V. MALA@INSKAS et al: EV ALUATION OF THE INSULATION CONDITION OF HIGH-VOLTAGE TRANSFORMERS ...
253–261
EV ALUATION OF THE INSULATION CONDITION OF
HIGH-VOLTAGE TRANSFORMERS BY DETECTING PARTIAL
DISCHARGES USING THE ELECTROMAGNETIC WA VE
RADIATION METHOD
OCENA STANJA IZOLACIJE VISOKONAPETOSTNIH
TRANSFORMATORJEV Z DETEKCIJO DELNE RAZELEKTRITVE
IN UPORABO METODE SEV ANJA ELEKTROMAGNETNEGA
V ALOV ANJA
Vilius Mala`inskas
1*
, Renaldas Rai{utis
1,2
, Alfonsas Morkvënas
1
,
Saulius Gud`ius
1
, Audrius Jonaitis
1
, Jonas Vai~ys
1
, Gediminas Dauk{ys
3
1
Department of Electric Power Systems, Kaunas University of Technology, Studentø str. 48, Kaunas, Lithuania
2
Ultrasound Research Institute, Kaunas University of Technology, K. Bar{ausko str. 59, Kaunas, Lithuania
3
Kaunas University of Applied Sciences, Pramonës avenue 20, Kaunas, Lithuania
Prejem rokopisa – received: 2020-09-07; sprejem za objavo – accepted for publication: 2020-11-12
doi:10.17222/mit.2020.175
High-voltage transformers are among the most important elements in an electric-power system. Each one of them is affected by
various external factors: overvoltage, partial discharge (PD), overheating, vibrations, etc., which are created by a strong electric
field, thermal effect, humidity, impurities, factory defects, dissolved water and gas in oil-type-transformer insulation. These and
other factors, caused by the environment, reduce the life of a device. Thus, the evaluation of the device condition is one of the
most important factors for a system-safety evaluation, which ensures a reliable and economical electrical-network operation.
This work reviews different contact and non-contact methods, used to evaluate the conditions of transformers by measuring the
level of PD. The selected method, i.e., the non-contact measurement of electromagnetic-wave radiation was used to evaluate the
voltage-transformer status. The experiment was performed at a 110 kV substation. The authors discuss the efficiency of the se-
lected method to evaluate the voltage-transformer insulation condition.
Keywords: insulation, partial discharge (PD), detection, transformers
Visoko-napetostni transformatorji so med najpomembnej{imi elementi sistemov za distribucijo elektri~ne energije. Vsak iz med
njih je pod vplivom razli~nih zunanjih faktorjev: napetostne preobremenitve, delnega razelektrenja, pregretja, vibracij, itd., ki so
posledica nastanka mo~nih elektri~nih polj, ogrevanja, vlage, ne~isto~, tovarni{kih napak, raztopljene vode in plinov v oljnih
transformatorjih. Ti vplivi in {e vrsta drugih okoljskih faktorjev skraj{a `ivljenjsko dobo transformatorjev. Zato je ovrednotenje
stanja dolo~ene naprave ena od najpomembnej{ih nalog sistema njenega varovanja, ki zagotavlja ekonomi~no in zanesljivo
obratovanje elektri~nega omre`ja. V ~lanku so avtorji obravnavali razli~ne kontaktne in brezkontaktne metode, s katerimi je
mo`no ovrednotiti stanje transformatorjev z merjenjem nivoja delne razelektritve. Za oceno stanja napetostnega transformatorja
so uporabili brezkontaktno meritev sevanja elektromagnetnega valovanja. Preizkus so izvajali na 110 kV pomo`ni postaji. V
~lanku so {tudirali tudi u~inkovitost izbrane metode za ocenitev stanja izolacije napetostnih transformatorjev.
Klju~ne besede: izolacija, delna razelektritev (PD), detekcija, transformatorji
1 INTRODUCTION
According to the data analysis presented in the
CIGRE transformer reliability review,
1
in 36.62 % of the
examined high-voltage transformers the failure was of
dielectric origin – partial discharges, corona, or electric
arc.
1,2
The listed phenomena are caused by insulation de-
fects, degradation of insulation, overvoltage, overheating
and other factors occurring during the service life.
The insulation of transformers is continuously af-
fected by partial discharges and the transformer lifetime
depends on the activity of PDs. Partial discharge (PD) is
an electric discharge that does not completely cover the
insulation between the conductors.
