Advanced Methods for Insulation Diagnostics on Power Transformers

Stephanie Uhrig; Martin Anglhuber; Ren Hummel, Michael Krueger,

OMICRON electronics GmbH

Abstract
Power transformers are the most expensive links in transmission networks. The demand for electrical
energy and also the average age of transformers is increasing. Low costs are now in focus of electrical
utilities in the liberalized energy market. The utilities try to postpone investments for replacing
transformers by using condition based maintenance instead of time based strategies. One crucial factor
determining the lifetime of a transformer is the insulation condition. Therefore reliable insulation
diagnostics of power transformers are highly demanded. Several tests can be performed to determine the
condition of power transformers. Routine tests often involve oil analysis, like measuring the breakdown
voltage and acidity of the oil or performing a DGA. Another routine test is the measurement of the
capacitance and dissipation factor (DF) at mains frequencies. However, modern insulation techniques are
able to deliver additional information to these conventional tests. Especially the combination of those
measurements helps to get a better picture of the insulation condition. The following paper discusses
different insulation diagnostic techniques, their applications and which information can be derived from
them.

1 Introduction
Nowadays we are facing a growing demand for electrical energy as well as an increasing average age of
power transformers in distribution networks [1]. Due to the liberalized energy marked, electrical utilities
focus on low costs. Since power transformers are the most expensive links in the transmission network,
utilities try to postpone investments for replacements. As a result the average age of a transformer is
increasing. Ageing of high voltage equipment typically means ageing of the insulation. If not recognized
early enough, this could lead to catastrophic failures and high follow-up costs. Therefore, it is necessary,
that critical defects are observed before a failure happens. New maintenance strategies based on the
transformer condition are introduced instead of using periodical measurements or refurbishments.
The result is a trend towards condition based instead of time based maintenance strategies. It reduces
maintenance effort, but reliable diagnostic methods are crucial to derive a complete picture of the
insulation condition. After routine tests - like DGA or capacitance and dissipation factor (DF) measurement
at mains frequency - indicate a possible defect, modern insulation test techniques can be used to gain
more information about a defect or ageing of the insulation. Especially the combination of some
measurements delivers additional information about the insulation condition and helps identifying a
possible problem.

2 Dielectric response measurements

The insulation of power transformers consist of solid insulation and oil. Water can harm this insulation
system. It causes a decreased dielectric strength, measurable in the oil. Also the bubbling effect can
occur, which induces higher field strengths at the surface of the water vapor bubbles and might lead to
partial discharges. Furthermore, water causes hydrolysis in the cellulosic insulation. This is a chemical
process deteriorating the cellulose. Together with oxygen it can form additional water, which can lead to a
kind of snowball effect and accelerates the ageing of the solid insulation. For most cases ageing and the
water content in the solid insulation correlate, so that the water content can be used as an indicator for the
ageing of the cellulose.
However, if the decrease in dielectric strength is severe, the oil can be exchanged, whereas it is a high
effort to refurbish the solid insulation. Therefore, it is important to determine the water content for an
evaluation of the insulation condition.
The measurement of the dielectric response helps to determine moisture in cellulose, the conductivity of
oil or the influence of conductive ageing products inside a power transformer. The measurement can be
performed in time domain or frequency domain. In time domain, the polarization and depolarization
current is measured continuously up to e.g. 10 000 s after applying a DC voltage (PDC method). In
frequency domain, the dissipation factor is measured in a wide frequency range, e.g. 100 Hz - 1 kHz
(FDS method). Both measured curves contain information about polarization effects, their time
dependence and conduction processes inside the insulation. So both methods can be used for moisture

1/6
determination in power transformers [2]. The advantage of the PDC method is the shorter measurement
time, but times below 1 s correlating to frequencies above 1 Hz cannot be measured. Whereas the
FDS method can be used up to higher frequencies, but the time need for the measurement is increased. It
is possible to combine both methods in order to increase the measurable frequency range and additionally
to decrease the measurement time.
The measurement setup for the methods is the same as for typical dissipation factor measurements
(Figure 1). In a transformer, the insulation between high and low voltage windings contains the main part
of the cellulose and is therefore recommended for the dielectric response measurements. Measurements
on bushings are also possible using the high voltage connection and the measuring tab. The main
difference to conventional dissipation factor measurements at mains frequency besides the larger
frequency range is the lower output voltage of 200 V or below.

