OLTC cannot
perform its primary task, i.e. to select another tap without interrupting the load
current. Contact degradation was modelled using equations taken from the
literature and corrected with the aid of experimentally fitted correction parameters
to match the results of dynamic resistance measurements on tap changers. OLTC
condition diagnosis is discussed in the next chapter.
30
CHAPTER 3 - ON-LOAD TAP CHANGER DIAGNOSIS
3
On-Load tap changer diagnosis
The previous chapter discussed the way on-load tap changers degrade, which
affects the transformer reliability. However, what are the possible options for an
early detection of these irregularities, and what are their limitations? This chapter
considers possible methods employed to assess the condition of an OLTC. There
are two basic kinds of methods which can be used: applying on-line monitoring or
performing off-line diagnosis.
On-load tap changer monitoring entails the continuous checking of parameters
that can be acquired on-line by a fixed, mounted monitoring device. A monitoring
system logs operation parameters and safeguards the normal operation of the tap
changer. Monitoring devices can generate alarms directly when a defect arises.
Most on-line OLTC monitoring systems also generate maintenance advice based
on the total switched current, the tap position history and operation temperatures.
In contrast to on-line monitoring, off-line diagnosis is characterised by periodic
rather than continuous measurements. In this way, more accurate condition
information can be obtained because more advanced measurement systems are
normally used. Diagnostic measurements can be used to supplement on-line
monitoring, for instance when the monitoring system indicates a defect, but can
also be applied separately. The focus of condition assessment techniques is to
identify defects at the earliest stage, before significant damage is caused. Early
discovery and remedy of defects avoids expensive consequences [40].
Table 3.1 shows which of the OLTC diagnostic measurements discussed in this
chapter can be applied on-line. The tap changer is subdivided into the oil /
insulation and the mechanical part. The applicability of the diagnostic methods to
these subsystems is also shown in Table 3.1. The methods that are suitable for
those parts will be discussed separately in Sections 3.1-3.3.
31

Table 3.1 Overview of OLTC diagnostic measurements discussed in this thesis and their
applicability to five subsystems.
mechanical
before
after
springs springs
Inspection / functional check
offline
Dissolved gas analysis
online offline
Vibration / Acoustic fingerprint
online offline
Temperature measurement
online
Motor power measurement
online offline
Axis position measurement
offline
Static resistance measurement
offline
Dynamic resistance measurement
offline
Transformer turn ratio test
offline
electrical contacts
arcing
switch
change-over
selector
dielectric
2V
1III
1VI
1I
1I
IV
0 not suitable
suitable as a functional
check for a part of the
1
subsystem or for some
defects
suitable for a part of the
2 subsystem or very
suitable for some defects
3 very suitable
1I
II
2II
VI
III
IV
Diagnoses (excessive)
friction
Only w hen the arcing
sw itch insert is not
removed
Rough diagnoses of
coking
Diagnoses only the
current tap position
Only arcing sw itch
accessible
Diagnoses discharges
and isolation aging
3.1 Oil and insulation diagnosis

3.1.1 Dissolved gas analysis
A frequently used method for transformer diagnosis is the analysis of transformer
oil. This method is called dissolved gas analysis (DGA). A sample of oil is taken
and the gasses which are dissolved in it are analysed, typically hydrogen, methane,
ethane, ethylene, acetylene, carbon monoxide and carbon dioxide [37]. The method
can be applied to oil from the arcing switch compartment [41] and to the oil of the
(coarse) tap-selector. The method is based on the fact that oil decomposes into
different gases depending on the temperature at the location of a defect (see Figure
3.1). A conclusion about the type of defect can thus be drawn. Detectable defects
include contact coking, overheated contacts and insulation degradation. Duval [4142] shows that gas development due to defects is temperature dependent (Figure
3.1 and Table 3.2). Some OLTC defects can also be found with DGA. The oil
from the arcing switch compartment can provide valuable knowledge about the
switching performance. [43] gives some examples of DGA results for an OLTC
which suffers from coking, overheating or arcing. Specific gas concentrations can
be used to draw global conclusions about the type of defect (Figure 3.2).
32
Figure 3.1 The concentration of dissolved gases is dependent on the fault temperature [44].
Table 3.2 Specific key gases are formed when transformer insulation oil decomposes into
different gases. This process is dependent on the temperature at the location of a defect [45].
Key Gas
H2
Characteristic Fault
Partial Discharge
C2H6
Thermal Fault <300 C
C2H4
Thermal fault 300 C to <700 C
C2H2, C2H4 Thermal Fault>700 C

C2H2, H2
Discharge of Energy
DGA provides on-line OLTC diagnosis. However, dissolved gas analysis does not
offer precise information about change-over selector contact wear when it shares
its oil with the transformer oil of the main tank. For example, DGA is effective
when the contact temperature exceeds 400-500 C due to a low rate of gas
generation below the boiling point [36]. Due to the small heating surface (a few
cm2) the rate of gas generation could be only a few ppm a month; gassing could be
a symptom of severe contact deterioration [36]. Localisation of the defect is always
difficult. In addition, looking at the trend in specific gases only provides a warning
about failure in serious cases, such as overheating and thermal runaway [46].
Moreover, regular leakage of dirty oil from the arcing switch compartment to the
change-over selector makes an unambiguous analysis of oil in the main tank
difficult. To conclude, defects inside the OLTC take some time to show in a DGA.
Oil samples for DGA can be analysed in a laboratory, but on-site devices for
DGA are also available, for example the ones described in [47-49].
33
Zone Identification
N
T3
T2
Recommended actions
Normal operation
Severe thermal fault T3 (T>700 C), heavy
coking.
Change the oil. Inspect the LTC for coking of
contacts
Severe thermal fault T2 (300<T<700 C),
coking
Fault T3 or T2 in progress, usually with light

Test or inspect the LTC for signs of light coking
X3 coking or increased resistance of contacts.
or resistance of contacts, or of severe arcing
Or, severe arcing D2
D1 Irregular arcing D1 (outside zone N)
Inspect the LTC for small signs of arcing
Irregular arcing D1 or thermal fault in

X1
Area still under investigation
progress
Figure 3.2 The Duval Triangle determines the source of the gas by looking at OLTC gas
ratios [41].
3.1.2 Study of the agreement between DGA and DRM results

It is of interest to know if there is a relationship between the gases that are
produced at degraded change-over selector contacts and the results of dynamic
resistance measurements. Both methods are sensitive to coking [43], which occurs
during the long-term aging effect on the OLTC contacts. In this situation, the
gases CH4, C2H6 and C2H4 are formed. C2H2 and H2 are expected to form in the
advanced stages (at higher temperatures). In addition, the contact resistance will
increase, and therefore a relationship with DRM is expected.
34

Figure 3.3 shows a possible relationship between DGA and DRM results.
Increasing gas concentrations were measured in the transformer main tank oil,
which triggered the Buchholz alarm three times (indicated with arrows in Figure
3.3). Analysis of the gas concentrations and ratios indicated a serious thermal fault
with coking. DRM results showed increased contact resistance on a few tap
positions, an indication of a tap-selector contact problem. Inspection of the tapselector contacts indeed showed overheated contacts as predicted by DGA and
DRM (the overheated and normal contacts are shown in Figure 3.3-inset).
3000
450
2500
400
350
2000
300
1500
250
200
1000
150
CH4
C2H6
C3H8
C2H4
100
500
Gas concentration (ppm) for C2H4 and C3H6
Gas concentration (ppm) for CH4, C2H6 and C3H8
500
50
C3H6
0
0
28-10-95 11-03-97 24-07-98 06-12-99 19-04-01 01-09-02 14-01-04 28-05-05 10-10-06 22-02-08 06-07-09
Date
Figure 3.3 Example of DGA results for an OLTC with overheated tap-selector contacts
(inset). Rising gas concentrations are visible, which caused Buchholz alarms at the dates
indicated by an arrow.
To determine whether there is a general similarity between DRM and DGA results,
a population of OLTCs was selected. Oil samples were taken from the main tank
and a dynamic resistance measurement was performed on all transformers.
The dataset selected contained 72 transformers. The power rating of the
transformers was between 18 MVA and 30 MVA. Most of the transformers had a
voltage ratio of 50 kV/10 kV and have selector switch type tap changers with a
change-over selector. Each change-over selector was measured with DRM:
31 were in good condition

26 were in moderate condition
15 were in severe condition
This information on the condition of the tap changer, together with the results of
DGA tests, was used for statistical analysis.
First, the correlation between the gas produced (or gas ratio) and the condition
of the tap changer is calculated. The result reveals that C2H4 and C3H6 and the gas
35
C2H2/C2H4 ratio
ratios C2H4/T (where T represents [CH4 + C2H4 + C2H2]) and C2H2/C2H4 have
the best correlation with the information about the condition of the OLTC. Gas
ratios were only calculated if one of the gases exceeds the Duval limits (for
example 3 ppm for C2H2 and 75 ppm for C2H4). For example, C2H2/C2H4 has the
highest correlation of 0.25 with the DRM results, which means that the gas ratio is
expected to decrease as the condition of the contacts deteriorates.
Scatter plots of the gases which have the highest correlation have been made.
Figure 3.4 shows this scatter plot for C2H2/C2H4. There is a large correlation
between the dissolved gases and the three contact conditions as determined by
DRM (good, moderate and severe).
A probability density plot was then made to determine the best fit of the
regression line onto the gases, using the Weibull distribution. Again, a large
correlation was found between the probability density functions of the C2H2/C2H4
conditions, which was expected regarding the scatter plot.
Using this dataset, only a very weak relationship between the condition of the
OLTC change-over selector contacts (as measured with DRM) and a certain gas
amount or gas ratio was found. A C2H2/C2H4 ratio could not be related to the
DRM results. This can be explained firstly by the large amount of oil in the main
tank: different degradation mechanisms can produce various gases at the same time
and distort the conclusions. Secondly, the main tank exchanges air with the
surrounding environment through breathers, and thus gases formed by coking can
be dissipated. Therefore, it is concluded that DRM can be used for finding contact
problems that did not appear in DGA.
Figure 3.4 C2H2/C2H4 scatter plot for the DRM conditions good, moderate and severe.
36
3.2 Mechanical diagnosis

3.2.1 Acoustic and vibration signature
The moving parts of the drive mechanism produce acoustic waves and vibrations
when the OLTC is in operation. Tap changers with mechanical or electrical
problems produce different signals compared to healthy tap changers. The number
of bursts, the time between them [13], the main frequencies [50] and the strength
of the signal can contain valuable information. Comparing the signal measured
with a known signature will indicate these problems. For this comparison it is
necessary to know the signature of a healthy tap change operation. A large number
of measurements have been used to determine a normal operation for each type of
tap changer. One can also compare the measurements with previous
measurements to see if the tap changer is damaged or whether excessive arcing
occurs. These measurements can be performed periodically or continuously and
can be done on-line. Problems in the arcing switch and the change-over selector
can be distinguished because they operate at another time. Monitoring devices [51]
compare the signatures at every tap change, in this way looking at the development
of wear.
Figure 3.5 shows an example of the acoustic waves that are recorded when an
OLTC operates. Hilbert transform was used to present this envelope. The tap
changer was measured when the transformer was in an energised and de-energised
state. Comparing the recordings of an energised and de-energised transformer can
separate the electrical (arcing) and mechanical sources.
Figure 3.5 Example of acoustic waves that can be recorded on an energised and deenergised OLTC, after Hilbert transform [52].
To obtain the desired information from the complex measurement data, a

transformation can be applied to the data, for instance a Fourier transform or
continuous wavelet transform [53]. References [13],[51] and [54-56] discuss this
37

diagnostic method in-depth. An example of a method using acoustic and vibration
signatures is given in [57].
3.2.2 Motor power measurement
Defects or friction in the drive mechanism and the OLTC can also be detected by
the measurement of the active power consumption of the drive motor [58]. The
active power consumption is proportional to the delivered torque of the motor.
Irregularities in power consumption can therefore indicate defects in the
mechanical system that is connected to the motor. The springs inside the drive
mechanism act as an energy accumulator and separate the motor drive from the
OLTC itself. Measuring motor power will therefore allow diagnosis of defects in
the drive from the motor to the springs. This measurement can be done both onand off-line.
Figure 3.6 provides an example of an off-line measurement of a motor power
curve of an OLTC drive motor. Power variations can be caused by moving the
selector contacts or by loading the springs.
Some tap changer monitoring devices use the motor operation time in
combination with average or maximum active power consumption. The start-up
time of the motor can also provide information about the status of the drive. The
curve can also be compared with earlier measurements. The location and
amplitude of the power peaks can indicate their cause.
Another example of a motor power measurement is described in Ref. [59].
Vb.emf
Figure 3.6 The motor power of a tap changer drive motor shows the inrush current followed
by a stable operation. Power variations can be caused by moving the selector contacts or by
loading the springs.
38

3.2.3 Position measurement
Section 3.2.2 showed that the active motor power consumption of the drive motor
can reveal defects and friction in the mechanical system up to the spring system.
Another method is necessary to diagnose that part of the drive mechanism and the
tap changer that is operated after releasing the springs. Measuring the position of
the remote end of the drive shaft is one way, although the drive shaft is not always
accessible at the remote end. Each OLTC operation results in a movement of the
shaft (see Figure 3.7, the DRM-current is also plotted). It can be clearly seen that
the shaft oscillates at the end position.
A position measurement such as this can be used to derive the graphs of velocity
and acceleration. Defects, friction, torsion or synchronisation problems between
the phases alter these graphs, so a comparison with earlier measurements is
indispensable.
3,9
1,4
3,8
1,3
3,7
1,2
3,6
1,1
3,5
3,4
0,9
3,3
0,8
3,2
0,7
3,1
0,6
3
2,9
0,5
2,8
0,4
2,7
403,7
403,75
403,8
403,85
403,9
403,95
Time [s]
404
404,05
404,1
404,15
0,3
404,2
Figure 3.7 The movement of the remote end of the drive shaft can show defects or friction in
the OLTC.
3.3 Contact status diagnosis

In addition to mechanical and insulation condition assessment, diagnosis of the
contact status can be performed. Three methods for contact diagnosis will be
discussed in this chapter. Apart from the methods discussed here, [60] mentions
that the exciting current and power factor in a transformer turn ratio measurement
and a sweep frequency response analysis can also indicate contact problems.
39

3.3.1 Temperature difference measurement
Damaged contacts inside the tap changer, for instance due to coking of the
change-over selector contacts, can lead to increased contact resistance. This is
accompanied by an increased power loss. These power losses may run up to
hundreds of watts for each contact at a nominal current [11]. At OLTCs which are
mounted against the transformer tank, the temperature of the oil can increase
significantly due to power losses on the contact interface. A temperature difference
between the tap changer oil and the main transformer oil will then develop [50]. A
healthy tap changer will produce very little heat and its temperature will be lower
than the main tank. Table 3.3 shows an example of how the temperature difference
can be interpreted.
In the case of an external OLTC, temperature difference measurement can be
used as a diagnostic or monitoring tool to detect bad contacts. This method has
been applied to reactor type OLTCs. Resistor-type OLTCs may warm the oil in the
external tap changer tank through the heat developed in the transition resistors.
However, as only one tap position is measured, the problematic contacts that are
not a part of the current circuit and do not produce heat will not be detected.
Table 3.3 Example of how the temperature difference between the tap changer and the main
transformer tank can be used to diagnose the contacts [45].
CLASSIFICATION
Attention
TEMPERATURE EXCESS
09 C
Intermediate
1020 C
Serious
2149 C
Critical
>50 C
18
The monitoring of the temperature difference can be done on-line. Fixed sensors
inside the transformer can be used and should be positioned on the same side at
the same elevation [61]. Other heat sources such as sun or a changing load should
not cause warning signals when the OLTCs oil heats faster than the oil in the main
tank or vice versa [50]. The same problem can occur when the cooling fans switch
on and off [50].
External tap changers can also be diagnosed by infrared thermography, see
Figure 3.8.
Examples of tap changer diagnosis using temperature measurements include
[62] and [63].
40
Figure 3.8 Infrared image of an OLTC, where the tap changer temperature is about 7 C
higher than the transformer main tank due to contact problems [64].
3.3.2 Static resistance measurement

OLTC contact degradation can be measured by means of a resistance
measurement. Additional information is obtained compared to the temperature
measurement because all tap positions are measured. These resistance
measurements are performed when the tap changer is motionless and are therefore
referred to as static resistance measurements.
Static resistance measurement can be performed using a simple DC power
source or a dedicated measurement device. Static resistance measurements can be
used to:
Determine the resistance of each tap position, see Figure 3.9. A straight line is
seen when the tap changer is in good condition. This method is also called a
transformer DC winding resistance measurement because the tap changer is
measured in series with the transformer windings. As the windings have a large
inductance, this measurement takes a long time: the resistance is measured when
the current through the winding is completely stable. Equation 3-1 shows this
relationship. Resistance is measured when dI/dt=0.
Resistance()
0,4
0,3
0,2
0,1
0
1
26
Tapposition
Figure 3.9 Graph of the static resistance measurement of all tap positions.
41
dI
dt
3-1
(TS TK )
[65]
(TM TK )
3-2
U I *R L*
The results obtained can be compared with earlier resistance measurements, but
one should be careful to correct the readings for changes in the winding
temperature because values depend on the latter, according to Equation 3-2.
RS RM
Where:
RS = resistance at reference temperature TS
RM = measured resistance
TS = reference temperature
TM = temperature at which resistance was measured
TK = Temperature coefficient, 234.5 for copper and 225 for aluminium
Determine the resistance of one tap position when the current is increased and
decreased. A considerable memory effect is measured when the contacts are
degraded (see also Appendix C). New contacts barely show this effect. The
resistance of the carbon and the surface film is highly dependent on the current
[11]; therefore a resistance measurement with high measurement currents can
diagnose the degradation at currents that resemble the operation conditions. In
this case, the removable part of the arcing switch is taken out and the
measurement current is directly applied to the change-over selector. In this way,
the main transformer windings are not measured, so a higher current can easily
be applied to the tap changer. A DC power source can be used to measure the
resistance (for example up to 100 A). An example of such a measurement on
degraded contacts is shown in Figure 3.10 [11]. The initial resistance can be in the
range of 500 m. When the current is increased, the resistance of the carbon
usually decreases (for example to about 20 m). A memory effect is present
when the current is decreased, as the resistance does not rise to the initial value
again. Section 4.2.2 will discuss this memory effect in more detail. A coarse tap
winding can be located in series with the measurement in some tap positions.
This value has to be subtracted from the measured contact resistance.
Detect discontinuity of the OLTC. The tap changer can be operated between
two static resistance measurements and some transformer ohmmeters can detect
discontinuity in the measurement current.
Resistance measurements on tap changers can be performed using transformer
ohmmeters such as [66] and [67].
42

3.3.3 Dynamic resistance measurement (DRM)
Dynamic resistance measurement is suitable for detecting irregularities in the
OLTC contact resistance, as are static resistance measurements. The main
difference between dynamic measurements and static measurements is that the tap
changer is operated during the measurement. Important information about the
importance and location of the long-term effect on the change-over selector
contacts can be extracted from the measurement. The contacts do not remain at
each position long enough to allow the measurement current to stabilise
(stabilisation is attained after a relatively long time constant caused by a high
inductance of transformer windings). Despite the fact that DRM can be considered
less accurate than static resistance measurements, it may provide more information
about the type and location of defects inside the OLTC. For example, a typical
dynamic measurement does not allow enough time to stabilise the measurement
current and cannot reveal contacts with resistance increased by 100 , whereas
static resistance measurement can do this, but cannot show irregularities in the
transition time. Chapter 4 deals with DRM in more detail.
Examples of diagnostic instruments that are able to show the dynamic
resistance during operation are [12] and [68-70].
Resistance (mOhm)
250
200
150
Increasing
Decreasing
100
50
0
0,00
20,00
40,00
60,00
80,00
100,00 120,00
Current (A)
Figure 3.10 Typical resistance measurement example [11]. When the current increases, the
resistance will decrease. In this example the resistance dropped from 237 m to 14 m
when the current was increased from 3.5 A to 100 A. At a decreasing current the memory
effect means the resistance cannot rise.
3.4 Discussion
In Table 3.4 we give an overview of the diagnostic methods for OLTC as
discussed in the previous sections. The following can be concluded:
43
44
Visual inspection can be used to reveal degradation in the accessible parts of

the OLTC and drive mechanism. Inspection intervals are determined by the
number of operations or by time intervals, whichever comes first [71].
However, not all parts are accessible for inspection. Change-over selector
contacts, for example, are not accessible in in-tank type OLTCs. Other
diagnostic methods can be used to diagnose these parts and to detect
maintenance errors.
Dissolved gas analysis is in principle sensitive to the irregular operation of
the contacts due to coking and overheating. However, oil samples taken from
the main tank of two-compartment type OLTCs provides information about
all components in the main tank, which masks the sought information about
contact conditions. Other methods can be used to localise the defect. DGA is
insensitive to mechanical defects.
Acoustic vibration measurements on the OLTC are useful for detecting
mechanical and electrical defects. Both types of defects can be distinguished by
comparing measurements of an energised and a de-energised OLTC and by
looking to dominant frequencies in the signal. On-line and off-line
measurements can be performed: on-line measurements can be used to detect
defects and degradation at an early stage, which can limit possible damage to
the transformer.
Different types of OLTCs can produce different vibration signals, the
interpretation of vibration signals can therefore be complex. The method is
insensitive to dielectric degradation.
Temperature measurements on the OLTC can also be useful to detect
increased contact resistances. This measurement is performed on-line but does
not diagnose the tap positions that have not been used. Defects that do not
generate heat are not detected. This method is applicable to reactor type
OLTCs installed in a separate tank.
Motor power and shaft rotation measurements can be used to find
problems in the drive mechanism: active motor power consumption in the
mechanical system up to the spring system and drive shaft rotation for parts
behind the springs.
Static and dynamic resistance measurements can determine the condition
of the contacts. For a dynamic measurement it must be possible to short
circuit the transformers secondary windings, because the effect of the
resistance changes can then be recorded with high time-accuracy. Both
methods use off-line measurements. Section 3.3.3 concluded that dynamic
resistance measurements provide more information about the condition than
static measurements because a larger contact surface is measured. This is an
added value of dynamic resistance measurements.
Other methods than resistance measurements must be used to obtain a full
picture of the insulation condition and mechanical condition (although DRM
is able to detect defects in the drive mechanism that affect the operation of the
contacts).

DRM results were compared with the results of DGA in Section 3.1.2: statistical
analysis showed that there is no significant correlation between defects that are
defected with the DGA and the DRM method on the basis of the database and
parameters used. Resistance measurements and DGA can be considered to
supplement each other.
Resistance measurements and ratio measurements are common detection
methods for power transformer defects [72-73] and these methods, together with
inspection and gas analysis, account for 61% of the popular power transformer
measurements [72]. These two methods are particularly suitable for detecting
defective windings. In addition, resistance measurements are also suitable for
detecting contact degradation, especially when applied while the OLTC is in
operation.
Table 3.4 Overview of OLTC diagnostic measurements. Each method diagnoses particular
areas of the OLTC: the practicability is indicated for each method.
Visual inspection
Mechanical
(drive)
Electrical
contacts
Up to springs
The drive can be inspected for wear and defects
Behind springs
The drive can be inspected for wear and defects

Accessible for inspection: contact wear can be recognised,
misaligned contacts could be visible.
Arcing switch
Change-over selector and

In-tank contacts are not accessible for inspection
tap-selector
Dielectric
Arcing switch compartment is accessible for inspection
Dissolved gas analysis

Mechanical
(drive)
Electrical
contacts
Up to springs
Mechanical defects usually do not influence DGA results
Behind springs
Mechanical defects usually do not influence DGA results
Arcing switch
Contact coking, contact wear leading to excessive arcing or

overheated contacts can be detected by DGA, but is not considered
reliable since the products of oil decomposition by arcing masks the
effect of contact degradation.
Contact coking, discharges or overheated contacts can be detected

by DGA. In case of two compartment type OLTCs, the oil sample
Change-over selector and resembles the condition of all components in the main transformer
tank and this method is not considered reliable since the products of
tap-selector
oil decomposition by arcing may leak from the arcing switch
compartment.
Insulation degradation due to high contamination and surface
discharged (treeing) can be detected.
Dielectric
Acoustic vibration measurements

Mechanical
(drive)
Electrical
contacts
Dielectric
Up to springs
Mechanical defects can cause deviations in the vibration signature

compared to recordings from the same type of OLTC.
Behind springs
Mechanical defects can cause deviations in the vibration signature

compared to recordings from the same type of OLTC.
Arcing switch
Excessive arcing and friction can be visible in the vibration signature.
Change-over selector and

Excessive arcing and friction can be visible in the vibration signature.
tap-selector
Dielectric defects and degradation usually do not show in vibration
signatures.
45

Temperature measurements
Mechanical
(drive)
Electrical
contacts
Dielectric
Up to springs
Mechanical defects do usually not generate heat.
Behind springs
Mechanical defects do usually not generate heat.

Contact coking or overheated contacts that carry the load current can
Arcing switch
generate heat and can be detected.
Change-over selector and Contact coking or overheated contacts that carry the load current can
generate heat and can be detected.
tap-selector
Dielectric defects and degradation usually do not generate heat.
Motor power measurements

Mechanical
(drive)
Electrical
contacts
Dielectric
Friction, contact binding and wear can change the active motor power
needed for the OLTC operation.
The drive mechanism behind the spring system is not powered by the
Behind springs
drive motor but by the springs.
The contacts after the spring system is not powered by the drive
Arcing switch
motor but by the springs.
Change-over selector and Friction and contact binding can change the active motor power
needed for the OLTC operation.
tap-selector
Dielectric defects and degradation usually do not affect the active
motor power.
Up to springs
Shaft angular position measurement

Up to springs
Mechanical
(drive)
Electrical
contacts
Dielectric
After springs
Defects and degradation in the drive system usually do not affect

shaft rotation.
The rotation of the remote end of the drive shaft can give condition
information about the mechanical properties of the shaft, friction and
synchronisation problems between phases.
Arcing switch
Friction of the contacts can influence the shaft rotation.
Change-over selector and The change-over selector and tap-selector are usually not accessible
for position measurements.
tap-selector
Dielectric defects and degradation usually do not affect the axis
rotation.
Static resistance measurements

Mechanical
(drive)
Up to springs
After springs
Arcing switch
Electrical
contacts
Dielectric
46

contact resistance.
contact resistance.
One contact position is diagnosed, defects and degradation such as
contact wear, coking and misalignment can be detected.
Change-over selector and One contact position is diagnosed, defects and degradation such as
tap-selector
Dielectric defects and degradation usually do not affect contact
resistance.

Dynamic resistance measurements
Mechanical
(drive)
Up to springs
After springs
Arcing switch
Electrical
contacts
Dielectric
Defects and degradation in this part of the drive system usually do

not affect dynamic resistance.
Defects and degradation in the drive system can influence OLTC
transition time and can be detected with DRM.
All contact positions are diagnosed, defects and degradation such as
Change-over selector and All contact positions are diagnosed, defects and degradation such as
tap-selector
Dielectric defects and degradation usually do not affect contact
resistance.
The overview in Table 3.4 can be summarized, as in Table 3.5. From Table 3.5 it
can be concluded that none of the methods provides a full OLTC diagnosis.
Mechanical condition diagnosis can be obtained by:
Inspection
Acoustic/vibration measurements
A combination of:
o Motor power measurement (for defects up to the spring system)
o Position measurement on the remote end of the drive shaft (for
defects behind the spring system)
Electrical contacts that are not accessible for inspection may benefit from
resistance measurements. Static and dynamic resistance measurements are
both performed on all tap positions, but DRM provides more information
about the condition of the contacts than static resistance measurements
because the contact surface for the entire cycle of operation is measured. In
addition, DRM can detect a wider variety of maintenance errors because
contact timing is also checked.
Temperature measurements can reveal serious defects in the OLTC, but
are less sensitive to increased contact resistances compared to resistance
measurements. Defective contacts that are not selected are not recognised by
temperature measurements.
In addition, DGA can also detect arcing and contact coking, but
determining the location where coking occurred is more difficult compared to
resistance measurements.
Dielectric degradation can be detected by dissolved gas analysis. The arcing
switch compartment insulation can be examined during inspection.
47

Table 3.5 Overview: suitability of OLTC diagnostic measurements as described in Table 3.4
for finding defects and degradation in particular parts of the OLTC.
Suitability of the discussed methods to diagnose specific OLTC parts
Mechanical
(drive)
Electrical
contacts
Inspection, acoustic vibration analysis, active motor power

measurement.
Inspection,acoustic vibration analysis, shaft rotation, dynamic
After springs
resistance.
Inspection, static and dynamic resistance, temperature
Arcing switch
measurements, acoustic vibration analysis.
Change-over selector and Static and dynamic resistance, temperature measurements, DGA,
acoustic vibration analysis.
tap-selector
Before springs
Dielectric
DGA, Inspection.
3.5 Conclusions
This chapter evaluated the various methods that can be used to diagnose the
condition of an OLTC, thereby detecting irregularities at an early stage, especially
defects that are not seen during inspection and due to maintenance errors. The tap
changer was first subdivided into three parts: oil and insulation, mechanics and the
contacts. Arcing switch contacts were distinguished from change-over selector
contacts (including tap-selector contacts).
Each diagnostic method was shown to be sensitive to defects in particular parts
of the tap changer. Every method also proved to have its limitations. Table 3.4
provided an overview of the methods and their suitability to each of the OLTC
parts. Table 3.5 showed that none of the methods provides a full OLTC condition
diagnosis. Therefore, a combination of different methods would be worthwhile:
Diagnosis of the mechanical system of the OLTC can be obtained by

inspection, acoustic vibration measurements or through a combination of
motor power and shaft position measurements.
Electrical contacts that are not accessible for inspection may benefit from
DRM. The following chapter will examine DRM in more detail.
DGA is a popular method for diagnosing the OLTC dielectric.
Table 3.5 showed that parts of the tap changer can be diagnosed by more than one
method because there is some correlation in their results. Section 3.1.2 elaborated
on the relationship between DRM and DGA results; however, no relationship
between the condition of the change-over selector contacts (as measured with
DRM) and a certain gas amount or gas ratio was found for the dataset selected.
48
CHAPTER 4 DYNAMIC RESISTANCE OF ON-LOAD TAP CHANGERS
4
Dynamic resistance of on-load
tap changers
As mentioned in the previous chapter, static resistance measurements on a
transformer indicate increased contact resistances in the OLTC. These increased
contact resistances can be measured by means of a dynamic resistance
measurement (DRM). However, what is the additional value of this measurement
and what are the important parameters that should be considered when
performing dynamic resistance measurements? This chapter discusses the different
aspects of dynamic resistance measurements, for example the measurement setup,
the influence of the amplitude of the measurement current and voltage, the effect
of circuit resistance and the secondary short circuit.
DRM runs a current through the OLTC to detect increased contact resistances
and interruptions in the current while operating the contacts. The difference from
static resistance measurements is that the tap changer is switched through all its tap
positions during the measurement. The current has no chance to fully stabilise due
to the inductance of the transformer windings. Dynamic resistance measurements
are therefore less accurate in resistance value than static measurements but provide
more information about the type and location of defects. This chapter will show
that DRM finds more defects on OLTCs than can be found using static resistance
measurements.
DRM was originally used for circuit breaker analysis, in which a high current,
typically 100 A or higher, is applied to the closed circuit breaker and the voltage
across the circuit breaker is measured. The circuit breaker is then opened. Valuable
information about the arcing contacts inside the circuit breaker can be extracted
from the contact resistance of the moving contacts and a high resolution can be
obtained by the high measurement current.
49
4.1 Measurement setup

When DRM is applied to a power transformer, valuable information about the
condition of the tap changer is obtained. Several options are possible:
Send a measurement current through the transformer and measure the

secondary voltage while the OLTC is in operation. The tap changer is
measured indirectly by measuring the voltage induced on the side of the
transformer without a tap changer (see Section 4.1.1).
Send a measurement current through the transformer and measure the total
current while the tap changer is in operation. The effect of irregularities in
OLTC resistance is directly measured (see Section 4.1.2).
Send a measurement current through the transformer and measure the
impedance [74] while the tap changer is in operation (with or without extra
resistance). The effect of irregularities in OLTC resistance is directly measured
(see Section 4.1.2).
Send a measurement current through the transformer and calculate the
transformer resistance while the tap changer is in operation (compensate for
the dynamic behaviour of the current). The effect of irregularities in OLTC
resistance is directly measured (see Section 4.1.2).
Use an oscilloscope to measure the measurement current or voltage, adjust the
trigger level to trigger on the arcing switch operation.
Use a dedicated recorder to measure both measurement voltage and current.
Use these parameters to calculate impedance, transition times and to show
close-up plots. A dedicated resistance recorder may be more user-friendly than
an oscilloscope.
The accuracy of these options differs in terms of time and amplitude. However,
the basic principle is the same: the measurement of electrical quantities that show
the effect of irregularities in resistance while the tap changer is in operation.
4.1.1 Indirect measurement of the OLTC
Most listed measurement possibilities concern measurements on the windings that
include the tap changer. Only the first option takes a measurement on the
windings without the tap changer. In this case, the voltage induced in the
secondary windings is measured while the tap changer is in operation on the
primary side. The dynamic resistance of the tap changer is indirectly measured.
Figure 4.1 shows an example of this secondary voltage during tap changer
operation. The plot shows a fast response to the transition resistor inside the
OLTC. It can be seen that the voltage decreases because DC does not pass
through the transformer, which is a drawback of the indirect measurement
method.
50

Information about the dynamic contact resistance can be obtained with better time
accuracy compared to direct resistance measurements (as will be discussed in
Section 4.1.2), especially when the effect of the transformer ratio is corrected. For
example, Figure 4.1 shows details of the resistance at about 0.1ms, which would
not be visible when the measurement is performed on the primary side, due to the
slow effect of the transformer leakage inductance.
Rapid voltage
changes of 0.1ms
are measured
Figure 4.1 DRM applied to a transformer with OLTC. The secondary voltage is recorded
while the OLTC switches.
Measurement of the secondary voltage is dealt with in Appendix B. The following

section will elaborate on the options for the direct measurement of the effects of
OLTC resistance. The effect of the transformer ratio is circumvented while the
decreased time accuracy is acceptable. In addition, a secondary short circuit
provides safety in the case of the primary measurement circuit being accidently
interrupted.
4.1.2 Direct measurement of the OLTC
Figure 4.2 shows a schematic representation of the setup for direct measurements
of the OLTC. A DC test voltage is applied to the transformer and the voltage and
current are measured while the transformer is switched through all of its tap
positions. As the basic function of an OLTC is to change from one winding tap to
another without interrupting the load current, DRM can therefore be considered a
direct method of checking the basic functionality.
The resistance and inductance of a power transformer with a tap changer is
significantly higher compared to a closed circuit breaker because of the presence of
transformer windings. For example, transformer windings can have a resistance of
500 m, while a closed circuit breaker may have a resistance of 100 . A
dynamic measurement of transformer resistance always measures some of the
transformer inductance because the current has no time to fully stabilise. The
resistance measurement is performed while the tap changer is moving, so the
inductance of the transformer still influences the impedance reading.
51
Figure 4.2 DRM applied to a transformer with OLTC. The voltage and current are measured
while the OLTC switches through all its tap positions.
Due to the inductance of the transformer (for example, a leakage inductance of 50

mH), the current through the transformer cannot change quickly. The time
constant of the changing current is given as:
L
R
4-1
When the DC voltage is connected to the transformer, two effects occur. First, the
current increases rapidly. The time constant can be calculated by entering the
circuit parameters in Equation 4-1, resulting into Equation 4-2, where Rcircuit is the
resistance of the measurement circuit, R1 the resistance of the primary transformer
windings, k the coupling factor between the windings, N the transformer ratio and
L1 the primary winding inductance. A current is induced in the short circuited
secondary windings.
2(1 k ) L1
Rcircuit R1 N 2 R2
[75]
4-2
As the current flows through the secondary short circuit the main inductance has
no effect and the current is able to change relatively quickly. Figure 4.3 shows the
voltage source connected to the transformer, represented as the T-equivalent
circuit, with a short circuit on the secondary side. The time constant is in the order
of tens of milliseconds. This rapid response is used during the measurement.
Figure 4.3 Schematic of a measurement voltage that is applied to a transformer, depicted as a

T-equivalent circuit, with a secondary short circuit.
52

The rapid response is followed by a slower effect, with a time constant which is in
the order of tens of seconds. The current now flows through the main inductance,
while the current through the secondary windings decays to zero. The input
impedance of the winding which is examined is then dominated by high
magnetisation-inductance (for example 200 H). The energy stored in the magnetic
field increases until the secondary current decays to zero, and then the
measurement current stabilises. Equation 4-3 is valid after the first effect has
disappeared (so the leakage inductance can be neglected). Figure 4.4 shows the
path of this slowly varying current. This effect can be largely eliminated from the
dynamic measurement by charging the main inductance before the measurement
starts. The energy stored in the inductance slowly increases during this charging
process, which uses a DC voltage source, and the dynamic measurement begins
when the current through the transformer reaches its equilibrium [75].
( Rcircuit R1 N 2 R2 ) L1
[75]
( Rcircuit R1 ) N 2 R2
Rcircuit
R1
4-3
N2R2
(1-k)L1
(1-k)N2L2=(1-k)L1
+
Utest
-
kL1
Figure 4.4 Schematic of a measurement voltage that is applied to a transformer, depicted as

a T-equivalent circuit, with a secondary short circuit.
The current response when a stepwise change in the measurement voltage or in the
circuit resistance is applied can therefore be written:
I C1 C2 e
C3e
[75]
4-4
The constants C1-C3 can be determined when the boundary conditions are known.
For example, when a step voltage is applied to the transformer, i(t=0) = 0 and i(t=)
= Utest/(Rcircuit + R1). Since 1<< 2, the current through the main inductance is still
zero after a few periods of 1, and therefore:
U test I ( Rcircuit R1 N 2 R2 ) [75]
4-5
Ref. [75] uses this effect to calculate the primary resistance and R2 without a
measurement on the secondary side of the transformer:
Rcircuit R1
U test
i(t n*1 )
N 2 R2 [75]
4-6
53
R2
U
1 U test
test ) [75]
(
2
N i(t n*1 ) i(t )
4-7
Ref. [75] also uses the difference between the two time constants to calculate the
value of an incremental resistance (dR) that exists in series with the primary
windings to explain the change in the current (di) due to the fast effect (1):
di ( Rcircuit R1 N 2 R2 )
[75]
dR
i( t 0) di
4-8
Conversely, when dR is known di can be calculated from the expression:
di
I (t 0) dR
4-9
Rcircuit R1 dR N 2 R2
Since 1<< 2, Equation 4-4 can be simplified to:
I C1 C2 e
4-10
for fast changes in the OLTC resistance, so that adding a series resistor at t=0
results in a current response of:
I I (t 0) di (1 e
4-11
In practice the measurement is performed on three-phase transformers. When the

transformer windings with the OLTC are star-connected, the three phases can be
measured simultaneously. Three currents are separately recorded and all three
phases are short-circuited on the secondary side. Figure 4.5 shows the setup,
consisting of three times the T-equivalent circuit of a transformer.
Rcircuit
R1
(1-k)L1
Utest
N2R2
(1-k)N2L2=(1-k)L1
N2R2
(1-k)N2L2=(1-k)L1
N2R2
(1-k)N2L2=(1-k)L1
kL1
Wye
measurement
R1
(1-k)L1
kL1
R1
(1-k)L1
kL1
Figure 4.5 Dynamic resistance measurement on a star-connected transformer winding.