3
This phenomenon
occurs at the weakest points of solid and liquid insula-
tion. This is due to the defects in the insulation, which
are affected by the connected voltage. Regarding their
properties, partial discharges are of a different nature
than electric discharges – the process that happens dur-
ing a partial discharge is unipolar and no free charges are
created.
While other phenomena of dielectric nature can be
observed with the naked eye (electric arc) or identified
even by smell (the nitrogen smell during the corona ef-
fect), PD is impossible to be detected without additional
equipment. During this phenomenon, acoustic and elec-
tromagnetic waves are emitted into the surroundings, the
temperature increases where a PD occurs and, in addi-
tion, methane and hydrogen gases are formed in the insu-
lation oil.
3
Due to the high variation speed of the param-
Materiali in tehnologije / Materials and technology 55 (2021) 2, 253–261 253
UDK 620.1:621.314:621.3.027.3 ISSN 1580-2949
Original scientific article/Izvirni znanstveni ~lanek MTAEC9, 55(2)253(2021)
*Corresponding author's e-mail:
vilius.malazinskas@ktu.edu (Vilius Mala`inskas)

eters that define the PD, various detection methods are
available such as the vibroacoustic, electromagnetic,
temperature-observation methods, chromatographic-gas
analysis and some other methods.
4–8
Each of these meth-
ods has its advantages and disadvantages, and their accu-
racies are also different. Therefore, it is necessary to de-
termine the general PD level and location mathematical
model, with which the advantages and utilization of the
above-mentioned methods can be completely used while
operating diagnostic devices based on different physical
principles. After determining certain correlations be-
tween different methods, the remaining operation life of
a transformer can be identified better and can allow plan-
ning the maintenance or repair of electric-system ele-
ments. At the same time, financial losses and a possible
risk to human life can be avoided, both of which might
arise due to a device’s malfunction.
The objective of this study is to assess the trans-
former status at 110 kV substations, using non-contact
measurements of electromagnetic-wave radiation.
2 REVIEW OF MEASUREMENT METHODS
FOR THE DETECTION OF PARTIAL
DISCHARGES
During the presence of partial discharges, the follow-
ing secondary phenomena of the insulation material are
observed:
9
• Generation of electrical impulses;
• Emission of electromagnetic waves;
• Emission of ultrasonic waves;
• Generation of acoustic waves;
• Release of heat;
• Decay of the insulating material.
Each of the above phenomena can be evaluated with
appropriate methods such as the recording of electrical
parameters (analysis of the emission level of high-fre-
quency (HF) waves or the partial discharge detection
approach), recording of mechanical waves (the vibro-
acoustic method), an analysis of the chemical composi-
tion and thermo-vision analysis.
The measurement methods listed above have certain
advantages and disadvantages and can be applied differ-
ently. This information is further analyzed in the article.
2.1 Electrical methods
The electromagnetic energy measurement method is
based on the measurement of the current, induced during
the partial discharge, and flows through the grounding
conductor, using a current transformer.
6
Two key devices
are used with this approach: a PD-signal recording de-
vice and a device for data analysis. This approach en-
ables us to record the PD signal by connecting measure-
ment devices (a voltage transformer and a current
transformer) directly to the object under investigation (a
power transformer, cable, etc.).
4
A typical diagram of the
equipment connection for an electrical partial discharge
measurement is shown in Figure 1.
With this approach, the recording of partial discharge
can be performed without stopping the operation of the
power transformer. Therefore, it allows an online evalua-
tion of the present condition of the power transformer.
Afterwards, the acquired information is analyzed and the
time-dependent deviations of the PD parameters are ob-
served to evaluate the condition of the device insulation.
4
Even though the electrical method of partial-dis-
charge measurement is very accurate, it has several dis-
advantages. The main disadvantage is the sensitivity of
V. MALA@INSKAS et al: EV ALUATION OF THE INSULATION CONDITION OF HIGH-VOLTAGE TRANSFORMERS ...
254 Materiali in tehnologije / Materials and technology 55 (2021) 2, 253–261
Figure 1: Electrical partial discharge measurement set-up:1–P Dequipment for data processing,2–P Dcalibrator,3–d e vice for connecting the
PD measuring equipment to personal computer systems,4–P Cwith data processing software,5–s ensor for coupling HF PD signals,6–m e a -
surement data splitting box,7–P Ddecoupler
10

the measuring device to ambient electromagnetic noises.
6
High-voltage and power equipment (power transformers,
cables, insulators, etc.) emit electromagnetic noises in
narrow and wide spectral bands, which can affect the re-
corded data. In some cases, it is difficult to distinguish
between the presence of PD and ambient electromag-
netic noise due to the very short signal duration of a PD
pulse. In such a case, the recorded data should not be
used for the evaluation of the equipment-insulation con-
dition.
