The 20th International Symposium on High Voltage Engineering, Buenos Aires, Argentina, August 27 – September 01, 2017

AN INVESTIGATION ON THE RELATION BETWEEN PRPD PATTERNS

ACQUIRED BY CONVENTIONAL AND UHF NONCONVENTIONAL PD
MEASURING TECHNIQUE FOR POWER TRANSFORMERS

M. Akbari Azirani1*, P. Werle1, A. Akbari2, H. Jahangir2 and J. Szczechowski3

1
Leibniz Universität Hannover, Institute of Electric Power Systems,
Division of High Voltage Engineering and Asset Management, Schering-Institute,
Callinstr. 25A, 30167 Hannover, Germany
2
K. N. Toosi University of Technology,
Department of Electrical Engineering, Tehran, Iran;
3
ABB AG, Transformers Service, Trafoweg 4, 06112 Halle, Germany
*Email: <[email protected]>

Abstract: Partial discharge (PD) signal propagation in the ultrahigh frequency (UHF) range
is one of the aspects that can be used to characterize PD inside high voltage equipment.
The UHF PD measuring technique has some advantages in comparison to the conventional
method according to IEC 60270 including the possibility of online application, localization
and improved immunity against external noises. Phase resolved partial discharge (PRPD)
patterns method, in which the PD pulses are displayed with respect to the phase of power
frequency of the AC cycle is a widely used and effective technique to analyze the PD data.
Based on the phase and charge magnitude of PD occurrences, a pattern is obtained, which,
according to theory, can be used to distinguish the types of PD. However, practical
experiences, especially for onsite measurements, indicate, that the PRPD pattern
technique is not always successful in identifying the cause of PD, particularly in case of
using nonconventional PD measuring techniques, such as the UHF method.

In this contribution, the alternative approach of applying the PRPD technique to the
captured UHF signals generated by PD in order to obtain an UHF PRPD pattern as a
complementary PD cause analysis is investigated. For this purpose, different fault
scenarios were created inside a specially designed transformer tank. The UHF signals were
captured via different antennas using several measuring setups to determine how the
position and the surrounding material of the PD source affects the sensitivity of the UHF
measuring technique using a PD fault model with a constant electric PRPD pattern.
Furthermore, the correlation between the captured PRPD patterns using the electrical
measurement according to IEC 60270 and the UHF measuring technique were analyzed
and compared. Finally, the results are presented and the advantages and flaws of the PD
cause identification via the UHF PRPD technique are discussed.

1 INTRODUCTION external interferences, this feature can be used to

develop a complimentary tool for the analysis of the
Partial Discharge (PD) electromagnetic wave PD data, in order to gain a more accurate
propagation in the ultrahigh frequency (UHF) range identification of the PD cause.
offers the possibility of localizing the source of PD.
Its advantages and limitations in characterizing the The effect of PD fault model position inside a
type of PD for evaluating the condition of power transformer tank was already investigated [5],
transformers is the subject of many however, the two antennas for capturing the UHF
investigations [1, 2, 3]. For this purpose, one of the PRPD pattern were not identical. Moreover, the
most applied methods is the Phase Resolved Partial electrical PRPD pattern was not constant for each
Discharge (PRPD) pattern recognition analysis. position due to the arbitrary nature of PD, the
Hereby, the amplitude as well as the phase of the measurement is repeated with an improved test
AC power cycle of each individual PD pulse is used setup. Furthermore, the investigation was extended
to acquire a pattern [4]. By using the PRPD to determine how the surroundings of the PD source
technique, the PD data can be interpreted to identify affects the UHF PRPD pattern.
the PD cause, leading to better preparations for the
repair by maintenance engineers. 2 PRINCIPLE OF MEASUREMENT

However, in a noisy environment, e.g. onsite 2.1 UHF Probe

measurements, external disturbances may lead to
false interpretation of the PD data. Since the UHF For the UHF measurements, the OMICRON
measuring technique is more robust against UVS 610 with a broad frequency range as shown in
Figure 1 was used. This conical monopole antenna model used for the experiments is a special design
is inserted into the transformer through the vent of and has six DN 80 oil drain valves, which are used
the oil drain valve, while maintaining the oil- to insert the OMICRON UVS 610 into the tank.
tightness via its DN 80 flange connector. This
antenna is designed for the frequency range up to For the measurements, the two identical OMICRON
1 GHz according to the manufacturer [6]. UVS 610 antennas are slided into the tank through
the lowest oil drain valve on two perpendicular sides
of the transformer tank model, as shown in Figure 3.
Hence, the dependency between the sensitivity and
position of the antenna inside the tank can be
investigated by carrying out the measurement with
both antennas simultaneously.

Figure 1: UHF Probe OMICRON UVS 610

Since the established electrical PD measurement

can be calibrated, the absolute value of the
magnitude of the PD pulses can be compared. This
is not the case with the UHF measuring technique,
since the possibility of calibration for this method is
still under investigation. Therefore, in order to make
them comparable among themselves, a reference
signal is defined for each used antenna.