Three phases can be measured simultaneously.
54

Transformers with the OLTC mounted in a
delta-connected winding must be measured by
means of three separate measurements. Only
the phase that is measured needs a secondary
short circuit (shown in Figure 4.7). This short
circuit allows for the rapid response of the
current. The other two phases lack this rapid
response, so irregularities in these phases do
not directly influence the current measured.
Rcircuit
R1
(1-k)L1
Figure 4.6 Only 2/3 of the current

flows through the tested windings.
N2R2
(1-k)N2L2=(1-k)L1
N2R2
(1-k)N2L2=(1-k)L1
N2R2
(1-k)N2L2=(1-k)L1
kL1
Utest
R1
(1-k)L1
kL1
Delta
measurement
R1
(1-k)L1
kL1
Figure 4.7 Dynamic resistance measurement on a delta-connected transformer winding.

Each phase has to be measured individually.
4.2 Measurement current

4.2.1 Winding configuration
This section discusses the influence of the winding configuration on the
measurement results. Chapter 4.1 already discussed the flow of the measurement
current of star and delta-connected transformer windings. It was shown that the
response of the current consists of a fast and a slow part in both cases. It can be
seen that the fast response of the current is comparable in both configurations.
However, there is a difference when the slow response is complete and the current
is totally stable. Figure 4.6 shows the division of the current when the slow
response takes over. The delta connection results in a division of the current. Only
2/3 of the total stabilised current flows through the OLTC phase that is diagnosed.
Because of the lower measurement current, contact resistances will be higher and
irregularities in this phase will be measured with a higher sensitivity.
The measurement current through the contacts should be equivalent when
comparing the results. Therefore, the measurement current should be 50% higher
55

at delta-connected windings. For example, when charging the measurement current
through a delta-connected phase to 1 A, the total current has to be stabilised to 1.5
A. In this case, the phase of the OLTC is diagnosed with the same current as a
star-connected transformer winding, namely 1 A.
4.2.2 Amplitude of the measurement current
4.2.2.1 Non-linear effect during static resistance measurements
The effect of the measurement current is best demonstrated using static resistance
measurements. Experiments show [11][76] that the resistance of degraded OLTC
contacts is highly dependent on the measurement current and voltage that is used
to determine the contact resistance. This section deals with the relationship
between this effect and DRM.
Figure 3.10 showed a typical example of this current dependency on serviceaged tap changer contacts: the resistance dropped from 237 m to 14 m when
the current was increased from 3.5 A to 100 A. Figure 4.8, which shows more
examples of the static resistance of a service-aged OLTC (the contacts are moved
before but not during the measurement) when the current is increased to 100 A.
Again, the contact resistance decreases when the measurement current is increased
and even at 100 A resistance did not reach a stable reading. Section 2.3.5 already
fitted a correction factor CDRM on results from static resistance measurements, in
order to correct measurement results for this effect. Ref. [77] comes with similar
plots for the dynamic resistance of circuit breaker arcing contacts measured up to
875 A and shows that the contact resistance at 2800 A still decreases. Appendix C
provides additional examples of static resistance measurements on service-aged
contacts using increasing measurement currents.
Another example is shown in Figure 4.9. This figure shows typical results of
static resistance measurements on clean and degraded OLTC contacts. Clean
contacts can have contact resistances in the range of hundreds of micro ohms. The
degraded contacts in this case had contact resistances in the range of hundreds of
milliohms (measured at 2.5 A). As can be seen in the plots, it is easier to
distinguish clean and degraded contacts from each other when the measurement is
performed at low currents. The difference in contact resistance becomes smaller
when the current is increased. However, contact degradation is measured with
higher resolution when the measurement current is low, which makes it possible to
distinguish contact resistance from winding resistance. Variations in the winding
resistance (due to temperature variations) can be higher than the contact resistance
at high currents, making it impossible to draw a conclusion about the contacts.
56
R esistan ce (m )

400
300
200
100
0
3,00
5,00
10,00
25,00
50,00 100,00
Current (A)
Figure 4.8 Static resistance measurements of service-aged OLTC contacts. The
measurement current was increased to 100 A. A strong dependency of the contact resistance
on the measurement current is revealed.
Static resistance of new and degraded contacts
Resistance (m)
100
10
Clean
Aged
1
0,1
0,00
20,00 40,00 60,00 80,00 100,00
Current (A)
Figure 4.9 The contact resistance of clean and degraded contacts when the current is
increased from 2.5 A to 100 A. Degraded contacts show high resistances at low currents,
while the difference between clean and degraded contacts becomes smaller at higher
currents.
The decreasing contact resistance at higher currents can be caused by several

effects [34]. Firstly, the higher source voltage tends to break the thin insulating film
on the contacts. Secondly, the high current can cause higher hotspot temperatures
in the contact interface, melting the contact material locally. This improves the
contact because the contact surface increases.
In general, it can be seen that the resistance measuring sensitivity to surface
film on the contacts increases when the measurement current is decreased. This
dependency can be a few decades in amplitude. A current that is too low (and
therefore a low measurement voltage) might allow the thick surface film to
interrupt the measurement current. The effect of the surface film decreases when
the current or voltage is increased.
57

4.2.2.2 Non-linear effect during dynamic resistance measurements
The effect described above should also be kept in mind when one performs
dynamic measurements: the magnitude of the measurement current influences the
outcome directly. This effect can easily be demonstrated by performing the DRM
on a service-aged OLTC using five different measurement current levels. Figure
4.10 shows the result. The current is plotted as a function of time at the moment
the OLTC operates. The contact resistance may be larger when the current is low,
in this example the current is almost interrupted at 200 mA, while the current
profile looks undisrupted at 20 A. Chapter 7.1.2 explores the effect of the surface
film in combination with low measurement currents in more depth using
laboratory experiments.
Current [A]
24
20
10
0
20 A
0
40
80
120
160
Time [s]
200
240
280
300
Current [A]
6
4
2
0
5A
0
40
80
120
160
200
Time [s]
240
280
320
360 370
Current [A]
Current [A]
1.2
0.8
0.4
0
1A
0
40
80
120
160
Time [s]
200
240
280
320
360
0.7
0.6
0.4
0.2
0
0.5 A
50
100
150
200
Time [s]
250
300
350
400
Current [A]
0.3
0.2
0.1
0
0.2 A
40
80
120
160
200
Time [s]
240
280
320
360
400
Figure 4.10 Dynamic resistance measurement on the same service-aged tap changer that
suffers from the long-term effect on the change-over selector. Five different measurement
current levels are used to demonstrate the influence of the amplitude.
Figure 4.11 shows an example of two dynamic resistance plots of the same serviceaged change-over selector that moves over a surface film. The movement of the
contact can be seen as spikes before the arcing switch operation. Two different
source voltages are used. The higher voltage of 1.8 V will break the surface film,
with only small irregularities being visible. The lower source voltage of 1 V is
58
Current [A]
Current [A]
obtained by removing the transformer windings from the measurement circuit.

This results in frequent interruptions of the measurement current. This effect is
also seen when the voltage is decreased by decreasing the measurement current.
This effect of the measurement current on degraded contacts is shown in
Figure 4.12. OLTCs in three different aging stages were examined using DRM with
five values for the measurement current and the deviation of the current from the
expected value in each tap position was calculated. The expected value is the
current that would be measured when there are no defects present in the tap
changer. Two tap changers were measured in each category; Figure 4.12 shows the
averaged maximum current deviation of those two OLTCs. A distinction between
good, aged and bad contacts can be made at low measurement currents (<5 A),
which is not possible at higher measurement current (e.g. 20 A). In addition, the
correction factor CDRM from Section 2.3.5 depends on the aging stage of the tap
changer.
Transformer ohmmeters that perform static resistance measurements [67-68]
also use low currents for resistance measurements. The ANSI/IEEE C57.12.901987 test code for power transformers advises that the current used should not
exceed 15% of the rated current of the winding whose resistance is to be
measured. Larger values may cause inaccuracy by heating the winding and thereby
changing its temperature and resistance [65]. The selection of a suitable
measurement current amplitude is complicated: on the one hand, sensitivity to
surface film on the contacts increases when the measurement current is decreased,
so a low measurement current, for instance below 10 A, may be desired if one
wants to assess the importance of the long-term effect. On the other hand,
interruptions of the measurement current due to the long-term effect may be
considered undesirable because the surface film itself does not directly endanger
proper OLTC operation and will not cause interruptions of the load current when
the transformer is in service. The interpretation of DRM results may be simpler if
only serious defects cause current interruptions. A current of at least a few
hundred milliamps can be selected to prevent interruptions of the measurement
current due to the thin film.
Within this margin, that is, a few hundred milliamps to 10 A, there will be large
resolution differences in which the degradation can be measured. At low current,
more false positive conclusions will be drawn from DRM. It will then indicate a
defective situation, although no serious defect is confirmed by visual inspection or
by additional measurements. Unnecessary time and money will then be spent on
Figure 4.11 DRM of a change-over selector moving over a surface film. Left: standard DRM
with a source voltage of 1.8 V. Right: The same movement when the OLTC is measured
separately. The source voltage is only 1 V because the transformer windings are not in series.
59

the OLTC. On the other hand, a high current will result in more false negative
conclusions, where a defective OLTC is not recognised by the DRM, which is a
very undesirable situation. These two situations should be balanced carefully by
choosing the optimal measurement current. However, the location where the
degradation appears in the plots and the shape of the resistance irregularities will
be the same. For that reason, this thesis discusses dynamic resistance
measurements using a fixed measurement voltage with an amplitude such that the
initial measurement current at the start of a measurement is 1 A. The current is not
easily interrupted by the surface film at this amplitude, but the measurement is very
sensitive to small defects and degradation. The chance of making false positive
conclusions is then higher than of making false negative conclusions, which is
desired because a transformer failure is more expensive than a diagnostic
measurement.
Figure 4.12 The deviation of the measurement current from the expected value decreases
when the measurement current is increased. This plot shows averaged DRM results from two
OLTCs with good contacts, two with aged contacts (a surface film) and two with bad
contacts (with contact coking).
Due to this non-linear relationship between the current and the resistance, the
resistance measured at low measurement currents does not resemble the contact
resistance at nominal currents. Contact resistances measured at different current
levels cannot be compared absolutely. When measuring with a low current, for
example 1 A, it is unjustified to extrapolate the resistances measured to the load
current level. For this reason dynamic resistance measurements at low current are
considered less reliable [78]. The current level should be the same when comparing
measurements (for example, from the same type of OLTC, between different
winding configurations or with earlier measurements). Dynamic resistance
measurements on high-voltage circuit breakers normally use currents of 100 A or
higher [77-78]. Circuit breakers are designed to have a very low resistance, so a
high measurement current is necessary to obtain an accurate voltage reading. A
power transformer has a much higher resistance and a high current would result in
a large power supply for the measurement device. The high inductance of the
60

transformer windings in combination with a high direct current would also make
the measurement more risky.
However, the results can be very useful when the importance of the long-term
effect on the change-over selector contacts is of interest. When taking the current
dependency into consideration, the long-term effect can be measured with a higher
sensitivity compared to measurements at currents of 100 A or higher. Again, the
measured ohmic values do not resemble the contact resistance at operating
currents.
4.2.3 Sensitivity
During a dynamic resistance measurement the contacts can move over degraded
contact areas and it is possible that the measured resistance only deviates from
normal values over a few milliseconds. The sensitivity of the dynamic resistance
measurement to these brief irregularities in contact resistance is also influenced by
the time that is needed for a defect to change the measurement current such that it
exceeds the noise level. Defects with a high resistance force the measurement
current to change rapidly. Therefore it is possible to detect defects that appear very
briefly if they have a high resistance. Defects with a low resistance need much
more time to change the measurement current more than the noise level.
Therefore, contact degradation with a low resistance that is only measured briefly
may remain below the detection limit of DRM. Figure 4.13 shows an example of
the detection limit. This limit is dependent on the circuit parameters and the time
constant of the transformer. Defects under the detection limit will not exceed the
noise level during the dynamic resistance measurement.
101
100
Resistance of the defect (Ohm)
10 - 1
10 - 2
10 - 3
10 - 4
0
0,01
0,02
0,03
0,04
Time of the defect s
0,05
Figure 4.13 OLTC degradation that causes increased contact resistance needs time to
influence the measurement current due to the circuit inductance. This graph gives an
example of the detection limit of defects that appear over a certain time (x-axis) and have a
specific amplitude (y-axis). Defects under this line lie below the noise level of this DRM
setup. Simulation parameters: 1 circuit resistance; 1 V test voltage; 20 mH leakage
inductance; 5 mA noise on current measurement.
61
4.3 Circuit resistance

The measurement circuit also influences the sensitivity of the measurement [79].
Differences in circuit resistance have an effect on:
1) The time constant of the measurement circuit. Equation 4-1 shows that a
higher circuit resistance allows for rapid changes in the current. This
increases the accuracy for short irregularities and the influence of the
inductance is less dominant in the plots (the current follows changes in the
resistance more rapidly).
2) The drop in current in terms of percentage. An unwanted resistance in the
circuit causes a smaller drop in current when the total resistance is larger.
3) The power supply that is needed.
4) The power losses in the circuit.
In relation to selecting measurement circuit parameters, the first effect is in conflict
with the other three. On the one hand, a high resistance allows briefly registered
defects to be detected; on the other hand, this results in a smaller signal that is
barely distinguishable from the noise. Power losses in an additional resistor should
also be considered.
A simulation has been carried out to obtain greater insight into these
contradictory effects. The circuits shown in Figure 4.5 and Figure 4.7 are used for
these simulations. The simulation parameters are based on a 50/10 kV 10 MVA
transformer. A resistance measurement on the primary and secondary windings
yields:
R1
= 0.45
R2
= 0.0575
N2R2 = 1.44
The inductances are based on the time constants of the measurement current
measured:
(1-k)L1 = 55 mH
The value of the main inductance could be calculated by the no load current and
losses of the transformer and yielded:
L1 = 197 H (rounded to 200 H for the calculations)
When the OLTC is in operation, the transition resistors are switched into the
circuit for a short time. The four effects of the circuit resistance that were
mentioned above are now explained by looking at the effect of the transition
resistors.
A series of tap change operations were simulated on a delta-connected OLTC
[80]. The transition resistor of 2 was in the circuit for 50 ms each time the
OLTC was operated. An extra resistor was then added to the circuit. The effect of
this extra resistor is listed in Table 4.1. It can be seen that the current drop due to
62

the transition resistor decreases from 25% to 2% when the resistance is increased
to 64 . A measurement voltage of about 100 V is needed in that case. The time
constant of the circuit also changes, as is depicted in Figure 4.14.
Table 4.1 The effect of an additional resistor on the current drop, response time and the
measurement voltage.
Timecontact(ms)
Currentdrop(%)
0
0.25
0.5
1
2
4
8
16
32
64
Current drop
-25%
-24%
-23%
-21%
-19%
-15%
-11%
-6%
-4%
-2%
(ms)
23.5
22.5
21
19.5
16.5
12.7
9.2
5.6
3.1
1.8
Utest
2.05
2.45
2.85
3.63
5.2
8.35
14.62
27.2
52.35
102.7
30
25
20
15
10
5
0
Timeconstant(ms)
Currentdrop(%)
20
40
60
Extracircuitresistance()
Figure 4.14 The effect of the transition resistors on the DRM current decreases when the
circuit resistance is increased, but the response also becomes faster (the time constant also
decreases).
A close-up view of the transition resistor operation is shown in Figure 4.15. A

complete flag-cycle is simulated. Normally, the measurement circuit has a
resistance of about 1 . When an additional resistor is added to increase this value
to 8 , it can be seen that the response becomes faster but has a smaller amplitude.
Figure 4.16 shows the effect of the transition resistors in the case of 64
resistance in the measurement circuit. The response is almost direct.
63
Figure 4.15 The response of the DRM current to the transition resistors. The amplitude of
the response is larger at a circuit resistance of 1 compared to 8 .
Figure 4.16 The response of the DRM current to the transition resistors at a circuit
resistance of 64 . The current follows the resistance almost directly, so the effect of the
inductance almost disappears.
When the circuit resistance is increased further, the current will directly react to
changes in the contact resistance. However, the low amplitude will cause problems.
Figure 4.17 shows a measurement with 75 added resistance. It can be seen that
the noise level is higher than the amplitude of the signal. It is impossible to detect
defects in the OLTC when the signal is so small. The method discussed in
Appendix B should be considered if a high level of time accuracy is desired.
The spikes in Figure 4.17 can be identified as noise because the current through
the transformer is not able to change quickly. Peaks in the dynamic resistance can
be identified as noise when they disappear much faster than the time constants of
the transformer. Single peaks in the dynamic resistance are therefore caused by
noise in the measurement circuit. The same effects can be measured or simulated
in the case of the OLTC being applied to a star-connected winding.
64

1,15
1,1
1,05
1
0,95
0,9
63,65
63,7
63,75
63,8
Time [s]
63,85
63,9
Current [A]
Figure 4.17 The response of the DRM current to the transition resistors at a circuit resistance
of 75 . Noise on the measured signal makes it almost impossible to detect defects in the
OLTC.
1.2
1
0.5
Current [A]
Current [A]
100
150
50
100
150
50
100
150
50
100
150
Time [s]
200
250
300
350
200
250
300
350
200
250
300
350
200
250
300
350
0.5
4
0
1.2
1
Time [s]
0.5
0
Current [A]
50
1.2
1
2
0
1.2
1
Time [s]
0.5
0
Current [A]
8
0
1
0
1.2
1
Time [s]
0.5
0
0.2
0
50
100
150
Time [s]
200
250
300
350
Figure 4.18 Dynamic resistance measurement on the same service-aged tap changer that
suffers from the long-term effect on the change-over selector. Five different series resistors
are used to demonstrate the influence of the circuit resistance.
Figure 4.18 shows an example of a DRM result of a service-aged OLTC with

contact degradation using five different values of circuit resistance. A series
resistance was added to the circuit for this purpose. It can be seen that the current
deviates more from the expected value when a low circuit resistance is used. Inset
A and B of Figure 4.18 show a close-up of the current when the tap changer
65
Deviation (%)
operates, from which it can be seen that the current reacts faster on changes in the
resistance when a high circuit resistance is used (B), which corresponds with the
simulations described in this section.
This effect can also be seen when the deviation of the current from the
expected value is calculated for each tap position. DRM results from OLTCs in
three different aging stages (good, aged and bad) were used, see Figure 4.19. Two
tap changers were measured in each category; Figure 4.19 shows the averaged
maximum current deviation of those two OLTCs. Five different series resistors
were added to the measurement circuit. This example shows that the difference
between the aging stages is higher at low series resistance (<1 ).
Another consequence of the circuit resistance dependency concerns the
comparison of data between different DRM sets. Different measuring instruments
can perform a dynamic resistance measurement [12][69][81-82], and may have
other circuit resistances. Thus, the difference in the circuit resistance should be
compensated for when comparing the resistances measured. An extra resistance
can be introduced into in the measurement circuit for this purpose.
Figure 4.19 The deviation of the measurement current from the expected value decreases
when the circuit resistance is increased. This plot shows averaged DRM results from two
OLTCs with good contacts, two with aged contacts (a surface film) and two with bad
contacts (with contact coking).
4.4 Secondary short circuit

Another issue concerns the effect of a secondary short circuit of the transformer
when carrying out a dynamic resistance measurement. This short circuit has two
functions. Firstly, safety is increased by the short circuit: the secondary short circuit
provides a path for the current in the case of the primary side being interrupted.
Without the short circuit, the current decreases and the steep dI/dt causes a high
voltage over the inductance, causing an arc to occur.
66

More important is the second function of the secondary short circuit. It allows the
measurement current to react quickly to changes in the supply voltage or in the
circuit resistance. Without this fast time constant, short irregularities in the contact
resistance cannot be measured. Section 4.1.2 elaborated on this time constant.
OLTCs in a delta-connected winding have to be measured three times, one
measurement for each phase. In each case, the secondary short circuit is only
applied to the phase(s) being measured (as depicted in Figure 4.20) and not to the
other phases. In other words, the two phases which are not measured lack the
secondary short circuit. In this way, the measurement current through the other
phases is not able to change during the measurement and only a defect in the
tested phase is measured and localisation possible. The secondary short circuit has
to be changed for each measurement, which is a disadvantage (for example, due to
safety).
Rmeasure
+
U
-
w
n
primary
HV
Transformer
secondary
LV
Short circuit
Figure 4.20 DRM setup for a transformer with the OLTC in a delta winding. The short
circuit is applied to the measured phase only.
Another way to perform DRM is to apply a secondary short circuit to three phases
at once. This situation is shown in Figure 4.21. However, the current through the
two non-tested phases is also able to change quickly. A defect in the tested phase
results in an increased current in the other two phases (due to current division on
parallel paths). In such case, the measurement current supplied by the
measurement system will not change as usual. Short irregularities in the contact
resistance of other phases are also measured and crosstalk between the phases
occurs.
Rmeasure
+
U
-
w
n
primary
HV
Transformer
secondary
LV
Short circuit
Figure 4.21 DRM setup for a transformer with the OLTC in a delta winding. The secondary
short circuit is applied to all phases.
67

A simulation was used to investigate the effect of the secondary short circuit. An
increased circuit resistance due to degradation causes an irregular current in the
dynamic resistance measurement. To simulate this degradation, extra resistors can
be introduced into the simulation. The irregularity can appear in one phase only or
simultaneously in three phases. Therefore, three different resistances which create
disturbance resistances are simulated:
An increased resistance in the measured phase

An increased resistance in one of the non-measured phases only
An increased resistance in all three phases
Again, a 2 resistor is switched in the circuit for 50 ms. The other circuit
parameters are as described in Section 4.3. By doing so, the simulation gives the
same time responses as a measurement on a real power transformer.
Table 4.2 gives an overview of the effect of these three irregularities on both
short circuit conditions. The decrease in the measurement current is determined in
terms of a percentage of the initial current. From the viewpoint of completeness,
the effect of the irregularity in a star-connected measurement is added.
Winding configuration
No of sec. short circuits
Configuration
Table 4.2 Overview of simulation results. The change in current is presented for three types
of irregularities and three measurement configurations. A resistance of 2 was in the circuit
for 50 ms.
Three phase
-24%
-24%
0%
-33%
-22%
-11%
-28%
-38%
14%
Location of defect (50 ms, 2 )

Measured phase
Non-measured phase
From these simulations it can be concluded that a dynamic resistance measurement

on a delta-connected transformer winding is best performed with one secondary
phase short-circuited. In that case, a three-phase defect and a one-phase defect
both give the same response (24% decline in the current), while a defect in a nonmeasured phase does not influence the measurement (0%). A three-phase short
circuit would give an inferior result.
The current through one phase can rise at a star-connected tap changer when
another phase has increased contact resistance (see Table 4.2). Figure 4.22 shows
the simulation result in this situation, where the current rises 14%.
68
4.5 Determination of the contact resistance
Current
DRM is a direct method for measuring defects on the OLTC contacts. The effect
of variations in the contact resistance is measured quickly (by switching the OLTC
through its cycle of operation). Corrections in the source current and voltage
measured may be needed before the resistance can be displayed.
Section 4.1 discussed the time constants of a transformer in cases where the
voltage and current measurements are performed on the same side of the
transformer. Essentially, the impedance is measured because DRM does not
provide long stabilisation times. Resistance
can only be read after a long period.
Corrections are needed in order to
determine the real contact resistance. The
same applies when the voltage response is
measured on one side of the transformer
and current is injected on the other side. A
correction or calculation is needed to obtain
the resistance values in both cases.
Firstly, the other circuit resistances need
to be corrected, for instance the resistance
of the measurement cables. A 4-point
measurement makes this correction Figure 4.22 Simulation of an OLTC
increased resistance in the
unnecessary. The resistance of the with
adjacent phases.
transformer windings should also be
subtracted from the value measured in order to obtain the contact resistances (note
that these values are temperature and frequency dependent).
Secondly, the time constant of the transformer windings to a DC voltage needs
to be corrected. The response of the measurement current to a resistor with a
known value is a way to avoid this: KEMA developed [68] and patented [83] this
method. The resistance is calculated after the effect of the fast time constant
disappears but before the slow response starts, as it is assumed that the current
through the windings has not yet changed. The ohmic value obtained by this
calculation is very close to the static resistance at that particular amplitude of the
current. This value should not be compared to values obtained at other levels of
the measurement current because the contact resistance can be strongly dependent
on the current.
Finally, compensating for differences in the measurement setup could make the
readings obtained less prone to variations in the setup, resulting in more readily
reproducible readings.
4.6 Conclusions
This chapter discussed aspects of dynamic resistance measurements, such as the
effect of the secondary short circuit, circuit resistance and measurement current
level. It was shown that a short circuit has to be applied to the secondary side of
69

the transformer (one short circuit at delta transformers). A four-wire resistance
measurement directly on the transformer bushings can be used to compensate for
the resistance of the test leads. We showed that a high circuit resistance results in
smaller changes in the measurement current, making the readings more prone to
noise, but the dynamic resistance that is measured shows fewer slow effects of the
winding inductance (because = L/R).
An important topic is the amplitude of the measurement current (and therefore
also the measurement voltage). We showed that when the measurement current is
decreased the sensitivity to surface film on the contact increases (see Figure 3.10).
This variation in the resistance can be a few decades in amplitude. A current that is
too low (and therefore a low measurement voltage) will allow the surface film to
interrupt the measurement current. Because the formation of the surface film on
the contacts is the first stage of the long-term aging effect that leads to coking, it is
of interest to determine if this degradation mechanism has started. The
dependency of the contact resistance and the amplitude of the measurement
current on each other can be used to detect the long-term aging effect at an early
stage. In addition, studying contact degradation by DRM at different current
amplitudes provides information about both the very early stage of contact
degradation (at low currents) and information about the behaviour of contact
degradation at a nominal current during OLTC service (at high measurement
current).
When doing so, a few issues have to be considered. Above all, it is impossible
to determine the real contact resistance of degraded contacts. Measurements at
both high current and low current do not indicate a contact resistance that is
independent of the measurement current. Results even fluctuate with the current
when the measurement is performed with 100 A or higher. Examples were given
in this chapter. This current dependency of the contact resistance during static
resistance measurements should also be kept in mind when performing dynamic
measurements as there was no stable reading of the resistance even at 100 A.
To conclude, performing a dynamic resistance measurement on an OLTC with
a low measurement current has important advantages: it is able to detect a large
variety of defects and degradation. The long-term aging effect can be discovered at
an early stage when a low current is used for the measurement. The method
provides more information about the tap changer condition than static resistance
measurements because resistance can be measured at more contact locations.
Using DRM with low measurement current also has disadvantages: it is difficult
to compare results with other measurement results (from other OLTCs or with
earlier measurements on the same unit). Results measured at different currents
should never be compared. One should be careful not to use a measurement
current that is too low: extreme resistances can be measured in such cases, even
causing total interruption of the current.
70
CHAPTER 5 POWER TRANSFORMER DIAGNOSTIC SYSTEM
5
Power transformer diagnostic
system
The diagnostic methods described in Chapter 3 do not cover all of the dominant
degradation mechanisms when applied separately. Thus, in order to measure all of
them, a combination of diagnostic measurements is recommended. This chapter
proposes a power transformer diagnostic system that combines a set of diagnostic
measurements which can diagnose the technical condition of the main power
transformer subcomponents, focusing on the dominant failure modes. The
diagnostic system discussed in this thesis focuses on condition assessment of
power transformers with a high-speed resistor-type OLTC. No distinction is made
between testing diverter and selector switch type OLTCs and between in-tank and
compartment type OLTCs.
The subcomponents that are not accessible to visual inspection during regular
maintenance, for example because they are enclosed in the transformer tank, are of
particular interest. The diagnostic measurements used are therefore selected on the
basis that they can reveal defects and degradation without disassembling the power
transformer as they measure quantities that are accessible externally. Examples of
such quantities include terminal voltage, excitation current, acoustic vibrations and
drive motor power.
Certain aspects of two important power transformer subcomponents primarily
determine the application of the transformer diagnostic system. First there is the
OLTC, which is normally found in power transformers installed in transmission
networks and in distribution networks, typically at voltages of 25 kV and above.
The off-line measurements using the proposed diagnostic system are
recommended on-site after regular tap changer maintenance or inspection. In this
way, safe operation of the power transformer is guaranteed when it is re-energised.
Contact degradation and defects that adversely affect the OLTC main function can
be checked using the diagnostic system described in this chapter, which offers a
direct way to detect whether the tap changer is able to select another winding tap
without interrupting the load current.
71

The second important transformer subcomponent is the winding. Due to the
diversity of transformer types, the winding configuration must be considered
during diagnostic measurements. Common winding configurations (Figure 5.1) are:
a.
b.
c.
d.
Delta-connected windings
Star (Wye)-connected windings
Autotransformers
Zigzag-connected windings
Although the measurement of winding resistance and transformer turn ratio on

these types of winding configurations are comparable, important differences exist:
1. The measurement voltage has to be applied on different terminals. Delta
windings are measured phase-to-phase while star-connected windings and
zigzag-connected windings with a neutral point can be tested phase-to-neutral
(Figure 5.1, see also Section 4.1.2).
2. Current and voltage signals used for the ratio or resistance calculation must be
measured on different terminals. The transformer vector group (e.g. Yd11)
describes which windings are magnetically coupled and which windings are
connected to which terminals. Therefore it determines which secondary
terminals should be used for the measurement (Figure 5.1).
3. Resistance of the three transformer phases can be measured simultaneously at
star and zigzag-connected windings with a neutral point (Figure 5.1). This not
only affects the measurement circuit, but also the way results can be displayed
(e.g. three phase results can be shown on the same timescale).
4. Section 4.4 described the secondary short circuit during a dynamic resistance
measurement (DRM), which is dependent on the winding configuration.
5. Delta-connected windings offer an alternative path in the case of a singlephase interruption of the measurement current. This should be kept in mind
when interpreting results.
A
1U
1U
1U
1U
2U
2V
1W
DC winding
resistance
measurement
1V
Three separate
measurements
Correction factor
needed
1W
1V
One measurement if
neutral point
accessible
1W
2W
1V
One measurement if
neutral point
accessible
1W
1V
One measurement if
neutral point
accessible
Dynamic
resistance
measurement
See above and

See above and
See above and
See above and
vector group
vector group
vector group
secondary short circuit
determines secondary determines secondary
determines secondary
on same phase
short circuit
short circuit
short circuit
Transformer
turn ratio
measurement
Vector group
Vector group
determines secondary determines secondary
voltage reading
voltage reading
Voltage reading on
same phase
Vector group
determines secondary
voltage reading
Figure 5.1 Common three-phase transformer winding configurations include a) delta, b)

star, c) auto and d) zig-zag.
72

Taking the above considerations into account, a transformer diagnostic system that
configures the measurement circuit according to the winding configuration can be
beneficial.
This chapter firstly discusses the international testing standards that apply to
power transformer testing (Section 5.1) and proposes a diagnostic system (Section
5.2). Secondly, the chapter elaborates on the aspects of this proposed diagnostic
system by describing: 1) the typical measurement procedure, including practical
aspects of using the diagnostic system (Section 5.3), 2) the hardware configuration
of the proposed diagnostic device and data processing and analysis (Appendix D).
The measurement data from this diagnostic system contains condition
information about the various transformer subsystems and requires interpretation
by a transformer expert. The interpretation of typical diagnostic data will be
discussed in Chapter 6.
5.1 International standards for power transformer testing

International standards are used to measurement different power transformer
types in a consistent way. They advise several checks and measurements be
performed on power transformers, but do not specify exactly how these checks
should be performed. Relevant international standards for power transformer
testing include:
1. ANSI/IEEE C57.12.90-1987: American national standard / IEEE Standard
Test Code for Liquid-Immersed Distribution, Power, and Regulating
Transformers.
2. IEC 60076-1: Power Transformers Part 1: General.
3. IEEE Std 4-1995, IEEE Standard Techniques for High-Voltage Testing.
4. IEEE Std 62-1995: IEEE Guide for Diagnostic Field Testing of Electric
Power Apparatus. Part 1: Oil Filled Power Transformers, Regulators, and
Reactors.
5. IEEE Std C57.140-2006: IEEE Guide for the Evaluation and Reconditioning
of Liquid-Immersed Power Transformers.
6. IEC 60214, Tap-changers Performance requirements, test methods and
application guide.
7. IEEE Std C57.131-1995: IEEE Standard Requirements for Load Tap
Changers.
These standards require that
mechanism, together with
maintenance, see Table 5.1.
film/carbon development is
during regular maintenance.
the condition of the OLTC contacts and the drive

contact timing, are determined during regular
In particular, checking the contact alignment and
advised. Turn ratio measurements are also advised
73

Table 5.1 Summary of general test requirements of international standards for power
transformer testing, describing which tests and checks should be performed during
maintenance or inspection.
As part of a transformer upgrade and service life extension,
the contact assemblies should be inspected for evidence of
poor performance, such as discoloration and/or material
flow (melting) of contacts and also film or carbon buildup
(coking) in conducting contact area.
IEEE Std C57.140-2006

IEEE Guide for the Evaluation and
Reconditioning of Liquid-Immersed Power
The internal tap assemblies and their operating
Transformers
shafts/gears should be inspected carefully for alignment
and physical condition.
Contact alignment should be checked and corrected as
needed.
IEEE Std 62-1995
IEEE Guide for Diagnostic Field Testing of
Electric Power Apparatus. Part 1: Oil Filled
Power Transformers, Regulators, and
Reactors.
The ratio in particular is checked during regular (e.g.

annual) inspections and it is recommended that it is also
checked before returning the transformer to service if the
unit has been subjected to a through-fault.
In addition to recommendations concerning what should be checked and

measured, the standards also set general requirements on how to perform the
measurements. The requirements for resistance measurements on power
transformers must be considered in dynamic resistance measurements on OLTCs,
which are summarized in Table 5.2. Transformer turn ratio measurement
requirements are summarized in Table 5.3. It can be seen that these international
standards do not specify precise measurement parameters for the advised
measurements; the most practical parameters can be chosen within the boundaries
described.
Table 5.2 Summary of test requirements of international standards for resistance
measurements on power transformers, describing in general how these resistance
measurements should be performed.
Measurement is made with direct current and simultaneous
readings of current and voltage are taken.
ANSI/IEEE C57.12.90-1987
The voltmeter leads shall be independent of the current
American national standard / IEEE
leads and shall be connected as closely as possible to the
Standard Test Code for Liquid-Immersed
terminals of the winding to be measured.
Distribution, Power, and Regulating
Transformers.
The current used shall not exceed 15% of the rated current
of the winding whose resistance is to be measured. Larger
values may cause inaccuracy by heating the winding and
thereby changing its temperature and resistance.
IEC 60076-1
Power Transformers - Part 1: General.
74
Tests shall be made at any ambient temperature between

10C and 40C.