4
To solve this issue, manufacturers recommend using
additional filters for electromagnetic noises. Another
possible way to solve this problem is to disconnect the
equipment from the power grid, using an external power
supply. However, the latter case is not practical because
the disconnection of the device from the power grid can
be very costly due to the resulting reduced reliability of
the grid.
Generalized information about the electrical method
for the PD measurement is given in Table 1.
Table 1: Advantages, disadvantages and application areas of the elec-
trical partial-discharge measurement method
4,6,7
Advantages: • Wide range of applications,
• High sensitivity,
• Suitable for on-line recording of PD,
• Possible to record multiple parameters of
PD (charge, voltage, discharge current,
etc.).
Disadvan-
tages:
• Measurement results are very sensitive to
the electromagnetic interference generated
in the environment,
• Complicated application at the substation,
• Costly measuring equipment as compared
with the other methods.
Applications • Power transformers,
• Cables,
• Insulators,
• etc.
2.2 Ultra-high frequency method
The ultra-high frequency (UHF) partial-discharge de-
tection method is based on electric resonance and elec-
tromagnetic-wave radiation, caused by PD in a frequency
range of 50 MHz to 1.5 GHz.
6
Electromagnetic waves of
the highest amplitude are registered near their source. In
this case, it is in the defective region that the PD occurs.
This is the reason why the partial-discharge location can
be determined according to the intensity of the electro-
magnetic wave.
12
The UHF method has been mostly used to detect PD
in gas-insulated substations (GISs).
4,13,14
Therefore, this
method is relatively new in determining the insulation
condition for other devices. Despite that, for GISs, the
UHF method is one of the main ones for determining
faults. Due to each substation’s specific parameters, even
a sufficiently weak PD can be critical. Due to the advan-
tages associated with the UHF method, PD can be regis-
tered even from 5 PC or lower.
13,14
Since the introduction of the ultra-high frequency
method for determining the insulation condition of
power transformers and other high-voltage electric de-
vices, this method has greatly improved. Various tools
have emerged, which help determine a PD more accu-
rately. Measurement devices based on this method be-
came essential in order to protect high-voltage compo-
nents and maintain their reliable operation mode.
13
Using a UHF registering device, it is possible to de-
termine the PD activity. Another possible use of UHF
registering devices is the determination of a partial-dis-
charge location, as presented in Figure 2. This principle
is used to analyze particular time moments needed to de-
tect electromagnetic waves generated by a PD at differ-
ent spatial points of UHF receiver antennas. Based on the
time difference between signal registering and the wave
propagation speed in a medium, the location of PD can
be determined.
15
Due to the complicated structure of the transformer
and its unconventional shape, calibration of the measure-
ment equipment for electromagnetic wave radiation is
mandatory. Neglecting the influence of different media,
internal structures and the shape of high-voltage compo-
nents, the obtained measurements would have a high er-
ror.
Generalized information of PD registration using the
UHF electromagnetic wave method is presented in Ta-
ble 2.
V. MALA@INSKAS et al: EV ALUATION OF THE INSULATION CONDITION OF HIGH-VOLTAGE TRANSFORMERS ...
Materiali in tehnologije / Materials and technology 55 (2021) 2, 253–261 255
Figure 2: Detection of an UHF partial-discharge location

Table 2: Advantages, disadvantages and applications of the UHF elec-
tromagnetic wave radiation method for the registration of partial
discharges
4,6,7
Advantages:
• Easy application when measuring in a
substation,
• Instantaneous data analysis available,
• Defect localization,
• Measurements are performed while a
high-voltage component is connected to
the power grid.
Disadvan-
tages:
• Measurements are very sensitive to the
ambient electromagnetic noises,
• Difficult to determine the remaining oper-
ation life of a high-voltage component
based on instantaneous measurements.
Applications
• Transformers,
• Gas-insulated substations (GIS),
• Cables,
• Insulators,
• etc.
2.3 Vibrodiagnostic method
During a partial discharge, the resulting energy heats
the adjacent insulating material to a possible explosion
of its fraction. This tiny explosion emits electromagnetic,
optic and acoustic waves.
11
The vibrodiagnostic method
for detecting PDs is based on on-line registration of
acoustic waves.