Voltage source Current sense 1

= A
Instrument

Instrument
Current sense 1 Voltage source
A =
Current sense 2
A A

Guard Guard

CL CHL CH CT CLT CHL

LV HV LV MV HV
(a) (b)
Figure 1: Measurement setup for dissipation factor measurement
(a) at a two winding transformer
(b) at a three winding transformer

The dielectric response analysis is usually displayed in the frequency domain. The curve of oil-paper-
insulated power transformers consists here of three components superimposing each other: The dielectric
response of the cellulose insulation (paper, pressboard), of the oil and the interfacial polarization effect.
The superposition of these three components follows in a typical s-shaped curve for an oil filled power
transformer (Figure 2).
conductivity

Pressboard
Pressboard

Geometry
Insulation

DF
Oil

1.00

0.30 0.20
0.10
0.10

0.03
0.02

0.01 0.01
DF 0.001 0.010 0.100 1.000 10.000 100.000
f/Hz f/Hz
0.001 0.010 0.100 1.000 10.000 100.00

(a) (b)
Figure 2: Dielectric response for different transformers at room temperatures
(a) dry transformer
(b) wet and aged transformer

In Figure 2 (a) the frequency range of 1000 Hz 2 Hz is dominated by the pressboard. Oil conductivity
causes the steep slope at 2 Hz - 0.01 Hz. The insulation geometry (ratio of barriers, spacers and oil)
determines the interfacial polarization effect. Interfacial polarization is typical for non-homogeneous
dielectrics with interfaces between materials of different permittivity and conductivity. This kind of
polarization determines the local maximum or "hump" at 0.003 Hz. Finally, the properties of pressboard
appear again at frequencies below 0.5 mHz. These ranges are typical for a dry transformer. An example
for a wet and aged transformer is given in Figure 2 b). The frequency limits vary with temperature,

2/6
 moisture, oil conductivity and amount of conductive aging by-products. To achieve a reliable result from
the dielectric response curve only the temperature must be known. The values for water content, oil
conductivity, geometry as well as the influence of conductive ageing byproducts can be derived from a
measured dielectric response by using the automatic analysis algorithms delivered in recent software. The
software contains a data base with material properties, e.g. for cellulose samples of different ageing
condition and with different water content. Together with oil properties a whole curve can be modeled for
each configuration, i.e. per water content, oil conductivity and insulation geometry. If the automatic
analysis is started, the model curve is found which fits the measured curve best. It can be assumed, that
the configuration of the model curve represents the values of the measured transformer as well.
Even though this method is still rather new, it is already accepted internationally [3] to give reliable values
for the water content of the solid insulation. In general, changes in the material structure can be seen, e.g.
changes in polar character, conductivity, water content and ageing. The dielectric response is an overall
measurement and only changes in the overall insulations are detectable. Single weak spots, like small
inclusions, cavities, or spots of reduced dielectric strength often do not significantly influence the dielectric
response.

3 Partial discharge measurements

Partial discharges (PD) are localized electrical discharges that only partially bridge the HV insulation [4].
PD measurement is a worldwide accepted tool for quality control of high voltage (HV) insulation systems
[5] [6]. Partial discharges usually appear prior to a complete breakdown in inhomogeneous fields, which
makes it possible to detect defects related to local weak spots before the failure happens. Manufacturers
are advised to perform partial PD measurements for quality assurance according to some standards such
as IEC 60076. Therefore, it is advisable to perform PD test during the design process on single parts, on
components of the transformer during manufacturing and later on the entire transformer.
A PD measurement can detect defects in the insulation (e.g. cavities, inclusions) due to improper
production processes or elevated stress in the insulation due to poor design. The partial discharge type as
well as the location can deliver important information about the defects.
One example for a measurement circuit according to IEC 60270 is given in Figure 3 (a). A voltage source
applies high voltage on the test object Ca. In parallel, a coupling capacitor C k is installed as well as a
coupling device CD, which is connected to the measurement device. A partial discharge inside the test
object will cause a current flow from the coupling capacitor towards the test object. This current flow is
measured, weighted in correspondence to a prior determined calibration factor, and related to the phase
angle of the testing voltage. This gives the possibility to create a phase resolved partial discharge diagram
(PRPD), also called fingerprint. Using these PRPD, the type of partial discharge can be determined.