2.2 Test Setup

The test circuit used for the measurements is shown

in Figure 2. The electrical measurement for DN 80
obtaining the PRPD pattern is carried out in Oil Valve 1
DN 80
accordance to IEC 60270 [7], while simultaneously
applying the UHF measuring technique with two Oil Valve 2
identical OMICRON UVS 610 antennas.

Figure 3: Air-filled transformer tank model with

the two OMICRON UVS 610 antennas

2.3 PD Model

The fault model used for the generation of PD

consists of two Rogowski electrodes on a non-
perfect aramid paper disc, as depicted in Figure 4.

Figure 2: Test circuit for evaluation of relation

between electrical and UHF measure-
ments

For the electrical measurement, the detection

frequency is set from 100 to 400 kHz, while the UHF
measurement covers two frequency bands: from
210 to 280 MHz (medium band) and from 100 MHz
to 2 GHz (wide band). The air-filled transformer tank Figure 4: Internal discharge PD fault model
3 EFFECT OF PD SOURCE POSITION ON respectively, as depicted in Figure 6. This concept
THE UHF PRPD PATTERN is used throughout this contribution.

In order to determine the influence of the position of

the PD source on the UHF PRPD pattern, the
obtained conventional electrical PRPD pattern
based on IEC 60270 must remain constant
regardless of the position. The chosen PD source
fulfills this requirement. The electrical PRPD pattern
for the used internal discharge model is shown in
Figure 5.

(a)

Figure 5: PRPD pattern of the internal discharge

model by conventional PD measure-
ment according to IEC 60270

Three positions inside of the transformer tank

model, as shown in Figure 6, were investigated. (b)

(c)

Figure 7: Effect of position on PRPD patterns for

Position 1 (a), Position 2 (b) and
Figure 6: Top view of the test setup to evaluate the Position 3 (c)
effect of PD source position on the UHF
PRPD Pattern It is evident, that the position has little to no effect
on the UHF PRPD pattern. This is due to the fact
These three positions have different angles and that the emitted electromagnetic waves travel with
distances to the two antennas and enable a nearly the speed of light c inside the air-filled tank
thorough investigation on the effect of the position model with no restriction. Thus for the short
of PD fault on the captured UHF PRPD pattern. distances inside the tank, the difference in the
arrival time of the signals is negligible.
The results are depicted in Figure 7 for the medium
band UHF measurement. For each result, the top It is also noticeable that the antenna placed in the
row shows the conventional electrical PRPD corner of the tank possesses a lower sensitivity.
pattern, while row 2 and 3 show the UHF PRPD This is due to the antenna radiation pattern at
pattern for the antennas placed in oil valve 1 and 2 different positions inside the tank. This issue was
already addressed in [9]. The investigation was also and its major part travels through the pressboard
carried out using a wide band UHF measurement. cylinder and reaches the antennas.
The results were similar to the medium band
measurement, although both antennas showed a
higher sensitivity. This is generally a positive point,
but for on-site measurements, the signal-to-noise
ratio decreases, since there are usually several
noise sources present. Thus, it is necessary to find
a suitable frequency range for each individual case
depending on the prevailing general conditions.

4 EFFECT OF BARRIERS ON THE UHF

PRPD PATTERN

The location of the PD fault model in regards to its

surroundings is another factor that has to be
considered. In order to do so, two scenarios that are
probable in case of PD failures are investigated. It
is known that a pressboard barrier out of aramid
paper has nearly no effect on the UHF PRPD Figure 9: Top view of the test setup to evaluate the
pattern due to its low reflection factor [5, 10]. This effect of different barriers on the UHF
phenomenon was reinvestigated by placing the PRPD Pattern
pressboard barrier in Figure 8 (a) around the PD
fault model. Furthermore, the effect of metal barrier
on the UHF PRPD pattern was studied via two
cylinders with the dimensions of the first one being
the same as the pressboard barrier as shown in
Figure 8 (b) and the second one two times higher
with the same diameter. These setups were utilized
in order to simulate PD sources in transformers
surrounded by either insulation or windings.

(a)

(a) (b)
Figure 8: Pressboard barrier (a) and metal
barrier (b)