During the application of the vibroacoustic method,
information-processing equipment (a computer with cor-
responding software), a data-acquisition system and sen-
sors (piezo element, accelerometer, etc.) operating within
a frequency range of 10–300 kHz are used to record me-
chanical vibrations generated during a PD.
4,6,8
This mea-
surement set-up can be used to determine the spatial lo-
cation of a partial discharge in the high-voltage
component being inspected.
Defects of the internal insulation of cables, anomalies
in insulators, circuit breakers, oil-immersed power trans-
formers and gas-insulated electrical-system components
can be identified by recording mechanic waves propagat-
ing through the elements of a power system, using the
vibrodiagnostic method.
9
By recording mechanic waves,
not only a PD can be identified but also conditions of the
insulation material such as changes in it, fractures or im-
purities.
16
When the vibrodiagnostic method is used for detect-
ing a partial discharge, we can determine the location of
the defect by examining the decrease in the amplitude of
the acoustic waves emitted from the defective location or
the time difference between the propagated waves. One
of the possible vibroacoustic sensors is the piezoelectric
transducer (PZT). When mounting the PZT to an outdoor
power-transformer tank, it captures the interference in
the environment, which can cause additional measure-
ment uncertainties. Moreover, because of different de-
signs of the power-transformer tank, measurements may
vary so that additional measurements are required to ad-
just the results to a particular design of the
power-transformer tank. To reduce the influence of am-
bient factors on the measurement results, a PZT trans-
ducer can be mounted inside a power transformer. Detec-
tion of a PD location with this method is based on the
measurement of the difference in the mechanical-wave
propagation time recorded by a transducer pair to esti-
mate the exact location of the PD. This method is more
resistant to the influence of ambient factors than other
on-line measurements (electric or high-frequency ones).
6
The data recorded with the vibrodiagnostic method
are unaffected by electromagnetic noise and, therefore,
contrary to the situation involving electric PD-detection
methods, the high-voltage component under investiga-
tion does not need to be disconnected from the power
system to obtain more accurate data. However, this
method also has drawbacks. Since power transformers or
other equipment under investigation are not homoge-
neous – they consist of layers of different materials – the
propagation of acoustic waves is difficult to predict in
advance. This makes it extremely difficult to precisely
locate the spatial position of the PD. Another problem is
caused by partial discharges themselves and the parame-
ters of detection devices – PDs emit acoustic waves of
very low amplitudes while measuring devices require an
extremely high sensitivity to register such signals.
As a result, the vibroacoustic method detects the PDs
that already strongly affect the insulation and the compo-
nent under investigation may already be in a critical con-
dition. In this case, the component must be disconnected
from the power system before a critical failure occurs.
4
A summary of the vibrodiagnostic method used for
detecting PDs is given in Table 3.
Table 3: Advantages, disadvantages and application areas of the
vibrodiagnostic method
4,6,7
Advantages: • Resistant to electromagnetic interference,
• Defect localization.
Disadvan-
tages:
• Measurement results are very sensitive to
external vibrations,
• Not suitable for early detection of PD,
• Not suitable for evaluation of equipment
condition,
• Not suitable for continuous device diag-
nostics.
Applications • Transformers,
• Gas-insulated substations (GIS).
2.4 Chromatographic analysis
The chromatographic method used for detecting par-
tial discharge in a high-voltage transformer is based on
an analysis of the insulating oil and gas of the trans-
former for the detection of particular changes during the
PD process.
6
There are two important chromatographic-analysis
methods: high-performance liquid chromatography
(HPLC) and dissolved gas analysis (DGA).
6
HPLC anal-
V. MALA@INSKAS et al: EV ALUATION OF THE INSULATION CONDITION OF HIGH-VOLTAGE TRANSFORMERS ...
256 Materiali in tehnologije / Materials and technology 55 (2021) 2, 253–261

yses the resulting products of PD, such as decomposed
glucose, released during the breakdown of solid insula-
tion. DGA analyses the amount of gas produced during
PD in a test substance.
The selection of the oil sample, gas separation and,
finally, chromatographic analysis give rise to a number
of uncertainties and errors that affect the accuracy of the
results. This method is not appropriate for detecting sud-
den changes in transformer insulation as it is applied
over scheduled periods.
5
Summarized information on the chromatographic
analysis is given in Table 4.
Table 4: Advantages, disadvantages and applications of chromato-
graphic analysis for a detection of partial discharge
5,6
Advantages:
• Suitable for identifying events that occur
in a device,
Disadvan-
tages:
• The set amount of gas does not correlate
with the functionality reserve of the com-
ponent being inspected,
• Not suitable for instantaneous evaluation
of the high-voltage component condition,
• Large uncertainties due to the measure-
ment principle.