Table 5.3 Summary of test requirements of international standards for transformer turn ratio
measurements, describing in general how these ratio measurements should be performed.
ANSI/IEEE C57.12.90-1987
American national standard / IEEE
Standard Test Code for LiquidImmersed Distribution, Power, and
Regulating Transformers.
The ratio test shall be made at rated or lower voltage and

rated or higher frequency.
Transformers that have Y diametric connections but do not
have the neutral of the Y brought out shall be tested for
ratio with three-phase power.
The test voltage shall be an alternating voltage having a
frequency in the range of 45-65 Hz.
IEEE Std 4-1995

IEEE Standard Techniques for High- The voltage wave shape should approximate a sinusoid
with both half cycles closely alike, and it should have a
Voltage Testing
ratio of peak-to-rms values equal to the square root of 2
within +5%.
If the transformer has load taps, the turn ratio should be

determined for all of these taps.
IEEE Std 62-1995
IEEE Guide for Diagnostic Field
Testing of Electric Power Apparatus.
Part 1: Oil Filled Power Transformers,
Regulators, and Reactors.
A meaningful ratio measurement may be made using only

a few volts of excitation. The transformer should be excited
from the highest voltage winding in order to avoid possibly
unsafe high voltage.
The voltmeters used should have accuracies
commensurate with the requirements of a 0.5% ratio
calculation.
5.2 Transformer diagnostic system

To obtain the desired information about the condition of the transformer
subcomponents as mentioned in the previous section, the following off-line
diagnostic methods are combined into a single diagnostic system:
-
Resistance measurements (dynamic resistance of the OLTC and DC

winding resistance)
Transformer turn ratio
Mechanical condition of the drive system
These measurements can be applied during regular tap changer maintenance and
diagnose the contacts and mechanical drive system, see Table 3.1. Concerning the
latter, to diagnose the mechanical condition and OLTC drive mechanism a
selection of measurements can be made based on the considerations of Section 3.2.
For example, information about the mechanical condition can be gained by
measuring:
a. The active drive motor power
b. The remote end drive shaft rotation for rotor-type OLTCs with an accessible
drive shaft (e.g. selector switch type tap changers)
c. The timing of protective contactors, for example an excessive switch-time
protective contact
75

Table 3.1 shows that these methods, in combination with a visual inspection,
diagnose defects in all the OLTC subsystems from Table 3.1. Figure 5.2 illustrates
the diagnostic methods selected for the proposed transformer diagnostic system.
The condition of the power transformer subsystems, depicted left in Figure 5.2, is
assessed using the four basic diagnostic measurements used in the diagnostic
system (see also Chapter 3). In addition to the measurements included in the
transformer diagnostic system, gas analysis of the transformer insulation oil, visual
inspection of the transformer tank and accessories are also advised in order to
determine the optimal maintenance strategy.
Figure 5.2 The transformer subsystems (left) are diagnosed by the transformer diagnostic
system (centre). Additionally, gas analysis and inspection are advised. As a result, the
optimal maintenance strategy can be determined.
The goal of applying this combination of diagnostic measurements is threefold:

1) To diagnose the power transformer windings and tap changer in cases where
protective devices have switched off the transformer.
2) To assess the contact condition and contact timing after regular tap changer
maintenance or repair, in particular for pre-failure detection of maintenance
errors and to find defects and
degradation that remained
undiscovered
during
inspection.
3) To detect contact degradation
at an early stage and thereby
determine
the
optimal
maintenance strategy.
In this thesis, the suitability of
DRM on OLTCs is studied using
the
Transformer
Diagnostic
System described in [12][84] and as
shown in Figure 5.3.
76
Figure 5.3 Transformer diagnostic system

described by [12] during DRM field testing.
The system is remotely controlled by a laptop.
5.3 Measurement procedure

Figure 5.4 shows the measurement procedure of the diagnostic system. Prior to the
off-line measurements, the power transformer must first be prepared for the
measurement. The transformer is de-energised, a safety ground is applied, the
high-voltage
connections
are
uncoupled
(Figure
5.4.1) and the clamps
and
sensors
are
connected
(Figure
5.4.2). Figure 5.4.3a-e
lists
the
general
measurement
procedure
of
the
proposed
diagnostic
system, which applies
for
all
diagnostic
methods
in
the
proposed
diagnostic
system. As a first step
in the measurement,
Figure 5.4 Measurement procedure for measuring power
transformers with the proposed transformer diagnostic
transformer parameters
system.
are entered (Figure
5.4.3a) in order to:
III.
IV.
Store measurement data in a consistent way

Define the transformer terminals that are used during the measurement
and configure the relays in the output stage accordingly
Make correct secondary short circuits during DRM
Correct the measured turn ratio
Subsequently, the desired measurement

methods are selected (Figure 5.4.3b). The
exact procedure involved in the measurement
itself (Figure 5.4.3d) depends on the
diagnostic method selected.
5.3.1 Dynamic resistance of the OLTC
Performing DRM with the proposed
diagnostic system involves the setup of the
measurement circuit, followed by charging,
recording and discharging in succession, see
Figure 5.5.
I
Configure setup by relays
II
Apply stable test voltage
III
Perform DRM (switch OLTC)
IV
Discharge
Repeat 3x for delta connected windings
I.
II.
Figure 5.5 General measurement

procedure for DRM.
77

Setup configuration
Configuration of the measurement circuit (Figure 5.5I) is done by connecting the

DC power source to the transformer windings. Which transformer terminals are
connected is determined by the winding configuration. The secondary short circuit
is applied according to Sections 4.1.2 and 4.4 and also depends on the winding
configuration. Figure 5.6 and Figure 5.7 give examples of a measurement circuit
configuration for an YNd-connected winding. The proposed transformer
diagnostic system has direct connections to all transformer terminals and
configures the setup with relays.
Rmeasure
Laptop
Embedded
control and
data
acquisition
Short circuit
U
-
Dynamic Resistance Instrument
Primary
HV
Transformer
Secondary
LV
Figure 5.6 Dynamic resistance measurement applied to an YNd-connected power

transformer. The diagnostic system is connected to the windings with the OLTC while the
secondary side is short-circuited.
Transformer connections
Diagnostic system
Figure 5.7 The transformer diagnostic system connected to the primary and secondary
transformer terminals.
Charging
Firstly, during the charging process (Figure 5.5II) a DC voltage must be applied
using a controllable DC voltage source. This voltage can be increased quickly (for
example within 1 s) until the desired current is reached.
78

Secondly, the voltage needs to be regulated to a lower value to keep the
measurement current at the desired level, due to the saturation of the transformer
core and the effects discussed in Section 4.1.2. An example of a charging process is
shown in Figure 5.8. During field measurements it was seen that, in general, a
stable measurement current can be assumed when the voltage applied remains
within 10% for 90 seconds. During this charging process, attention should be paid
to the following:
1. Do not charge windings without a correct secondary short circuit, which can
be recognized by a slowly rising current during the charging process.
2. Beware of an incorrect transformer connection, which can be recognized as a
negative measurement current or zero current in some phases.
3. Beware of corrosion of the transformer terminals, which endangers a good
electrical contact with the measurement clamps. This can be recognized by
incorrect voltage readings.
4. Beware of saturation effects in the transformer core caused by the DC
measurement current. If not eliminated during the charging process prior to
the measurement, saturation might lead to overcurrent conditions during the
measurement.
It should be kept in mind that transformer windings have a very high inductance
(in the range of hundreds of Henry) and can therefore store a considerable amount
Test voltage
3x test current
Figure 5.8 Charging the transformer windings with a DC current. Initially a higher source
voltage is needed.
79

of magnetic energy. The diagnostic system charges and discharges the transformer
windings safely and determines when the charging effect has disappeared and a
stable reading is obtained. Attention should be paid to the current amplitude when
testing delta-connected transformers, see Section 4.2.2.
Measurement
During the resistance measurement itself (Figure 5.5III) the source voltage is kept
constant and the tap changer is operated. The current of the tested phases is stored
together with the voltage difference of the corresponding transformer bushings.
Again, attention should be paid to not overload the power supply, because the
measurement current can increase due to changes in the winding resistance and
contact resistance. During the recording phase, OLTC control can be applied by
the transformer diagnostic system. Therefore the diagnostic system can be
equipped with OLTC control contactors which are designed to control the drive
mechanism. These contacts bridge the control buttons in the control box of the
drive and this way the tap changer can be automatically switched through all its tap
positions.
Discharge
During the discharge process (Figure 5.5IV) the DC power supply is switched off
and a series resistor added to the circuit in order to dissipate the energy that is
stored in the windings. Interrupting a small current through the transformer
windings can still damage sensitive electronics inside the measurement system.
Therefore, when the current reaches safe levels, a short circuit should be applied
by the diagnostic system for safety reasons.
View results
When all phases are measured, results should be displayed in such a way that a
decision about re-energising the transformer, without risk of failure, can be made
on-site (Figure 5.4.3e). For that purpose, critical parameters from the DRM results
should be clearly visible. The following results should be shown:
1. An overview of the measurement current or impedance during the complete
cycle of operation. Its presentation depends on the winding configuration.
Figure 5.9 provides an example of a DRM current overview graph for an
OLTC in a Y-connected winding (the three phases are measured
simultaneously) in good condition. Appendix D shows an example of an
overview graph of a delta-connected OLTC with contact degradation (three
phases are measured on a different timescale) and an example of the
impedance overview graph.
2. Close-up view of the measurement current or impedance for each tap changer
operation (see Appendix D).
3. Transition times of each operation (for example in a table or a graph, see the
right side of Figure 5.9).
4. The voltage and current that were used for the measurement: to allow
comparison of measurement results, which can be dependent on these
measurement parameters (indicated in the right bottom side of Figure 5.9). It
80

can be seen that a decreasing current shows as an increasing impedance
because the measurement voltage is fixed.
Figure 5.9 DRM results for OLTCs in star-connected windings can be displayed on the
same timescale (A), results can be displayed as current.
5.4 OLTC drive system condition

Signals recorded for the mechanical diagnosis of the OLTC by the proposed
diagnostic system can be displayed as in Figure 5.10. By default (see Section 5.2)
these signals consists of drive motor voltage, drive motor current, drive shaft
angular position and contact voltage of the excessive switch-time protective
contact. Figure 5.10 shows the motor power results, consisting of an overview plot
and a close-up plot. Graphic presentation of motor operation time and average
motor power can give a clear indication of irregular taps. The close-up of the plots
for each tap change operation reveal defects more readily.
In order to create the active motor power plots, the voltage and current
recordings need to be processed. Instantaneous power p3 ( t ) of the three-phase
drive motor is defined by:
p3 (t ) v A(t )iA( t ) vB (t )iB (t ) vC ( t )iC (t )
5-1
81
Figure 5.10 Example of active motor power results on an OLTC drive mechanism.
where v A ( t ) , vB ( t ) and vC ( t ) are the line-to-neutral voltages of the three-phase

drive motor and iA( t ) , iB ( t ) and iC ( t ) its three currents. Double net frequency
components should be filtered out before results are analyzed. If RMS values are
used, the power factor needs to be considered:
P3 VA I A cos A VB I B cos B VC I C cos C
5-2
To reduce the number of measured signals, balanced operation of the drive motor
can be assumed:
VA VB VC VPhase
I A I B I C I Phase
5-3
cos A cos B cos C cos

The active power can then be calculated based on RMS values of the phase voltage
and the measured single-phase current:
P3 3VPhase I Phase cos

82
5-4

Figure 5.11 shows the OLTC drive shaft position results, where derivatives of the
shaft rotation resulted in graphs of velocity and acceleration. Appropriate digital
filters (e.g. low pass filters) are applied by the proposed diagnostic system to
reduce noise on these signals. Irregularities in the shaft rotation, velocity and
acceleration can be more readily detected when a graph is plotted with
characteristic values for each tap changer operation. Therefore, the mechanical
switching time, overshoot and maximum values of speed and acceleration are
shown by the proposed diagnostic system. The close-up plots for each tap changer
operation are useful to study faulty operations in more detail.
The timing of an excessive switch-time protective contact can be checked by
plotting changes in the recorded contact voltage (AC or DC) in the dynamic
resistance plots, see Appendix D. Incomplete or defective protection settings can
then be recognized.
Figure 5.11 Example of drive shaft rotation results on an OLTC drive mechanism.
5.5 Hardware and data processing

Design considerations for such a diagnostic system are discussed in Appendix D.
The diagnostic system is subdivided there into four basic units. The hardware
demands are determined for each unit, on the assumption that the diagnostic
system is designed to perform automated measurements. Appendix D also
83

discusses the data processing and analysis required as part of the embedded control
of the diagnostic system. Examples of how the measurement results can be
displayed are also given. Special emphasis is laid on the need to present DRM
results clearly, for example by means of:
a dynamic resistance overview plot covering all tap positions

a dynamic resistance close-up plot for each operation
details of the transition times for each operation
values of the voltage and current used for the measurements
This should provide an adequate basis for the decision as to whether the power
transformer can be re-energised without the risk of failure. Chapter 6 deals with
the interpretation of the DRM results.
5.6 Conclusions
Firstly, this chapter explains the effect of the transformer winding configuration on
different diagnostic measurements.
Secondly, it was shown in this chapter that international standards on power
transformer testing do not specify precise measurement parameters for the advised
measurements; the most practical parameters can be chosen.
Thirdly, this chapter discussed a combination of diagnostic measurements that
is sensitive to assess the dominant OLTC degradation mechanisms and defective
power transformer windings:
Resistance measurements (dynamic and static)

Turn ratio measurements
Mechanical condition of the OLTC and drive
The condition of different transformer subsystems is evaluated using these

proposed diagnostic measurements, after regular maintenance or after protective
relays removed the transformer from service. In particular, the focus is on
transformer subcomponents that are not accessible for visual inspection during
regular maintenance because they are enclosed in the transformer tank. Practical
aspects of using the proposed diagnostic system were discussed.
In the following chapters, the proposed diagnostic system will be used to show
the suitability of DRM to detect the dominant OLTC degradation mechanisms. In
particular, the following chapter will be concerned with the structured evaluation
of the diagnostic data obtained by DRM on OLTCs.
84
CHAPTER 6 - EVALUATION OF TYPICAL DRM DIAGNOSTIC DATA
6
Evaluation of typical DRM
diagnostic data
Dynamic resistance plots obtained by measuring tap changers that are in good
condition look similar. Many different kinds of irregularities in the results are
expected when DRM is applied to defective or degraded OLTCs. This chapter
presents an overview of possible measurement results and relates them to the
physical condition of the tap changer. The variety of defects and degradation
mechanisms that can be detected before failure occurs [85], shows that DRM is a
valuable tool for the assessment of OLTC condition.
Four basic types of measurement results can be distinguished:
a.
b.
c.
d.
DRM on OLTCs in good condition

DRM on OLTCs with increased contact resistance
DRM on OLTCs with open contacts (infinite resistance)
DRM on OLTCs with contact timing problems (phase synchronism, contact
wear and irregular transition times)
DRM results obtained in this project could all be assigned to one or more of these
categories, which are discussed in turn below. This subdivision allows the different
types of defects to be analyzed separately, as illustrated in Figure 6.1. The location
where the defect shows in the current curve and the shape of this irregular current
are used in this thesis to find the cause and importance of the defect. DRM plots
can be compared with a reference graph, which may be a plot obtained from an
OLTC without any defects or signs of contact degradation. Finally, a condition
index may be determined as described in Chapter 8.
Most dynamic resistance plots presented in this chapter were recorded with the
Transformer Diagnostic System described in the previous chapter and were
obtained by measuring service-aged OLTCs in the Dutch utility grid. However
some of the dynamic resistance plots were determined during experiments in a
high-voltage laboratory.
85
Figure 6.1 Condition assessment of an OLTC with the aid of DRM. Firstly, the results can
be subdivided into:
no signs of defects or degradation
increased contact resistance
open contacts (infinite resistance)
contact timing problems
Secondly, the location and cause of the defect is determined, resulting in assessment of the
importance of the defect (which can be summarized in terms of a condition index).
6.1 Regular DRM plots

If the DRM curve shows no signs of defects or degradation, only two sources of
resistance variations are visible in the recorded measurement current:
Transition resistors
Resistance of the tapped windings
Figure 6.2 shows a typical example of the measurement current recorded on a

healthy tap changer. The tap changer was switched through its cycle of operation
(from the lowest to the highest tap position and back again) to obtain this
recording. The effects of the two above-mentioned resistances may be clearly seen.
The winding resistance changes when tapped windings are selected or deselected
by the tap changer, resulting in a slope in the recorded current.
The close-up plot of Figure 6.2 reflects the operation of the transition resistors,
while Figure 6.3 shows the relationship between the rotation of a selector switch
(see Appendix A for a more extensive description) and the measurement current.
The transition resistors cause the measurement current to decrease. This fast
changing current is possible due to the short circuit on the secondary side of the
transformer.
86

Winding resistance
Current [A]
1.4
1
tap 11
tap 1
tap 1
0.5
tap n
tap n+1
0
0
100
200
Time [s]
300
350
Figure 6.2 Typical DRM current in a healthy OLTC.
First, only one transition resistor is in the circuit and the current given by Equation
4-11. Only the fast response of the transformer, discussed in Section 4.1, is taken
into account here.
Half-way through the OLTC operation, both the transition resistors are in
parallel in the measurement circuit as shown in Figure 6.3. The resistance of the
windings and the transition resistance may then be written:
Rw
Rt ( Rt Rtw )
2 Rt Rtw
6-1
where Rw is the winding resistance, Rt the transition resistance and Rtw the tapped
winding resistance. This situation generally causes a less steep decrease in current
(the current may also increase, depending on the contact timing and circuit
Figure 6.3 The relationship between the transition resistors and the DRM measurement
current. The transformer windings are in series with the OLTC.
87

resistance). Finally, the transition resistor is removed and the measurement current
stabilizes again. The current during OLTC operation can then be written as
follows, using Equation 4-11 and the values of the different resistances in the
circuit (see also Figure 6.4):
I0 for t<0
I0 Rt
I0
t ( Rcircuit R1 Rt N 2 R2 )
2(1 k ) L1
(1 e
Rcircuit R1 Rt N 2 R2
R (R R )
I1 t t tw
2 Rt Rtw
I1
(1
R (R R )
Rcircuit R1 t t tw N 2 R2
2 Rt Rtw
It
I2
R (R R )
( t t1 )( Rcircuit R1 t t tw N 2 R2 )
2 Rt Rtw
2(1 k ) L1
I 2 ( Rtw Rt )
Rcircuit R1 Rtw Rt N 2 R2
I 4 ( I 4 I3 ) e
) for 0t<t1
6-2
) for t1t<t2
(1
( t t2 )( Rcircuit R1 Rtw Rt N 2 R2 )
2(1 k ) L1
( t t3 )( Rcircuit R1 Rtw N 2 R2 )
2(1 k ) L1
) for t2t<t3
for tt3
where:
I0
Utest
Rcircuit R1
I0 Rt
I1
(1 e
Rcircuit R1 Rt N 2 R2
t1 ( Rcircuit R1 Rt N 2 R2 )
2(1 k ) L1
R (R R )
R (R R )
( t2 t1 )( Rcircuit R1 t t tw N 2 R2 )
I1 t t tw
2 Rt Rtw
2 Rt Rtw
2(1 k ) L1
(1 e
I 2 I1
)
R (R R )
Rcircuit R1 t t tw N 2 R2
2 Rt Rtw
I 2 ( Rtw Rt )
I3 I 2
(1 e
Rcircuit R1 Rtw Rt N 2 R2
I4
88
Utest
Rcircuit R1 Rtw
( t3 t2 )( Rcircuit R1 Rtw Rt N 2 R2 )
2(1 k ) L1
Figure 6.4 Definition of the times and currents used in Equation 6-2.
Diverter switch type OLTCs show DRM plots that are similar to those for selector
switch type OLTCs; the plot of Figure 6.3 is typical for flag-cycle type OLTCs.
The timing of the flag cycle (as shown appendix A) can be read off from the plot.
The effect of the transition resistors is made clearly visible when the tap
changer is measured separately, without the effect of the transformer windings
(Figure 6.5). The time constant of the current is no longer influenced by the
leakage inductance of the transformer windings under these circumstances, and the
current can change rapidly with circuit resistance changes.
Combining the results of Figure 6.3 and Figure 6.5 shows the effect of the
leakage inductance clearly, see Figure 6.6. The current slowly changes towards the
new equilibrium, but the transition resistors are operated again before the new
equilibrium has been reached.
0,4
0,35
0,3
0,25
0,2
0,15
15,34
15,36
15,38
15,4
Time [s]
15,42
15,45
Figure 6.5 The relationship between the transition resistors and the DRM measurement
current. The transformer windings are removed from the measurement circuit.
89

Various ways of performing DRM
on a tap changer were explained in
Section 4.1, where it was pointed
out that defects appear in the
measurement results in comparable
ways.
The
present chapter
discusses the recorded current
flowing through the OLTC during
the measurement in detail.
An impedance plot like that of
Figure 6.7 is another way of
displaying the results of a DRM.
The voltage across the transformer
bushings can be used to calculate
the impedance. Impedance plots
are mirror images of the current
recordings, since the voltage is kept
constant during the measurement.
Figure 6.6 The DRM results of Figure 6.3 and

Figure 6.5 are combined in this graph. The
measurement current directly changes when the
OLTC is measured separately, but slowly
changes towards the equilibrium when the
leakage inductance slows the current down.
2.5
2
1.5
1
0
100
200
Time [s]
300
350
Figure 6.7 Impedance plot of a healthy OLTC. This plot corresponds to the current plot in
Figure 6.2.
Another effect that can be found in normal dynamic resistance plots is related to
the design of the change-over selector. Some OLTCs are designed in such a way
that the arcing switch operates two times extra when the change-over selector is
operated [86]. When operating the change-over selector, the arcing switch is
operated twice to change one voltage step; the effect of the transition resistors can
be seen in the plots at the moment when the change-over selector is operated (as
indicated in Figure 6.8).
Figure 6.9 shows another effect that occurs at OLTC designs with a reversing
switch: the tapped windings are reversed, allowing the tap changer to subtract the
voltage of the tapped windings from that of the main windings. The polarity of the
tapped windings is not measured with DRM, since the reversal does not affect the
DC resistance. This gives the plot of Figure 6.9, where it looks as if the OLTC is
switched through its cycle of operation twice but this is the normal way this type
of OLTC works.
90

The tapped winding that is reversed by the reversing change-over selector was
charged in the opposite direction prior to the measurement. Reversal can result in
a situation where the effect of the inductance is still measured. The central portion
of the dynamic resistance plot may therefore look different from that shown in
Figure 6.9. Figure 6.10 gives an example of a plot measured on a delta-connected
OLTC with a reversing change-over selector showing this effect: the central
portion of the plot is flat here, which is not caused by defects or contact
degradation.
Measurement errors can also change the measurement plots, for example due
to irregular switching, changing the measurement circuit or measurement voltage
during the measurement, poor connections to the transformer terminals and
starting the measurement before the current is stable. Figure 6.11 shows an
example of a DRM result with a measurement error: the measurement was started
before the charging phase had finished, resulting in an unstable current during the
measurement. Such measurement errors typically lead to a higher current at the
end of the measurement and a curved instead of a linear current pattern.
Irregularities in the plots due to measurement errors are not due to defects and
degradation and the measurement can be redone to get better results.
1,1
1
0,9
0,8
0,7
0,6
0,5
0,4
0
100
200
300
Time [s]
400
500
630
1,1
1
0,9
0,8
0,7
0,6
0,5
0,4
425
515
Time [s]
Figure 6.8 DRM results from an OLTC with three centre positions: two extra arcing switch
operations are visible when the change-over selector operates (the arcing switch operates
twice in succession to perform one voltage step).
91

1
0,9
0,8
0,7
0,6
0
50
100
150
200
Time [s]
250
300
360
Figure 6.9 DRM plot of the cycle of operation of an OLTC with reversing switch.
1,3
1
0,75
0,5
0,25
0
0
20
40
60
80
100 120 140 160 180

Time [s]
200 220
Figure 6.10 DRM plot of the cycle of operation of a delta-connected OLTC with reversing
switch.
1,8
0
0
100
200
300
400
500
Time [s]
Figure 6.11 DRM plot of an OLTC affected by a measurement error: the measurement was
started before the charging phase was completed. This caused the current to stabilize during
the measurement instead of during the charging phase.
92
6.2 Irregular DRM plots

It is useful to interpret every deviation from the normal DRM plot (as discussed in
the previous section) directly after the measurement, before taking the transformer
back in operation again [87]. Plots with signs of defects or degradation can be
resolved into different components [88], as illustrated in Figure 6.13:
Figure 6.13-A:
Figure 6.13-B:
Figure 6.13-C:
Figure 6.13-D:
Figure 6.13-E:
Measured plot, equal to the sum of components B+C+D+E

Effect of winding resistance and measurement errors (this can
be used as the expected value, see 6.1)
Effect of the transition resistors (see 6.4.6)
Effect of contact resistance
Effect of open contacts (current interruptions)
The first source of irregularities in dynamic resistance results which is related to

degradation or defects (Figure 6.13D) is caused by increased contact resistance;
this is not always a serious defect. In general, these irregularities can be caused by:
1
2
the change-over selector (discussed in Section 6.3), for example changes in

contact resistance due to different stages of long-term aging
the arcing switch (discussed in Section 6.4), for example due to contact
resistance or contact timing problems
During a DRM, the change-over selector is always measured in series with the
arcing switch. Defects in both parts of the OLTC will show up in the results.
These will be discussed in detail in the next two sections. First, Section 6.3 will give
an overview of irregularities in the change-over selector. Section 6.4 will then
review the effect of degradation and defects in the arcing switch.
Complete interruption of the measurement current due to open contacts is in
general a more serious problem.
It is important to know the location of the defect when interpreting the
measurement results. All the irregularities occur at specific locations on the current
profile. Figure 6.12 gives a schematic representation of a typical measurement
current recorded on a healthy tap changer as shown in Figure 6.2. This will be used
as a basis for indicating where the defect is expected in DRM plots of the OLTCs
cycle of operation.
Figure 6.12 Schematic representation of a typical DRM current recorded on a healthy

OLTC.
93
B:
Effect of winding resistance and
measurement errors.
This value can be used as the expected
value, which is measured when the
OLTC is in good condition. An
unstable test current also belongs in this
category.
OLTC defects and degradation are not
seen here.
C:
Effect of the transition resistors.
Irregular transition times,
synchronisation problems between
phases and contact timing problems can
be seen here.
D:
Effect of contact resistance.
Contact resistance problems such as
long-term aging can be revealed in this
part of the DRM plot.
E:
Effect of open contacts. The current
drops to zero due to an infinite
resistance.
Contact timing problems can be seen
here.
Figure 6.13 Example of a DRM plot of an OLTC (A) which can be separated into the effect
of winding resistance (B), the effect of the transition resistors (C), the effect of contact
resistance (D) and the effect of open contacts (E).
94
6.3 Irregularities in the change-over selector

More than 600 dynamic resistance measurements have been performed on OLTCs
within the framework of this thesis project. Almost all change-over selector
irregularities observed in the Dutch utility grid were due to the long-term aging
effect discussed in Section 2.3. The present section deals with DRM where longterm aging is observed:
1) when the change-over selector selects another tap in combination with a
selector switch (6.3.1),
a) at an early stage (6.3.1.2)
b) at an advanced stage (6.3.1.3)
2) in the parking position and when the tap changer is at the end of its regulating
range (6.3.1.4)
3) in the change-over selector at the moment of operation of the arcing switch
(6.3.1.5)
4) in the tap selector when used in combination with a diverter switch (6.3.1.6)
Irregularities in the measurement plots caused by the change-over selector appear,
change or disappear while it is operated. A typical irregularity here is the surface
film, which can be recognized when the change-over selector moves. An example
is given in Figure 6.14, which shows a close-up view of the operation of an arcing
switch preceded by spikes in the contact resistance due to contact movement in
the change-over selector. To study this effect, it is necessary to determine when the
change-over selector is moved by the drive mechanism.
Figure 6.14 Moving change-over selector contacts cause spikes in the dynamic resistance
just before the arcing switch is operated.
6.3.1 Long-term aging in selector switch type OLTCs

6.3.1.1 Movement of the coarse change-over selector
Depending on the number of fine taps and the desired number of tap positions,
one or two coarse taps are used for expanding the regulating range. The coarse
change-over selector only selects another coarse tap a few times during each cycle
95

of operation. This selection is much slower
than the operation of the arcing switch.
Figure 6.15 shows the corresponding
locations in the recorded current for an
OLTC with one (Figure 6.15A) or two
(Figure 6.15B) coarse taps. The coarse
change-over selection can cover two tap
change operations of the selector switch.
The contacts of the change-over selector
typically move a few seconds before the
6.15
Positions where the
operation of the selector switch. Valuable Figure
coarse-tap selector moves its contacts
information about long-term aging can be to a next tap in an OLTC with one (A)
extracted from the DRM results when the or two (B) coarse taps.
change-over selector contacts are carrying
current when they move. This section takes
this type of tap changer as an example to explain the movement of the changeover selector.
The change-over selector contacts can move beyond these two tap positions
due to the design of the drive mechanism. In the case that the drive is designed
that way, the drive mechanism also causes the change-over selector to move a
slightly outside the two positions of the coarse tap selection, to prepare for the
coming coarse tap change. These small movements can occur two positions before
and one position after the mid-position of the change-over selector, depending on
the drive mechanism used. It follows that the total movement of the coarse
change-over selector covers four tap change
operations as indicated in Figure 6.16.
During the coarse tap change operations,
the coarse change-over selector makes
contact with both coarse taps, as already
shown in Figure A.5C. This position will be Figure 6.16 Positions where the
called the mid-position of the coarse-tap coarse-tap selector moves its contacts
selector. It does not always coincide with the to a next tap.
parking position of the selector switch (the
k contact).
For example, Figure 6.17 shows the DRM plot of a service-aged coarse changeover selector. The contact resistance increases just before the operation of the
arcing switch and lasts several seconds until the operation of the arcing switch. The
Geneva wheel driver can cause a step in the measurement current, while the
movement of the change-over selector to and from the mid-position gives rise to a
spike, as can be seen in Figure 6.17. In general, moving contacts are visible as
spikes in the contact resistance curve.
Figure 6.17 also shows the relative timing of the change-over selector and the
arcing switch: the change-over selector (or tap-selector) starts to move a few
seconds before the arcing switch operates. The tap change ends with the operation
of the arcing switch, which is recognizable by the effect of the transition resistors.
The OLTC and drive are motionless after the tap change, giving a constant
resistance. This information can be used to recognize the operation of the arcing
96

switch in heavily distorted DRM plots. The plot can be read backwards to detect
tap changes: the first resistance change then normally corresponds to operation of
the arcing switch.
1,4
1,2
1
0,8
0,6
0,4
163 165
A
B
C
D
170
175
180
Time [s]
185
190
194,5
Geneva wheel driver hooks into notch, causing a slight movement.

Movement of coarse change-over selector to mid-position.
Movement of coarse change-over selector to next stator block.
Geneva wheel driver leaves the wheel, moving the tap-selector slightly.
Figure 6.17 Typical coarse change-over selection covering four tap positions. The contact
resistance is measured during operation of the change-over selector. The driver of the
Geneva wheel causes irregularities A and D, while irregularities B and C originate
in the movement of the change-over selector to and from the mid-position.
The movement of the change-over selector can depend on the direction the OLTC
moves through all tap positions, making the range of motion of the change-over
selector asymmetrical. This asymmetry is simply the result of the change-over
selector design: no current switching is allowed. One set of contacts is without
current and switches first, but this depends on the direction in which is switched.
Figure 6.18 gives an example of this asymmetry for a design of selector switch type
OLTC that is commonly used in the Netherlands. When the tap position is
increasing, the irregularities are present at tap positions 8-11. The parking position
(or k contact) is used in tap position 9 (indicated by P), while the mid-position of
the change-over selector occurs at tap position 10 (indicated by M). With a
descending tap position (shown on the right), a higher contact resistance is also
observed at tap positions 8-11 but the parking and mid-positions are both at tap
position 9 in this case.
In general, these examples show that changes in the contact resistance caused
by the change-over selector can be expected two positions before the mid-position
and one position afterwards it.
97

1,2
Current [A]
17
0,8
0,4
0
0
100
Current [A]
285
1,15
8
65
10
75
11
85
Time [s]
12
95
Current [A]
1,25
0,8
200
Time [s]
P+M
0,9
11
0,7
195
10
205
8
215
7
225
Time [s]
Figure 6.18 Example of DRM on a service-aged OLTC suffering from change-over selector
contact degradation. Asymmetry in the change-over selector movement is observed. When
the tap position is increasing (left), the irregularities are observed in tap positions 8-11. The
parking position is at position 9 (indicated by P), while the mid-position of the changeover selector occurs at position 10 (indicated by M). With a descending tap position (right),
the irregularities also occur at positions 8-11, but the parking and the mid-positions are both
at position 9.
Finally, in some types of OLTCs the drive system can prepare for a coarse tap
change at the end of the regulating range, but no coarse tap change is actually
made because the end of the regulating
range has been reached. The slight
movement of the coarse change-over
selector can be detected in dynamic
resistance results as a rapid change in contact
resistance, made visible as a stepwise change Figure 6.19 All positions where the
in the measurement current. This effect only coarse-tap selector can move its
shows up in measurement results in contacts.
combination with long-term aging of the change-over selector contacts. To sum
up, all places where the change-over selector can move are indicated in Figure 6.19.
6.3.1.2 The surface film layer
The surface film layer on change-over selector contacts, which is caused by longterm aging, results in increased contact resistance. This early stage of the long-term
aging effect is measured when the contacts are moved over the film while the
contacts carry the measurement current. This less conductive layer described in
Section 2.3 grows slowly, so when DRM is performed during regular tap changer
98

maintenance normally once every few years the increase in contact resistance
can be observed in different stages.
The early stage of long-term aging (i.e. the surface film) can be recognized by
small changes in the contact resistance at the moment the change-over selector
moves. Figure 6.20 shows an example. It is clearly visible here that the change-over
selector is moved twice during the measurement. The recorded current curve
remains linear after the coarse tap selection has been completed. The increased
resistance is only observed in this example at tap positions which are infrequently
used (no overheated contacts, as described in Section 2.3.2). It is clear that the
irregularities originate in the change-over selector and not in the arcing switch.
1,4
1,25
1
0,75
0,5
0,25
0
0
Figure 6.20
selector.
50
100
150
200
250
Time [s]
300
350
400
450
DRM diagnostic data showing contamination of the coarse change-over
Figure 6.21 shows an example of a DRM plot of a service-aged OLTC with 8 fine
windings and another plot for a tap changer with 5 fine windings. The latter makes
two coarse tap changes instead of one to reach the same number of tap positions.
The tap change operations indicated by rectangles involve the change-over
selector. Both tap changers also show irregularities in contact resistance at the end
of the regulating range. Once again, the arcing switch does not give rise to any
irregularities in this example: contact degradation is only observed when the
change-over selector moves, and thus only involves the contacts used at the few
tap positions where the change-over selector operates. Since the change-over
selector contacts in this example are infrequently used, they will be coated with a
surface film (as discussed in Section 3.2.2.1). Several inspections reveal that these
contacts are normally not degraded. As long as contact resistance is only present
on change-over selector contact areas that are used briefly during OLTC operation,
these contacts will not be pitted or covered with pyrolytic carbon. In general, this
early stage of long-term aging should not be considered as critical, but the abovementioned surface film could lead to coking in the long-term. It may be regarded
as harmless in itself, in the absence of contact overheating leading to carbon
formation.
It should be noted that this change-over selector contamination is only
observed when the change-over selector contacts move to another position while
carrying the measurement current, which is not always the case (as discussed in
99

appendix A). For example, small irregularities from normality will remain
undiscovered in most diverter switch type OLTCs.
1,4
1
0
0
100
200
240
Time [s]
1,6
0
0
100
Time [s]
200
280
Figure 6.21 Typical irregularities caused by a moving coarse change-over selector. Top:
DRM plot for a tap changer with 8 fine windings, using one coarse winding. Bottom: DRM
plot for a tap changer with 5 fine windings, using two coarse windings.
6.3.1.3 Contact coking

When long-term aging develops further, the
early change-over selector degradation
discussed in the previous section will also
lead to higher contact resistances when the
change-over selector is motionless. The
advanced long-term aging is defined as a
state where a higher contact resistance is also Figure 6.22 The positions where
present in tap positions where the change- advanced long-term aging is observed.
over selector is motionless. This situation
allows contact coking to occur. High contact
temperatures are expected, as described in Section 2.3.3. The tap changer reached a
part of its lifetime where the resistance increases rapidly and coking occurs, as
shown in Figure 2.17 [24]. The measurement current will thus drop to lower values
during significant parts of the regulating range. This stage of long-term aging can
be recognized when the DRM diagnostic data shows stable changes in the contact
resistance when the contacts are motionless. The high contact resistance leads to
the deposition of low-conductivity carbon at the tap positions where the OLTC is
100
Current [A]
Current [A]
often positioned when transformer is in service. We observed during overhaul that

this phase of long-term aging is accompanied by the presence of a thick surface
film, pyrolytic carbon and deep pitting of the contact surface.
A frequently observed irregularity of the recorded current plot is found when
the change-over selector is moved and a stable resistance for that particular coarse
tap is formed. Figure 6.23 shows two typical examples of measurement results
from service-aged OLTCs. One of the coarse taps is in better condition, but extra
resistance is introduced in the circuit when another coarse tap is selected. The extra
resistance remains in the circuit until the change-over selector operates again. This
is typical of change-over selector contact degradation.
Figure 6.23 DRM diagnostic data showing increased contact resistance in one of the coarse
change-over selector contacts.
6.3.1.4 The parking position

Another frequently observed irregularity of
service-aged OLTCs with advanced longterm aging of the change-over selector is a
rise or fall in the resistance when the changeover selector moves. We say that the parking
position is visible, as illustrated in the DRM
plot of Figure 6.25. The change-over Figure 6.24 Typical location of the
selector now uses parts of the stator contact parking position.
blocks without carbon formation or pitting.
This contact area is not thermally degraded
but a layer with low conductivity exists on the contact, as described in Section
2.3.2. The irregular contact resistance may also be observed away from the parking
position, since the movement of the change-over selector can cover up to four tap
positions (see Figure 6.24).
Such defects look local, and hence not severe, at first sight but it should be
borne in mind that the contacts at the parking position are usually in better
condition than those at other tap positions, because they are only used at a few tap
positions and are therefore not likely to suffer from severe contact coking. In this
case, the other contacts of the change-over selector will be more liable to longterm aging.
101
Figure 6.25 DRM diagnostic data showing a stable resistance in all positions, but less in the
parking position.
6.3.1.5 Vibrations in the change-over selector due to operation of the arcing switch
Long-term aging of the change-over selector
does not only show up when its contacts are
moved by the drive mechanism. Small
vibrations due to the arcing switch operation
can also affect the contact resistance of
heavily degraded change-over selector
contacts. The change-over selector then
causes changes in the contact resistance at Figure 6.26 Locations where the effect
of vibration of the change-over
points where it is normally motionless. This selector due to operation of the arcing
is the area where arcing switch defects are switch can be observed.
normally observed, giving the plot a chaotic
appearance.
A trial was performed to investigate the effect of arcing switch vibrations on
long-term aging of a change-over selector. An OLTC with advanced long-term
aging was removed from the transformer tank and mounted on a drive
mechanism. A DRM plot was recorded for the change-over selector of this device
without the arcing switch. Vibrations due to the mechanical operation of the drive
mechanism and the arcing switch were found to have an effect on the contact
resistance of the change-over selector. Figure 6.27 gives an example of the DRM
plot of such a change-over selector, which is kept stationary. The contact
resistance changes at the moment when the arcing switch is operated (indicated by
* in the plot).
Irregularities in the change-over selector resistance due to operation of the
arcing switch can be recognized by:
1
2
3
4
102
The presence of advanced long-term aging of the change-over selector

contacts (see 6.3.16.3.1.3).
A time of inception that is related to operation of the arcing switch.
Contact resistance variations of an amplitude that is not related to the
position of the arcing switch (see 6.4.1.2.
Irregularities in the current curve at points where the change-over selector
is normally motionless.