For this purpose, the test setup shown in Figure 9

was realized. The PD fault model is placed in
position 2 according to Figure 6. Since the captured
UHF PRPD patterns did not depend on the (b)
coordinates of the PD fault model, position 2 is used
for the investigations of this section. Figure 10: Electrical and UHF PRPD patterns for
barrier free (a) and with pressboard
The result for a medium band UHF PD barrier (b) around the PD fault model
measurement is depicted in Figure 10. In measurement
comparison to the barrier free measurement shown
in Figure 10 (a), the fact stated in [5] regarding the The effect of metal barriers on the UHF PRPD
negligible effect of pressboard barrier is again pattern for the barrier free measurement is shown in
confirmed as seen in Figure 10 (b). The reason is Figure 11 (a). The result for the metal cylinder
the high transmission factor of pressboard. Thus, barrier and the two times higher metal cylinder
only a small fraction of the incident wave is reflected
barrier are depicted in Figure 11 (b) and (c) context is turned into a cylindrical waveguide. Since
respectively. the electromagnetic waves cannot travel through
conductor material, they are reflected from the walls
The first metal barrier damps the propagated and continue to travel up the barrier and leave
electromagnetic signals. This effect is observed for through the opening at the top. Then they reach the
the antenna in the corner of the tank more antennas, but have a lower amplitude, since
significantly. The higher metal barrier damps the through every incident with the walls of the
signal further, affecting the electromagnetic waves cylindrical metal barrier, they have lost a fraction of
propagated in the negative half cycle more than the their energy. Thus, the amplitude of the captured
positive half cycle. UHF PRPD pattern is decreased. The longer the
metal barrier, the lesser the magnitude of the
individual received electromagnetic signal. This
phenomenon is more severe for the negative half
cycle, which can be used as an additional analysis
tool.

5 CONCLUSIONS

The effect of position inside the transformer on the

captured UHF PRPD patterns was investigated. It
was shown that due to the traveling speed of the
electromagnetic waves, the short distances
between the PD source and the antennas are
negligible. However, the investigations reconfirm
(a) that due to different reflection paths, sensors that
are mounted in corners of the tank are usually much
less sensitive than sensors that are mounted in the
middle of a tank wall, which is therefore preferable.
Furthermore, the influence of the surroundings of
the PD source was studied. In this regard,
pressboard and metal barriers were utilized to
simulate the insulation and windings respectively. It
was concluded, that the pressboard barrier had little
to no effect on the captured UHF PRPD patterns,
due to its insignificant reflection factor. By
investigating two metal barriers, it was shown that
the incident waves, emitted by the PD fault model
inside the cylindrical metal barrier, were reflected
(b) from the walls, until they reached the opening of the
cylinder on its top. The received signals via the
antennas had a lower amplitude in direct proportion
to the height of the cylindrical metal barrier. The
effect on the signals propagated in the negative AC
power half cycle was more significant, which can be
exploited as an additional analysis tool. Moreover, it
explains why PD inside field-graded transformer
bushings often cannot be detected using UHF
techniques.

ACKNOWLEDGMENTS
The authors would like to express their gratitude
towards the ABB AG, Germany for the financial
(c) support of this research.

Figure 11: Electrical and UHF PRPD patterns for REFERENCES

barrier free (a), with metal barrier (b) [1] A. Akbari, P. Werle, M. Akbari and H. R. Mirzaei,
and two times higher metal barrier (c) "Challenges in calibration of the measurement
around the PD fault model measure- of partial discharges at ultrahigh frequencies in
ment power transformers." IEEE Electrical Insulation
Magazine 32.2 (2016): 27-34.
By this test set-up, the PD fault model acts as the [2] H. R. Mirzaei, A. Akbari, E. Gockenbach, M.
stimulator of the metal cylinder barrier, which in this Zanjani and K. Miralikhani, "A novel method for
 ultra-high-frequency partial discharge
localization in power transformers using the
particle swarm optimization algorithm." IEEE
Electrical Insulation Magazine 29.2 (2013): 26-
39.
[3] N. Shirdel, A. Akbari, H. R. Mirzaei and M. S.
Abrishamian, "Three-dimensional simulation of
UHF signal propagation in transformer using
FDTD method." Power Engineering, Energy and
Electrical Drives (POWERENG), 2011
International Conference on (pp. 1-6). IEEE.
[4] H. Jahangir, A. Akbari, P. Werle, M. Farahani
and J. Szczechowski, “UHF PD Measurements
on Power Transformers – Sensitivity
Investigations”, VDE High Voltage Technology
2016; ETG-Symposium, 2016.
[5] M. Akbari Azirani, P. Werle, A. Akbari, H.
Jahangir and J. Szczechowski, "An
Investigation on PRPD Patterns Generated by
the UHF Measurement Technique for Power
Transformers." 25th Iranian Conference on
Electrical Engineering (ICEE), 2017.
[6] OMICRON electronics GmbH, MPD600 Product
Manual, 2017.
[7] International Electrotechnical Commission
(IEC), "High Voltage Test Techniques - Partial
Discharge Measurements," Geneva,
Switzerland, IEC International Standard 60270
2000.
[8] H. Jahangir, A. Akbari and P. Werle, "Possibility
of PD Calibration on Power Transformers Using
UHF Probes," to be published in IEEE Electrical
Insulation Magazine (2017).
[9] H. Jahangir, A. Akbari, P. Werle, M. Farahani
and J. Szczechowski, “UHF PD Measurements
on Power Transformers – Sensitivity
Investigations”, VDE High Voltage Technology
2016; ETG-Symposium, 2016.
[10]Y. Sun, Y. Wang, G. Xu, F. Cheng and J. Hu,
"Dielectric properties of paper-based
composites at microwave frequency." Textile
Research Journal 86.14 (2016): 1533-1542.