Applications:
• Transformers.
2.5 Thermovision analysis
Every body with a temperature greater than absolute
zero emits infrared radiation.
11
These rays and their in-
tensity can be recorded using a thermal imager.
This method is described as mobile, non-invasive,
contactless, wide-ranging and it provides fast measure-
ment results.
5
However, the results of thermographic
analysis using a thermal imager are easily influenced by
environmental factors and, in the case of polished sur-
faces of the transformer housing, the results are also af-
fected by errors. The thermal imager can only provide
the temperature of the external surface of the transformer
housing, which rarely reflects the actual condition of the
transformer. Also, heat dissipation is only one of many
parameters that are measured to detect the existing de-
fects, making it difficult to determine particular defects
(except couplings or joints) and having a high degree of
uncertainty.
5
A summary of the thermovision analysis is given in
Table 5.
Table 5: Advantages, disadvantages and applications of the thermo-
vision analysis
5,11
Advantages: • Mobile,
• Instantaneous data.
Disadvan-
tages:
• Inappropriate for setting up a device oper-
ation resource,
• Only the condition of the outer layer can
be assessed,
• Does not represent the actual condition of
a transformer.
Applications: • Transformers;
• Cables.
3 APPLICATION OF THE UHF METHOD FOR
REGISTRATION OF THE ELECTROMAGNETIC
WAVE RADIATION OF HIGH-VOLTAGE
TRANSFORMERS
In order to investigate the applicability of the UHF
method on site (at real substations), a DFA300 Doble
Lemke device is used for measuring electromagnetic
wave radiation. This device is intended for the detection
of insulation or mechanical defects in gas-insulated sub-
stations or in open-source substations. DFA300 registers
the radio frequency interference (RFI) of radiated elec-
tromagnetic waves due to PD from 50 to 1000 MHz and
acoustic emissions (AEs) from 10 to 300 kHz.
17
3.1 Experimental diagram and measurement principle
Using the DFA300 measuring device, 110-kV substa-
tions equipped with high-voltage Pfiffner current (EJOF
123) and voltage (EOF 123) transformers were subjected
to a condition analysis. The object under investigation is
presented in Figure 3.
More than two hundred different high-voltage trans-
formers at over sixty different substations were tested in
the study.
Measurement principles:
• A plan for the installation of the substation’s measur-
ing equipment is drawn up;
V. MALA@INSKAS et al: EV ALUATION OF THE INSULATION CONDITION OF HIGH-VOLTAGE TRANSFORMERS ...
Materiali in tehnologije / Materials and technology 55 (2021) 2, 253–261 257
Figure 3: Object under investigation: a current and voltage trans-
former

• A baseline background spectrum analysis near the
substation is conducted, more than 20 meters away
from the high-power components;
• A baseline background spectrum is captured with the
DFA300 to allow an instant evaluation of the compo-
nent condition;
• A spectrum analysis of the received electromag-
netic-wave signals of each investigated component is
performed and compared to the baseline background;
• Doble Lemke recommends an analysis of a spectrum
range of 600–900 MHz to detect partial discharge;
• If spectral mismatches are detected within the speci-
fied frequency range, an additional analysis of the
component under investigation is performed by re-
cording radiation intensities during the wave period,
level values with the alternating phase and level mea-
surements;
• Device failures are identified.
During the measurements, the background spectrum
within the range from 50–1000 MHz, more than 20 m
away from the nearest interfering component, was
scanned. At the points shown in Figure 4, the measure-
ment was repeated in order to compare the results with
the background spectrum. The measurements close to the
component under investigation were made at a distance
of 1–1.5 m, depending on the height of the component
installation. The measuring results at these measuring
points were considered to be scarcely affected by the
other transformers. The spectral analysis of each compo-
nent was performed twice.
The results of the stored spectrum were further ana-
lyzed using the MATLAB software. The spectral-analy-
sis values recorded for each component were compared
with the baseline background. The results were plotted,
showing all the spectral analyses (background and com-
ponent) and providing a differential graph that subtracts
the spectral-analysis values of the background from the
spectral values of the component. The differential graphs
of each phase transformer were compared to determine
which phase transformer was most affected by the insu-
lation degradation. Examples of graphs are provided in
Section 4 of this article.
3.2 Advantages of the selected measurement method
Using DFA300, the measurements are performed un-
der real operating conditions. The components under in-
vestigation are connected to the power grid. The power
grid does not suffer from the extra line loads as the com-
ponent does not need to be disconnected.