0,703
0,6
0,5
0,4
0,262
46,154038 50
60
Time [s]
*
70
74,22396
Figure 6.27 Close-up view of the DRM plot of a change-over selector. Vibrations in the drive
mechanism and the arcing switch led to changes in the contact resistance.
These irregularities are easily confused with long-term aging of the arcing switch
contacts themselves. In both cases, the resistance changes stepwise when the
arcing switch operates. In general, the arcing switch is likely to cause irregularities
in the dynamic resistance if the contact resistance when the change-over selector is
motionless is out of proportion with the changes found when the change-over
selector is operated. To put it the other way round, severe long-term aging of the
change-over selector is likely to cause contact resistance changes when the OLTC
operates. Contact degradation located in the arcing switch may hence remain
undiscovered in this case.
6.3.1.6 Diverter switch type tap changers
The previous section considered a change-over selector design where the contacts
were moved (not switched) while the measurement current flowed through them.
The surface film on the contacts showed clearly in the DRM graphs at these types
of OLTCs. However, diverter-switch type OLTCs normally move their tapselector contacts without current. In that case, the surface film on the briefly-used
contact area is not measured and only the stationary contact locations will be
diagnosed. Early long-term aging of the change-over selector contacts may remain
undiscovered. If the aging progresses further, a step change in the resistance can be
observed when the diverter switch selects the next position of the coarse tapselector (as shown in Figure 6.28). Diverter
switch type tap changers normally use snap
contacts (also called tap selector contact
terminals, see Figure 6.29) to pre-select the
next winding tap. These contacts have a
strong mechanical wiping effect on the
surface film and are thus less prone to long- Figure 6.28 The locations where
term aging. An example of the DRM plot defects are observed in a coarse tapselector in combination with a diverter
for such an OLTC is given in Figure 6.30.
switch.
103
1,25
0,75
0,5
0
Figure 6.29 Change-over selector

contact of a diverter switch type
tap
changer.
Two
contact
terminals for (pre) selecting a
transformer winding are shown.
The bottom terminal has been
selected.
50
100
150
Time [s]
200
250
300
Figure 6.30 A typical current profile of a tap changer

which moves its coarse tap-selector contacts at zero
current. Only the stationary resistances are visible.
Another typical example is shown in Figure 6.31, which presents the plot of a
diverter switch type OLTC with a number of overheated tap-selector contacts.
Several tap positions show increased contact resistance. These tap-selector contacts
did not have any surface film, and inspection showed that they were already heavily
pitted.
1,2
1
0,5
0
0
50
100
Time [s]
150
210
Figure 6.31 Example of a DRM plot of a service-aged OLTC with a tap-selector using snap
contacts. Increased contact resistance was observed at several tap positions. Overheated tapselector contacts were found during inspection.
6.3.2 Current interruptions due to the change-over selector

The second source of changes in the dynamic resistance measurement results that
is related to degradation or defects (Figure 6.13E) occurs when the measurement
current is completely interrupted by the tap changer. In that case, problems can be
104

expected when the OLTC is in service. The contacts are not designed to handle
interruptions in the load current: the tap changer and the transformer surrounding
it can be seriously damaged when the OLTC tries to interrupt the load current.
Several factors can cause the measurement circuit to be totally interrupted during
diagnostic measurements. These factors may be classified according to the location
of the interruption (arcing switch or change-over selector).
The duration of the interruption and its location in the cycle of operation
provide information about the site of the interruption. The influence of the
amplitude of the measurement current can provide further information about the
cause of the interruption.
Current interruptions caused by the change-over selector will appear and
disappear when the change-over selector is operated, and typically last longer than
the transition time of the arcing switch. Interruptions longer than 200 ms that
appear in the area of the DRM plot where the change-over selector is moved are
typically caused by the change-over selector. These interruptions may take place
only when the change-over selector moves (Figure 6.32A) or may cover complete
coarse taps (Figure 6.32B). Two examples of these defects as measured in the
Dutch utility grid are shown in Figure 6.33 and Figure 6.34. The first concerns an
OLTC with contact timing problems in the change-over selector due to a
maintenance error. Wrong timing due to incorrect reassembly caused incorrect
contact timing between the selector switch and the change-over selector. As a
result, the change-over selector switched before the current was transferred to
another contact. The interrupted current was recognized as a serious defect and
the error was corrected before the transformer was re-energised. The example of
Figure 6.34 concerns a change-over selector that did not establish a correct
connection in one of its coarse taps due to advanced long-term aging. This
transformer was damaged beyond repair when it was energized directly after the
taking of the DRM of Figure 6.34 (note: this plot was recorded with the system
described in [81]). Both cases involved complete interruption of the current: such
current interruptions due to the change-over selector or tap-selector should be
considered to be critical.
In general, interruptions of the current due to the change-over selector can be
caused by:
Figure 6.32 Current interruptions due to the change-over selector can occur when the
change-over selector moves (A) or can cover larger parts of the cycle of operation (B).
105

1
2
3
Long-term aging (leading to a thick surface film or coking). DRM results

recorded during previous OLTC maintenance will already show signs of
long-term aging. The contact resistance will depend on the measurement
voltage and current amplitude. Interruptions of the measurement current
may disappear when the current amplitude is increased.
Wrong contact timing; such interruptions are independent of
measurement voltage and current amplitude.
Mechanical damage to the OLTC; these interruptions are also independent
of measurement voltage and current amplitude.
1,2
1
0,8
0,6
0,4
0,2
0
0
20
40
60
80
100
Time [s]
120
140
170
Current (A)
Figure 6.33 Dynamic resistance measurement on a tap changer with a change-over selector
which interrupted the current due to incorrect contact timing.
Figure 6.34 Dynamic resistance measurement on a service-aged tap changer with a changeover selector which interrupted the current due to long-term aging.
6.4 Irregularities in the arcing switch

It was demonstrated in the previous section that defects in the change-over
selector show up when the change-over selector contacts move. In general,
irregular values of the contact resistance appear, change and disappear when the
contacts move. This also applies to the arcing switch: irregularities in the contact
resistance caused by the arcing switch appear, change or disappear when it is
operated. It should be noted that the arcing switch operates throughout the entire
cycle of operation. Hence, the arcing switch can give rise to changes in the contact
resistance that are more frequent but shorter-lived than those caused by the
change-over selector.
106

The arcing switch is also operated more frequently during the service life of the
power transformer than the change-over selector, typically 1-10 times a day. Longterm aging gets less chance to develop on the moving contacts under these
conditions, because the surface film is wiped off during each operation of the
arcing switch.
This section will elaborate on how defects in the arcing switch can lead to
irregularities in the DRM plot. The following defects are considered:
1
2
3
4
Long-term aging (see 6.4.1)

Contact wear (see 6.4.2)
Timing difference between phases (see 6.4.3)
Maintenance errors (see 6.4.4)
Defects and contact degradation in the arcing switch can be checked during regular
maintenance.
6.4.1 Long-term aging in the arcing switch
Despite the frequent movement of the moving contacts in the arcing switch, we
observed that pyrolytic carbon deposition can occur here. This slow degradation
effect can develop between fixed contacts
(interfaces) that often carry the load current,
and is similar to the long-term aging of the
change-over selector contacts. It shows up
in the contact resistance as a disturbance of
the linear profile (Figure 6.35) during
operation of the arcing switch. In general, Figure 6.35 The positions where step
the presence of a linear pattern in the plot changes in the current due to the
points to an arcing switch without long-term arcing switch are observed.
aging. Two types of arcing switches (those
used in selector-switch type OLTCs and in diverter-switch type OLTCs) are
discussed in turn below.
6.4.1.1 Selector switch type OLTCs
The effect of long-term aging on this type of arcing switch was investigated by
plotting the DRM current curve for a separate arcing switch to ensure that contact
degradation of the change-over selector did not influence the readings. A selectorswitch type OLTC which had been in service for 50 years was used for this trial.
Inspection of the contacts showed signs of coking due to long-term aging.
For this experiment, the selector switch was measured separately by
disconnecting the arcing switch from the change-over selector and the transformer
windings. Figure 6.36 shows the current recorded while the tap changer was
switched through all tap positions. It may be clearly seen that the resistance is
increased by long-term aging of the contacts of the selector switch. Variations in
the resistance occur at the moment when the selector switch operates. The
107

1,2
1
0
15
50
100
135
Time [s]
Figure 6.36 Step changes in the resistance are visible in this DRM plot of a separate arcing
switch.
resistance remains constant between selector switch operations, because the switch
is then motionless.
Figure 6.37 shows a
close-up view of the current
plot during arcing switch
operation. The effect of the
transition
resistors
is
measured during the interval
200-300 ms. The spikes
caused by moving contacts
over degraded fixed contacts
are visible before and after Figure 6.37 Close-up view of Figure 6.36.
the transition resistors are
used. It will be recalled that the selector switch contacts can oscillate after each tap
change operation (see Figure 3.7) and the spikes in the dynamic resistance are
measured during these oscillations.
The small increase in contact resistance when the arcing switch is installed in
the power transformer is also visible in the DRM plots. Figure 6.38 gives an
example: the increase in resistance shows up here as a drop in current. This effect
is also visible in the DRM plots of circuit breakers [78].
0,8
0,7
0,6
0,5
0,4
0,3
401,2
401,25
401,3
401,35
Time [s]
401,4
401,45
401,5
Figure 6.38 Close-up view of the measurement current variation during a dynamic
resistance measurement on a selector-switch type OLTC. The contact resistance increased
just before the contacts separated.
108

6.4.1.2 Diverter switch type OLTCs
Current [A]
Unlike a selector-switch type OLTC, a diverter-switch type OLTC has only two
sets of contacts available for tap selection. One is selected by the diverter switch
and is under load. The second set of contacts is used to select the next tap (preselected tap) at zero current. The diverter switch switches from the selected to the
pre-selected tap each time the tap changer is operated (see also Appendix A). A
typical irregularity of the current appears when one of these contact sets has an
increased resistance. The measured resistance will then alternate. An example of
such a dynamic resistance measurement is shown in Figure 6.39, which was
obtained by measuring a tap changer with a defect that had been artificially applied
in order to investigate this effect.
Figure 6.39 Dynamic resistance measurement on a diverter switch type OLTC with one set
of deteriorated contacts.
6.4.2 Contact wear of the arcing switch

Wear of arcing switch contacts also changes the DRM plots because it influences
the timing of the transition resistors. This effect is short-lived and shows up at
throughout the entire cycle of operation where the arcing switch operates (Figure
6.40).
Figure 6.41 shows a typical plot for a
service-aged arcing switch (from a diverterswitch type OLTC) which uses one of the
transition resistors a little longer. This causes
the measurement current to drop to lower
values. Trigger points are added to the figure
to show up the minima more clearly.
This effect can be caused by contact wear Figure 6.40 The positions where
or by an irregular switching speed. contact wear of the arcing switch
Experiments have been designed to show shows up.
how the contact timing can be affected. The
results will be presented in Section 7.2.
Heavily worn contacts can also cause short-lived current interruptions during
DRM. These interruptions will be discussed in Section 6.4.5.
109

1
0,95
0,9
0,85
0,8
0,75
50
100
150
200
Time [s]
250
300
350 380
Figure 6.41 The depth of the current drop during the tap change operations alternates due
to changes in the switching speed or contact wear in one set of diverter switch contacts.
6.4.3 Timing difference between phases

The three phases of the OLTC are designed to switch at the same time. Defects in
the drive mechanism or in the drive shaft can cause loss of synchronism between
the phases. The difference in the timing of the transition resistors can be revealed
by DRM. When there is a loss of synchronism, the transition resistor is introduced
in one phase earlier than in another phase. The measurement current during an
OLTC operation will then briefly increase in the path with the lowest resistance
(i.e. the phase without transition resistors). Figure 6.44 gives an example: the
current through the three phases is shown at the moment when the defective
OLTC operates. Two phases switch prematurely, causing the current through the
third phase to increase. A simulation of this defect has already been shown in
Figure 4.22.
The effect can be seen most clearly when the three phases are measured
simultaneously. When only one phase is measured, for example in the case of an
tap changer in a delta-connected transformer
winding, only the increasing measurement
current is observed, thus indicating when the
other phases switched. The increasing
current typically appears very close to the
point of operation of the arcing switch, and
can show up throughout the entire cycle of Figure 6.42 The location where
operation, as illustrated in Figure 6.42.
contact timing problems are observed.
6.4.4 Maintenance errors
In addition to typical degradation-related DRM results, atypical irregularities in the
plot may be caused by errors in the maintenance of the arcing switch. These errors
may affect the transition times, cause interruptions of the current or cause irregular
contact resistances. An example is shown in Figure 6.43. The irregularities
observed in this DRM plot could not be ascribed to severe long-term aging
110

(contact coking), since they would then appear at several locations in the plot. Nor
could they be attributed to other forms of long-term aging or to other contact
degradation mechanisms. In fact, they were due to a maintenance-induced problem
leading to interruption of the measurement current in one phase. The recorded
current was not interrupted because the OLTC was mounted in a delta-connected
winding so that an alternative path for the current was still available. A current
interruption in a delta-connected transformer leads to a drop of current in another
phase. In this case, the defect resulted in incorrect operation of the arcing switch
and total loss of the power transformer after it was energised after taking this
DRM. If follows that inspection (or re-inspection) of the arcing switch is advisable
when atypical irregularities appear in the plot.
Figure 6.44 Example of a recording of the DRM current in an OLTC with synchronisation
problems between the phases.
Current [A]
The example of Figure 6.43 shows clearly that a critical situation can arise due to
an irregular dynamic resistance in only one tap position. While a tap position can
be used for a long time, a defect on one particular tap position can cause a
catastrophic failure of the
1,25
transformer. Once again,
1
it may be said that a chain
is only as strong as its
weakest link.
0,5
0
0
100
200
300 330
Time [s]
Figure 6.43 Dynamic resistance measurement on a tap

changer with a defect in the arcing switch.
111

6.4.5 Current interruptions due to the arcing switch
Interruptions of the measurement current
that originate in the arcing switch are
characterized by their short duration
normally only part of the transition time
(typically 10-200 ms) or the time
corresponding to one complete tap
position. Such irregularities in the DRM
plot can be attributed to the arcing switch Figure 6.45 Interruptions due to the
switch are short and can occur
when they appear at moments when the arcing
throughout the entire cycle of operation.
change-over selector is motionless. Longer
interruptions are unlikely to originate in the arcing switch because it is operated
more frequently than the change-over selector.
All interruptions appear when the arcing switch is operated, as indicated in
Figure 6.45. Three causes are discussed in this section: oil between the contacts,
contact wear and damaged transition resistors.
6.4.5.1 Short interruptions due to a surface film
Short current interruptions when the arcing switch operates can be caused by a
thin layer of oil that has to be pushed away by the moving contacts before a lowresistance contact can be made. This phenomenon occurs at low measurement
currents in arcing switches that use contact blocks. The oil layer will not cause
current interruptions in normal operation, because the open-circuit voltage is then
higher (for instance 300 V instead of 3 V) and the current is also higher. Short
interruptions of this type are therefore not critical, but if they occur the OLTC
should be checked to ensure that it is indeed subject to this effect.
6.4.5.2 Short interruptions due to wear
Wear of arcing switch contacts can lead to incorrect timing, causing an arcing
contact to take over the current too late. This effect accelerates contact wear even
more. OLTC types using roller contacts are particularly prone to this effect. Figure
6.46 shows a typical DRM plot for a service-aged selector switch type OLTC with
1,3
Current [A]
0,5
0
0
100
200
300
Time [s]
Figure 6.46 DRM plot of an OLTC with roller contacts. Current interruptions are visible due
to selector switch contact wear.
112

roller contacts where current interruptions due to the arcing switch are visible. It
may be clearly seen that only specific selector switch contacts are involved. Visual
inspection inside the OLTC showed that excessive arcing during normal operation
had already damaged the contacts. A close-up view of the relevant portion of this
DRM plot reveals one or multiple interruptions in succession (see Figure 6.47).
The current expected in the absence of interruptions is shown by a dotted line.
Figure 6.47 Close-up view of a DRM plot of an arcing switch with contact wear. The current
is interrupted one or more times in succession.
6.4.5.3 Short interruptions due to damaged transition resistors

The measurement current can also be interrupted due to damaged transition
resistors. The resistors are not designed to carry the load current for a long time.
Damage to the transition resistors can lead to an open measurement circuit. The
pressure relay of the arcing switch immediately switches the transformer off the
moment the resistor fails. The open circuit can also be detected with the aid of
DRM, and shows up in the plot at locations where the arcing switch normally uses
the damaged resistor. An example can be seen in Figure 6.48, where one of the
damping resistors caused an open circuit. The total recorded current plot is shown
on the left and a close-up view of one operation on the right. The interruption of
the measurement current is observed each time the OLTC operates. The current
falls to zero at the moment the resistor should
take over the current.
Figure 6.48 DRM plot (left) and close-up view (right) of a slowly-switched OLTC with one
damaged transition resistor in the arcing switch. The current falls to zero each time the
OLTC operates.
6.4.6 OLTC transition times

The third source of changes in DRM plots that is related to degradation or defects
OLTC transition times was already mentioned in Figure 6.13C. The transition
113

time is determined for each operation in the cycle of operation. Mechanical defects
or friction can result in longer transition times. This can accelerate arcing contact
wear because the switching arc will persist for longer. Irregular transition times can
be seen at locations in the DRM plot where the arcing switch operates (Figure
6.49), and may be detected by comparing the
results with earlier measurements or with
standard transition times for that type of
OLTC.
Figure 6.50 shows three examples of the
use of DRM to detect irregular transition
times due to a defective drive mechanism. In
each case, the transition times are shown for Figure 6.49 Irregular OLTC transition
all tap change operations in the cycle of times show up each time the
transition resistors are used.
operation. The three cases considered are:
1. OLTC with 2 broken energy-accumulation springs in the drive mechanism
(transition time between 80 and 90 ms),
2. The same type of OLTC with 1 broken energy-accumulation spring in the
drive mechanism (transition time between 50 and 60 ms),
3. The OLTC of example 2 after repair of the defect (transition time between 35
and 40 ms).
100
Transition time (ms)
90
80
70
60
50
40
30
20
10
34
32
30
28
26
24
22
20
18
16
14
12
10
OLTC switch operation
Figure 6.50 Three examples of the use of DRM to measure OLTC transition times. All
transition times in the cycle of operation are displayed for the three phases. The cases shown
are: an OLTC with two broken energy-accumulating springs (1), an OLTC with one broken
energy-accumulating spring (2) and the same OLTC after repair of the drive mechanism (3).
It will be seen that longer transition times are observed when less mechanical
energy is available to move the contacts.
Figure 6.51 shows close-up views of the current during the operation of the
arcing switch for the examples of Figure 6.50. The transition resistor is used longer
due to the slower OLTC operation.
114

It can be concluded that the transition time can be interpreted by comparing
results with earlier measurements. Ref. [89] considers a transition time deviation
more than 10 ms as unacceptable and recommends visual inspection.
Current [A]
1,1
1.
2.
0,9
1
3.
0,7
OLTC with two broken

energy-accumulating springs.
OLTC with one broken
energy-accumulating spring.
Same OLTC after repair of
drive mechanism.
0,5
0
200
400
600
Time [ms]
Figure 6.51 Close-up view of the current through the arcing switch for the three examples of
Figure 6.50.
6.5 Defects in both parts of the OLTC

Exact localization of the defects and degradation mechanisms can be difficult
when long-term aging causes resistance changes in all parts of the DRM plots. An
example is given in Figure 6.52. All that can be concluded from this plot is that:
the long-term aging is severe (thermally degraded contact exist, as described in

Section 2.3.3, and the tap changer is in an aging stage where the resistance can
increase rapidly, as shown in Figure 2.17.
the contact resistance of the change-over selector has increased
Figure 6.52 Dynamic resistance measurement on an OLTC with different types of long-term
aging.
The enormous amount of contact degradation on the change-over selector

contacts in this example makes it difficult to say whether the arcing switch also has
higher resistance. Inspection of the arcing switch can give the decisive answer here.
A trial of a service-aged selector switch type OLTC provides a good example of
this approach (see Figure 6.53). Measurements were first carried out on the OLTC
115
Figure 6.53 DRM plots of a service-aged selector switch type OLTC when the selector
switch and the change-over selector were measured in series (top) and separately (bottom)
as a whole (top). The low measurement voltage (0.75 V) in combination with the
degraded contacts caused short interruptions of the measurement current. The
results of separate measurements on the selector switch and the change-over
selector showed clearly that both parts of the OLTC contributed to distortion of
the contact resistance. The change-over selector distorts the DRM when its
contacts are moved. The selector switch, which uses the same contacts every 9 tap
positions, shows the same distortion twice. The degradation of the arcing switch
contacts did not show up when both parts were measured in series.
This example shows clearly that increased resistance of change-over selector
contacts can mask arcing switch contact degradation in DRM results; the absence
of a linear current pattern can be caused by the arcing switch and the change-over
selector.
6.6 Conclusions
This chapter dealt with typical DRM results for high-speed resistor-type OLTCs.
Firstly, measured plots were subdivided into:
1) Plots showing no effects of defects or contact degradation
2) Plots showing current interruptions, which can lead to OLTC damage in
service. The current interruption may be located:
a) on the change-over selector, or
116

b) on the arcing switch.
3) Plots showing current changes caused by contact resistance, leading to higher
local contact temperatures and sometimes ultimately to thermal runaway,
a) on the change-over selector, or
b) on the arcing switch.
4) Plots showing irregular transition times due to mechanical defects.
It follows that the change-over selector, the arcing switch and the drive system can
be diagnosed separately with the aid of DRM. The location of the defects can
generally be determined on the basis of the moment when the resistance changes.
Secondly, it was shown that many defects and degradation mechanisms lead to
changes in resistance, which makes DRM a useful tool for OLTC condition
assessment. In particular, it was shown that DRM is sensitive to a variety of defects
and degradation such as:
1) Long-term contact aging on the contacts
a) early (local effects of the surface film)
b) advanced (widespread effects of surface film and coking)
2) Wear of the arcing switch contacts
3) Timing differences between phases
4) Maintenance errors
5) Irregular transition times
Typical DRM plots revealing these various defects were presented and it was
explained how these defects can be recognized in different types of OLTCs. As
will be clear from the complicated relationship between the OLTC design, the
degradation mechanisms and the DRM method, these results need to be
interpreted by a specialist. This specialist needs to have expert knowledge of tap
changer operation and aging, and of the DRM method applied. This way, the
containing information on the OLTC condition can be extracted from the
measurement results.
117
CHAPTER 7 - LABORATORY VERIFICATION OF SELECTED FAILURE

MECHANISMS
7
Laboratory verification of
selected failure mechanisms
It is important to be able to assess the condition of tap changer parts which are not
accessible for inspection during regular maintenance (which only examines the
arcing switch). An example being contacts of the change-over selector, at which
the long-term aging effect is a major source of contact irregularities observed in the
DRM results. However, defects or wear in the arcing switch are also of interest
because they may be overlooked during visual inspection. This chapter discusses
experiments with typical tap changer degradation, using artificial degraded contacts
mounted in a physical tap changer model. The following topics are studied in
depth:
Measurability of the surface film using resistance measurements

Correlation between DRM and contact degradation
Measurability of arcing contact wear
(7.1)
(7.1.1)
(7.2)
7.1 Measurability of the surface film using DRM

7.1.1 Correlation between DRM and contact degradation
The exact localisation of contact irregularities provides a better understanding of
the stage of development of contact degradation in the OLTC. However, a direct
relationship between dynamic contact resistance and the contact condition is not
observed.
Chapter 6 revealed that it is possible to determine whether the irregularities in
contact resistance originate in the arcing switch or the change-over selector by
examining the moment they appear. This section is dedicated to the localization of
119

MECHANISMS
contact degradation on the change-over selector contacts. The movement of the

change-over selector contacts is related to the physical location and condition of
the contacts. The contact status was examined during OLTC overhaul and using
overhaul reports. This section will demonstrate that a higher (static or dynamic)
resistance does not necessarily mean that the contact surface at that location is in a
bad state.
Figure 7.1 This section elaborates on the location of contact degradation in the change-over
selector.
The change-over selector moves several times for each coarse tap change, thereby
using different contact areas. These dynamic movements of the change-over
selector can provide valuable information about the stage of development of the
long-term aging effect. This section takes a closer look at how these change-over
selector contact areas are measured with DRM.
Different types of tap changer are available to study the long-term aging effect
using DRM. A typical case study is discussed here, using a selector switch type
OLTC with a change-over selector. The type discussed in this section moves its
roller contacts while they carry the load current (more information about this type
of tap changer can be found in Chapter 2.1.1). The movement of the change-over
selector, covering four tap positions, as discussed in Section 6.3.1, can be divided
into ten sub-positions, as shown in Figure 7.2. A different DRM result was
observed for each sub-position. The mid-position of the change-over selector
forms the centre, recognisable by spikes in the current which are observable on
both sides of the mid-position (indicated by M). These spikes are caused by the
moving roller contacts of the change-over selector. Resistance deviations begin
two positions before the mid-position (M-2), in terms of a stepwise change in the
current. A stepwise change in the current was also observable one position after
the mid-position (M+1).
120

MECHANISMS
Figure 7.2 A typical deviation originating from the change-over selector covers four tap
positions. The change-over selector moves at these tap positions. This movement can be
divided into ten parts.
Figure 7.3 shows the stator of a coarse change-over selector with one coarse tap
winding, which can be selected by placing the rotor contacts on contact block CT.
The load current flows through contact block A, except in the parking position,
where block A0 is used. Visual inspection of these contacts revealed an advanced
long-term aging effect, see Figure 7.3.
To selector
switch
CT
M
A0
Coarse tapselector
Figure 7.3 Stator contacts of a coarse tap selector. The coarse tap is selected when the rotor
selects contact block CT. Block A0 is used in the parking position.
121

MECHANISMS
The ten change-over selector positions of Figure 7.2 are related to the position of
the rotor. The amount of pitting and carbon is determined for all rotor positions
and expressed as a condition index using the condition indices of Table 7.1.
Condition indices for the DRM results are also determined on the basis of Table
7.1, resulting in the overview in Figure 7.4, which shows the contact status (stator
and rotor average) as seen during inspection and the corresponding dynamic
resistance results. All three phases of the transformer and all ten change-over
selector positions are considered. Figure 7.4 also shows a linear trend line through
these 30 data points. A direct relationship between the contact status and the
corresponding irregularity in the DRM plots was expected. Figure 7.4 shows the
opposite: an increased contact resistance is measured when the contacts are in
good condition, while there is a decrease in resistance when the contacts are in
poor condition. The following section investigates this important issue in more
detail.
Table 7.1 Overview of condition indices for assessing change-over selector degradation.
Condition code
Contact status
DRM result
No pitting and no carbon
No additional contact resistance
Light pitting and carbon
Light additional contact resistance
Deep pitting and carbon
Considerable additional contact resistance
9
7
6
5
4
2
nt
ac
ts
ta
tu
1
0
0
1
DRM res
ult
0
2
Co
Number of contacts
Figure 7.4 Relationship between the condition of 30 contact positions as measured with
DRM and as seen by inspection. A linear trend line through these 30 data points is plotted.
122

MECHANISMS
7.1.2 Effect of the surface film on DRM

A more detailed investigation shows that the presence of the surface film deviates
more clearly the DRM plots than the more dangerous pyrolytic carbon. Figure 7.5
shows our assessment on the relationship between the OLTC contact surface and
the contact resistance. The plot shows a decreased measurement current caused by
the contact resistance of one of its coarse taps.
Figure 7.5 Localisation of change-over selector degradation. There is carbon present in the
measurement circuit at places where the DRM plots are linear. Irregularities appear in the
DRM plots when the slightly aged stationary contacts (right) are used.
It can be observed that the contact resistance of stationary contact CT is higher

than the contact resistance of stationary contact M. Visual inspection showed that
the contact surface of stationary contact M was affected by coking (as described in
Section 2.3.3), while stationary contact CT only showed a thick surface film (due to
infrequent use, see Section 2.3.2). When the contact resistance is related to the
contact condition, the location with higher contact resistance relates to the surface
film on the contacts.
Figure 7.6 shows our assessment of the relationship between the DRM plot and
the physical condition of the contacts at the moment the change-over selector
moves. Spikes occur in the dynamic resistance when the contacts move over a
surface film. This effect is typical for this type of service-aged OLTC. A higher
resistance is present at these contact locations which, being less frequently used,
123

MECHANISMS
are covered with an insulating surface film. This film layer causes a relatively high
contact resistance.
To conclude, this case study, as well as results obtained from other service-aged
OLTCs, confirms that DRM at 1 A measures an increased contact resistance due
to the surface film. This layer can be measured with a higher amplitude, compared
to contact locations subject to coking and pitting. From these observations it
follows that the conductivity of pyrolytic carbon is better than the conductivity of
contact film.
Figure 7.6 Relationship between the DRM plot (top) and the physical condition of the
change-over selector contacts. Moving contacts, in combination with a thick surface film,
cause spikes in the dynamic resistance.
7.1.3 Accelerated degradation tests

The measurability of the surface film is important because it indicates the onset of
the long-term aging effect that leads to coking. In addition to visual inspection of
service-aged OLTCs, accelerated degradation tests on a test model [90] can be used
to show in a controlled way that the surface film can be measured with DRM.
7.1.3.1 Setup: the test model
Several degraded tap changer contacts were examined in the laboratory to
determine the influence of a surface film on their dynamic resistance. A model was
developed to test artificially degraded contacts at different stages of degradation.
The test model (Figure 7.7) consists of one phase of a 60 kV/630 A selector switch
124

MECHANISMS
type OLTC with two 2.4 transition resistors. The regulation winding was
simulated by eight series-connected 43 m resistors that can be selected using 9
stationary contacts. The model was connected in series with a power transformer
to provide the required circuit inductance. The transformer (three phase
50 kV/14 MVA) was equipped with a tap changer that would not cause
irregularities in the dynamic resistance plots and which was not operated during
these experiments. The test model was hand-driven and switched in 45-70 ms, a
little slower than normal high-speed resistor-type OLTCs. Tap selection was
performed with roller contacts using the contact pressure of the original design
(108 N). The tap changer model was not submerged in
Figure 7.7a Schematic representation of the

OLTC test model.
Figure 7.7b The OLTC test model

used for experiments with artificial
degradation.
transformer oil because the contacts tested already showed the same resistive
behaviour as contacts in oil (see [11]). A constant DC measurement voltage was
applied to the test model (and the series-connected transformer) and regulated in
such a way as to provide 1 A initial measurement current. As soon as the current
was stable the test model was
switched through all its tap
positions and the current was
recorded. A sequence of reference
DRMs showed that this test
model did not cause irregularities
in the contact resistance when
clean contacts were used, as
shown in Figure 7.8. In total, 20 Figure 7.8 The reference current pattern recorded
laboratory experiments were on the test model with clean contacts using the
performed, each consisting of 3- DRM technique. The transition resistors reduce
the current when the OLTC moves to the next tap.
16 measurements.
125

MECHANISMS
7.1.3.2 Accelerated degradation tests

Thermally accelerated degradation tests were carried out on the test model using an
artificially grown surface film to investigate its effect on DRM. Service-aged Shell
Diala transformer oil was used for accelerated degradation on nine contact blocks
of a selector switch. The bulk oil was heated, resulting in a surface film on the
entire surface of the contacts. The formulae of [24] were used to calculate the test
time (see also Chapter 2): submerging the contacts for 6 hours in oil of 165 C
simulated 40 years of service at 60 C, which is the average age of the OLTC
population evaluated in [11] and tap changer with an age of 25-35 years show
increased contact resistances in DRM at 1 A. The thickness of the surface film that
is formed during these tests was not verified by ellipsometry or other methods, but
visual inspection of the contact surface confirmed the presence of surface film.
These accelerated degradation tests caused 300 additional contact resistance
during a static resistance measurement at 100 A, which could not be measured
with DRM due to the noise level of the setup. Ref. [11] describes these
measurements in more detail, with no increased contact resistance being measured.
Section 7.1.2 showed that a surface film is expected to have a significant effect
on contact resistance. Therefore, the accelerated degradation tests as described
above were continued, with the objective of generating the plots as measured in
the field. During this accelerated degradation test both the contact blocks and the
roller contacts were submerged in hot oil, in contrast with [11]. This doubled the
thickness of the surface film (note: tap changers in service can have two of such
contact sets in series).
The first accelerated degradation test concerned contacts that were aged at
180 C (temperature controlled between 176-182 C) for 5.5 hours. Those contacts
showed increased contact resistances at only a few of the DRM plots at 1 A. The
effect is small and appears only when the contacts are moved. Figure 7.9 provides
an example of a plot measured for one of these contacts. In other measurements,
the thin film broke and was not detected with DRM.
The final experiments artificially aged the contacts for 14.7 hours at an oil
temperature of 201 C (temperature controlled between 199-202 C). This time all
the contacts tested showed an increased contact resistance in the plots, causing up
Figure 7.9 DRM plot of an artificially aged OLTC contact (5.5h at 180 C oil). The contact
resistance shows a few spikes when the contacts are moved (t=7-10 s) just before the OLTC
is switched (t=18 s).
126

MECHANISMS
to 8% variation in the 1 A measurement current (Figure 7.10). This effect is still

small compared to the contact resistances measured during field tests (for example
50% variation in the measurement current). The thin film breaks easily and is
quickly wiped away when the contacts are continuously moved for some time
(Figure 7.11). However, the typical spikes in the contact resistance correspond with
the results of field tests.
Figure 7.10 DRM plot of artificially aged

OLTC contacts (14.7 h at 201 C oil). The
measurement current drops by 8%.
Figure 7.11 The surface film is wiped away

when the contacts are continuously moved
for 25 s.
Figure 7.12 shows the results of 16 dynamic resistance measurements on the

contacts that were aged for 14.7 h at 201 C. All of the measurements show
increased contact resistances above the noise level.
These accelerated degradation tests show that a surface film on contacts can
indeed cause a considerable increase in the contact resistance as measured at 1 A.
This confirms findings from DRM on service-aged contacts. A thin layer is not
detected in every measurement because it often breaks and the conductive path is
restored. However, a thick layer is observable as increased contact resistance in
every measurement [91]. High local temperatures are needed to grow an insulating
layer that is thick enough to be observed in every DRM plot. Service-aged OLTCs
8
6
4
2
Noise level = 0.3%
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Change in the resistance (%)
10
Experiment
Figure 7.12 Sixteen experiments were performed with degraded OLTC contacts to show that
the surface film can be measured. All measurements show increased resistances above the
noise level.
127

MECHANISMS
in the utility grid can be exposed to such high local temperatures, as suggested by
the carbonisation of the oil observed in some service-aged OLTCs.
Therefore, it can be argued that the maximum contact resistance that is
measured with DRM can inform us about the thermal degradation stage of the
contact surface. DRM may be useful in finding the cause of increased contact
resistance: the long-term effect can be recognised due to the current spikes in the
plot as shown in Figure 7.6. Static resistance measurements could not have
assessed this contact area because it is not used in static situations. A thick surface
film on the contacts indicates an advanced long-term aging effect and coking may
have occurred on locations where the load current heats the contacts. In order to
draw a conclusion about the presence of pyrolytic carbon due to coking, the aging
mechanism and the length of time that the thick surface film has been present
should be considered. Therefore, previous DRM results for the same OLTC
should be considered in order to reach a solid conclusion about the presence of
pyrolytic carbon.
7.2 Arcing contact wear

Contact wear can be determined by visual inspection during regular maintenance
but can also be measured by DRM because it affects contact timing. Irregularities
in the transition time can be measured by DRM when contacts are worn, with even
short interruptions of the measurement current appearing in the plots. Section
6.4.5.2 presented an example of DRM where arcing contact wear influenced
contact timing. This section discusses two further ways to detect arcing contact
wear:
1) determine the size of the arcing contact using DRM in combination with a
position sensor
2) reveal contact timing differences by comparing contact timing with a
reference
Figure 7.13 shows how wear of the arcing contact is defined: a new arcing contact
has 0% wear, while the contact wear is 100% when the contact material of the
arcing contact is abraded until the main contact. The example shown in the picture
is therefore 65% worn.
7.2.1 Arcing contact dimensions
Dynamic resistance results provide information about the arcing contact
dimensions, in particular when they are combined with a contact position
measurement using a position sensor on the remote end of the drive shaft (see also
128

MECHANISMS
Direction of wear
0% wear
Arcing contact
65% wear
100% wear
Main contact
Vanished
material
contact
Figure 7.13 An arcing contact about 65% worn.
Figure D.8). Figure 7.14 shows a combination of the measurement current (black
line) and the position measurement (grey line) during an arcing switch operation.
The contacts of the arcing switch start moving at T1 and the main contact moves
to the end of the stator contact (X2). During this movement, the main contact
(without transition resistor) carries the measurement current, with the latter (black
line) indicating an increase in contact resistance. The main contact leaves the
contact block at T2, thereby introducing a transition resistor. The main contact
Figure 7.14 The combination of dynamic resistance measurements (black line) and the
position measurements of the OLTC contacts (gray line) during an arcing switch operation.
The rotor starts moving at T1 and the transition resistor is used at T2. The main contact
reaches the next tap at T3.
129