Ck
ut(t)
v(t)

Ca
CD

(a) (d)
(c)
(b)

Figure 3: PD measurement circuit according to IEC 60270 with blocking impedance Z, coupling capacitor Ck; coupling device CD
and test object Ca
(a) coupling capacitor and the coupling device in series (b) photo of coupling capacitor next to a bushing
(c) using the measuring tap of a bushing (d) photo of connection of measuring unit to the bushing tap

On power transformers, two test setups are common. Either a coupling capacitor is used (Figure 3 (a) and
(b)), or the capacitance of the bushing is used, if a measuring tap is installed (Figure 3 (c) and (d)). The
main advantages of using of the measuring tap are fewer components, which act as antennas for external
disturbances. Furthermore, the installation is easier and faster. The disadvantage of this procedure is the

3/6
limited capacitance of the transformer bushing in contrast to a coupling capacitor, which leads to a
diminished sensibility. The advantages and disadvantages of both procedures have to be weighted in
every case, depending on the type of bushings and the external noise.

500pC Center Phase

850pC

900pC Right Phase

Left Phase

Figure 4: Development of a 3PARD (Three Phase Amplitude Relation Diagram) diagram from a measurement on a
power transformer

For measurements on power transformers, de-noising techniques are very important to achieve reliable
results. For example a simultaneous measurement of electric impulses in all three phases can be used to
develop a so called Three Phase Amplitude Relation Diagram (3PARD) [7]. Impulses on one phase
couple to the other phases as well. By performing a multi-synchronous measurement at all phases at the
same time, these coupled pulses can be detected and correlated to each other. The amplitudes of these
pulses are represented by a vector in the 3PARD. After vector addition and marking each impulse in a
star diagram, clusters of higher intensity will appear for each partial discharge source (Figure 5).

(a) (b)
Figure 5: PRPD (a) and 3PARD (b) showing different internal PD and external interferences

The density-dependent colour coding of the resulting clusters reflects the frequency of the incoming
pulses. Internal PD pulses are mainly located within clusters next to the axes of the diagram. Statistic
noise will appear in all three phases at the same amplitude and therefore the vector addition delivers a
cluster in the star centre. Single clusters can be chosen and clusters can be analysed separately
(Figure 6). Using this technique, noise and outer partial discharge sources can be suppressed for further
analysis.

(a) (b) (c)

Figure 6: PRPD pattern of external disturbance (a), background noise (b) and internal PD (c) separated with 3PARD

4/6
Another de-noising technique uses UHF sensors, which are brought into the transformer tank, additionally
to the electrical measurement. Since the UHF sensor recognizes only partial discharges inside the tank, it
can be used to separate inner partial discharges from outer noise [8].
PD measurements help to detect localized weak spots, e.g. small inclusions, cavities, or spots of reduced
dielectric strength. General changes in the material structure, like changes in polar character, conductivity,
water content and ageing cannot be recognized.

4 Frequency and voltage sweep of capacitance and dissipation factor

The above mentioned methods are surely no routine measurements on transformers. Typically they are
performed after an indication, e.g. due to DGA results. If the (time) effort for the above mentioned
methods is too high, also the advanced measurement of the capacitance and dissipation factor using a
multi-functional device can help to narrow down a possible defect.
One method is the measurement of the capacitance and dissipation factor in a frequency range of e.g.
15 Hz - 400 Hz. The measurement setup is the same as for the dielectric response testing (Figure 1). The
measurement can be performed at higher voltages compared to the dielectric response measurement, for
example at 2 kV. The measured curve is just a small part of the dielectric response (red rectangle in
Figure 7). However, this method provides already much more information compared to conventional
measurements at mains frequencies due to the visible slope of the dissipation factor.
In cases of new and moderately aged 5.000

0.1Hz

50Hz
DF
transformers, the dissipation factor at mains 2.000
frequency can be very similar. It is not 1.000
possible to see if the insulation is in new 0.500
condition or if it has already aged. Using this
0.200
small frequency sweep it can be seen if the 0.100
curve slope is decreasing or increasing. 0.050
New Moderate Aged

During lifetime the curve is shifted to the right.

0.020
Therefore, the slope can tell whether the
0.010
values are on the right side of the minimum
0.005
(typically new and very good condition) or on
the left side of the minimum (slightly or 0.0001 0.001 0.010 0.10 1.0 10.0 Freq/Hz 1000
moderately aged). For a highly aged Figure 7: Frequency dependent dissipation factor of a new,
insulation, the absolute dissipation factor moderately and highly aged power transformer
values are significantly increased.
This method can help to discern between new, slightly aged and highly aged insulation. Therefore it is
suitable to follow the ageing process of a transformer fleet, e.g. for long term investment planning. It
indicates high water contents, high oil conductivity and/or ageing, even, but cannot give a number for both
parameters. Also, it cannot be used to distinguish between the influence of the oil and the cellulosic
material. However, this measurement is faster compared to the dielectric response and can be performed
at higher voltages, e.g. 2 kV.
Another quick check for serious problems can be a voltage sweep of the capacitance and dissipation
factor measurements. Such a voltage sweep delivers as well additional information about a possible
defect in the insulation. A typical application would be the measurement using voltage in the range of e.g.
400 V - 12 kV. It is known, that a healthy insulation typically does not show significant voltage
dependence up to its rated voltage. If such a voltage dependence is observed, it can be a sign for partial
discharges or partial breakdown between grading layers. Also poor contacts can be a reason.
A measurement in the range of 400 V - 12 kV help to detect defects like mentioned above earlier
compared to conventional capacitance and dissipation factor measurements. The measurement is
relatively fast, but can already deliver an indication for weak spots with a lower dielectric strength or bad
contacts.