The measuring device provides various tools for
monitoring and assessing the condition of the facility be-
ing measured, such as:
• "Spectrum analyzer": during this measuring mode,
the device registers radiation of electromagnetic
waves at different frequencies in a range of 50 MHz
to 1000 MHz;
• "Time resolved": during this measuring mode, the de-
vice synchronizes with the frequency of the power
grid via a wireless synchronization adapter. In this
way, the distribution of the electromagnetic-wave in-
tensity over the corresponding period is recorded at
the selected fixed frequency. This type of measure-
ment can accurately represent the type of fault exist-
ing in the component under investigation;
• "Level versus Phase": these measurements detect the
periodic distribution of peak-wave magnitudes at a
given fixed frequency. When the measuring device is
synchronized with the main frequency of the power
grid, the selection of the number of different peaks
per measurement (1000, 2000, 5000) shows the pat-
tern of peak distribution. According to this regularity,
it is possible to characterize the defect;
• "Level meter": it records the intensity level of the
electromagnetic wave at a certain fixed frequency.
This measurement can be used to estimate which
high-voltage component emits electromagnetic waves
V. MALA@INSKAS et al: EV ALUATION OF THE INSULATION CONDITION OF HIGH-VOLTAGE TRANSFORMERS ...
258 Materiali in tehnologije / Materials and technology 55 (2021) 2, 253–261
Figure 4: Partial-discharge measurement plan

of a higher intensity than the nearby ones (thus deter-
mining which particular component requires an addi-
tional analysis) or which component is already de-
fected.
The measuring device is mobile, user friendly and
convenient for daily measurements when visiting a num-
ber of different substations.
3.3 Disadvantages of the selected measurement method
During the measurements, an environmental impact
on the measuring-device operating characteristics was
encountered. Depending on the humidity of the air, the
accuracy of the measurements changes. In the morning,
when the humidity level in the environment was rela-
tively high, the device was not able to register the distri-
bution of the peak values of emitted electromag-
netic-wave intensity versus phase at a fixed frequency
value ("Level versus Phase" mode). By the middle of the
day, when the air humidity dispersed, the measurements
were again performed smoothly.
The registered spectrum values of the emitted electro-
magnetic waves (background or particular high-voltage
component) during the spectrum analysis had a signifi-
cant error of about 4–9 %. This error was calculated by
comparing the results of two consecutive measurements
of the same component.
The intensity values of the emitted electromagnetic
waves registered during a period at the selected fre-
quency cannot be stored in a time interval for a later
analysis. The only way to save data is to stop the mea-
surement and capture the current image. The disadvan-
tage of this is that during the recording period, the anom-
aly that could identify a malfunction of the component is
not always recorded. This causes additional difficulty
during the analysis of the measurement results.
Different measurement times were estimated when
measuring the level of electromagnetic wave in the
"Level versus Phase" measuring mode. As a result, dur-
ing the investigation of high-voltage components, for
some components, 1000 peak values were accumulated
in a few seconds and for others, in a few minutes. There-
fore, it is not possible to determine the total duration dur-
ing a measurement.
The device cannot compare the results of certain
measurements. For this reason, additional measurements
are required when analyzing the recorded data using a
mathematical model and in case of doubt concerning the
obtained results.
4 RESULTS
During the analysis of the data of each phase of the
transformer, the baseline background spectrum and the
particular phase transformer spectrum (Figure 5), the
difference between these spectra (Figure 6) and a com-
parative graph of the differences between all the phases
of the transformer (Figure 7) were provided. The data
below is from one of the 110-kV substations’ voltage
transformers "ÁT-T101" (grade EOF-123).
According to the recommendations of Doble Lemke,
the manufacturer of the DFA300 instrument, partial-dis-
charge signals are analyzed in the 600–900 MHz range
where the maximum spectral divergence is determined.
17
Based on this assumption, an anomaly is detected in a
frequency range of 780–820 MHz – the intensity of the
electromagnetic wave is higher than the background in
the whole spectrum. This effect, although small, is attrib-
uted to PD. Based on the information provided in Fig-
ure 7 for the comparison of the difference of all three
phase spectra with the baseline background spectrum, it
can be stated that the Phase-B voltage transformer shows
V. MALA@INSKAS et al: EV ALUATION OF THE INSULATION CONDITION OF HIGH-VOLTAGE TRANSFORMERS ...
Materiali in tehnologije / Materials and technology 55 (2021) 2, 253–261 259
Figure 7: Comparison of the differences of all three phase spectra of
"ÁT-T101" with the baseline background spectrum
Figure 5: Baseline background and Phase B spectra of transformer
"ÁT-T101"
Figure 6: Difference between "ÁT-T101" Phase B spectrum and base-
line background spectrum

the highest PD level and Phase A shows the lowest PD
level.