MECHANISMS
then moves to the next tap, reaching it at T3 and comes to a standstill at position
X4, after some mechanical oscillations. Comparing these two measurements
provides information about:
X2-X1: the size of the arcing contact before the main contact disconnects
X3-X2: the distance between the contacts
X4-X3: the size of the arcing contact after the main contact connects
DRM current (A)
Tap n+2
Tap n+1
Tap n
OLTC axis position
The degree of contact wear of an arcing contact can then be determined in terms
of the total distance that the main contact travels over the contact block, which can
be read from the combined plots. The width of the main contact and the length of
a new arcing contact should be considered in this case. Figure 7.15 shows that the
moment the main contact arrives at the stationary contact of the diverter switch
(indicated by T3) and the moment it leaves (T4) can be read from the current curve.
In this example, this yields a contact length of XDRM. Contact wear is then
calculated as XNEW-XDRM. This example demonstrates that contact wear does not
affect the results of the position measurement, but the timing T3 and T4 of the
transition resistors does change. Contact wear can then be determined with an
accuracy of a few millimetres when DRM is performed with a 1 ms resolution.
Visual inspection will be more accurate.
Figure 7.15 The length of the OLTC contact can be determined when the dynamic
resistance and the shaft angular position measurements are combined. The main contact of
the diverter switch arrives at moment T3 and leaves the contact again at T4, resulting in a
contact length of XDRM=X4-X3.
130

MECHANISMS
7.2.2 Arcing contact timing

Contact wear can be revealed without determining the exact contact dimensions if
the timing of the arcing switch contacts is compared with a reference. A previous
DRM result for the OLTC in question or a typical result for that type of OLTC
can be used as a reference. The effect of contact wear on the transition time is
small. However, these small variations change the shape and depth of the DRM
plot when the arcing switch operates, due to the large dI/dt of the measurement
current (as already demonstrated in Figure 6.41).
Measurements of artificially worn arcing contacts were performed with the test
model (as described earlier in this chapter) to confirm the effect of contact wear on
DRM. Figure 7.16 provides a typical example of the results of these experiments.
Figure 7.16a shows a close-up view of the OLTC operation using arcing contacts
with 0% wear. The plots change significantly when arcing contacts that were 80%
worn are used (Figure 7.16b). The length of time that both transition resistors are
used in parallel (TP) is shortened, while the length of time that only one transition
resistor is used is extended (TR1 and TR2). The current will drop (from 0.6 A to
0.55 A in Figure 7.16) because one resistor causes a higher dI/dt than two parallel
resistors.
To eliminate variations in the transition time of the test model, measurement
results are displayed as the percentage of the transition time in which only one
transition resistor is used:
CW
TR1 TR 2
*100%
TR1 TP TR 2
7-1
Figure 7.16a Close-up view of a DRM plot at

the moment the OLTC switches. The arcing
contacts are 0% worn.
Figure 7.16b Close-up view of a DRM

plot at the moment the OLTC
switches. The arcing contacts are 80%
worn.
Figure 7.17 shows this percentage for each tap change operation in the OLTCs
cycle of operation. Contact 3 was 80% worn, which is clearly indicated at each tap
change operation involving contact 3. This example demonstrates that small
changes of the contact timing at worn arcing contacts can be measured with DRM.
131

MECHANISMS
60
50
45
40
35
30
2>1
3>2
4>3
5>4
6>5
7>6
8>7
9>8
8>9
7>8
6>7
5>6
4>5
3>4
2>3
25
Variationsdueto
imperfectionsin
thetestmodel
55
1>2
Percentage of transition time

with one resistor Cw (%)
65
Tap change operation
Figure 7.17 Graph of the percentage of the transition time in which only one transition
resistor is used. The average and standard deviation of 6 measurements are displayed for
every tap change operation in the cycle of operation. The contact of tap 3 is 80% worn,
which is clearly indicated in the graph.
Percentageoftransitiontimewithone
resistor,Cw(%)
Equivalent effects were also observed in service-aged OLTCs. Three examples are
provided in Figure 7.18.
Another experiment revealed the effect of extremely worn arcing contacts.
Figure 7.19 shows an example of a roller contact from a selector switch about to
leave a stator contact block. At this moment, two contacts are on the stator block:
the main roller contact (which is about to leave) and the arcing roller contact
(which has just arrived at the stator block). The stator contact block is wide
enough for both roller contacts. When the arcing contacts are heavily worn, this
contact is endangered. This can result in a short interruption of the measurement
current due to the main roller contact leaving the contact block too early.
100
90
80
70
OLTC #1
OLTC #2
Reference
60
50
1
11
13
15
17
19
Tapchangeoperation
Figure 7.18 Three examples of the transition resistor timing in service-aged OLTCs: the
percentage of the transition time in which only one transition resistor is used is plotted. The
transformer windings have two coarse taps, therefore in this example the selector switch uses
the same arcing contacts three times.
132

MECHANISMS
IL
V+V
Figure 7.19 A roller contact at the moment it leaves the stator contact block. The contact
block is wide enough for both roller contacts, but contact wear can endanger good contact
timing.
Experiments with 100% worn arcing contacts mounted in the tap changer test
model generated such interruptions of the measurement current. Figure 7.20
shows a recorded current plot of such an experiment. The arcing contact of tap 3
was artificially worn, resulting in short current interruptions when that contact was
involved in the tap change operation. Figure 7.20b provides a close-up view of the
current interruption.
Figure 7.20a DRM plot of the OLTC test

model. The arcing contacts of tap position 3
are 100% worn.
Figure 7.20b Close-up view of a DRM

plot at the moment the OLTC switches.
The arcing contacts are 100% worn.
Another experiment shows that there is no significant effect of the measurement

current on the transition time, see Figure 7.21. DRM can be used at any current to
determine the timing of the transition resistors.
It should be noted that the wear of the arcing contacts is a slow process which
can be recognised in an early stage during normal inspection of the arcing switch.
There is no specific need for arcing contact diagnosis with DRM for the current
generation of OLTCs; however, the effect of wear may be indicated by DRM and
should be recognised in order to determine the condition of the tap changer.
133
Transition time (ms)
MECHANISMS
36
34
32
30
0
10
15
20
Test current (A)

Figure 7.21 OLTC transition times are determined for an OLTC in good condition, using
different measurement currents. A trend line is added. No significant change in the OLTC
transition time was seen.
7.3 Conclusions
Firstly, this chapter closely examined the relationship between DRM and the
physical condition of tap changer contacts. Dynamic contact resistance was shown
to increase significantly when contacts move over a surface film. This can conceal
more significant contact degradation.
Secondly, this chapter discussed experiments on a tap changer test model with
worn arcing contacts and with a surface film. A measurement current of 1 A was
used. Experiments with surface film on the contacts showed that the early stage of
the long-term effect sometimes causes small irregularities in the contact resistance,
but may not always appear, because a thin layer of oil associated with the early
stage of the long-term effect often breaks down. However, the effect of the surface
film is observed in all measurements when the layer grows thicker. The artificially
grown surface film only causes increased contact resistances when the OLTC is
operated. Service-aged OLTCs with a more advanced long-term effect show that
the static contact resistance also changes significantly when the tap changer is
motionless. From experiments it can be concluded that the long-term aging effect
influences the dynamic resistance. The surface film is measurable with DRM at 1 A
and can be recognised by spikes in the plots when the current-carrying contacts
move over it (which is not the case in every type of OLTC). It was also shown that
a thin surface film can be wiped away when the contacts move over it, reducing
contact resistance. For that reason, reproducible irregularities in plots are more
severe than irregularities whose resistance decreases when the OLTC is switched a
few times though its cycle of operation. It was demonstrated that the maximum
contact resistance measured with DRM can indicate the stage of development of
the surface film. A thick surface film on the contacts indicates an advanced longterm aging effect, while coking may have occurred on locations where the load
134

MECHANISMS
current has heated the contacts. Previous results measured on the same tap
changer should be considered in order to estimate the length of time that the
advanced long-term effect has been present, leading to the development of
pyrolytic carbon on the contacts. Historical information of the OLTCs tap
positions being used can also be used to estimate the amount of carbon and film
on the contacts, using the experience that infrequently used contacts are likely to
have a contact film and that frequently used contacts are sensitive to contact
coking. Further study may include research into the effect of coking on contact
resistance.
Thirdly, the experiments confirmed that arcing contact wear influences contact
timing, changing the shape and amplitude of the effect of the transition resistors in
the DRM plots. The timing of the transition resistors showed no significant
dependency on the amplitude of the measurement current. Interruptions of the
current can appear in the plots when excessively worn contacts are tested. Contact
wear of the arcing contacts can also be detected by:
1
2
comparing the timing of the transition resistor during DRM with

earlier measurements on the same OLTC
DRM in combination with a shaft angular position measurement
135
CHAPTER 8 - DRM KNOWLEDGE RULES FOR DECISION SUPPORT
8
DRM knowledge rules for
decision support
Tap changers which are in critical condition and will best profit from efforts to
reduce failure (e.g. maintenance) can be selected on the basis of the dynamic
resistance measurement results. DRM is particularly suitable for assessing OLTC
degradation processes and defects that potentially lead to failure, and therefore on
this basis can be used to reduce the failure rate.
Such huge amounts of diagnostic data are acquired with the aid of DRM that it
is difficult to pick out relevant information about the condition of the tap changer.
The diagnostic data must therefore be transformed into a statement about the
condition which can be done using derived quantities and boundary values. In
other words, these results require interpretation before they can be used for the
strategic prioritisation of maintenance or the overhaul of the OLTC. For that
purpose, a simple way of expressing the results of DRM will be beneficial. An
example of the application of this process will also be presented.
8.1 Condition assessment

This chapter elaborates on the technical aspects of a specific class of asset
management decisions, namely condition assessment based on DRM at the
component level. Selecting the optimal maintenance strategy on the basis of
information about the reliability of the tap changer and its life-cycle costs involves
more than implementing the diagnostic method described in this chapter, as
information about an OLTCs technical condition is only one input in the
maintenance decision process. Figure 8.1 shows that an OLTC which is subject to
degradation mechanisms and defects can be diagnosed by performing
measurements with a diagnostic instrument. A selection of diagnostic methods can
be applied in such a way that the symptoms of degradation which occur most
frequently are diagnosed (see Chapter 5).
137

OLTCs in critical condition can
be selected for overhaul or repair
using the results of a single
measurement or from successive
measurements over time. In this
way, the transformer population
can be maintained cost effectively
and with a certain reliability.
Economical, societal and
safety aspects are not considered
in this chapter. In particular,
additional information such as
operational data (temperature,
load profile), failure statistics and
the results of visual inspection
must also be considered, see
Figure 8.1. Furthermore, the cost
of failure, the consequence of this
failure for the surrounding power
transformer and the continuity of
power delivery should also play
an important role.
8.2 DRM knowledge rules

Residual life estimation and
decisions
for
intrusive
maintenance actions are often
based on fuzzy information [92].
The interpretation of diagnostic
data, as introduced in Figure 6.1,
can be structured by using
knowledge rules to extract
information about degradation
and defects from the diagnostic
data (see Figure 8.1). Criteria for
interpretation are described as
knowledge rules [93], which are
derived in order to support the
condition-based
maintenance
process [93]. Parameters from the
DRM plots that reflect the OLTC
condition (see Figure 8.2) are
used as the basis for this process.
Boundary values can then be
138
Figure 8.1 Assessment of OLTC degradation for

the determination of an optimal maintenance
strategy. Symptoms of degradation are measured
and interpreted, which results in a selection of
higher risk OLTCs. Knowledge rules are used
during the localisation of the defect to find its
cause and determine its significance.

applied to categorize results in a condition index.
In general, knowledge rules are based on degradation mechanisms on the
different OLTC parts and the effect of these failure modes on the DRM results.
Knowledge rules are based on the following facts:
findings from visual inspection and DRM on service-aged OLTCs during

regular on-site maintenance
laboratory studies with artificial degraded OLTC components
post-mortem investigation of tap changer degradation processes
All these sources of knowledge were considered in this thesis, resulting in the
application example of Section 8.5. In particular, collecting and analyzing
measurement data is important in the process of knowledge rule generation.
In order to determine the parameters that reflect the OLTCs condition, the
overview of Table 8.1 is generated, which summarises the directives for
interpretation that were discussed in this thesis and the pages where they were
introduced or discussed, structured according to the condition assessment
flowchart of Figure 6.1.
Figure 8.2 Parameters that reflect the OLTC condition can be used to extract condition
information from dynamic resistance measurement results. Boundary values are needed
for this purpose.
Table 8.1 Overview of major DRM knowledge rules from this thesis and the page where they
are discussed.
Directives for DRM interpretation
General
1.1
Deviation from the standard plots can be related to degradation or defects and
should be investigated before re-energising the transformer.
OLTC failure can be prevented this way.
1.2
The interpretation of results should not only consider the maximum resistance:
the location and nature of the resistance variation should also be taken into
account. The location and shape of the irregularity can be used to determine the
cause and importance of the defect.
Degradation mechanisms have different effects on contact resistance; separate boundary
values should be used.
1.3
The current pattern can be read backwards to recognise tap change operations.
This helps to determine the defective tap position(s).
Localization
2.1
The change-over selector, the arcing switch and the drive system can be diagnosed
separately.
Because DRM provides information about transition times, contact timing, contact
resistance, open contacts and location (based on the location in the DRM plot).
Page
93
Page
85
Page
97
Page
85
139

2.2
Irregularities in the plots that are caused by the change-over selector appear,
change or disappear when it is operated.
2.3
Irregularities in the plots that are caused by the arcing switch appear, change or
disappear when it is operated.
2.4
The change-over selector is likely to cause changes in the plots that are less
frequent but persist longer than those caused by the arcing switch.
Because the change-over selector is used less frequently.
2.5
The absence of a linear current pattern can be caused by the arcing switch and the
change-over selector.
Because the arcing switch is used during every tap change, which can influence the resistor of
the change-over selector due to mechanical vibrations.
2.6
The interruption time and its location in the cycle of operation provide
information about the location of the interruption.
Each contact is used at particular locations in the OLTC cycle of operation and is used a
particular time.
2.7
The arcing switch is likely to cause problems if the irregularities in the plot that
occur when the change-over selector is motionless are out of proportion
compared to the irregularities that occur when the change-over selector is
operated.
2.8
A current interruption at a delta-connected transformer can manifest as a
decreasing measurement current in another phase.
Because delta-connected windings offer an alternative path for the current if one phase is
interrupted.
Cause (degradation and defects)
3.1
The presence of a linear pattern in the plot points to an arcing switch with no
long-term aging effect.
The arcing switch is operated every tap change, so it will change the contact resistance
during the entire DRM.
3.2
Change-over selector contact areas that are only used briefly during OLTC
operation will not be pitted and are not covered with pyrolytic carbon.
See the following knowledge rule.
3.3
Frequently used contacts that carry the load current can be subject to contact
coking.
Contact pitting and coking occurs when the load current flows through the contacts for an
extended period.
3.4
The surface film is measurable at 1 A and can be recognised by spikes in the
resistance when the current-carrying contacts move over it.
See page 134.
3.5
Carbonised oil on the contacts is measurable.
As observed during on-site measurements.
3.6
Contact resistance can decrease during the long-term aging effect: the surface film
breaks and pitting can occur.
A low contact resistance is not enough to conclude that the OLTC is in good condition: the
location of the defect and the degradation mechanism should also be taken into account.
3.7
A thin surface film can be wiped away when the contacts move over it, reducing
the contact resistance.
This can be used for testing the development stage of contact degradation and to clean the
contact surface.
3.8
The surface film grows thicker as the contacts age and this oil-film growth rate is
strongly dependent on the surface temperature of the contacts.
Therefore a thick film with high resistance indicates a medium or advanced stage of contact
degradation.
3.9
A thin surface film is not detected in every DRM, because the thin layer breaks
and the conductive path is restored, while a thick layer will be detected.
Therefore a thick film appearing in every DRM indicates a medium or advanced stage of
contact degradation.
140
Page
95
Page
106
Page
106
Page
116
Page
105
Page
103
Page
111
Page
107
Page
99
Page
100
Page
134
Page
35
Page
23
Page
134
Page
181
Page
127

3.10
Low measurement voltages and low currents will allow a very thick surface film to
interrupt the current. The interruptions disappear when the current is increased.
This can be used for testing the development stage of contact degradation.
3.11 A short increase in the measurement current during OLTC operation can be
caused by asynchronous operation of the OLTC phases.
Importance
4.1
The distinction between increased contact resistance and open contacts is
important: in general interruptions are more critical.
Because an OLTC with open contacts during its cycle of operation may not be able to select
some taps, an interruption of the load current may be destructive.
4.2
Single peaks in the measurement current are caused by noise in the measurement
circuit.
Rapid current changes cannot flow through the windings due to the high inductance.
4.3
The worst tap position of all phases determines the condition index of the tap
changer.
Because one tap can be used for a long time and one defective tap can cause failure of the
OLTC.
4.4
The importance of contact degradation can be investigated by performing DRM
with different measurement currents.
Contact resistance becomes smaller when the current is increased and sensitivity to surface
film on the contacts increases when the measurement current is decreased. A low current
may be desired if one wants to assess the importance of the long-term effect.
4.5
Short measurement current interruptions when the arcing switch operates are not
always critical.
They can be caused by a thin layer of oil that has to be pushed away by the moving contacts
before a low-resistance contact can be made. The oil layer will not cause current
interruptions in normal operation.
4.6
The way in which measured results change with time can provide valuable
additional information about the importance of contact degradation.
A lower condition index can be assigned when the irregularity in the DRM plots increases
in time.
4.7
Reproducible irregularities in plots are more severe than irregularities whose
resistance decreases when the OLTC is switched a few times though its cycle of
operation.
Because light contact degradation can be wiped away by moving contacts.
4.8
An advanced long-term effect on the OLTC contacts will appear in every DRM.
Because advanced contact degradation cannot be easily wiped away by moving contacts.
Page
57
Page
110
Page
93
Page
64
Page
111
Page
59
Page
112
Page
151
Page
134
Page
127
8.3 Quantities reflecting the OLTC condition

In general, diagnosing the condition of high-voltage components can be done
using characteristic quantities which reflect the condition of the component,
thereby reducing the complex measurement result to a set of numbers. For
example, discharge measurements on insulating systems use a partial discharge
inception voltage (PDIV) as a quantifier, and dissolved gas analysis on insulation
oil uses gas ratios to summarise the results of gas analysis. Boundary values can be
applied to these quantities, see also Figure 8.2. These quantities should be carefully
selected because defects that do not influence these quantities will no longer be
revealed after quantification. In the case of DRM, the following quantities will
cover the most typical OLTC defects and degradation mechanisms (see also Table
8.2):
141

1 OLTC transition time
Measured as the time that the transition resistors are used. This
quantity is sensitive to contact misalignment, mechanical defects and
friction, see also Section 6.4.6.
2 Timing of the transition resistors
This quantity expresses the timing of the transition resistors in relation
to the transition time and is therefore also sensitive to contact timing
differences (for example due to contact wear, see also page 131).
3 Timing difference between phases
This quantity can be used to check synchronism between the three
phases. The final time of the OLTC operation can be used because it is
easy to recognize in dynamic resistance plots.
4 Time that the measurement current is interrupted
This quantity is defined as the time that the measurement current is
below 10% of its initial value, see also Section 8.5.1. It is sensitive to
misaligned contacts, defective transition resistors, a thick film on the
contacts, advanced long-term aging, mechanical defects and to contact
wear that causes incorrect contact timing.
5 Maximum contact resistance when using infrequently used contacts
An example of such a contact occurs when the change-over selector is
in operation. A film can develop on these contact locations, as
described in Section 2.3.2, but thermal degradation described in Section
2.3.3 is not present here; this parameter represents the thickness of this
layer and therefore the stage of development of long-term aging, in
particular for light contact degradation.
6 Maximum contact resistance when OLTC is using frequently used contacts
Contacts through which the load current flows are more sensitive to
contact coking and less subject to the development of a surface film.
Therefore these contact locations provide information about the
advanced stage of the long-term aging effect, which involves pyrolytic
carbon, as described in Section 2.3.3.
Table 8.2 shows how these quantities are calculated and which knowledge rules
from Table 8.1 should be considered. In addition, Figure 8.3 shows how the
quantities can be calculated from a DRM plot. These quantities can be used as
input for condition assessment to reveal typical OLTC defects, as depicted in
Figure 8.4. DRM results can be checked for problems that affect the transition
time, contact timing, phase synchronisation, contact resistance or that interrupt the
current. This way, the condition of the tap changer can be determined using
boundary values, as will be discussed in the next section.
142

Table 8.2 Quantities derived from DRM which represent the OLTC condition, including the
OLTC defects to which they are sensitive, the way they are calculated and the knowledge
rules from Table 8.1 that must be considered.
Quantity:
Sensitive to:
1 Transition time
Contact misalignment
Defective drive mechanism
Friction
Contact wear
Timing of the transition Contact misalignment
2
resistors
Damaged transition
resistors
Timing difference
3 between
phases A, B and C
between phases
Contact wear
Time that the
Damaged transition
4 measurement current is
resistors
interrupted
Advanced long-term aging
effect
Maximum contact
resistance when the
5 OLTC uses
infrequently used
contacts
Maximum contact
resistance when the
6
OLTC uses frequently
used contacts
Long-term aging effect:

surface film
Long-term aging effect:

pyrolytic carbon
Calculated by (see
Figure 8.3):
Knowledge
rules:
TTR (ms)
1.1 1.3 2.3 2.4

2.8 3.10 4.2
4.3
TR1 +TR2
*100 (%)
TTR
1.1 1.3 2.3 2.4

2.8
4.1-4.3 3.10
3.11
max TEND-A ,TEND-B ,TEND-C

min TEND-A ,TEND-B ,TEND-C
(ms)
1.1
2.1-2.4 2.8 3.4
3.10
4.1-4.5 4.7 4.8
TI<10% (ms)
Iexp -Imin
Iexp
I high -Ilow
I high
*100
*100
(%)
(%)
1.1-1.3
2.1 2.2 2.4-2.7
3.1 3.2 3.4 3.6
3.7 3.9 3.10
4.1-4.4 4.64.8
1.1-1.3
2.1
2.3-2.7
3.1-3.6 3.8
4.1-4.3 4.64.8
8.4 Condition indexing

It is important that condition assessment is as objective and consistently applied as
possible. Ideally, a scoring system would be used [40] to model the continuous
degradation process as a discrete process. Condition indexing can be used for this
purpose, based on the derived quantities described in the previous section.
Condition indexing is the process of assigning limited index numbers to the
condition of the tap changer. These condition indices provide fast, simple and
systematic information about the condition to the asset manager. For example, the
OLTC subsystem condition can be described in terms of Good, Aged and Bad.
Ref. [94] uses the categories Not aged, not worn out, Aged and Worn out
(noticeably) to classify the condition.
143

Quantity 5:
Infrequently used contact surface
Quantity 6:
Frequently used contact surface
Iexp
Ihigh-Ilow
Imin
Ihigh
Ilow
100%
TEND-X
(X=A,B,C)
10%
0%
TI<10%
Quantity 4:
Test current interruptions
TP
TR1
TR2
TTR
Quantity 1-3:
Transition time and transition resistor timing
Figure 8.3 Quantities that reflect the OLTC condition can be used to extract condition
information from DRM results. This figure shows a set of quantities that can be used to
reveal the most common OLTC defects.
To apply condition indexing, boundary values are needed to convert the selected
continuous parameters from DRM (see Section 8.3) into a few categories. Section
8.5 describes an application example of these boundary values, based on the
relationship between DRM results and visual inspection of the OLTC contacts.
Statistical analysis on derived quantities could also be used to derive norm values,
see [95]. If the boundaries are too conservative, more false positive conclusions
will be drawn, unnecessarily high maintenance expenditure may result and the early
replacement of the OLTC components may occur. However, if the boundaries are
too tight, unreliable tap changers may remain undetected (false negative
conclusions), which can result in serious damage to the transformer. It is very
difficult to decide when a fault is serious enough to justify radical action [40].
144
Figure 8.4 The quantities from Table 8.2 can be used to check DRM results for typical
irregularities in a structured way. Transition times, contact timing, phase synchronism, open
contacts and contact resistance can be checked this way.
Degree of truth, (D)
Information about condition is lost during the process of condition indexing.

However, this final step of condition assessment is advantageous because of its
simplicity: the condition index allows different OLTCs to be easily compared,
critical OLTCs to be selected and decisions about the maintenance strategy to be
made.
The relationship between DRM results and condition indices can be expressed
as a membership function. Membership functions have a value between 0 and 1
that expresses the degree that a continuous parameter from the diagnostic data is a
member of a linguistic variable. This value is called the degree of truth. The
linguistic variables that assess the condition are by definition vague, for example
Good, Aged and Bad. This allows the condition of the OLTC to be described
by a more human classification system, which makes it suitable to present the
results to an expert. Figure 8.5 shows an example in which numerical condition
indices are used. The condition is described by condition Good, Aged and Bad.
Fixed boundaries are used, based on the deviation of the measurement current in
the DRM results. Limits for condition indexing can be obtained by:
Figure 8.5 Example of deriving condition indices from the deviation in the DRM plots,
using fixed boundaries for the membership function.
145
A study of service-aged components (see Section 8.5)

o during overhaul
o after failure
A statistical analysis of historical diagnostic data (see [95])
Less information is lost during condition indexing when the result is expressed as a
fuzzy membership function. Another advantage of fuzzy membership functions is
apparent when DRM is performed on large populations of OLTCs. Many
engineers are involved in the process of taking these measurements, and therefore
different interpretations of measurement data are inevitable when no strict
flowcharts for condition indexing are used. If a comparison of condition indices is
required, these decision flowcharts should be based on precisely prescribed
quantities.
Figure 8.6 provides an example of a fuzzy membership function, where the
condition is defined by fuzzy logic using trapezoidal membership functions. It
gives the degree of truth of the deviation in dynamic resistance results for the three
linguistic variables (in this example Good, Aged and Bad). For example, a
measurement result plot with a certain irregularity can have a degree of truth of 0.7
to the variable Aged and 0.3 to Bad.
Figure 8.6 Example of a membership function of the deviation in DRM plots for the
linguistic variables Good, Aged and Bad.
8.5 Application example

The condition indexing of DRM results for OLTCs is type dependent, as each type
has a different sensitivity to degradation mechanisms. In general, a faulty condition
is expected when the contact resistance is increased 5-10 times the initial value of
contact resistance [40].
This section describes the application of DRM at a Dutch utility [96], where
condition indexing was used to assess the contact condition of a large number of
OLTCs. In this example, parameters from DRM which reflect the tap changer
condition are combined with experience from laboratory experiments, failure
studies and knowledge about the degradation processes (as shown in Figure 8.2).
The condition indices A, B and C are used to indicate the condition,
corresponding with Bad, Aged and Good as described earlier. To begin with, a
distinction is made between:
146

1) OLTCs with open contacts
2) OLTCs with an increased contact resistance
(8.5.1)
(8.5.2)
8.5.1 Open contacts

The tap changer has open contacts when the measurement current is interrupted
during DRM. The duration of the interruption and the location in the plot are used
to determine its source (see Section 8.3). The boundaries from Table 8.3 are set as
follows:
200 ms is the maximum transition time of the OLTCs in the population

Short interruptions as observed in this population are normally less than 3 ms,
10 ms is taken as the boundary
20 s is the maximum expected time between two operations and depends on
the motor operation time of the drive mechanism and time delay relays
An open contact is defined here as a measurement current that drops below
10% of the initial current. This limit is required because the zero current of
short interruptions is not always sampled at a 1 kHz sampling rate. The slowly
increasing current will be sampled before it reaches the 10% limit again.
Condition indexing is uncomplicated in the case of open contacts: all contacts

which remain open for longer than 10 ms are given condition index A
(bad/critical). Shorter interruptions of the measurement current are given
condition index B aged) when roller contacts are used and condition index C
(good) when contact blocks are used. Figure 8.7 can be used in the case that fuzzy
logic is preferred.
Figure 8.7 The relationship between the interruption time of the DRM measurement current
and the OLTCs condition can be displayed as a fuzzy membership function.
147

Table 8.3 The interruption time measured with DRM versus the location of the interruption.
Boundary values are applied to find the cause of the interruption.
Interruption time
tI<10% < 10 ms
10 ms tI<10% < 200 ms

200 ms tI<10% < 20 s
TI<10% 20 s
Location of the interruption

Arcing switch operates
Change-over selector moves
A harmless layer of oil can be The long-term effect is measured;
measured when the arcing switch most likely a thick surface film.
uses contact blocks to establish a
connection.
When roller contacts are used,
the arcing contacts of the stator
can be worn.
Defective transition resistors.
The long-term effect is measured;
most likely a thick surface film.
Change-over selector problems are
measured covering a single tap
position.
Open contacts inside the changeover selector are measured
covering multiple tap positions.
8.5.2 Contact resistance

Tap changers with an increased contact resistance can show irregularities in the
dynamic resistance results; the location of the degradation or defect is determined
by observing the moment the irregularity does (or does not) occur. The timing of
resistance changes can be related to the operation of specific contacts, a process
requiring knowledge of the OLTC construction and mechanical operation.
Answering the following questions can determine the location:
A. Does the resistance change when the change-over selector moves?
B. Does the OLTC have higher contact resistance when the change-over selector
is motionless?
C. Is there a linear pattern in the dynamic resistance plot?
Table 8.4 shows the relationship between contact resistance and the location and
type of defect. Again, answering three questions can help:
148
Do the contacts on that location carry current during normal service

conditions and are therefore prone to contact coking?
Are the contacts on that location switched often and therefore less prone to
contact coking?
Is the resistance constant for the duration of the tap position, or are spikes
present in the measured resistance?

Table 8.4 Irregularities in the DRM current pattern can be localised by 1) examining at
which moment it occurs and 2) whether there is a linear section in the DRM plot.
A
Resistance deviation
when change-over
selector moves?
No
Yes
Yes
Yes
Yes
B
Resistance deviation
when change-over
selector is
motionless?
Yes
Yes
Yes
Yes
No
Resistance
deviation
B A?
Yes
Yes
No
-
C
Linear
pattern
present?
No
Yes
-
Indication of the
location
Arcing switch
Change-over selector
Depending on the location of the defect, the aging mechanisms and the type of
OLTC, different boundary values are required. As an example, a flowchart for the
condition indexing of change-over selector contact degradation has been made, see
Figure 8.8. Table 8.4 should be used prior to this flowchart, thereby taking into
account the location of the defect. This example uses condition indexing based on
the deviation from the expected value, using the maximum of parameters five and
six from Table 8.2. The four questions from Table 8.4 are included in this
flowchart (top). The expected value in a DRM plot cannot be determined when
the results are extremely distorted. In such cases maximum and minimum values
are used as an approximation. Figure 8.8 shows the flowchart for the condition
indexing of the change-over selector using DRM at 1 A. Boundaries are
determined on the basis of the selectors physical condition as seen during
overhaul or after failure [11], when the contact condition can be compared with
the measurement results.
The 10% boundary is used to exclude noise, and the 40% limit to distinguish
between aged and critical contacts. This 40% boundary is based on observations
made during the overhaul of the OLTC or after failure: coking and critical pitting
are present in the population under consideration when the deviation due to the
long-term effect exceeds 40%. To determine this boundary value, eight case studies
were selected, see Table 8.5. These boundary values were used for all contact
materials: silver-plated contacts were diagnosed
Table 8.5 Eight case studies of OLTCs at an advanced stage of development of the longterm aging effect. The maximum deviation of the DRM measurement current with respect to
the expected value is given, together with the contact status as observed during visual
inspection of the change-over selector contacts.
Case 1
Case 2
Case 3
Case 4
Case 5
Case 6
Case 7
Case 8
DRM current deviation

25%
35%
40%
45%
50%
60%
80%
100%
Visual inspection of OLTC contacts

About to be critical
Not critical
Critical
About to be critical
Critical
Critical
Critical
Failed
149
150
Does the current

deviate from the
expected value?
DRM result
(measured at 1ADC)
No
Yes
condition index C
Yes
Yes
condition index B
No
Deviation <10% of
the initial value?
Use deviation from

expected value
No change-over selector
defects measured
Deviation when
change-over selector
is motionless?
No
Yes
No
Deviation when
change-over selector
is moving?
Yes
condition index C
No
Yes
Linear pattern
measured?
Yes
condition index A
No
Deviation <40% of
the initial value?
No
No
No
Trending?
Yes
condition index B
Use maximum minimum value
Deviation <10% of
the initial value?
Use deviation from

expected value
Trending?
Yes
Figure 8.8 Example of the derivation of condition indices for the change-over selector from
the irregularities in the DRM plots at 1 A, based on the questions and boundaries discussed
in this section.

using the same decision rules as non-silver-plated contacts. The way in which the
results change over time can also provide valuable additional information [40].
Therefore, the contact deterioration was compared with previous results and a
lower condition index was assigned when the irregularity in the plots increased
over time.
Condition indexing based on fuzzy membership functions can be applied to
this application example as well, using the same boundary values; in that case
contact degradation in the change-over selector can be assessed by the
membership function of Figure 8.9 for degradation that is not changing over time.
The membership function shifts when degradation increases over time, see Figure
8.10.
Figure 8.9 Example of a fuzzy membership function for assessing the condition of the
change-over selector in case the degradation is not increasing over time, based on DRM at
1 A.
Figure 8.10 The membership function of Figure 8.9 shifts when the degradation increases
over time.
If there is insufficient failure data available for a particular type of OLTC, it is

impossible to make a good prediction of its end-of-life [97]. In such cases, a
statistical analysis using the DRM database can be used to determine boundary
values [95], because degradation and aging mechanisms are statistical in nature [97].
If the cause of irregularities in the DRM results is not immediately clear, additional
measurements can be performed:
to determine if the results are reproducible

to determine if the contact degradation in the form of a surface film can be
wiped away, by switching the OLTC through its cycle of operation a few times
to determine whether the resistance due to degradation is dependent on the
current (attention should be paid to not heating the contacts), by using
different measurement current amplitudes
151

8.5.3 Flowchart verification
2000
1995
1990
1985
1980
1975
1970
1965
1960
1951
Number of OLTCs
To verify the flowchart of Figure 8.8, it was applied to the OLTCs of a Dutch
utility. The utility used DRM as a method to pre-select critical tap changers, which
would then undergo additional measurements, and to detect maintenance errors.
This population of tap changer, 455 in total, consisted of 33% diverter switch type
OLTCs from manufacturer A and 52% selector switch type OLTCs from
manufacturer B (the remaining 15% is from different producers, mostly diverter
switch type OLTCs). Most of the transformers were 50/10 kV and the average
power rating of the population was 39 MVA. Figure 8.11 shows the condition
indices of the total population in relation to the year of installation. It shows that
the flowchart of Figure 8.8 correctly does not classify the relatively new OLTCs as
B (aged) while tap changers older than 20 years did show irregularities in the
DRM plots and were assigned a condition index B (aged) or A (bad).
Figure 8.11 DRM results (after condition indexing using the flowchart of Figure 8.8) for the
total population of OLTCs, sorted by the year of installation.
Figure 8.12 shows the subpopulations selected and their average ages. OLTCs are
grouped according to their condition, their type of contact and contact material. A
special subpopulation was created to compare OLTCs with roller contacts to
OLTCs using snap contacts (old snap contacts), ensuring that the average age of
152

both subpopulations was equal. Differences in design are seen when the flowchart
of Figure 8.8 is applied to the DRM results from this population, see Figure 8.13.
It can be seen that tap changers using roller contacts at the change-over selector
show more effects of long-term aging than tap changers using snap contacts at the
same location. Change-over selectors using roller contacts also have the highest
number of critical condition indices.
Figure 8.12 Overview of the year of installation of each selected OLTC subpopulation.
The most striking contrast is seen in Figure 8.13 when comparing the type of
contact material used at the change-over selector. According to the flowchart of
Figure 8.8, the copper-brass contacts had aged to an extreme degree compared
with the silver-plated contacts. The flowchart successfully assessed this effect
because:
OLTCs using copper-brass contacts were older (average year of construction

1968 as opposed to 1981)
OLTCs using copper-brass contacts had no overhaul, while some tap changers
with silver-plated change-over selector contacts were silver-plated during their
service life (and therefore contact degradation was removed)
OLTCs using copper-brass contacts are subject to a more rapid long-term
aging effect on the change-over selector [24]
The flowchart from Figure 8.8 takes into account contact deterioration over time
by looking at the difference between latest measurements and those taken
previously. It was seen that about 50% of the tap changers had degraded since the
previous measurement and on this basis were assigned a lower condition index.
153

Condition indices
100%
75%
Not measured
Condition 'A'
50%
Condition 'B'
Condition 'C'
25%
0%
Roller
contacts
Old snap
contacts
Copper
brass
Silver
plated
Subpopulation
Figure 8.13 Overview of the condition index as determined by the flowchart of Figure 8.8 for
each OLTC subpopulation selected from the population studied.
When the flowchart of Figure 8.8 is applied to the OLTCs of Table 8.5 it can be
seen that 7 out of 8 cases is categorised according to results from visual inspection.
Only case 4 is categorised as A (bad) because it has a measurement current
deviation of 45% while the boundary value of Figure 8.8 is 40%, which is incorrect
because visual inspection showed that the contacts were about to be critical, but
the OLTC could remain in service for some more years (a false positive
conclusion). The fuzzy logic approach of Figure 8.9 and Figure 8.10 circumvents
this issue when relating DRM results to visual inspection results.
The flowchart of Figure 8.8 was verified in more detail by selecting:
one OLTC that was categorised as C (good, 0% measurement current

deviation)
one categorised as B (aged, 24% measurement current deviation, not
trending)
one of category A (bad, 48% measurement current deviation, not trending)
The contact resistance of the change-over selector was determined at 100 A for
each position of the change-over selector. DRM was performed and the deviation
of the current from the expected value was calculated at the same tap positions as
for the static measurement. Figure 8.14 shows the static resistance versus the
dynamic resistance for the investigated tap position of these OLTCs. The trend
line is added, including 80% confidence bounds. It can be seen that the flowchart
assesses the good OLTC correctly: the contacts with no irregularities in the DRM
results also have a low static contact resistance (~5 m) and group nicely near
zero. Good contacts are also easily distinguished from aged and bad contacts,
which have higher static and dynamic contact resistance. Finally, the trend line
shows a light statistical relationship between static and dynamic resistance, but this
154

relationship is too weak to predict the static resistance based on the dynamic
resistance and vice versa.
Figure 8.14 Static resistance at 100 A versus dynamic resistance at 1 A. Five tap positions
were measured at three OLTCs in different aging staged. The trend line and the 80%
confidence interval are drawn.
8.6 Conclusions
This chapter discussed how to deal with dynamic resistance measurement results
from OLTCs and demonstrated how these results can be represented as condition
indices using a flowchart that categorises them using boundary values. It was
shown how the technical condition can be determined by applying boundary
values to derived parameters from DRM graphs. By doing so, a set of critical tap
changers can be selected that will benefit from failure-reducing efforts (e.g.
maintenance). This chapter also concluded that DRM alone does not provide
enough information to determine an optimal maintenance strategy for a population
of OLTCs, other sources of information, such as results from visual inspection,
should be considered as well.
Firstly, knowledge rules were derived from different sources, such as field
measurements, laboratory experiments and failure investigation. On this basis a list
of DRM knowledge rules was provided.
Secondly, the chapter explained which quantities that are determined from
dynamic resistance results will cover the most common defects.
Thirdly, it was shown that with the use of a condition indexing flowchart it is
possible to convert complex diagnostic information into simple condition indices,
which depends on the type of tap changer being measured. The categories
resulting from a condition indexing flowchart can be expressed in vague linguistic
variables or as clear numbers using fixed boundaries. Information is lost during
this process of condition indexing, but the simplified representation can be an
advantage for the asset manager when making decisions about OLTC maintenance
or replacement.
155