5 Concluding comparison of the mentioned methods

The insulation condition can be influenced on one hand by single weak spots or defects and on the other
hand by an overall material change or degradation. Both are important to observe, but the action would be
completely different. An aged transformer might not have single defects causing strong partial discharges,
but it needs to be replaced due to the highly deteriorated insulation. A new transformer might have a very
good general insulation condition, but an inclusion or metallic particle is causing partial discharges, which

5/6
can lead to a catastrophic failure. In such a case only the specific part of the insulation needs to be
repaired.
Typically it is not possible to single week spots and deliver information about the general insulation
condition by using only one method. Thus, especially the combination of diagnostic measurements can
help to get a clear picture of the insulation condition. The previously described methods can help to
diagnose the insulation condition. Generally like for most approaches the faster methods deliver an
indication, but for a full picture (e.g. to measure a water content or localize a partial discharge location)
more time has to be invested to use the highly specialized methods. The following table gives an overview
about the mentioned advanced diagnostic methods and the achievable information.

Dielectric Response DF Sweep C and DF Partial Discharge

(0.1 mHz 1 kHz) (15 Hz 400 Hz) Voltage Dependence Measurement

Diagnosis of overall material changes like: Diagnosis of weak spots like:

- ageing - spots of reduced dielectric strength
- water content (e.g. inclusions or cavities)
- general material changes (degradation) - partial breakdowns
Not able to detect single weak spots. Not suitable to detect overall material changes.

Distinguishes between No differentiation No differentiation of Distinguishes between

influence from oil and between oil and defect cause. type of discharge, e.g.
cellulose. cellulose. cavity or surface PD.
Localization possible.

Measures: Indication for: Indication for: Measures:

- water in cellulose, - water in cellulose - partial discharges or - frequency and
- oil conductivity and - high oil conductivity or - partial breakdown of - intensity
- geometry values - ageing grading layers (bushing) of partial discharges
- influence of ageing - bad contacts
byproducts

6 Literature
[1] M. Gernandt: Auswertung von Strungen und Gas-in-lanalysen bei Hochspannungs-
Transformatoren, ETG Fachbericht 104, Kassel 2006
[2] Cigre Brochure 254: Dielectric Response Methods for Diagnostics of Power Transformers; 2002
[3] Cigre Brochure 414: Dielectric Response Diagnoses For Transformer Windings; 2010
[4] IEC 60270, "High-voltage test techniques Partial discharge measurement", Version 2000, 3rd
Edition
[5] D. Knig and Y. N. Rao: "Partial Discharges in Electrical Power Apparatus", VDE 1993
[6] J. C. Montanari: Insulation diagnosis of high voltage apparatus by partial discharge
investigation, Liu-Yeda Memorial Lecture, in Proc. IEEE ICPADM, Bali, Indonesia, 2006
[7] K. Rethmeier: "Separation of Superposed PD Faults and Noise by Synchronous Multi-Channel
PD Data Acquisition", 17th Biennial IEEE International Symposium on Electrical Insulation,
Vancouver, Canada, 2008
[8] J. Fuhr, Procedure for Identification and Localization of Dangerous PD Sources in Power
Transformers, IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 12, No. 5, pp.
1005-1014

6/6
OMICRON is an international company serving the electrical power industry with
innovative testing and diagnostic solutions. The application of OMICRON products
allows users to assess the condition of the primary and secondary equipment on
their systems with complete confidence. Services offered in the area of consulting,
commissioning, testing, diagnosis and training make the product range complete.

Customers in more than 140 countries rely on the companys ability to supply leading-
edge technology of excellent quality. Service centers on all continents provide a broad
base of knowledge and extraordinary customer support. All of this together with our
strong network of sales partners is what has made our company a market leader in the
electrical power industry.

For more information, additional literature,

and detailed contact information of our
worldwide offices please visit our website.

www.omicron.at | www.omicronusa.com OMICRON