These measurements were made on April 4, 2019. In
May, an accident occurred at one of the 110-kV substa-
tions – a Phase B voltage transformer "ÁT-T101" ex-
ploded. Prior to the accident, the parameters of the power
grid did not exceed the normal-mode limits.
On July 26, 2019, additional measurements were car-
ried out with the replacement of the Phase B voltage
transformer "ÁT-T101" (grade EOF-123). The compari-
son of the differences between the three phase (A, B and
C) spectra of the replaced "ÁT-T101" with the baseline
background spectrum is shown in Figure 8. The compar-
ative spectra of the first and second measurement in ab-
solute values for Phase-B voltage transformer "ÁT-T101"
are shown in Figure 9.
According to the information from Figure 8,a n
anomaly of spectral differences for Phase B similar to
that before the accident with the transformer is observed
in the frequency range of 780–820 MHz. Therefore, such
a difference could be used as a reliable parameter for
identifying the condition, in which the level of PD is in-
creased and the transformer is under risk of being af-
fected by failure.
Comparing the spectra in Figures 7 and 8, it can be
concluded that during the second measurement the sub-
station was much less loaded. For this reason, it is possi-
ble to see the difference in Figure 9 within the range of
780–820 MHz. When analyzing the comparative graph
(Figure 8), it can be stated that Phase B of the replaced
transformer "ÁT-T101" is no longer different from other
phases (A and C).
This method comparing the background and
high-voltage component spectra differences is based on
the accuracy of the results recorded. If the results are
false, frequencies of a possible PD may not be deter-
mined, which may result in an incorrect estimation of the
insulation condition of the component under investiga-
tion.
5 CONCLUSIONS
This study reviews different contact and non-contact
methods, used to evaluate the condition of operating
transformers that are not disconnected from the electrical
grid, by measuring the levels of PDs. The selected
non-contact UHF measurement method using electro-
magnetic-wave radiation was used to evaluate the voltage
transformers’ status. The experimental investigations
were performed at a 110-kV substation.
The spectrum of the electromagnetic waves emitted
was registered by a DFA300 device with a 4–9 % error
tolerance; therefore, during unfavorable conditions the
frequency needed for the detection of partial discharges
could be undetermined.
It makes the most sense to perform a condition evalu-
ation of a component when the substation has the highest
power load.
Measurements performed at one of the 110-kV sub-
stations could have helped to evade the failure; however,
the registered differences between the phase spectra
within the frequency range of 780–820 MHz were quite
small, making it difficult to identify the presence of a
high-level failure of "ÁT-T101" Phase B. However, the
difference we detected shows that the level of PD in-
creased and the transformer was under risk of being af-
fected by failure.
We need to capture a broader parameter, combining
electromagnetic and vibrodiagnostic partial-discharge
measurement methods (if allowed by the measuring-de-
vice type), additionally considering the system’s voltage,
frequency, current and other surrounding factors such as
temperature and humidity.
6 FUTURE WORK
As routine inspections are performed at certain time
intervals, they are not completely accurate due to the un-
defined status of the device between predefined intervals
– the set examination intervals may be too long to pre-
vent the device’s failure or too short and inaccurate.
Therefore, future research should concentrate on opti-
mizing the inspection procedure for the insulation condi-
tion of the transformer and its variations due to operating
V. MALA@INSKAS et al: EV ALUATION OF THE INSULATION CONDITION OF HIGH-VOLTAGE TRANSFORMERS ...
260 Materiali in tehnologije / Materials and technology 55 (2021) 2, 253–261
Figure 8: Comparison of the differences of all three phase spectra of
the replaced "ÁT-T101" with the baseline background spectrum after
the Phase-B accident
Figure 9: Comparative analysis of the absolute values of spectra of
the first measurements (defected "ÁT-T101", before the replacement)
and second measurements (after the replacement of "ÁT-T101") of the
Phase-B voltage transformer

factors. We need to propose a method that would allow
us to schedule the device’s repair time and perform the
examination according to the condition of the insulation,
using the obtained data.