An example of a condition indexing flowchart was also presented and its
application to a population of OLTCs discussed. Clear differences between tap
changer contact materials and transformer age were seen; it was shown that type of
contact under consideration has a relationship with the condition indices.
156
CHAPTER 9 - CONCLUSIONS & RECOMMENDATIONS
9
Conclusions & recommendations
This thesis has explored contact resistance measurements for the diagnosis of
high-speed resistor-type on-load tap changers (OLTCs), extensively used in highvoltage power transformers. The feasibility of using dynamic resistance
measurements (DRM) for the pre-failure detection of frequently observed
degradation mechanisms and defects has been investigated. Also, the additional
value of performing DRM after maintenance has been demonstrated in relation to
its capacity to detect the different defects.
9.1 Conclusions
In line with the research goals, the conclusions of this thesis are summarised as
follows:
1. Tap changer degradation
The on-load tap changer showed to be responsible for a major part of the
transformer failures. Dominant failure modes include contact degradation,
mechanical defects and maintenance errors.
Contact resistance models from literature have been discussed and corrected
to make them suitable for resistance measurements on tap changers, thereby
compensating for the contact operation and for the amplitude of the
measurement current.
2. Diagnostic methodology for on-load tap changers
No relationship between the electrical condition of the change-over selector

contacts and dissolved gas-in-oil amount (or gas ratios) has been found for the
dataset selected.
157
A transformer diagnostic system has been derived by combining several

diagnostic measurements for power transformers, i.e. measurement of the
transformer turn ratio, static and dynamic resistance, drive motor power and
drive axis rotation. The proposed measurements are advised in combination
with visual inspection and dissolved gas analysis.
Dynamic resistance measurements assess a larger contact area than static

measurements and more degradation mechanisms are detected, e.g. mechanical
defects and contact wear.
This thesis discussed dynamic resistance measurements on the transformer

windings with OLTCs (a direct measurement) and on the windings without
OLTC (an indirect measurement). Indirect measurements are more sensitive to
rapid changes in the contact resistance but the results are dependent on the
transformer ratio. Direct measurements are safer because a secondary shortcircuit is applied.
3. Sensitivity verification
DRM has been found to be a dedicated method for :

i. Assessing the basic function of an OLTC in a direct and nondestructive way, because it is sensitive to current interruptions
that may lead to failure.
ii. Discovering contact resistance caused by different kinds of
contact degradation that may reduce the reliability of operation.
iii. Assessing the condition of parts which are not accessible for
inspection during regular maintenance.
Therefore, this method can be used to detect a large variety of defects and
degradation mechanisms before failure can occur, in particular in combination
with regular tap changer maintenance.
DRM can also be used for localisation of defects. Localisation is important for
risk assessment.
Laboratory experiments with artificially degraded tap changer contacts show

no sensitivity for non-critical thin contact films at 1 A. A thick film can be
detected. The measurement of medium contact degradation is inconsistent at
1 A and the film may be wiped away by the moving contacts.
4. Dynamic resistance measurement parameters
158
Resistance measurements are more sensitive to surface film and light contact
degradation if a low measurement current is used. The initial stage of longterm contact degradation, leading to contact coking, is then revealed, which
makes DRM suitable for pre-failure diagnostics. Results measured at low

current can vary due to the tendency of the contact film to break. Therefore, a
trade-off should be made.
Contact resistance measured at different current amplitudes should be

corrected before they are compared, due to the non-linear relationship
between current and resistance, in particular at degraded contacts. Contact
resistance does not resemble the values at nominal current. A correction factor
has been introduced to compensate for this effect.
The measurement current at a tap changer in a delta-connected winding

configuration also flows through the parallel phases. A 50% higher
measurement current should be used to compensate for this effect in cases
where results must be compared with, for example, star-connected tap
changers.
A measurement circuit with higher circuit resistance results in smaller changes

in the measured current, and therefore a lower signal-to-noise ratio, but a
higher sensitivity to rapid resistance changes is obtained.
International standards for power transformer measurements do not specify

the exact measurement parameters for resistance measurements: most practical
parameters can be freely chosen, but standardisation would be beneficial
because contact resistance depends on the amplitude of the measurement
current.
5. Interpretation of DRM results
The measurement results can be subdivided into those showing:

i. no irregularities
ii. the effects of contact resistance
iii. open contacts
iv. irregularities in contact timing (e.g. phase synchronisation,
transition times or timing differences due to wear).
Due to this categorisation a more systematic interpretation is possible.
DRM plots can be interpreted by comparing them with reference

measurements. The following procedure provides a more advanced
interpretation:
i. Localisation of the defect the arcing switch, change-over
selector and drive system can be separately diagnosed
ii. Determining the cause the degradation mechanisms on the
specific location should be evaluated
iii. Determining the importance interpretation of results is type
dependent
159
Results can be summarised as a condition index, as is demonstrated in this

thesis. On the basis of the dynamic resistance results, criteria which can be
used to reveal the most important defects were determined; boundary values
can be applied to these quantities during interpretation. These boundary values
depend on the type of tap changer and the measurement parameters.
Condition indexing using flowcharts is beneficial due to its simplicity and
objectivity in supporting management decisions.
The following characteristics of dynamic resistance results should be

interpreted to ensure that the transformer can be re-energised safely:
i. A dynamic resistance overview plot of all tap positions
ii. A dynamic resistance close-up plot of each tap changer operation
iii. The transition times of each tap changer operation
All these characteristics should be in correct state, considering the voltage and
current that were used for the measurements
It is recommended that dynamic resistance measurements should be carried

out on all on-load tap changers directly after maintenance and that the results
of these measurements should be interpreted before the tap changer is
returned into operation.
9.2 Suggestions for future work
Additional accelerated degradation experiments with tap changer contacts

should be performed to confirm the boundary values for the condition
indexing of DRM results, as were derived in this thesis using the relation
between contact resistance and the observed degree of degradation. Criteria
should be derived for each type of tap changer and degradation mechanism.
The distinction between coked contacts and contacts with a thick contact film
should be further investigated to make a better distinction between aged and
critical contacts. An overhaul could be used to obtain better insight into the
location of excessive contact resistance: the behaviour of contact resistance at
different aging stages can be further investigated by cleaning the change-over
selector step by step, measuring each phase of degradation separately. A
transformer which failed due to a winding defect and having aged tap changer
contacts would be most suitable for these tests.
A correction factor for variations in the measurement current amplitude was

determined for 2.5 A and higher (CDRM, as introduced in Section 2.3.5).
Additional experiments could be conducted to derive a correction factor for
lower values of the measurement current (e.g. from 0.5 A and higher) in order
to make more sophisticated corrections for contact resistance at lightly aged
160

contacts. Furthermore, different correction factors can be derived for tap
changers in different aging stages.
A database with DRM case studies can be used to compare results directly
after a measurement with previously measured results; most similar results
from the database can then be shown on-site to the test engineer. These case
studies provide more detailed information about the tap changer condition.
An on-line monitoring system can be developed to detect overheated tap

changer contacts when the power transformer is in service. Different sensors,
for example measuring local contact temperatures in the change-over selector
or measuring contact voltages, can be applied to detect severely aged contacts.
161
REFERENCES
References
[1]
R. Moxley, A. Guzmn, Transformer maintenance interval management,

Schweitzer Engineering Laboratories, Cigre Study Committee B5
Colloquium, paper 118, September 2005.
[2]
F. Roos, S. Lindahl, Distribution system component failure rates and repair

times an overview, Nordic Distribution and Asset Management
Conference, Espoo, Finland, 23-24 August 2004.
[3]
D.J. Allan, A. White, Transformer design for high reliability, Second

International Conference on Reliability of Transmission and Distribution
Equipment, Conference Publication No. 406, p. 66-72, 29-31 March 1995.
[4]
Cigr WG 12.05, An international survey on failures in large power

transformers in service, ELECTRA No. 88, p. 21 48, 1983.
[5]
M.S.A. Minhas, J.P. Reynders, P.J. de Klerk, Failures in power system

transformers and appropriate monitoring techniques, Eleventh
International Symposium on High Voltage Engineering, London, UK,
Conference Publication No. 467, paper 1.94.S23, p. 94-97, 1999.
[6]
R. Jongen, P.H.F. Morshuis, E. Gulski, J.J. Smit, A statistical approach to

processing power transformer failure data, 19th CIRED International
Conference on Electricity Distribution, Vienna, Austria, paper 546, May
2007.
[7]
G. Spence, A.C. Hall, Users experience and perspective, European seminar

on-load tap changers-current experience and future developments,
Regensburg, Germany, 9 November 1995.
[8]
H.F.A. Verhaart, L.J.C. Wakkerman, Tussenrapport over faaloorzaken van

distributie- en koppeltransformatoren op basis van de onvoorziene nietbeschikbaarheid, KEMA, rapport 43613-T&D 94-102278, 1994 (Dutch).
[9]
H.U. Schellhase, R.G. Pollock, A.S. Rao, E.C. Korolenko, B. Ward, Load
tap changers: investigations of contacts, contact wear and contact coking,
Proceedings of the Forty-Eighth IEEE Holm Conference on Electrical
Contacts, p. 259- 272, 2002.
163
REFERENCES
[10] G. Breen, Essential requirements to maintain transformers in service,
presented at the International Council on Large Electric Systems (CIGR),
Paris, France, paper 12-103, August 30September 5 1992.
[11] J.J. Erbrink, Tap changer diagnostics on high voltage power transformers
using dynamic resistance measurement, Graduate thesis Delft University of
Technology, Delft, the Netherlands, 2007.
[12] P.P. Seitz, B. Quak, J.J. Erbrink, E. Gulski, R. Leich, Advanced on-site
diagnosis of power transformers, International conference on condition
monitoring and diagnosis, Beijing, China, paper 148, April 2008.
[13] P. Kang, D. Birtwhistle, J. Daley, D. McCulloch, Non-invasive on-line
condition monitoring of on load tap changers, IEEE proceedings on
generation, transmission and distribution, Vol. 3, p. 2223-2228, January 2000.
[14] J.J. Erbrink, E. Gulski, P.P. Seitz, R. Leich, Advanced on-site diagnosis of
transformer on-load tap changer, IEEE International Symposium on
Electrical Insulation, Vancouver, Canada, p. 252-256, June 2008.
[15] A. Krmer, On-load tap-changers for power transformers; Operation
principles, applications and selection, Maschinenfabrik Reinhausen GmbH,
Regensburg, Germany, ISBN 3-00-005948-2, 2000.
[16] A. White, Tapchanging the transformer designers perspective, IEE
European Seminar on Developments On-Load Tap Changers: Current
Experience and Future, Regensburg, Germany, p. 4/1-4/6, 9 November
1995.
[17] IEC 60214-1, International standard, tap changers Part 1. Performance
requirements and test methods, first edition, 2003.
[18] V.V. Sokolov, Considerations on power transformer condition-based
maintenance, EPRI Substation Equipment Diagnostic Conference VIII,
New Orleans, USA, 20-23 February 2000.
[19] R. Crutcher, D. Hanson, L. Savio, New equipment and performance design
review LTC management course materials, Report 2-4 in New
Equipment and Performance Design Review LTC Management Course
Materials, EPRI, Palo Alto, USA, Technical Report ID 1012350, December
2006.
[20] P.J. Hopkinson, Electrical contacts for off-circuit tap changers for oil
immersed transformers, IEEE/PES Transformers Committee, DETC
working group, October 11, 2005.
164
REFERENCES
[21] R. Holm, Electric contacts: theory and application, 4th edition, Springer
Verlag, New York, 1967.
[22] J. A. Greenwood, Constriction resistance and the real area of contact,
British Journal of Applied Physics, vol. 17, no. 12, p. 1621-1632, 1966.
[23] T. Sun, Fundamental study of contact resistance behavior in rsw
aluminum, dissertation at the Ohio State University, 2003.
[24] K. Lemelson, About the failure of closed heavy current contact pieces in
insulating oil at high temperature, IEEE transactions on parts, hybrids, and
packaging, Vol. Php-9, No.1, p. 50-52, March 1973.
[25] EPRI, Evaluation of load tap changer oil for susceptibility to coking, from
www.epriweb.com, viewed June 2010.
[26] EPRI, Development of load tap changer monitoring technique: mechanism
of coking, EPRI, Palo Alto, USA, Technical Report ID 1001946, 2001.
[27] A. Erk, M. Schmelzle, Grundlagen der Schaltgeratetechnik. Kontaktglieder
und Lascheinrichtungen Elektrischer Schaltgerate der Energietechnik,
Berlin, Germany, Springer-Verlag, ISBN 3540060758, April 1974 (German).
[28] T. Tamai, Growth of oxide films on the surface of Cu contact and its effect
on the contact resistance property, Electronics and Communications in
Japan, Part 2, vol. 72, no. 7, p. 87-93, 1989.
[29] J. Kosco, The effects of electrical conductivity and oxidation resistance on
temperature rise of circuit-breaker contact materials, IEEE Transactions on
Parts, Materials and Packaging, vol. 5, no. 2, p. 99- 103, June 1969.
[30] L. Boyer, Contact resistance calculations: generalizations of Greenwood's
formula including interface films, IEEE Transactions on Components and
Packaging Technologies, vol. 24, no. 1, p. 50-58, March 2001.
[31] S. Schoft, J. Kindersberger, H. Lbl, Joint resistance of busbar-joints with
randomly rough surfaces, Proceedings of the 21th International Conference
on Electric Contact Phenomena, Zurich, Switzerland, p. 230-237, 9-12
September 2002.
[32] M. Braunovic, V.V. Izmailov, M.V. Novoselova, A model for life time
evaluation of closed electrical contacts, Proceedings of the Fifty-First IEEE
Holm Conference on Electrical Contacts, p. 217-223, 26-28 September 2005.
[33] E. Takano, K. Mano, The failure mode and lifetime of static contacts,
IEEE Transactions on Parts, Materials and Packaging, vol. 4, no. 2, p. 51-55,
June 1968.
165
REFERENCES
[34] Z. Chen, K. Karasawa, K. Sawa, Effects on contact resistance of passing
electrical current through wiping palladium contacts, IEEE Transaction on
Components, Packaging, and Manufacturing Technology, Part A. Vol. 18,
No. 3, p. 693-700, September 1995.
[35] H. Schellhase, N. Dominelli, M. Lau, W. Horn, B. Ward, Evaluation of a
filtration system on onload tap changers a case study, Report 2-3 in New
Equipment and Performance Design Review LTC Management Course
Materials, EPRI, Palo Alto, USA, Technical Report ID 1012350, December
2006.
[36] Cigre working group 12.18, Guidelines for life management techniques for
power transformers, draft final report Rev 2, 22 June 2002.
[37] Z. Wang, Y. Liu, P.J. Griffin, Artificial intelligence in OLTC fault diagnosis
using dissolved gas-in-oil information, IEEE Power Engineering Society
Summer Meeting, Seattle, USA, Vol. 4, p. 2422-2427, 16-20 July 2000.
[38] V. Sokolov, J. Mak, A. Bassetto, D. Hanson, Transformer risk assessment
considerations, Proceedings of the EuroTechCon 2002, Birmingham, UK,
1-3 October 2002.
[39] EPRI, Development of a filter using adsorbent technologies for the
removal of coking precursors: laboratory evaluation, EPRI, Palo Alto, USA,
Technical Report ID 1002048, December 2003.
[40] Cigre, Life management techniques for power transformers, Working
Group A2.18, Technical Brochure 227, June 2003.
[41] M. Duval, The Duval triangle for load tap changers, non-mineral oils and
low temperature faults in transformers, IEEE Electrical Insulation
Magazine, Vol. 24, No. 6, p. 22-29, November-December 2008.
[42] M. Duval, A review of faults detectable by gas in oil analysis in
transformers, IEEE Electrical Insulation Magazine, Vol. 18, No. 3, p. 8-17,
May-June 2002.
[43] R. Zuijderduin, J.J. Erbrink, E. Gulski, J.J. Smit, R. Leich, Condition
assessment of power transformers OLTC by DGA and dynamic resistance
measurements, Proceedings of the 16th international symposium on high
voltage engineering, Johannesburg, South Africa, 2009.
[44] IEEE Std C57.104-1991, IEEE Guide for the interpretation of gasses
generated in oil-immersed transformers.
166
REFERENCES
[45] M. Wang, A.J. Vandermaar, K.D. Srivastava, Review of condition
assessment of power transformers in service, IEEE Electrical Insulation
Magazine, Vol. 18, No. 6, p. 12-25, November-December 2002.
[46] L.R. Lewand, P. Griffin, Condition assessment of oil circuit breakers and
load tap-changers by the use of laboratory testing and diagnostics, Doble,
published in Neta World, summer 2004.
[47] Kelman, Transport X, Portable Dissolved Gas Analysis unit and Moisture
in Oil, www.kelman.co.uk, viewed June 2010.
[48] BPL Global, Serveron, Online Transformer Monitor Model TM3,
www.bplglobal.net, viewed June 2010.
[49] GE Energy, Hydran M2, Online Transformer DGA Monitoring,
www.gepower.com, viewed June 2010.
[50] D. Chu, A. Lux, On-line monitoring of power transformers and
components: a review of key parameters, Proceedings of the IEEE
Electrical Insulation Conference and Electrical Manufacturing and Coil
Winding Conference, Cincinnati, USA, p. 669675, 26-28 October 1999.
[51] L.A.L. de Almeida, M. Fontana, F.A. Wegelin, L. Ferreira, A new approach
for condition assessment of on-load tap-changers using discrete wavelet
transform, Proceedings of the IEEE Instrumentation and Measurement
Technology Conference, p. 653-656, 16-19 May 2005.
[52] M. Foata, R. Beauchemin, C. Rajotte, On-line testing of on-load tap
changers with a portable acoustic system, Proceedings of the IEEE 9th
International Conference on Transmission and Distribution Construction,
Operation and Live-Line Maintenance, Montreal, Canada, p. 293-298, 2000.
[53] P. Kang, D. Birtwhistle, Condition assessment of power transformer onload tap-changers using wavelet analysis, IEEE Transactions on Power
Delivery, Vol. 16, No. 3, p. 394-400, July 2001.
[54] P. Kang, D. Birtwhistle, Condition monitoring of power transformer onload tap-changers. Part 2: Detection of aging from vibration signatures,
IEEE proceedings on Generation, transmission and distribution, Vol. 148,
No. 4, p. 307 311, July 2001.
[55] P. Kang, D. Birtwhistle, Condition monitoring of power transformer onload tap-changers. Part 1: Automatic condition diagnostics, IEEE
proceedings on generation, transmission and distribution, Vol. 148, No. 4, p.
301 306, July 2001.
167
REFERENCES
[56] F. Poza, P. Marino, S. Otero, F. Machado, Virtual instrument for condition
monitoring of on-load tap changers, Proceedings of the Third IEEE
International Workshop on Electronic Design, Test and Applications, p.
214-218, 17-19 January 2006.
[57] ABB, Tstat Transformer Monitor, Advanced monitoring for Transformer
Management, www.abb.com, viewed June 2010.
[58] S. Tenbohlen, D. Uhde, J. Poittevin, H. Matthes, U. Sundermann, P. Werle,
H. Borsi, Enhanced diagnosis of power transformers using on- and off-line
methods: results, examples and future trends, Cigre 2000, paper 12-204,
Paris, France, August 2000.
[59] S. Tenbohlen, T. Stirl, G. Bastos, J. Baldauf, P. Mayer, M. Stach, B.
Breitenbauch, R. Huber, Experience-based evaluation of economic benefits
of on-line monitoring systems for power transformers, Cigre Session 2002,
paper 12-110, Paris, 2002.
[60] P.J. Griffin, L.R. Lewand, R.C. Peck, N.A. Letendre, D. Frieze, J.L. Johnson,
Load tap changer diagnostics using oil tests a key to condition-based
maintenance, Minutes of Doble Engineering Company Conference, 2005.
[61] EPRI, Inspection and maintenance guidelines for distribution substations:
master-plan for content development, EPRI, Palo Alto, USA, Technical
Report ID 1002629, December 2004.
[62] Qualitrol, 509 ITM-300 Intelligent Transformer Monitor, Load Tap
Changer Monitor, www.qualitrolcorp.com, viewed June 2010.
[63] GE Energy, LTC-MAP 1525, Load Tap Changer-Maintenance Action
Planner, www.ge-energy.com, viewed June 2010.
[64] B. Turnbull, S. McConnell, How infrared thermography helps Southern
California Edison improve grid reliability, from www.goinfrared.com,
viewed June 2010.
[65] IEEE C57.12.90 1987, IEEE standard test code for liquid-immersed
distribution, power and regulating transformers and IEEE guide for shortcircuit testing of distribution and power transformers.
[66] Omicron, CPC 100, Multifunctional Primary Test System for Substation
Commissioning and Maintenance, www.omicron.at, viewed June 2010.
[67] Megger, MTO210 Transformer Ohmmeter, Transformer Winding
Resistance and Tap-changer Test Set, www.megger.com, viewed June 2010.
168
REFERENCES
[68] P. de Graaf, KEMA off-line regelschakelaardiaganostiek met aanvullende
diagnostieken. Handleiding van het bij KEMA ontwikkelde meetsysteem,
KEMA, report 44412-T&D 95-103450, 8 December 1995 (Dutch).
English version: P. de Graaf, Manual for the KEMA off-line tap changer
diagnostics, KEMA, report 94460525-TDP 98-02371A, 1998 (non vidi).
[69] KBK Power Solutions, TDRCA, Tapchanger Diverter Resistor and
Contact Analyzer, www.kbkps.co.za, viewed June 2010.
[70] DV Power / Ibeko Power AB, RMO60TC, Winding Ohmmeter & Tap
Changer Analyzer, www.dv-power.com, viewed August 2010.
[71] A. Kramer, J. Ruff, Transformers for phase angle regulation considering the
selection of on-load tap-changers, IEEE Transaction on Power Delivery,
Vol. 13, No. 2, p. 518-525, April 2008.
[72] X. Zhang, E. Gockenbach, Asset-management of transformers based on
condition monitoring and standard diagnosis, IEEE Electrical Insulation
Magazine, Vol. 24, No. 4, p. 26-40, July-August 2008.
[73] Doble Client Committee on Transformers, Analysis of replies to the 1998
technical questionnaire on power transformer failures and troubles,
Proceedings of the 65th Annual International Conference of Doble Clients,
p. 8-2.1-8-2.10, March 1998 (non-vidi).
[74] J.J. Erbrink, E. Gulski, J.J. Smit, R. Leich, P.P. Seitz, B. Quak,
Reproducibility of dynamic resistance measurement results of on-load tap
changers effect of test parameters, International Conference on Condition
Monitoring and Diagnosis, Tokyo, Japan, 6-11 September 2010.
[75] H.F.A. Verhaart, A diagnostic to determine the condition of the contacts of
the tap changer in a power transformer, Proceedings of CIRED, Brussels,
Belgium, paper 1.13, 1995.
[76] J.J. Erbrink, E. Gulski, J.J. Smit, P.P. Seitz, B. Quak, R. Leich, R. Malewski,
Diagnosis of onload tap changer contact degradation by dynamic resistance
measurements, IEEE Transactions on Power Delivery, Vol. 25, No. 4, p.
2121-2131, October 2010.
[77] M. Landry, O. Turcotte, F. Brikci, A complete strategy for conducting
dynamic contact resistance measurements on HV circuit breakers, IEEE
Transactions on Power Delivery, Vol. 23, No. 2, p. 710-716, April 2008.
[78] M. Landry, A. Mercier, G. Ouellet, C. Rajotte, J. Caron, M. Roy, F. Brikei,
A new measurement method of the dynamic contact resistance of HV
circuit breakers, IEEE PES Transmission and Distribution Conference and
Exhibition, Dallas, USA, p. 1002-1009, 21-24 May 2006.
169
REFERENCES
[79] J.J. Erbrink, E. Gulski, J.J. Smit, J. Aditya, L.A. Chmura, R. Leich, P.P. Seitz,
B. Quak, Test procedure and test circuit considerations for on load tap
changer dynamic resistance measurement, International Conference on
High Voltage Engineering and Application, New Orleans, USA, 11-14
October 2010.
[80] J.J. Erbrink, E. Gulski, J.J. Smit, R. Leich, P.P. Seitz, B. Quak, Effect of test
parameters on dynamic resistance measurement results from on-load tap
changers, IEEE International Symposium on Electrical Insulation, San
Diego, USA, 6-9 June 2010.
[81] H.F.A. Verhaart, Diagnostiek voor de toestandsbepaling van de contacten
in een omschakelaar van een regelschakelaar, KEMA, report 43151-T&D
93-4615, 24 November 1993 (Dutch).
[82] Vanguard Instruments Company Inc, LTCA-10, Load Tap Changer
Analyzer, www.vanguard-instruments.com, viewed June 2010.
[83] H.F.A. Verhaart, Werkwijze voor het meten van een weerstandswaarde,
NL Patent 9400221, 11 February 1994 (Dutch).
[84] J.J. Erbrink, R.A. Jongen, E. Gulski, J.J. Smit, Tap changer diagnostics on
high voltage power transformers using dynamic resistance measurement, 4th
IEEE young researchers symposium in electrical power engineering,
Eindhoven, the Netherlands, p. 1-5, February 2008.
On-load tap changer diagnosis an off-line method for detecting
degradation and defects: part 1, IEEE Electrical Insulation Magazine, Vol.
26, No. 5, p. 49-59, September - October 2010.
[86] Maschinenfabrik Reinhausen GmbH, Oiltap M, On-Load Tap-Changer
for Regulating Transformers Technical Data, www.reinhausen.com, viewed
June 2010.
[87] J.J. Erbrink, E. Gulski, J.J. Smit, L.A. Chmura, R. Leich, P.P. Seitz, B. Quak,
Condition assessment of on load tap changers using dynamic resistance
measurements, International Conference on High Voltage Engineering and
Application, New Orleans, USA, 11-14 October 2010.
Interpretation of dynamic resistance results of on-load tap changers,
International Conference on Condition Monitoring and Diagnosis, Tokyo,
Japan, 6-11 September 2010.
170
REFERENCES
[89] R. Samsudin, Yogendra, A. Berhanuddin, Y. Zaidey, M. Haneef, Experience
on dynamic contact resistance test on eroded and worn-out tap-changer
contacts, World Academy of Science, Engineering and Technology
(WASET), Singapore, 25-27 August 2010.
[90] J.J. Erbrink, E. Gulski, J.J. Smit, P.P. Seitz, R. Leich, Experimental model of
aging mechanisms of on-load tap changer contacts, International
conference on condition monitoring and diagnosis, Beijing, China, p. 247250, April 2008.
[91] J.J. Erbrink, E. Gulski, J.J. Smit, R. Leich, Experimental model for
diagnosing on-load tap changer contact aging with dynamic resistance
measurements, 20th International conference on electricity distribution,
Prague, Czech Republic, paper 135, June 2009.
[92] E. Gulski, S. Meijer, J.J. Smit, F. de Vries, R. Leich, H. Geene, R. Koning, L.
Lamballais, T.J.W.H. Hermans, E.R.S. Groot, J. Slangen, M.J.M. Boone, J.
Kanters, A. Bun, A. Janssen, Condition assessment and AM decision
support for transmission network components, Proceedings of the 2006
Cigre International Council on Large Electric Systems, Paris, France, paper
D1-110, 27 August 1 September 2006.
[93] E. Gulski, F.J. Wester, W. Boone, N. van Schaik, E.F. Steennis, E.R.S.
Groot, J. Pellis, B.J. Grotenhuis, Knowledge rules support for CBM of
power cable circuits, Cigre Conference Proceedings, Paris, France, Session
15, paper 104, 25-30 August 2002.
[94] M. Fischer, S. Tenbohlen, M. Schfer, R. Haug, Determining power
transformers sequence of maintenance and repair in power grids, IEEE
International Symposium on Electrical Insulation, San Diego, USA, 6-9 June
2010.
[95] E. Gulski, J.J. Smit, F. de Vries, The use of diagnostics data for condition
indexing of HV components, Proceedings of the 8th International Power
Engineering Conference IPEC, Singapore, December 2007.
[96] J.J. Erbrink, E. Gulski, J.J. Smit, R. Leich, P.P. Seitz, B. Quak, On-load tap
changer diagnosis: interpretation of dynamic resistance deviations, IEEE
2010.
[97] E. Gulski, O. Piepers, J.J. Smit, R.A. Jongen, P.P. Seitz, F. Petzold, F. de
Vries, New generation of on-site diagnosis for distribution power cables,
Proceedings of the CIRED 19th International Conference on Electricity
Distribution, Vienna, Austria, paper 178, 21-24 May 2007.
171
REFERENCES
[98] Smit Transformers, On-load tapchanging equipment, product catalogues
T59/01/E.
[99] V. Sokolov, Z. Berler, V. Rashkes, Effective methods of assessment of
insulation system conditions in power transformers: a view based on
practical experience, Proceedings of the Electrical Insulation Conference
and Electrical Manufacturing and Coil Winding Conference, Cincinnati,
USA, p. 659667, 26-28 October 1999.
[100] R.M. Schomper, Condition assessment model for power transformers,
Master thesis, Technical University of Delft, December 2004.
[101] R. James, W. Bartley, Transformer asset management, Weidmann-ACTI
Inc. Second Annual Conference, New Diagnostic Concepts for Better Asset
Management, November 2003.
[102] IEEE Std 62-1995, IEEE guide for diagnostic field testing of electric power
apparatus. Part 1: oil filled power transformers, regulators, and reactors.
[103] ANSI/IEEE C57.12.90-1987, American national standard / IEEE standard
test code for liquid-immersed distribution, power, and regulating
transformers.
[104] Megger, Instruction Manual AVTM830280 for Transformer Ohmmeter,
www.megger.com, viewed June 2010.
[105] Megger, Instruction Manual AVTM550503, for Three-Phase TTR
Transformer Turn Ratio Test Set, www.megger.com, viewed June 2010.
172
APPENDIX A OLTC TECHNOLOGY, DEGRADATION AND PROTECTION
Appendix A
A. (<- appendix numbering)
OLTC technology, degradation and

protection
This appendix takes a closer look at a tap change operation of the selector switch
and a diverter switch (Section A.1). In addition, OLTC degradation is discussed in
detail (Section A.2) and common ways of OLTC protection are discussed (Section
A.3).
A.1
OLTC technology
Chapter 2 already gave an introduction to the operation principles of an OLTC. It

was discussed that OLTC types can be subdivided into types with a diverter and a
selector switch. This section describes how these types of OLTC switch from one
winding tap to another by using an arcing switch, tap-selector and change-over
selector.
A.1.1 Selector switch type OLTC
The selector switch type OLTC as depicted in Figure A.1 consists of a selector
switch on top of the coarse change-over selector. The selector switch is accessible
at the top of the transformer, while the coarse change-over selector is mounted
deeper within the transformer tank, underneath the selector switch. The selector
switch and the coarse change-over selector both consist of a stator on which the
static contacts are mounted. These contacts are connected to the taps on the
transformer windings, shown in Figure A.1. The rotor is inside the stator and
rotates using a drive system. The rotor makes a connection between the stator
contacts. The OLTC has now selected the main winding on tap A8. The current is
brought to the selector switch stator by the rotor of the coarse change-over
selector (using contact A in Figure A.1). The selector switch has selected three fine
tap windings using tap A3. The current leaves the tap changer using a slip ring.
173

Selector switch cycle
A selector switch cycle is passed through when a

fine tap winding is (de-)selected. This cycle is
built up from a few small steps, shown in Figure
A.2. The rotor of the selector switch consists of
four sets of roller contacts (Figure A.3) for each
phase: the main roller contact, two transition
contacts which are in series with a low ohmic
transition resistor and a set of roller contacts
which connects to a slip ring. The roller contacts
of the arcing switch are made from tungsten to
prevent extreme wear due to arcing. The stator
contacts are also equipped with tungsten edges.
These tungsten contacts are called arcing
contacts.
When the selector switch moves from
contact block A3 to A4, the first transition
contact takes over the load current. The main Figure A.1 One phase of a
roller contact will leave contact block A3 (Figure selector switch type OLTC with
A.2-2). When the rotor moves a little further, the coarse change-over selector.
second transition roller contact will reach tap A4
(Figure A.2-3). The load current now flows through both transition resistors. The
transition resistors are necessary to prevent the short circuit of adjacent fine tapped
windings and to provide a non-interrupted current path for the load current during
the switching operation. The resulting circulating current is determined by the step
voltage and the resistance of the selector switch.
A4
V+V
A4
A3
A4
A4
V+V
R
IL
IL
R
A3
IL
A3
V+V
R
IL
V+V
A4
IL
A3
V+V
A3
Figure A.2 A selector switch cycle. Normally the load current flows through the main contact
(1). When the selector switch moves to the next tap, a transition contact takes the load for a
moment (2). Then, the second roller makes contact with the next tap causing a circulating
current to flow (3). The rotor moves further and the first roller leaves its contact block (4). The
switch cycle is completed when the main contact once again takes over the load current
(5).[98]
174

Following this, the main roller contact will
move to contact block A4 (Figure A.2-4) and
take over the complete load current (Figure
A.2-5). The selector switch has now selected
another tap and is ready for the next switching
cycle. The entire selector switch movement will
take about 30-200 ms.
This thesis will focuses on a diagnostic method
which assesses the timing of this selector switch
cycle. The corresponding vector diagram of this
tap change operation can be represented as in
Figure A.3 Roller contacts of a Figure A.4, with the output voltage and the
load
selector switch rotor.
curren
t that flows through the tap changer being
depicted. In general the current IL will
lag the voltage. The OLTC switches the
output voltage from V to V+V using
the five steps from Figure A.2. The
output voltage drops because of the
voltage over the transition resistor. The
output vector follows the shape of a flag
when switching from step 1 through to 5,
which is why this tap change operation is
called a flag-cycle operation. All OLTCs
that transfer the load current from the
main contacts before creating a
A.4 Vector diagram of a selector
circulating current show this vector Figure
switch type OLTC (flag-cycle operation).
diagram.
Coarse change-over selection
To lower the secondary voltage, the selector switch can select more fine tap
windings on the primary side by switching from A1 to A7 (as described in Section
A.1.1). When all of the fine tap windings are selected by the selector switch (see
Figure A.5A), a coarse tap winding has to be selected before the selector switch
can continue. The coarse change-over selector shown at the bottom of Figure A.1
makes a turn for that purpose. The steps of the coarse change-over selector
movement are depicted in Figure A.5.
Firstly, one contact of the coarse change-over selector moves to tap A9 (Figure
A.5B). Note that the roller contact switches no current because the load current
flows through contact A. The voltage on the k-contact is increased with 1 coarse
voltage step which is equal to 8 fine voltage steps. The position of the change-over
selector in Figure A.5B is called the mid-position. The two contact sets of the
change-over selector are on different coarse tap contacts.
175

Secondly, the selector switch now switches to the k-contact (Figure A.5C). As a
result, the selector switch experiences only one voltage step difference. The kcontact of the selector switch is connected directly to the coarse change-over
selector. When the selector switch is at the k-contact, the load current flows
through coarse change-over selector contact A0. The roller contact which normally
carries the load current is now free to move. This tap is called the parking position.
Finally, the second roller contact of the change-over selector is moved to tap
A9 (Figure A.5D). As a result, the first roller contact moves over contact A9 while
carrying current. After switching the coarse change-over selector, the selector
switch is ready to make another rotation.
Figure A.5 Switching the coarse change-over selector of a selector switch type OLTC.
176

A.1.2 Diverter switch type tap changer
A diverter switch type OLTC combines a diverter switch and a
tap selector. The tap selector has two sets of contacts available
for tap selection. One of the two contacts is selected by the
diverter switch and is under load. The other contact selects the
next tap (pre-selected tap) without switching current. The
diverter switch then switches from the selected to the preselected tap. Figure A.6 shows a schematic overview of a tap
selector and a diverter switch. The tap selector can be
combined with a change-over selector to expand the range of
the OLTC. Again, the change-over selector can be designed as
a coarse change-over selector (Figure A.7) or a reversing
change-over selector (Figure A.8). By switching the reversing
change-over selector one can add or subtract the selected
transformer windings, thus doubling its range. A coarse
change-over selector can add a larger winding to the main
winding, in this way changing the voltage of all tap selector
contacts.
Diverter switch type OLTCs can also use transition
A.6
resistors to limit the circulating current in the diverter switch. Figure
Over a short period, both taps of a different voltage are Diverter switch
with
transition
connected to the same output, causing a circulating current, as resistors (top) and
is the case with the selector switch type tap changer. The flag- a tap selector
cycle vector diagram of Figure A.4 also applies to this diverter (bottom).
switch type OLTC.
In contrast to the selector switch type tap changer, most diverter switch type
OLTCs have no rotor inside the arcing switch. A stationary insert is used instead.
The diverter switch can easily be maintained by removing the diverter switch
insert. The dirty oil can be replaced, the diverter switch inspected and its arcing
contacts replaced.
A.2
Degradation mechanisms
This section elaborates on OLTC degradation, as introduced in Chapter 2.

Mechanical defects and contact degradation on all OLTC contacts show to be
especially important and it is discussed in this section that maintenance errors also
contribute to a significant part of OLTC failure. Degradation can lead to failure, a
situation in which the OLTC cannot perform its primary task, i.e., to select another
tap without interrupting the load current.
Degradation and failure modes can be evaluated using a failure mode, effects,
and criticality analysis (FMECA). An example of such a FMECA, which is used at
a utility, is shown in Figure A.9. A score is assigned to each OLTC failure mode,
based on the pre-failure detection chance and the risk. The risk is calculated by
multiplying the failure chance and the consequence (which includes the repair time,
177

repair costs, safety and environmental factors). Using a FMECA, utilities can direct
their maintenance expenditure to prevent OLTC failure. In this example, a broken
OLTC drive shaft, contact coking by the long term effect and electrical treeing
show to be most important, this section discusses these failure modes.
Figure A.7 Diverter switch type OLTC with

coarse change-over selector.
Figure A.8 Diverter switch type OLTC

with reversing change-over selector.
A.2.1 Electrical treeing

In-tank tap changers are insulated from the ground by an insulating support
cylinder. The high electric field strength that is present over the cylinder can cause
partial discharges over the surface of the cylinder. Electrical trees (tracking) can be
the result. Electrical treeing is a slow process of creeping surface discharges along
the cylinder [99], creating a track that cannot be repaired. These tracks grow
slowly, weaken the insulating properties and can cause flashover. Tap changers
installed in wye-connected windings can be located at the neutral point and are
able to insulate only a part of the line-to-neutral voltage. Tap changers installed in
delta-connected windings have to cope with higher electrical stress, for they have
to insulate the line-to-line voltage. Tap changers that are installed in deltaconnected windings are therefore more prone to electrical treeing than tap
changers located at the neutral point of the transformer [100].
178
Mechanical
Defective
No or limited
functioning
Too sensitive
No or limited
functioning
Flow/pressure relay
arcing switch
Flow/pressure relay
arcing switch
Pressure relay
change-over
selector
Protection
Protection Protection
Protection
No tap change possible
Motor
Drive
Drive
Electric
Leakage
Gear
(perpendicular)
Blocked
Drive axis
Brake
Drive
Low switch speed
Loose
Drive
Springs
Drive
OLTC
Rupture
Mounting of
contacts
OLTC
Mechanical
Contamination
Insulation oil
Phase-to-phase fault
Drive axis arcing

switch
Rotor (outside)
OLTC
Dielectric
OLTC
OLTC
Stator
Rotor (inside)
OLTC
Contact interruption
Change-over
selector / tapselector
Interrupted
Sort of failure
Arcing contacts
Part
Primary
Subsystem
OLTC
OLTC
OLTC
System
Catastrophic failure of
transformer
Possible consequences of
failure
Open contact, transformer
de-energised by
pressure/flow relay
Defective
Settings out of
range
Defective
Wear / aging
Wear
Low seal quality
Broken
Weakening /
Broken
Mechanical
load/fatigue
Mechanical
load/fatigue
Switching arcs
Electrical treeing
delta OLTCs
Electrical treeing
delta OLTCs
Electrical treeing
delta OLTCs
Failure of transformer
Transformer de-energised by
Buchholz protection, heavy
damage when the arcing
switch fails
Unnecessary de-energised
transformer
Transformer de-energised by
Buchholz protection, heavy
damage when the changeover selector fails
No tap change possible
Too high or too low voltage
tap change not finished

Accelerated wear due to
mechanical oscillation
Functional
check
Functional
check
Functional
check
None
Visual
inspection
Dynamic
resistance
measurement
None
Visual
inspection
Visual
inspection
None
Visual
inspection
Accelerated wear and carbon

deposition on insulation
transformer
Blocked arcing switch,
transformer de-energised by
pressure/flow relay
Visual
inspection
Visual
inspection
Visual
inspection
Dynamic
resistance
measurement
None
A
Repair
time
Visual
inspection
Detection
possibility
transformer
transformer
transformer
Low switch speed, Transformer de-energised by

blocked half-way
pressure/flow relay
Contact
degradation (long
term effect)
Contact wear
Cause of failure
B
C
D
Corrective
Environ
Safety
costs
ment
E=A+B+C+D
16
16
15
15
15
15
15
Consequence
F
Failure
chance
32
32
10
10
21
10
16
30
16
45
45
45
16
45
16
Risk
G=E*F
H
Detection
chance
96
24
96
30
50
63
50
48
150
16
135
45
45
80
135
16
I=H*G
Total
score
Figure A.9 Example of an OLTC Failure Mode, Effects, and Criticality Analysis (FMECA).
179

A.2.2 Contact degradation
Chapter 2 already discussed that the arcing switch and the change-over selector
wear differently, because the operation frequency and the amplitude of the current
that is switched are different. Normally, an OLTC that is installed in a power
transformer in the utility grid will switch a few times a day to compensate for
voltage changes in the network. Therefore, the OLTC will stay in one position for
several hours and usually only a limited number of tap positions are used. In cases
where this limited number of tap positions involves the operation of the changeover selector, the latter may be used every day. However, the change-over selector
is not necessarily used on a daily basis and can be motionless for extremely long
periods.
Change-over selector contact degradation
Change-over selector contacts (including tap selector contacts) do not wear as fast
as arcing switch contacts that wear due to the switching of load currents. These
contacts will not switch significant currents, but can show pitting of the contacts
and the development of pyrolytic carbon [18-19]. This contact degradation is not
due to the arcs caused by switching the current but is caused by two other
degradation mechanisms.
Floating windings
The first cause of change-over selector contact

degradation discussed here is the small capacitive
current that is switched off when the changeover selector operates. The fine tapped windings
are temporarily disconnected from the main
windings when the tap changer is on the kcontact [15]. Figure A.10 shows this situation for
a selector switch type tap changer with a coarse
change-over selector. The floating windings take
a potential based on the capacitive coupling
between the transformer windings and earth
[71]. A small capacitive current in the order of
tens of milliamps is switched at a recovery
voltage of tens of kilovolts [15]. The changeover selector contacts can be slightly damaged by
the discharges that occur.
Long-term effect
The long-term effect can be initiated on

infrequently operated OLTC contacts. Three
basic stages of the long-term effect can be
distinguished and are shown in Figure A.11.
Section 2.3 already elaborated on this long term
aging effect.
180
Figure A.10 Floating fine tapped

windings (A-A7) in a selector
switch type OLTC.

The long-term effect starts with the formation of a thin layer of oil deposited on
the contacts [20][24], which reduces conductivity.
Literature describes experiments [18][27] showing that the surface film grows
thicker as the contacts age and that the growth rate of this layer is strongly
dependent on the surface temperature of the contacts. It can be described as:
s k0 * 03.862 * t 0.3559
A-1
where s is the thickness in of the film layer on copper contacts under Shell Diala
insulation oil, 0 the surface temperature in C at which the surface film grows, t
the duration of the aging test in hours and k0 = 1.88310-6 with an appropriate unit
to balance the equation.
The surface film can be reduced by switching the tap changer through all its tap
positions. By doing so, the film layer is partially wiped off and the long-term effect
is slowed.
Figure A.11 The long-term effect starts with the formation of a surface film on the contacts.
The increased resistance can cause coking. Thermal runaway can finally cause failure of the
OLTC due to open contacts.
Contact resistance and power losses on the contact increase as a result of the
surface film. At this early stage of the long-term effect, the surface film can be
wiped away by switching the OLTC through its cycle of operation: the structure of
the contact contamination is changed due to the moving contacts [20][34-35].
Due to the temperature dependence of the surface film growth rate, the longterm effect accelerates. The insulation oil dissociates because of discharges and the
high contact temperatures. This is the advanced stage of the long-term aging effect.
Decomposed transformer insulation oil results in carbon between the contacts
(Figure A.11B, [20][37]). This formation of pyrolytic carbon is called coking [19].
Oil cracking occurs at local temperatures over 300 C [20]. In addition to coking,
the contact material wears off locally and pitted spots become visible on the
contacts. The contacts are now damaged and cannot be repaired by switching the
OLTC through all its tap positions. An overhaul is needed to undo the pitting of
the contacts. Because of the infrequent movement of the change-over selector,
coking can occur on all stator contacts, rotor contacts and other movable parts
that carry the load current
The coking accelerates due to increased contact resistance and the thermal
resistance characteristics of the carbon [9]. Coking between moving parts of the
contacts and around the springs that provide the contact pressure can cause the
contacts to lose pressure. Finally, a considerable amount of contact material
vanishes and excessive arcing activates the tap changer protection mechanisms.
Figure A.12 shows an example of change-over selector stator roller contacts
damaged in the advanced stage of the long-term effect.
181

The long-term effect is accelerated by [24]:
Infrequent movement
o The surface film is not wiped off the contacts when the changeover selector is not operated.
High temperatures
o The formation of the surface film is highly temperature
dependent.
High load
o Power losses on the contact interface increase exponentially with
the load current and heat the contacts.
o Coking and pitting of the contacts will cause jammed contact
more easily when the contact pressure is low. The layer of
contamination on the contact surface is also more easily damaged
at high contact pressure.
Copper or brass contacts
o The growth rate of the surface film is higher on copper and brass
contacts compared to silver contacts [24]. Pitting and coking is
reduced by silver-coating of the contact surface; however, older
OLTC designs may use bare copper contacts.
Figure A.12 Change-over selector stator roller contacts are damaged in the advanced stage
of the long-term effect.
Three examples of failed OLTCs due to the long-term effect are discussed below.
Damaged contacts can be compared with contacts from another phase of the
OLTC, because it can be assumed that they are used and loaded equally but have
worn differently (one set of contacts failed in the following examples).
Example 1
One of the results of the long-term aging effect is that the spring system loses
resilience. The roller contacts may remain pressed in, see Figure A.13. This breaks
the contact and results in an electrical arc. Inspection of the change-over selector
during several OLTC overhauls shows that the sliders (and holders) and the
stationary contacts are both degraded.
182
Correct sliders
Blocked sliders
Figure A.13 Example of sliders which are blocked. Normally, the sliders reach beyond the
holders, as can be seen left (correct sliders). At blocked sliders (right), the sliders and
holders are at the same position. Contact pressure is threatened.
Example 2
Figure A.14 shows an example of a change-over selector contact with pitting and
carbon due to the long-term effect. The contact shown on the right already
exhibits the effect of extreme arcing. The left contact is from another phase and
represents the condition of the contact just before arcing occurred. In this case the
Buchholz protection relay detected the development of gas, the transformer was
switched off and further damage was prevented.
Figure A.14 Change-over selector stator contact blocks before and after failure. Gas
development due to arcing was detected by the Buchholz protection relay and the
transformer was switched off.
Example 3
When the Buchholz protection relay does not detect gas development from the
change-over selector in time, flashover occurs and the transformer can be switched
off by the Buchholz or the differential protection mechanism. Figure A.15 shows
an example of a change-over selector after a flashover. Again, a contact from an
undamaged phase was taken as a reference and is shown on the left. The long-term
effect which caused this failure, was diagnosed using dynamic resistance
183

measurements (pre-failure), and displayed considerable current interruptions. The
decision rules found in Chapter 8 will diagnose these last two examples as critical.
Figure A.15 Example of stator contact blocks before (left) and after (right) flashover. The
transformer was switched off by the differential protection and the Buchholz protection
relay.
Arcing switch contact degradation
Basically, the arcing switch has two different modes of contact degradation.
Contacts that switch the current can wear due to arcing, while contacts that never
switch current are liable to the long-term effect.
The circulating current can increase or reduce the current that is interrupted by
the arcing switch contacts depending on the direction of operation. This results in
different wear of the arcing contacts on both sides of the main contacts, but this
effect is largely compensated for by the bidirectional movement of the tap changer
rotor.
Before the current is switched to another winding tap, it is transferred from the
main contacts to the transition contacts that switch off the current. Both the main
contacts and the transition contacts will be pitted due to arcing. The arc is normally
quenched at the first zero current.
The arcing switch contacts are normally roller contacts or contact blocks. The
tungsten arcing roller contacts and accompanying stator contacts normally do not
show carbon on the contacts due to coking.
The long-term effect is also active at the arcing switch, especially on contacts
that carry the load current and that are not subject to arcing or wiping. The longterm effect at the arcing switch does not develop as fast as the long-term effect at
the change-over selector because of the more frequent movement of the arcing
switch contacts.
A.2.3 Mechanical degradation and failure
In addition to the electrical contacts of the tap changer, the mechanical
components of an OLTC are also liable to wear and are responsible for a
substantial number of OLTC failures [13]. Some frequently reported mechanical
problems are:
184
Loss of synchronism between the arcing

switch and the change-over selector due to a
broken or damaged shaft, see Figure A.16.
Operating too slowly or stopping when the
main contacts are not on a contact block,
resulting in damage to the transition resistors.
Leakage of the arcing switch oil onto the
change-over selector and into the transformer
tank.
Insufficient contact pressure due to broken or
damaged springs or due to excessive carbon
build-up between the sliders.
Defective braking mechanism of the Geneva
wheel or a broken lock system.
Too much play or broken components in the
drive mechanism due to wear and vibrations.
The spring energy accumulator may lose its
elasticity.
Figure A.16 Broken drive shaft

of the selector switch.
A.2.4 Maintenance-induced failures

Finally, another cause of OLTC failures can occur during normal maintenance.
The arcing switch compartment is opened, cleaned and provided with new
insulation oil. Incorrect maintenance is a key contributor to OLTC failure [13] due
to the mechanical complexity of the tap changer and the stringent requirements
that have to be met during and after maintenance. Several failure studies
[13][52][101] conclude that maintenance and reassembly-induced failures can
certainly not be overlooked. Figure A.17 shows an example of a study of the
causes of OLTC failures, where 33% of the failures are related to incorrect or
inadequate maintenance or reassembly. Therefore, diagnostic measurements that
check the quality of maintenance could significantly reduce OLTC failure [13].
A.3
On-load tap changer protection
When the OLTC fails, for instance due to a defect in the drive mechanism or
through the long-term effect, protective relays can switch off the transformer to
prevent damage. In addition to local damage to the OLTC, an OLTC failure can
also cause damage to the associated transformer. This section discusses protective
devices that are used to detect OLTC defects and limit the consequences of OLTC
failure.
185
Excessivemoisture
pollution
5%
Aged/deteriorated
components
23%
Other
8%
Design
10%
Manufacture
8%
Incorrectoperation
5%
Overdue
maintenance
8%
Incorrector
inadequate
maintenance/
reassembly
33%
Figure A.17 Poor maintenance and reassembly are responsible for a considerable number
of OLTC failures [13].
When a failure occurs within the arcing switch compartment, gas can be generated
by an electrical arc. Due to this gas, a flow is generated in the oil pipe connecting
the tap changer to the conservator (see Figure A.18). To deal with this, a flow relay
is mounted in the oil pipe to protect the arcing switch compartment. The changeover selector and, when present, also the tap selector, is indirectly protected by the
Buchholz relay of the transformer main tank, in the case of the change-over
selector sharing its oil with the transformer tank (see Figure A.19). This
configuration is called a two-compartment system. In a three-compartment
Figure A.18 The oil of the arcing switch (in this case a diverter switch) is separated from the
main tank. Flow relay d protects the diverter switch compartment. The change-over
selector shares its oil with the main tank (two-compartment configuration), thus Buchholz
relay f also protects the change-over selector.
186

configuration, the change-over selector also has a separate oil tank. In this case, the
change-over selector can be protected by a separate protective relay, as is the
arcing switch (see Figure A.19).
Figure A.19 Defects in the change-over selector can lead to the generation of gas, which can
be detected by a protective relay. The change-over selector can have its own relay (threecompartment configuration) or share with the main tank (two-compartment configuration).
As discussed earlier, the mechanical drive system of the OLTC is also a significant
source of tap changer failures. To detect problems in moving the rotor (Figure
A.20), the OLTC can be protected with a switching-time control relay (tap change
incomplete protection). If the switching procedure takes too long, the switchingtime control relay will switch off the transformer to prevent arcing inside the
OLTC or burnout from the transition resistors. For this purpose, a contactor for
switching-time control is mounted inside the drive mechanism to detect the
switching (also called the tap change in progress or tap change incomplete
contactor). This contactor activates the switching-time control relay. This relay will
operate the power switch and turn off the transformer. In the case of arcing, the
resulting oil flow can trigger a flow relay.
Figure A.20 Defects in the OLTC drive can cause slow operation or stop the operation of the
contacts. The switch-time protective relay can detect excessive operation times. Oil flow due
to heated insulation oil or short circuits can trigger a flow relay.
Other OLTC protection mechanisms are also used to protect against failure, for
instance systems that block the drive mechanism in the case of excessive load
currents, out-of-step protection in the case of parallel transformers being in a
different tap position, protection against an undesirable oil level, running-through
protection, limit switches to detect the end of the cycle of operation or switches
that block the drive motor when the OLTC is operated manually.
In addition, transformer protection relays can also detect OLTC failures, as can
a differential protection relay or an over current relay. Figure A.21 shows examples
that can cause failure of the arcing switch and trigger protection through an oil
flow or short circuit current.
187

Oil contamination
and dirt
Electrical treeing
Insufficient contact
pressure, play or
loose connections
Oil flow
Flow relay
Short circuit
current
Differential or over current

protection
Arcing switch failure
Worn arcing
contacts
Damaged bridging
resistors
Figure A.21 Examples that can lead to failure of the arcing switch. A resulting oil flow can
trigger the flow relay or a short circuit current can be detected by transformer differential
protection or over current protection.
The protective devices described above cannot guarantee the prevention of heavy
damage to the tap changer or pollution and damage to the adjacent transformer
windings in the case of a failure. Transformer protection mechanisms are used to
switch off the transformer in the case of failure, but pre-failure detection of OLTC
degradation can be useful in the prevention of damage. An example of damaged
transformer windings caused by an OLTC defect is shown in Figure A.22. In this
particular case, the OLTC switch operation was not completed due to a
mechanical failure, causing the transition resistors to heat the surrounding
insulation oil. A short circuit finally triggered the protective devices, but the
transformer was destroyed. Pre-failure detection of OLTC degradation using
diagnostic measurements could thus be useful here.
Figure A.22 Example of transformer windings that are destroyed due to OLTC failure,
showing that protective mechanisms do not always react quickly enough to prevent
transformer damage.
188
APPENDIX B INDIRECT MEASUREMENT OF THE OLTC DYNAMIC

RESISTANCE
Appendix B
B.
(<- appendix numbering)
Indirect measurement of the OLTC

dynamic resistance
This appendix deals with the measurement of the voltage on the secondary side of
the transformer when the OLTC switches a DC current. In this way, the OLTC
resistance is measured indirectly.
B.1
Introduction
The resistance of an OLTC changes abruptly when the transition resistors are
switched in the circuit. Results of a dynamic resistance measurement should show
this change in resistance directly. However, Figure 6.5 showed that the inductance
of the transformer windings prevents rapid changes in the current. The result is
that the dynamic resistance graphs also show effects of the inductance.
The first way to improve the response of the measurement current is discussed
in Chapter 4.3: connecting an extra resistor in series with the OLTC. This
improves the response time of the system (which is described by =L/R). A
disadvantage is that the amplitude of the signal is reduced, resulting in an
insufficient degree of accuracy (as shown in Figure 4.17).
A second way to reduce the effect of the winding inductance was introduced in
Section 4.1.1: the voltage over the transformer windings can change quickly when
the transition resistors are switched in contradiction to the current through the
windings. Measuring this voltage on the secondary side of the transformer at the
moment the OLTC switches a DC current will indicate the OLTC resistance
indirectly. The effect of the inductance is reduced while the signal-to-noise ratio is
kept high. This allows the OLTC resistance to be analysed with a high degree of
time accuracy.
189

RESISTANCE
B.2
Dynamic resistance measured on the secondary side
As discussed in the previous section, the advantage of performing a dynamic

resistance measurement on the secondary side of the transformer is the fast
response to changes in the circuit resistance. There is also a drawback of this
method: a DC current through the transformer windings is dangerous when it is
interrupted. There is no alternative path available for the current due to the open
secondary side. An attempt to interrupt the DC current will result in a high voltage.
These interruptions can be caused by defects in the OLTC or by operation errors.
To prevent these situations, the OLTC should firstly be checked for open circuits
before performing this measurement. This can be done by a dynamic resistance
measurement with the secondary side of the transformer short circuited. Secondly,
the measurement circuit should be completely discharged after the measurement.
Figure B.1 shows the procedure to perform this measurement in a safe way.
Figure B.1 Procedure for a dynamic resistance measurement with an open circuit on the
secondary side of the transformer. The circuit should be checked for open contacts during
OLTC operation before the measurement is performed.
A simple setup is used to measure the secondary voltage when the OLTC is in
operation (Figure B.2). A measurement voltage is applied to the primary side of the
transformer and a current begins to flow. The voltage over the inductances will be
zero during DC conditions. Therefore, no voltage is measured on the secondary
side when the measurement circuit is stable. In this case the applied voltage Utest
equals the voltage over the circuit resistance UR.
Figure B.2 Simple representation of the measurement setup for measuring the dynamic
resistance of the OLTC on the secondary side of the transformer.
The transition resistor (Rtransition) is added to the circuit when the OLTC is in
operation. This causes a change in the voltage on the transformer inductance that
can be measured on the secondary side of the transformer. Because the
measurement current through the transformer windings is not able to change
quickly, it can be assumed that the measurement current will not change as long as
the OLTC transition time remains less than the time constant of this LR-circuit.
190

RESISTANCE
Therefore the voltage over Rcircuit remains at Utest (and Utest has a fixed value). The
voltage UT = Itest*Rtransition will appear on the primary transformer inductance (so
UL = UT). This can be measured on the secondary side; however, the amplitude
will be different for each tap position due to the changing transformer turn ratio.
Figure B.3 shows an example of the secondary voltage that is recorded when
the OLTC on the primary side switches to another winding tap. The measurement
was performed on a 50/10 kV Ynd7 power transformer equipped with an OLTC
with one transition resistor of 2.4 (pennant-cycle operation). Figure B.3 shows
the effect of this transition resistor, which is used for about 35 ms. The
measurement was performed with 1 A and a series resistor of 23 .
Figure B.3 Example of a secondary voltage measurement of an OLTC operation. The

transition resistor is measured directly.
B.3
Effect of the circuit resistance
In the previous section it was assumed that the current through the measurement
circuit will not change due to the quick OLTC transition time. In practice, the
measurement current will change in reaction to the extra resistance during the
OLTC operation. The voltage that is measured on the secondary side will decrease.
This effect can be seen in Figure B.3 as a small decrease in the amplitude of the
secondary voltage during the OLTC operation.
Changes in the primary current can be reduced by using a series resistor or by
adjusting the measurement voltage during the measurement. The measurement
circuit can be calibrated for each position of the OLTC by briefly adding a resistor
with a known value in series with the voltage source and reading the change in the
secondary voltage.
The effect of permanently adding an additional series resistor in the
measurement circuit affects the time constant of the system. The time constant is
191

RESISTANCE
proportional to Ltotal/Rtotal. Figure B.4 shows an example of a secondary voltage

measurement of the same tap change operation: the deepest curve was measured
with a series resistor of 15 and the upper curve with 23 . The secondary
voltage returns to zero faster when the primary circuit resistance is higher. In
contrast, a small circuit resistance results in a slow system, so the inductance has
no chance to affect the results during the OLTC operation. This is desirable
because the inductance should not be measured; only the OLTC resistance is of
interest.
23
15
Figure B.4 Example of a secondary voltage measurement of an OLTC operation: the
deepest curve is measured with a series resistor of 15 and the upper curve with 23 .
B.4
Effect of the transformer ratio
A second variation in the secondary voltage measured is caused by the transformer

ratio. The value of the transition resistor is equal for each tap position, but on the
secondary side of the transformer it is measured with different amplitudes due to a
different transformer ratio. The effect of the transformer ratio on the amplitude of
the secondary voltage is dependent on the regulating range of the transformer and
is typically around 20%. Figure B.5 gives an example of a secondary voltage
measurement of an OLTC operation: three tap change operations are shown. The
different transformer ratio at these tap positions caused a variation of 25% in the
amplitude of the secondary voltage.
192

RESISTANCE
tap1->2
tap10->11
tap18->19
Figure B.5 Example of a secondary voltage measurement of an OLTC operation: three tap
change operations are shown. The different transformer ratio caused a variation of 25% in
the amplitude of the secondary voltage.
B.5
Discussion
This appendix discussed the secondary voltage at the moment the OLTC switches
a DC current. An advantage of this method is the fast response to changes in the
resistance on the primary side of the transformer. In this way, the effect of
inductance on the measurement is minimal. A disadvantage is the danger of
interrupting the DC current.
Two effects were discussed: the effect of the transformer turn ratio and the
effect of the primary circuit resistance (which determines the time constant of the
circuit). A high resistance is desired on the primary side to keep the changes in the
measurement current low. However, a low resistance is desired to keep the LRsystem slow (the time constant L/R is then large compared to the OLTC transition
time).
193
APPENDIX C EXAMPLES OF STATIC RESISTANCE MEASUREMENTS WITH

INCREASING TEST CURRENT ON SERVICE-AGED OLTC CONTACTS
Appendix C
C. (<- appendix numbering)
Examples of static resistance

measurements with increasing
current on service-aged OLTC
contacts
This appendix presents examples of static resistance measurements on service-aged
change-over selector contacts. It is shown how degraded contacts react to different
amplitudes of the measurement current. The contact resistance memory effect and
the power losses on the contacts are also shown. Typical values of static resistance
measured on clean and service-aged OLTC contacts are plotted as a function of
measurement current intensity. It can be seen that:
Resistance of clean silver-plated contacts is approximately constant and stays

within the hundreds of micro-ohm range (see the example of Figure C.2)
Resistance of degraded contacts measured at a low current of 2.5 A may be as
high as hundreds of milliohms (see the example of Figure C.3)
The difference between the clean and degraded contacts can be readily detected
when resistance is measured at a low current. However, such differences in contact
resistance become smaller when the current is increased.
Measurement at a low current allows for distinguishing between the contact
and winding resistance. The change of winding resistance caused by temperature
variations is negligible at a low current and does not mask the contact resistance of
degraded contacts.
Section 2.3.5 uses measurement results from static resistance measurements to
derive a correction factor CDRM to account for the effects described in this
appendix.
195

Contact resistance
Resistance (mOhm)
100
10
Clean
Aged
0,1
0,00
50,00
100,00
Current (A)
Figure C.1 Example of static resistance measurements on service-aged and silver-plated
change-over selector contacts.
0,8
Resistance (mOhm)
0,75
0,7
0,65
Up
0,6
Down
0,55
0,5
0,45
0,4
0,00
20,00
40,00
60,00
80,00
100,00
120,00
Current (A)
Figure C.2 Example of static resistance measurements on silver-plated change-over selector

contacts. The resistance is about 600-700 .
196

Resistance (mOhm)
350
300
250
200
150
100
50
0
3,00
5,00
10,00
25,00
50,00
100,00
Current (A)
Figure C.3 Typical example of static resistance measurements on service-aged change-over
selector contacts with a severe long-term aging effect. The initial resistance can be high, but
decreases when the current is increased.
Similar resistance variations in degraded circuit breaker contacts measured with a

current up to 2800 A were reported in [77]. Figure C.4 shows the breaker contact
resistance as a function of the measurement current, Figure C.5 on its main
contacts. The main circuit breaker contact resistance behaves similarly to clean
OLTC contacts, while arcing contacts of a circuit breaker are comparable with
OLTC contacts that suffer from the long-term effect. The current was increased to
875 A but the resistance readings remained current dependent ([77] shows that this
effect is still present at 2800 A).
Resistance measurement of windings and an OLTC without movable contacts
shows a very small resistance variation (in this case 0.6 m) when the current is
increased to 100 A (see Figure C.6). The changes in contact resistance discussed in
this appendix are therefore caused by movable change-over selector contacts (the
arcing switch is not measured).
197

2500
Ra ()
2000
1500
1000
500
0
0
100
200
300
400
500
600
700
800
900
1000
DC injected current (A)

Figure C.4 Example of static resistance on circuit breaker arcing contacts, as discussed in
[77].
300
250
200
150
100
50
0
0
100
200
300
400
500
600
700
800
900
1000
DC injected current (A)

Figure C.5 Example of static resistance on circuit breaker main contacts, as discussed in
[77].
198
Resistance (mOhm)

14,60
14,50
14,40
14,30
14,20
14,10
14,00
13,90
13,80
0,00
Up
Down
20,00
40,00
60,00
80,00 100,00
Current (A)
Figure C.6 Example of static resistance measurement on parts of the OLTC and tapped
windings without movable contacts. The resistance only changes 0.6 m when the current is
increased to 100 A.
Static resistance measurements on service-aged change-over selector contacts can

show a memory effect. Resistance decreases when the current is increased to 100 A
but remains low when the current is reduced again. A typical result is shown in
Figure C.7 and an example of DRM results for all change-over selector contact
positions is shown in Figure C.8.
Clean silver-plated contacts show very little memory effect, with the resistance
following almost the same line when the current is increased or decreased.
Figure C.9 gives a typical example of the power loss on change-over selector
contacts that suffer from the long-term effect. These losses were measured during
a static resistance measurement when the current was increased to 100 A. During
these measurements, the power loss on the contact can be as high as 150 W for
each phase. Power losses during normal operation can be much higher, resulting in
higher contact temperatures.
199

R e s is ta n c e (m O h m )
350
300
250
200
Up
150
Down
100
50
0
0,00
20,00
40,00
60,00
80,00
100,00
Current (A)
Figure C.7 Typical example of static resistance measurements on service-aged change-over
selector contacts with a severe long-term aging effect. The initial resistance can be high, but
decreases when the current is increased and remains low when the current is reduced.
R esistance (m O hm )
350
300
250
200
Up
150
Down
100
50
0
0
20
40
60
Current (A)
80
100
Figure C.8 Typical example of static resistance measurements on all contact positions of a
service-aged change-over selector contacts with a severe long-term aging effect. The initial
resistance can be high, but decreases when the current is increased and remains low when
the current is reduced.
200

160
140
Power (W)
120
100
PhaseA
PhaseB
PhaseC
80
60
40
20
0
0,00
20,00
40,00
60,00
80,00
100,00
Current (A)
Figure C.9 Example of power loss on change-over selector contacts during static resistance
measurements when the current is increased to 100 A. The contacts suffered from the longterm effect.
201
APPENDIX D DESIGN CONSIDERATIONS OF A TRANSFORMER DIAGNOSTIC

SYSTEM
Appendix D
D. (<- appendix numbering)
Design considerations of a
transformer diagnostic system
This appendix gives a more detailed description of the transformer diagnostic
system proposed in Chapter 5.
Firstly, some examples are given of a user interface that presents key aspects of
the DRM results to the user. Secondly, the method used to test the transformers
winding resistance and turn ratio is briefly described. Thirdly, the hardware design
of such a transformer diagnostic system is discussed in detail; and finally, ways of
processing the measurement data before it is presented to the user are explained.
D.1
Examples of a DRM user interface
It was mentioned in Section 5.3.1 that DRM measurement results should be

displayed in such a way that critical defects should be recognisable. An example of
one simple DRM user interface for presenting relevant measurement results has
already been shown in Figure 5.9. Three further examples of such user interfaces
are given below.
1. Figure D.1 shows an example of an overview plot for a delta-connected
OLTC with contact degradation (with the three phases measured on different
time scales).
2. Figure D.2 shows the same measurement data in the form of an impedance
overview plot. The voltage was held constant during the measurement so the
impedance graph looks like the current graph, only upside down.
3. Figure D.3 gives an example of a user interface that shows DRM close-up
plots for each OLTC operation, the timing of the excessive transition time
protection and a plot of the estimated resistance for each tap position.
The transition times of each OLTC operation are shown on the right of the user
interface, in the form of a Table (Figure D.1) or a graph (Figure D.2).
203

SYSTEM
Figure D.1 DRM results for the phases of delta-connected OLTCs are recorded on a
different timescale. Results can be displayed as current.
Figure D.2 Signals for mechanical diagnosis can be checked prior to the measurement to
reveal erroneous connections. Motor voltage is shown in this example.
204

SYSTEM
D.2
Procedure for OLTC drive system measurements
The measurement procedure for DRM has already been discussed in Chapter 5.
This section deals with the mechanical diagnosis of the OLTC and OLTC drive
mechanism, which depends on the diagnostic methods selected, the local situation
in the substation and the transformer type used. The transformer type and
implementation determine which diagnostic methods are technically possible. For
example, if the drive shaft is inaccessible the angular position of the shaft cannot
be measured and the OLTC drive may not be equipped with excessive OLTC
switch-time protection. In general, the procedure illustrated in Figure 5.4 applies.
Firstly, the transformer is de-energised (Figure 5.4.1) and the clamps and
sensors are connected to the test object (Figure 5.4.2). Secondly, a short signal
check, by switching the OLTC once, can reveal erroneous connections before the
measurement begins (Figure 5.4.3c). Figure D.4 shows a signal check screen for
motor power signals (motor voltage is displayed), and Figure D.5 a signal check of
the excessive switch-time contact voltage.
No charging or discharging is necessary when the OLTC drive system
condition is assessed separately from DRM, for example in the case of a drive
motor power measurement when the transformer is energised, because no DC
reveal erroneous connections. Excessive switch-time contact voltage is shown in this
example.
205

SYSTEM
reveal erroneous connections. Motor voltage is shown in this example.
reveal erroneous connections. Excessive switch-time contact voltage is shown here.
206

SYSTEM
current is applied to the transformer windings in such a case. As discussed in

Section 3.2, measurements of the OLTC drive system can also be performed in
combination with DRM. If all of the methods proposed in Section 5.2 are selected,
simultaneously recorded signals can be related to each other on the same timescale.
Although the mechanical properties of the drive system and drive shaft can be seen
when one tap change operation is recorded, friction may occur at specific tap
change operations. Therefore, measuring the complete cycle of operation is
advisable.
D.3 Winding resistance and turn ratio measurement

procedure
The procedure and transformer connection used in DRM can also be used in
winding resistance measurements (as showed in Figure 5.4 and Figure 5.5). Both
methods could also be combined into a single measurement. The primary and
secondary transformer terminals are connected to the diagnostic system and the
system configures the measurement circuit. Compared to DRM, the charging
process is prolonged and a longer period between OLTC operations must be
maintained because the current needs to be stable so that a high level of accuracy
can be attained. The stability of the resistance reading can be used to determine
when a pure DC situation has been reached. Each tap position should be measured
while discharging between measurements on the same phase is unnecessary.
Discharging is identical to DRM: a resistor is switched in added to the circuit
until the energy from the windings is dissipated.
The turn ratio is measured by applying a low AC voltage to the high-voltage
windings of the transformer using the output stage. Again the measurement
procedure of Figure 5.4 applies and all taps should be measured. A short
measurement time is sufficient to obtain high accuracy and a discharge process is
not needed when using AC voltages.
D.4
Hardware configuration
The hardware which is schematically depicted in Figure D.6 can be used to

facilitate the diagnostic measurements described in the previous sections. The
diagnostic system (Figure D.6B) forms the interface between the user (Figure
D.6A) and the transformer under test (Figure D.6C). Inside the diagnostic system,
the embedded control (Figure D.6.4) converts the demands of the user into
control signals for the power source (Figure D.6.1) and the output stage (Figure
D.6.2). The power source is needed to energise the windings during the off-line
resistance and ratio measurements.
The main function of the output stage is to configure the measurement circuit
by connecting the power source to the test object and to apply a secondary short
circuit. Meanwhile, the embedded control converts the diagnostic parameters, as
207

SYSTEM
measured by the signal conditioning block (Figure D.6.3), in such a way that they
can be displayed to the user (for example by scaling and filtering the signals). Its
four basic building blocks are discussed in this section for each of the selected
diagnostic measurements. Figure D.7 provides a more detailed schematic of the
output stage, which forms the interface between the voltage sources and the power
transformer. Its basic functions include configuration of the measurement circuit,
discharging and current measurements. The specific demands for this block
diagram are also discussed in this section.
D.4.1 Dynamic resistance of the OLTC
For each building block of Figure D.6.1-4, technical demands for performing
DRM are set up and summarized in Table D.1.
Power source
As a power source (Table D.1.1), a controllable DC voltage source is necessary to

supply a DC current. Due to the short circuit on the secondary windings, the
current rises quickly and a low source voltage can be used (for example, a few volts
are sufficient when measuring with currents 5 A). The diagnostic system used for
the work of this thesis is designed to take resistance measurements up to 8 A
and 30 V, ratio measurements are performed with 20 V at 0-300 mA magnetizing
current.
1
2
Power source Energize the test object

Output Stage Configure the test circuit:
- connect the power source to the test object
- apply secondary short circuit
3 Signal conditioning Scale measured signals
4
Embedded control Converts user demands into control signals for the
power source and the output stage
Scale and filter meaured signals for display to the
user
Figure D.6 Simple block representation of the proposed transformer diagnostic system and
the main function of the four building blocks.
208

SYSTEM
Output stage
The main functions of the output stage (Table D.1.2) during DRM include
configuration of the measurement circuit, discharging and overvoltage protection:
1. The output stage configures the measurement circuit using relays (Figure
D.7D).
2. During the discharge process, the discharge resistor is added to the circuit by
the output stage (Figure D.7B) to dissipate the energy stored in the
transformer windings safely. During discharge, the voltage on the transformer
bushings is dependent on the total measurement current and the value of the
discharge resistor. This voltage can exceed the maximum range of DC power
supply and the signal conditioning. The value of the discharge resistor should
be selected based on this discharge voltage.
3. The output stage also offers overvoltage protection (Figure D.7E). An
overvoltage is shunted to ground by overvoltage diodes or spark gaps, for
example in the case of a current interruption, thereby reducing the stress on
the signal conditioning block.
Table D.1 Hardware demands for DRM with the proposed transformer diagnostic system.
Demands for the basic building blocks of Figure D.6 are listed and discussed in this section.
1
Power source Stable DC source, controllable by embedded control
for automatic accurate charging
2
Output Stage Three current measurements
Relays to connect power source to primary
transformer connection
Relays to short-circuit transformer secondary side
Overvoltage protection
Discharge resistor
3
Signal conditioning Voltage measurement on primary transformer
connection
4
Embedded control Digital outputs for relay control of the output stage
Analogue output for voltage source control
Analogue inputs for voltage and current
measurement
OLTC control relay contacts
5 Transformer connection Four-point measurement to measure the voltage
separately from the injected current
Signal conditioning
The signal conditioning block (Table D.1.3) is responsible for the scaling of the
measured signals, filtering and for the electrical insulation between the embedded
control and the transformer.
209

SYSTEM
Figure D.7 Detailed schematic of the output stage, as shown in Figure C.6, for the proposed
transformer diagnostic system.
A voltage and current measurement are needed to record the dynamic resistance.
Three phase measurements can be supported when three current and voltage
measurements are implemented. These DC currents, for example in the range of 010 A, can simply be measured by means of a current shunt (Figure D.7C). Voltage
measurements, by means of voltage dividers, are needed on all terminals on the
OLTC side of the transformer. The maximum range is determined by the
maximum voltage of the DC voltage source or by the maximum current multiplied
by the value of the circuit resistance, including the discharge resistor.
Embedded control
The embedded control (Table D.1.4) selects the voltages and currents that are used
for the calculation based on the transformer vector group, the winding
configuration and the tested phase. Analogue measurement channels are included
to measure these signals at 1 kS/s (or higher when oversampling is desired). OLTC
control is added to the proposed transformer diagnostic system, which gives an
OLTC operation command in case the next tap can be measured. Relay contacts
are included to bridge the OLTC control buttons in the OLTC drive control box.
In this way, the results are easier to interpret because the OLTC is switched at
regular intervals.
Transformer connection
Four-point measurement must be used according to international test standards to

measure the voltage on the transformer terminals [102-103] (Table D.1.5).
210

SYSTEM
D.4.2 OLTC drive system condition

The hardware configuration for the diagnosis of the mechanical condition consists
of signal conditioning with filtering, amplitude adjustment and electrical insulation.
Its hardware demands for the building blocks of Figure D.6 are discussed in this
section and are summarized in Table D.2.
The proposed transformer diagnostic system is equipped with a multi-purpose
connector which can be configured to support the diagnostic measurements
described in Section 3.2.
Power source
As is standard, the transformer diagnostic system measures the active drive motor
power, the remote end drive shaft rotation and the timing of a protective contact
(see Section 5.2). Only an auxiliary power source is necessary (Table D.2.1). This
low-voltage auxiliary power source is available for measurement of the contact
voltage of de-energised protective contacts.
Output stage
Overvoltage protection of the output stage (Table D.2.2) should be designed to

manage the incorrect connection of this auxiliary voltage.
Signal conditioning
Inside the signal conditioning block (Table D.2.3), active motor power is measured
by recording the motor voltage (phase-to-neutral, see Section 5.4) and motor
current (of one phase, see also Section 5.4). Motor current is measured using a
standard current probe, and motor voltage by an electrical insulated voltage
divider. The remote end drive shaft rotation is measured using a position sensor
connected to a 0-10 V analogue input without specific hardware demands.
The timing of a protective contact is measured by recording the contact
voltage, for example at an OLTC-in-operation contact.
Embedded control
The embedded control (Table D.2.4) has analogue measurement channels available
to digitize the analogue signals from the signal conditioning block (at 1 kS/s or
higher when oversampling is desired). Control signals are sent to the power source
and output stage using digital outputs.
Transformer connection
An example of transformer connections (Table D.2.5) is depicted in Figure D.8.