7 REFERENCES
1
CIGRE WG A2.37, Transformer Reliability Survey, 2015
2
CIGRE WG A2.18, Life management techniques for power trans-
former, Brochure 227, Paris 2003
3
D. F. Akiyoshi, B. A. Castro, J. V . F. Leão, M. A. Rocha, J. A. A.
Rey, R. R. Riehl, A. L. Andreoli, Evaluation of Low Cost Piezoelec-
tric Sensors for the Identification of Partial Discharges Evolution,
Proceedings, 4 (2019)3 6
4
I. A. Soomro, M. N. Ramdon, Study on different techniques of par-
tial discharge detection in power transformers winding: simulation
between paper and EPOXY resin using UHF method, International
Journal of Conceptions on Electrical and Electronics Engineering, 2
(2014)1
5
R. Rao, Z. Li, H. Song, Y . Chen, D. Li, A New Kind of Transformer
Oil State Detection Method Based on Multi-Frequency Detection
Technique and Multivariate Statistics, Advances in Engineering Re-
search, 2016, doi:10.2991/iccte-16.2016.206
6
M. Yaacob, M. Alsaedi, J. Rashed, A. Dakhil, S. Atyah, Review on
partial discharge detection techniques related to high voltage power
equipment using different sensors, Photonic Sensors, 4 (2014)4 ,
doi:10.1007s13320-014-0146-7
7
A. S. Kumar, R. P. Gupta, K. Udayakumar, A. Venkatasami, Online
partial discharge detection and location techniques for condition
monitoring of power transformers: A review, 2008 International Con-
ference on Condition Monitoring and Diagnosis, Beijing, 2008,
927–931
8
M. Koziol, L. Nagi, M. Kunicki, I. Urbaniec, Radiation in the Optical
and UHF Range Emitted by Partial Discharges, Energies 12 (2019),
4334
9
M. Oda, Y . Fuchigami, Y . Sawabe, Development of diagnostic tech-
nologies for electrical facilities and portable instruments utilizing
these techniques, JFE Technical Report, (2016) 21, 153-160
10
S. Gudzius, A. Jonaitis, R. Miliune, A. Morkvenas, P. Valatka, V .
Malazinskas, Investigation of influence of short duration overvoltage
disturbances on partial discharge characteristics, 2017 IEEE 58th In-
ternational Scientific Conference on Power and Electrical Engi-
neering of Riga Technical University (RTUCON), Riga, 2017, 1–5
11
R. Yadav, S. Kumar, A. Venkatasami, A. M. Lobo, A. M. Wagle,
Condition based maintenance of power transformer: A case study,
2008 International Conference on Condition Monitoring and Diagno-
sis, Beijing, 2008, 502–504
12
M. Suzuki, S. Ota, R. Ikeda, M. Kawada, Internal defect position
evaluation of the ground coil by detecting the electromagnetic waves
from the partial discharge, Proceedings of 2011 International Sympo-
sium on Electrical Insulating Materials, Kyoto, 2011, 181–184
13
M. D. Judd, Experience with UHF partial discharge detection and lo-
cation in power transformers, 2011 Electrical Insulation Conference
(EIC), Annapolis, MD, 2011, 201–205
14
F. Tian, Y . Hao, Z. Zou, Y . Zheng, W. He, L. Yang, L. Li, An Ultra-
sonic Pulse-Echo Method to Detect Internal Defects in Epoxy Com-
posite Insulation, Energies, 12 (2019) 4804
15
M. Zhu, Y . Wang, Y . Li, H. B. Mu, J. Deng, X. J. Shao, G. Zhang,
Detection and localization of partial discharge in air-insulated substa-
tions using UHF antenna array, 2016 3rd Conference on Power Engi-
neering and Renewable Energy (ICPERE), Yogyakarta, 2016,
221–224
16
M. Kunicki, A. Cichoñ, Application of a Phase Resolved Partial Dis-
charge Pattern Analysis for Acoustic Emission Method in High V olt-
age Insulation Systems Diagnostics, Archives of Acoustics, 43
(2018) 2, 235–243, http://acoustics.ippt.gov.pl/index.php/aa/arti-
cle/view/2077, 24.11.2019, doi:10.24425/122371
17
DFA300 Dielectric Fault Analyzer User Guide, Doble Engineering
Company, USA, Rev. 1, 2013
18
DFA300 Training Course, Doble Engineering Company, 2015
V. MALA@INSKAS et al: EV ALUATION OF THE INSULATION CONDITION OF HIGH-VOLTAGE TRANSFORMERS ...
Materiali in tehnologije / Materials and technology 55 (2021) 2, 253–261 261