Clamps for the motor voltage and the excessive switch-time contact are connected
to the drive mechanism and a current probe is used to measure the motor current
(Figure D.8A). A rotation sensor for the drive shaft rotation is connected to the
remote end of the drive shaft (Figure D.8B).
211

SYSTEM
Table D.2 Hardware demands for diagnosis of the mechanical condition of the OLTC and
OLTC drive using the proposed transformer diagnostic system. Demands for the basic
building blocks of Figure D.6 are listed.
1
Power source Auxilliary voltage source to measure protective
contacts
2
Output Stage Overvoltage protection
3
Signal conditioning Insulated line voltage measurements of drive motor
voltage and protective contacts
Noise reduction on drive motor current and axis
rotation signal
4
Embedded control Digital output for relay control of auxilliary voltage
Analogue inputs for signal measurement from signal
conditioning
5 Transformer connection Current probe to measure drive motor current
360 rotation sensor for axis rotation
D.4.3 Winding resistance and turn ratio

Table D.3 lists the hardware demands for the building blocks of Figure D.6 for the
measurement of DC winding resistance, in addition to the demands for DRM. The
same voltage and current are measured: differences include a higher accuracy and
an extra current measurement of the DC current on the secondary side of the
transformer.
The winding resistance is relatively low compared to the inductance; values of
0.1-2 are typical for power transformers in the high-voltage distribution and
transmission network. Compared to dynamic resistance measurements, DC
Figure D.8 Example of a connection for motor voltage and current measurement, using
clamps and a clamp-on current probe (A), and an example of a rotation sensor connection
on the remote end of a selector switch rotor (B).
212

SYSTEM
winding resistance measurements can use a high output DC voltage to saturate the
transformer core faster, thereby achieving a DC situation more quickly.
Power source
The transformer turn ratio can be measured by adding an AC voltage source.
Using the same output stage as defined in Figure D.7, this low AC voltage is
applied to the high-voltage windings of the power transformer. Voltage is
measured on the primary and secondary transformer terminals and the voltage
difference between two terminals is used for ratio calculation.
Typical power transformer ratios are in the range of 1:2 to 1:15. The proposed
diagnostic system should be accurate in this range (a deviation in the turn ratio of
maximum 0.5% is considered acceptable).
Table D.3 Hardware demands for winding resistance testing and transformer turn ratio
measurements with the proposed transformer diagnostic system. Additional demands not
found in Table D.1 are listed for the basic building blocks of Figure D.6.
1
Power source low AC voltage, line frequency, 3 phase
2
Output Stage 1 more current measurement for secondary side
3
Signal conditioning no additional demands
4
Embedded control no additional demands
5 Transformer connection no additional demands
Output stage
In the output stage, the primary excitation current is also measured. This current is
in the range of milliamps and can be measured with the same hardware as is used
for winding resistance measurements.
D.5
Data processing and analysis
Measurement data must be analyzed immediately after each measurement or when

the stored measurement data is being opened. This section is dedicated to filtering,
processing and selection of relevant measurement data by the proposed
transformer diagnostic system, whereupon it is displayed to the user. Each of the
diagnostic methods of the transformer diagnostic system faces different issues,
which this section elaborates on.
D.5.1 Dynamic resistance of the OLTC
The measurement current during DRM has a typical time constant in the order of
tens of milliseconds. Dynamic resistance results can therefore be processed and
shown in 1ms resolution without losing information about the OLTC condition.
213

SYSTEM
Current [A]
Current [A]
In order to generate the results displayed in Figure D.1-3, the recorded voltages
and currents are filtered. In this way, undesired signals that are superimposed on
the recorded voltages and currents can be removed. In addition to hardware noise
filters, digital signal processing (e.g. low pass filters and sample averaging) and
oversampling can be used to optimize the accuracy.
Figure D.9 gives two examples of the most common noises in the DRM
signals. Firstly, spikes can be visible on the recorded signals (Figure D.9A, at
1100 ms), for example due to switching transients that are generated when
operating the OLTC motor drive. These current spikes cannot flow through the
transformer windings due to the slow time constant and must therefore be caused
by noise. A spike filter can be applied to the recorded measurement current to
remove these spikes.
Secondly, net frequency noise (Figure D.9B) can occur, especially when using
unshielded measurement cables, which makes the determination of the OLTC
transition time impossible and reduces the sensitivity of DRM to low-resistive
defects. Digital net frequency filters can be applied to reduce this noise, but care
should be taken, as excessive filtering can affect the determined OLTC transition
times.
Figure D.9 Example of a DRM current with A) a spike on the recorded current at 1100 ms
and B) net frequency noise during OLTC operation.
After filtering, the filtered signals can be used to calculate the impedance and to
determine the moments of OLTC operation. Close-up views of the plots of the tap
change operations are made (see Figure D.10A) to study defects in the arcing
switch in more detail. A close-up period of 300 ms is sufficient, because OLTCs
typically have operating times of 200 ms. Graphic presentation of tap position
characteristics can give a clear indication of irregular taps. The proposed diagnostic
system graphically presents:
a.
A median of the measurement current or impedance of each tap position and

for each phase (Figure D.10B).
b. The fluctuation of the measurement current or impedance (maximum minus
minimum value) in each tap position and for each phase (Figure D.10C).
214

SYSTEM
c.
OLTC transition times for each phase and for each tap transition (Figure
D.10D), based on the moment the main contact of the arcing switch breaks
and restores the contact.
Chapter 6 elaborated further on the analysis of DRM results.
Figure D.10 Analysis of a DRM file results in close-up plots of the current or impedance
during OLTC operation (A), plots of the median values (B) and fluctuation (C) of the current
or impedance for each tap position and a transition time plot (D).
D.5.2 Winding resistance and turn ratio

For winding resistance measurements, the proposed transformer diagnostic system
averages the measured current and voltage over multiple seconds to increase the
accuracy of the measurement. These average values are used to calculate the
resistance and should be stable before final resistance values can be determined.
In the case of a wye-connected transformer winding without a neutral point,
the current can flow from one phase to another while measuring the voltage on the
third phase. Values read from delta-connected transformer windings can be
multiplied by 1.5 to obtain the one-phase winding resistance in the case of equal
winding resistances, because the two other windings are measured in parallel.
Values obtained from delta-connected windings with unequal winding resistance
can be corrected using the formulae found in [104]. After this calculation and
215

SYSTEM
correction, values for each phase and tap position are displayed, together with the
accuracy that was reached by the diagnostic system.
Average values can also be used during turn ratio testing: the proposed
transformer diagnostic system averages the measured primary and secondary
voltages until the desired accuracy is reached. Ratio measurements on some
transformer vector groups are also subject to correction factors to convert the
measured ratio to the transformer turn ratio. A factor of 3 should be used when
line-to-line values are compared with line-to-neutral values. For example, an Yyn0
transformer indicates the turn ratio directly while the measured turn ratio at an
Ynd7 transformer needs a factor of 3 to obtain the line-to-line voltage ratio. In
cases where there is no accessible neutral, a factor of 0.5*3 can be applied [105].
In addition to the turn ratio, information about the presence of winding and
core defects is also obtained by the proposed diagnostic system through the
measurement of:
a. The primary excitation current
b. The phase angle between the primary measurement voltage and secondary
voltage
c. The phase angle between the primary measurement voltage and primary
current
216
Lists
List of abbreviations
AC
ANSI
DC
DETC
DGA
DRM
HV
IEC
IEEE
KEMA
LV
OLTC
RMS
Alternating Current
American National Standards Institute
Direct Current
De-Energised Tap Changer
Dissolved Gas Analysis
Dynamic Resistance Measurement
High Voltage
International Electrotechnical Commission
Institute of Electrical and Electronics Engineers
NV tot Keuring van Elektronische Materialen te Arnhem
Low Voltage
On-Load Tap Changer
Root Mean Square
List of gasses
C2H2
C2H4
C2H6
C3H6
C3H8
CH4
H2
T
Acetylene
Ethylene
Ethane
Propene
Propane
Methane
Hydrogen
Total amount of CH4 + C2H4 + C2H2
List of units
Degree
Degree Celsius
Micro Ohm
Ohm
Ampere
ngstrm
Farad
217
H
Hz
kg
kS/s
kV
m
mA
MVA
N
s
V
W
Henry
Hertz
Kilogram
Kilo samples per second
Kilo Volt
Milli Ohm
Milli Ampere
Megavolt Ampere
Newton
Second
Volt
Watt
List of symbols
c
m
E, V
T
0
, i, j
eff
(D)
, 1, 2
i1, i2
,
eff
A
a
ac
ai, aj
am
An
B
cos
cosN
C, C1, C2
CDRM
COLTC
218
Angle
Apparent contact radius (meter)
Bulk contact resistivity factor
Contact area correction factor
Film thickness (ngstrm)
Critical film thickness (ngstrm)
Maximum film thickness (ngstrm)
Electric potential difference (Volt)
Temperature difference (degree Kelvin)
Homologous temperature in relation to the melting point
Contact surface temperature (degree Celsius)
Surface resistivity of contact film (Ohm-square meter) at spot i and j
Effective thermal conductivity of contact film (Watt/Kelvin)
Degree of truth
Specific electric resistivity (Ohm-meter)
Time constant (Seconds)
Constants used to describe the effect of the current amplitude
Phase angle (Degree)
Effective electrical resistivity of contact film (Ohm-meter)
Cross sectional contact area (square meter)
Contact radius (meter), amount of coke precursor (g/cm3)
Critical contact spot radius (meter)
Contact spot radius of spot i of j (meter)
Initial maximum contact spot radius (meter)
Winding tap (n = 1,2,)
Temperature dependent diffusion parameter
Power factor
Power factor of phase N (N = A..C)
Constant
Correction factor for the contact resistance
Correction factor for the contact resistance
CT
CW
D
D0
di
dI/dt
dij
dR
E
Ea
f
f(x)
I, I0, In, i
I exp
I high
Ii
Ij
IL
I low
I min
IN
i N(t)
ioffset
I Phase
i(t=x)
Itest
K
k
k0, k1, k2, k3
K
kdyn1, kdyn2
L
L1
L1
L2
L3
M
N, n
n, n(t)
n0
Stator contact of a change-over selector (connected to the coarse

tap)
Coefficient of contact wear
Deviation in the DRM graph (% of expected value), distance
(meter), diameter (meter), diffusion coefficient of contact film
(m2/s)
Frequency factor of contact film growth (m2/s)
Current difference due to changes in circuit resistance (Ampere)
Derivative of the current (A/s)
Distance between centres of contact spots i and j (meter)
Dynamic resistance variation (Ohm)
Electric potential (Volt)
Activation energy of the coke deposition process (Joule per mol)
Frequency (Hertz)
Contact spot radius distribution
Current (Ampere)
Expected current in DRM plot (Ampere)
Trend line of the DRM current of a coarse tap with highest current
(Ampere)
Injected current (Ampere), current through contact spot I (Ampere)
Current through contact spot j (Ampere)
Load current (Ampere)
Trend line of the DRM current of a coarse tap with lowest current
(Ampere)
Minimum current in DRM within a tap position (Ampere)
RMS current of phase N (N = A..C) (Ampere)
Current of phase N (N = A..C) as function of time (Ampere)
Constant used to describe the effect of the current amplitude
RMS current in balanced operation (Ampere)
Current at time x (Ampere)
Test current (Ampere)
Coke deposition rate (cm/day)
Coupling factor, critical contact resistance increment
Coefficients
Coefficient of approximation derived by [32]
Constant used to describe the effect of the current amplitude
Inductance (Henry), Lorenz constant (Volt2/degree Kelvin2)
Primary inductance (Henry)
First transformer connection
Second transformer connection
Third transformer connection
Mid-position of the change-over selector contacts, stator contact of
a change-over selector (connected to the main winding)
Transformer ratio, transformer neutral connection, number
Number of contact spots, as a function of time
Initial number of contact spots
219
ny
P3
p3(t)
ppm
R
R(t)
r
R0
R1
R2
R2,5
r(t)
Ra
RB
Rc
Rc(t)
Rcircuit
Rclean
Rfilm
Ri
Rinteraction
RM
Rmeasure
Rny=0
Rp
Rparallel
RS
Rt
Rtunnel
Rtw
Rw
S
s
T
T, t1, tn
Tb, TB
Tc
TEND-X
TI<10%
TK
tl
Tm
TM
tm
Tn
220
Number of OLTC operations per year

Three-phase RMS power (Watt)
Three-phase instantaneous power (Watt)
Parts per million
Resistance (Ohm), gas constant (Joule per molKelvin)
Resistance as a function of time (Ohm)
Effective contact radius (meter)
Initial contact resistance (Ohm)
Primary resistance (Ohm)
Secondary resistance (Ohm)
Contact resistance measured at 2.5 Ampere (Ohm)
Effective average contact radius as a function of time (meter)
Contact resistance of a circuit breaker (Ohm)
Bulk resistance (Ohm)
Contact resistance (Ohm)
Contact resistance as a function of time (Ohm)
Circuit resistance (Ohm)
Contact resistance of a clean metal-to-metal contact (Ohm)
Resistance of a thick film (Ohm)
Contact resistance measured at i Ampere (Ohm)
Contact resistance due to interaction between contact spots (Ohm)
Measured resistance (Ohm)
Measurement resistance (Ohm)
Contact resistance when the OLTC is not operated (Ohm)
Contact resistance of a circuit breaker (Ohm)
Contact resistance due to parallel contact spots (Ohm)
Resistance at reference temperature TS (Ohm)
Transition resistance of the OLTC (Ohm)
Tunnelling resistance of a thin film (Ohm)
Resistance of a tapped winding (Ohm)
Winding resistance (Ohm)
Contact surface area (square meter)
Thickness of the contact film layer (ngstrm)
Temperature (degree Kelvin)
Time, duration (Seconds)
Bulk temperature (degree Kelvin)
Super temperature increment over the bulk temperature (degree
Kelvin)
End time, with X = A..C (Seconds)
Time that the current is below 10% of its expected value (Seconds)
Winding material constant
Contact lifetime (Seconds)
Melting point of the contact material (degree Kelvin)
Temperature at which resistance was measured (degree Celsius)
Maximum contact lifetime (Seconds)
Time (n = 1,2,) (Seconds)
TP
TR1, TR2
TS
TTR
U
UL
UR
UT
Utest
V
VN
v N(t)
VPhase
W
Wa
Wa0
X, X
Xn
YNd
Yyn0
Z
Z1
Time with two transition resistors in parallel (Seconds)

Time with one transition resistor in use (Seconds)
Reference temperature (degree Celsius)
OLTC transition time (milliseconds)
Voltage (Volts), first transformer connection
Voltage over winding inductance (Volt)
Voltage over a resistor (Volt)
Voltage over transition resistor (Volt)
Test voltage (Volt)
Second transformer connection, voltage (Volt)
RMS line-to-neutral voltage of phase N (N = A..C) (Volt)
Line-to-neutral voltage of phase N (N = A..C) as function of time
(Volt)
RMS line-to-neutral voltage in balanced operation (Volt)
Third transformer connection
Stator contact of a change-over selector
Length (meter)
Position (n = 1,2,...) (millimetre)
Transformer vector group: primary star with neutral, secondary
delta
Transformer vector group: primary star, secondary star with neutral,
zero phase shift
List of publications
[1]
J.J. Erbrink, Tap changer diagnostics on high voltage power transformers

using dynamic resistance measurement, Graduate thesis Delft University of
Technology, Delft, the Netherlands, 2007.
[2]
J.J. Erbrink, R.A. Jongen, E. Gulski, J.J. Smit, Tap changer diagnostics on
high voltage power transformers using dynamic resistance measurement, 4th
IEEE young researchers symposium in electrical power engineering,
Eindhoven, the Netherlands, p. 1-5, February 2008.
[3]
P.P. Seitz, B. Quak, J.J. Erbrink, E. Gulski, R. Leich, Advanced on-site

diagnosis of power transformers, International conference on condition
monitoring and diagnosis, Beijing, China, paper 148, April 2008.
[4]
J.J. Erbrink, E. Gulski, J.J. Smit, P.P. Seitz, R. Leich, Experimental model of
aging mechanisms of on-load tap changer contacts, International
conference on condition monitoring and diagnosis, Beijing, China, p. 247250, April 2008.
221
[5]
J.J. Erbrink, E. Gulski, P.P. Seitz, R. Leich, Advanced on-site diagnosis of

transformer on-load tap changer, IEEE International Symposium on
Electrical Insulation, Vancouver, Canada, p. 252-256, June 2008.
[6]
J.J. Erbrink, E. Gulski, J.J. Smit, R. Leich, Experimental model for

diagnosing on-load tap changer contact aging with dynamic resistance
measurements, 20th International conference on electricity distribution,
Prague, Czech Republic, paper 135, June 2009.
[7]
R. Zuijderduin, J.J. Erbrink, E. Gulski, J.J. Smit, R. Leich, Condition

assessment of power transformers OLTC by DGA and dynamic resistance
measurements, Proceedings of the 16th international symposium on high
voltage engineering, Johannesburg, South Africa, 2009.
[8]
J.J. Erbrink, E. Gulski, J.J. Smit, R. Leich, P.P. Seitz, B. Quak, Effect of test
parameters on dynamic resistance measurement results from on-load tap
changers, IEEE International Symposium on Electrical Insulation, San
Diego, USA, 6-9 June 2010.
[9]
J.J. Erbrink, E. Gulski, J.J. Smit, R. Leich, P.P. Seitz, B. Quak, On-load tap
changer diagnosis: interpretation of dynamic resistance deviations, IEEE
2010.
Reproducibility of dynamic resistance measurement results of on-load tap
changers effect of test parameters, International Conference on Condition
Monitoring and Diagnosis, Tokyo, Japan, 6-11 September 2010.
Interpretation of dynamic resistance results of on-load tap changers,
International Conference on Condition Monitoring and Diagnosis, Tokyo,
Japan, 6-11 September 2010.
On-load tap changer diagnosis an off-line method for detecting
degradation and defects: part 1, IEEE Electrical Insulation Magazine, Vol.
26, No. 5, p. 49-59, September - October 2010.
Diagnosis of onload tap changer contact degradation by dynamic resistance
measurements, IEEE Transactions on Power Delivery, Vol. 25, No. 4, p.
2121-2131, October 2010.
[14] J.J. Erbrink, E. Gulski, J.J. Smit, J. Aditya, L.A. Chmura, R. Leich, P.P. Seitz,
B. Quak, Test procedure and test circuit considerations for on load tap
changer dynamic resistance measurement, International Conference on
222
High Voltage Engineering and Application, New Orleans, USA, 11-14

October 2010.
[15] J.J. Erbrink, E. Gulski, J.J. Smit, L.A. Chmura, R. Leich, P.P. Seitz, B. Quak,
Condition assessment of on load tap changers using dynamic resistance
measurements, International Conference on High Voltage Engineering and
Application, New Orleans, USA, 11-14 October 2010.
223
Acknowledgements
Writing this PhD thesis and performing the accompanying research was not
something I could have done alone; the book you have in your hand would not be
there without you! Thank you!
If this is the first page you read (and perhaps the only page you plan to read) I
would like to invite you to read some of the chapters and find out what was
achieved with your help.
Firstly, I would very much like to thank Professor Johan Smit, my supervisor, for
the opportunity to work in the Department of High Voltage Technology and
Management, which is a wonderful and unique place to work!
Special thanks to my daily supervisor Edward Gulski for his enthusiasm and
trust, constructive criticism and coaching during this project.
Thanks to my former colleagues at Seitz Instruments AG, I really enjoyed working
with you all. I would like to thank Paul Seitz for giving me the opportunity to
spend time writing my thesis and for his trust in dynamic resistance measurements
as a diagnostic tool for tap changers. Ben, thank you also for critically thinking
through the issues with me, for your valuable comments on my papers and for
your assistance with LabView. Your ideas really inspired me.
I warmly thank the highly experienced people at Liandon for their hospitality and
for giving me the opportunity to explore the on-site activities there. The practical
aspects and on-site measurement experience were indispensible for this research.
The work would never have been completed by simulations and laboratory studies
alone. In particular, I would like to express my gratitude to Nico Sinnige, Ad van
der Giessen, Jur Beemer, Piet van der Gulik, Martin Snel and Bob Koene for
broadening my view of OLTC maintenance. Rory Leich, thank you for sharing
your thoughts with me and Philip Salverda, thank you for sharing your experience!
In addition, Im also very thankful to:
Ryszard Malewski, thank you very much for your editorial support; your
experience and opinion as an expert in transformer testing were crucial to me,
my professor and my daily supervisor
Roy Zuijderduin and Julian Aditya for your role in this research project as MSc
students and for your assistance during on-site testing
Frank Wester at TenneT, for allowing us to perform OLTC diagnosis in the
TenneT grid
225
Singapore Power Grid Ltd, for their hospitality and for their interest in
dynamic resistance measurements
Tim Noonan and Brendan Diggin from ESB International, as it was of
valuable assistance to also perform measurements in Ireland
Mr Wiesinger from Maschinenfabrik Reinhausen for his valuable comments
and for sharing his extensive experience in servicing OLTCs
Smit Transformers for offering me material to assist in the building of an
OLTC experimental model
Thanks to Ren Orler and colleagues from Spital Schwyz for their excellent
diagnostics and repair after I enjoyed Switzerland a bit too much; Im very happy
that I was able to continue my PhD work again.
My fellow MSc students must also be mentioned here. I have had a wonderful time
in Delft and would not have decided to stay there to work on this thesis otherwise.
Therefore, I want to express my particular appreciation to Rookenmeijer, Lisse,
Putter, LH Groenewoud, triple thanks to Bertus, Friesland, Zuijderduin, Bleeker,
Victor Charlie, Pernis and Rotterdam. Thanks not only for the great time, but also
for your trust in my capabilities as a PhD researcher, even before I decided to start
my PhD research.
At the University of Delft, I want to thank Paul van Nes, Bertus Naagen and
Wim Termorshuizen for their support, patience and ideas during my experiments
in the High Voltage Laboratory, and Aad van der Graaff for his extremely accurate
mechanical engineering. And of course I want to thank my fellow PhD students
with whom I worked day in, day out. Rick vKessel, mijn complimenten and my
sincere thanks for your inspiring advice.
Als laatste, maar zeker niet het
minste, wil ik graag mijn hartelijke
dank uiten aan mijn familie en
vrienden: jullie voortdurende interesse
in mijn voortgang, het luisterend oor
dat jullie me boden of gewoon de
gezellige afleiding buiten werktijd!
Anneke, jij liet me de drukte op mijn
werk direct vergeten zodra we bij
elkaar waren, ontzettend bedankt
daarvoor! Dank je voor de tijd die je
me gaf om rustig aan dit proefschrift
te werken en voor je geduld tot ik
eindelijk dit boekje afgerond had.
Jur Erbrink
226
Last, but certainly not least, its my

pleasure to express my thankfulness
to my family and friends: your
continued interest in my progress, the
hours that you spent listening or the
entertaining distractions you supplied
at the end of a hard days work!
Anneke, you enabled me to forget the
pressure at work as soon as we were
together, and I am immensely
grateful! Thanks for giving me the
time to focus on my dissertation and
for your patience until the end
product was finally complete.
Samenvatting
Regelschakelaardiagnose aan hoogspannings
vermogenstransformatoren gebruikmakend van
dynamische weerstandsmetingen
Het is een moderne tendens om toestandsafhankelijk onderhoud in plaats van
periodiek onderhoud uit te voeren. Op deze manier proberen nutsbedrijven
onderhoudskosten te reduceren, de levensduur van materieel te verlengen en
mogelijk catastrofale storingen te voorkomen. Toestandsafhankelijk
regelschakelaaronderhoud benodigd diagnostische metingen om de conditie van de
regelschakelaar (on-load tap changer, OLTC) te bepalen, om daarmee vast te
stellen wanneer en welk onderhoud nodig is. Daarnaast kunnen diagnostische
metingen gebruikt worden voor het detecteren van defecten die niet tijdens
onderhoud gezien zijn voordat ze een storing veroorzaken en om de conditie te
beoordelen van onderdelen die niet gemakkelijk voor visuele inspectie toegankelijk
zijn. Ondanks dat er veel verouderde regelschakelaars in bedrijf zijn en ondanks
dat de literatuur laat zien dat het verantwoordelijk is voor een groot deel van de
transformatorstoringen, is het ongebruikelijk om na onderhoud de conditie te
bepalen alvorens de transformator weer in bedrijf te nemen. Onderhoudsfouten,
contactdegradatie en mechanische defecten kunnen dus onopgemerkt blijven.
Dit proefschrift weidt uit over de conditiebepaling van elektrische contacten in
regelschakelaars met schakelweerstanden gebruikmakende van dynamische
weerstandsmetingen (DRM). Van oorsprong is het toegepast op
vermogensschakelaars maar het is ook geschikt om de weerstand te meten terwijl
de regelschakelaar schakelt. Dit proefschrift beschrijft hoe DRM uitgevoerd kan
worden en hoe de resultaten genterpreteerd kunnen worden.
Volgend op de introductie geeft Hoofdstuk 2 een overzicht van OLTC
technologie, veroudering en storing. Verschillende modellen voor
contactweerstand uit de literatuur worden besproken en de toepasbaarheid op
DRM is toegelicht. Vervolgens worden in Hoofdstuk 3 de mogelijke methoden
besproken die beschikbaar zijn om de technische conditie van regelschakelaars te
diagnosticeren. Het verband tussen gas-in-olie analyse (DGA) en contactweerstand
is verder onderzocht om te kijken of beide methoden overlappen. Een statistische
analyse is uitgevoerd op DGA en DRM resultaten van een onderzoekspopulatie
transformatoren maar er was geen relatie gevonden tussen de geselecteerde
gegevensverzamelingen.
Hoofdstuk 4 laat zien dat conditiebepaling op verschillende typen
regelschakelaars toegepast kan worden en dat DRM een grote variatie aan defecten
227
en verouderingsmechanismen kan detecteren door gebruik te maken van een lage

meetstroom. Extra informatie over de conditie van de regelschakelaar is verkregen
ten opzichte van statische weerstandsmetingen en belangrijke informatie over de
ernst en locatie van verschillende verouderingsmechanismen kan uit de resultaten
gehaald worden. Dynamische weerstandsmetingen worden in detail besproken,
waaronder het effect van de meetstroom en circuitweerstand.
Omdat elke diagnostische methode gevoelig is voor bepaalde defecten is er een
set methoden geselecteerd en gemplementeerd (Hoofdstuk 5) met een
diagnostisch meetsysteem als resultaat. De bouwstenen van een dergelijk
diagnostisch systeem zijn besproken en hardware-eisen voor elk deel vastgesteld.
Het hoofdstuk bespreekt welke resultaten na een meting duidelijk zullen worden
gepresenteerd zodat de transformator in bedrijf genomen kan worden zonder het
risico van een storing.
De verouderingsmechanismen die door DRM ontdekt kunnen worden zijn
geordend
in
Hoofdstuk
6.
De
verschillende
onderzochte
verouderingsmechanismen zijn: filmvorming op het contactoppervlakte,
koolvorming op plaatsen met hoge temperatuur veroorzaakt door de
belastingsstroom, putvorming en contacterosie door vlambogen. DRM resultaten
van ongeveer 450 regelschakelaars zijn gedurende dit project onderzocht en
meetdata van een aantal van deze regelschakelaars was vergeleken met bevindingen
van visuele inspecties. Een substantieel deel binnen deze onderzoekspopulatie
blijkt gedegradeerde contacten te hebben. Hoofdstuk 6 bewijst dat DRM een
effectief hulpmiddel is om de technische conditie van regelschakelaars te toetsen,
het blijkt hoogst gevoelig voor de dominante defecten, in het bijzonder voor
contactveroudering door het lange termijn verouderingseffect in de grofkiezer,
omschakelaar of voorkiezer, en voor defecten die conflicteren met de hoofdtaak
van de regelschakelaar (schakelen tussen aftakkingen zonder de laststroom te
onderbreken). Typische DRM grafieken die deze gevarieerde defecten onthullen
zijn gepresenteerd met uitleg over hoe deze defecten kunnen worden herkend in
verschillende soorten regelschakelaars, in het bijzonder in delen van de
transformator die tijdens regelmatig onderhoud niet toegankelijk zijn voor visuele
inspectie omdat ze zijn ingesloten in de transformatorbak.
Laboratoriumexperimenten (Hoofdstuk 7) en verschillende testcases zijn
vervolgens gebruikt om te verifiren voor welke verouderingsmechanismen DRM
gevoelig is. Naast DRM met 1 A zijn ook andere meetstroom niveaus gebruikt
tijdens experimenten (bijlage B). Het effect van kunstmatig verouderde contacten
op de meetresultaten was bestudeerd met een testmodel. Het onderzoek laat zien
dat filmvorming op het contactoppervlakte meetbaar is; de contactweerstand bleek
significant toe te nemen als de contacten over een oppervlaktefilm bewegen, wat
ernstigere contactdegradatie kan verbergen. Slijtage van de afbrandcontacten
benvloedt de timing van contacten en verandert daardoor de vorm en amplitude
van de weerstand grafieken. Stroomonderbrekingen kunnen in de grafieken
verschijnen wanneer buitensporig versleten contacten gemeten worden.
Dit proefschrift demonstreert ook hoe DRM resultaten gerepresenteerd
kunnen
worden
door
conditie-indices
gebruikmakende
van
een
interpretatiediagram die ze categoriseert door toepassing van grenswaarden
(Hoofdstuk 8). Complexe diagnostische informatie is omgezet in simpele conditieindices. Dit proces is afhankelijk van het gemeten type regelschakelaar.
228
Afwijkingen kunnen worden toegewezen aan de lastschakelaar, de voorkiezer,

grofkiezer of omschakelaar, en de aandrijving. Kwantiteiten waren afgeleid uit
DRM resultaten die gebruikt kunnen worden om de meeste gebruikelijke defecten
te ontdekken. In principe kunnen zes situaties ongescheiden worden: contacten in
goede conditie, contacten met een verhoogde contactweerstand, regelschakelaars
met open contacten, schakeltijd afwijkingen, versleten contacten of fase
synchronisatieproblemen. Een voorbeeld van een interpretatiediagram voor
beslissingsondersteuning was ontwikkeld en geverifieerd voor praktische
toepassing op een grote populatie van regelschakelaars. De afwijking van de DRM
stroom ten opzichte van de verwachte waarde was gebruikt en limieten waren
vastgesteld voor het maken van beslissingen. De technische conditie kan hierdoor
vastgesteld worden, wat het mogelijk maakt een selectie te maken van kritische
regelschakelaars die voordeel hebben van storingreducerende maatregelen.
229
Curriculum Vitae
Jur J. Erbrink was born in Arnhem, the Netherlands, on July 20, 1982. He studied
electrical engineering at the HAN University of Applied Sciences in Arnhem from
2000 to 2004. In 2007 he was awarded an MSc in electrical power engineering
from the Delft University of Technology, Delft, the Netherlands. He continued to
work as a researcher at TU Delft and from 2008-2009 worked as a product
specialist at Seitz Instruments AG, Niederrohrdorf, Switzerland. In 2007 he also
began working on his PhD thesis at the Delft University of Technology in the
Department of High Voltage Technology and Management, focusing on tap
changer diagnosis with dynamic resistance measurements, leading to this thesis.
Switching the OLTC during a dynamic

resistance measurement
231

Thesis Jur Erbrink

Uploaded by

Copyright:

Available Formats

Thesis Jur Erbrink

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Thesis Jur Erbrink

Uploaded by

Copyright:

Available Formats

On-load Tap Changer Diagnosis on HighVoltage Power Transformers using

Dynamic Resistance Measurements

On-load Tap Changer Diagnosis on HighVoltage Power Transformers using

ter verkrijging van de graad van doctor

Dit proefschrift is goedgekeurd door de promotor:

measurements are discussed in detail, including the effect of the measurement

3.4 Discussion ......................................................................................................... 43

7.2.1 Arcing contact dimensions...................................................................128

To link different voltage levels in the high-voltage power grid in such a

To keep the voltage at an acceptable level when the load changes.

To select another transformer tapped

Figure 1.2 Example of an in-tank

An on-load tap changer is driven by a complex

1.2 Failure aspects of on-load tap changers

1.3 Tap changer diagnostics

1.4 Objectives of this thesis

1.6 Outline of this thesis

Chapter 2 introduces the reader to the subject by explaining the operation

Chapter 3 describes the possible options for on-load tap changer

Chapter 4 provides more detail on dynamic resistance measurements (for

Chapter 5 presents a measurement device for transformer diagnosis based

Chapter 6 discusses the possible irregularities of the dynamic resistance

Chapter 7 deals with the measurability of a surface film that is formed on

Chapter 8 presents the main issues related to the assessment of the

Chapter 9 presents conclusions and recommendations for future research.

CHAPTER 2 OLTC TECHNOLOGY AND DEGRADATION

2.1 OLTC technology overview

CHAPTER 2 OLTC TECHNOLOGY AND DEGRADATION

CHAPTER 2 OLTC TECHNOLOGY AND DEGRADATION

CHAPTER 2 OLTC TECHNOLOGY AND DEGRADATION

2.2 OLTC degradation mechanisms

2.3 Contact degradation

CHAPTER 2 OLTC TECHNOLOGY AND DEGRADATION

2.3.1 Clean contacts

over the range X = 0 to D/2-C, so that two contacting

CHAPTER 2 OLTC TECHNOLOGY AND DEGRADATION

which can be approximated as:

CHAPTER 2 OLTC TECHNOLOGY AND DEGRADATION

Resistance (arbitrary units)

where n is the number of contact spots with radius a

CHAPTER 2 OLTC TECHNOLOGY AND DEGRADATION

Ac 5.16 106 7.27 107 e1.0510

CHAPTER 2 OLTC TECHNOLOGY AND DEGRADATION

+ 1.68x10-6 T)exp (-1.17x10-2 T)

2.3.2 Contacts with a surface film

CHAPTER 2 OLTC TECHNOLOGY AND DEGRADATION

CHAPTER 2 OLTC TECHNOLOGY AND DEGRADATION

Figure 2.12 Resistance of an oxidation film on

The contact resistance can now be written:

CHAPTER 2 OLTC TECHNOLOGY AND DEGRADATION

where D0 is the frequency factor, B the temperature-dependent diffusion

Ref. [32] determined contact lifetime of copper contacts using a coefficient K:

The contact resistance of this model can then be written:

CHAPTER 2 OLTC TECHNOLOGY AND DEGRADATION

It follows from this model [32] that Rc when

Equation 2-16 then becomes:

CHAPTER 2 OLTC TECHNOLOGY AND DEGRADATION

CHAPTER 2 OLTC TECHNOLOGY AND DEGRADATION

CHAPTER 2 OLTC TECHNOLOGY AND DEGRADATION