Content uploaded by Stefan Tenbohlen
Author content
All content in this area was uploaded by Stefan Tenbohlen on Nov 03, 2014
Content may be subject to copyright.
Conference Proceedings of CMD2010
Sensitivity Evaluation of Different Types of PD
Sensors for UHF-PD-Measurements
A. TROEGER1*, U. RIECHERT1, S. BUROW2 and S. TENBOHLEN2
1 1ABB Switzerland Ltd., Zurich, Switzerland
2 University of Stuttgart, Germany
*E-mail: alexander.troeger@ch.abb.com
Abstract — Ultra high frequency (UHF) partial discharge
detection is a common on-site insulation diagnosis technique for
gas-insulated switchgear (GIS) that has increasingly acquired
importance in recent years. Compared to other methods of
partial discharge (PD) measurement, it is less susceptible to
disturbances from outside the system. It does have the
disadvantage that the measured amplitude depends on a variety
of different factors which make it impossible to perform a signal
calibration as it is done for PD measurements according to
IEC 60270. It is possible, however, to perform the so called
sensitivity verification. This contribution describes the
measurement and evaluation of the influence of different
parameters like the type of sensor and the evaluation parameters.
PD sensors were developed to have a flat frequency response up
to and beyond 2 GHz which has been verified by laboratory
experiments. In addition to purpose-built PD sensors, it is
possible to use other components in the GIS for sensing UHF
partial discharges. This is namely the earthing switch shield
which is provided in all fast acting and maintenance earthing
switches. The presented frequency response and the sensitivity
check of these sensors prove the suitability for PD diagnostic and
monitoring in almost the same manner compared to the PD
sensors.
Keywords -- UHF, PD-Measurement, Sensors, Sensitivity
I. INTRODUCTION
The demand for reliable and economic medium- and high-
voltage installations is increasing due to the present rapidly
changing conditions in the power substation and distribution
markets. As a result, the diagnostics of gas-insulated
switchgear has also generally acquired importance.
In order to perform diagnosis or monitoring of PD which
occur inside a GIS, the UHF method is generally applied. Its
advantage in comparison to the “electrical” measurement
method described in IEC 60270 is that a bulky coupling
capacitance is not required. It is however difficult to use the
amplitudes of the measured signals for the interpretation of the
measurement results – they are dependent on the type of
defect, the architecture of the system and the distance between
the defect and the measuring sensor. The same defect at a
different position leads to a different PD amplitude, a
calibration of an UHF PD measurement is therefore not
possible. It is possible, however, to perform a sensitivity
verification test that is described in the report of the CIGRÉ
Task Force 15/33.03.05 [1]. The investigations which are
described in this paper focus on the first part of the suggested
sensitivity verification, the determination of the voltage
amplitude of the pulse, which has to be injected in order to
simulate a 5 pC defect (hopping particle) in the GIS.
One of the essential parameters which do have a significant
impact on the results of such a sensitivity verification test is
the type of sensor that is connected to the measuring system.
In this paper, the sensitivity of isolated electrodes in
disconnector and earthing switches is compared to the
sensitivity of standard partial discharge sensors. Also,
different methods for the comparison of the obtained results in
respect to their frequency spectrum are evaluated.
II. INTERNAL SENSORS
ABB utilizes specially designed electric field sensors
(hereafter referred to simply as ‘PD sensor’) for detecting PD
in GIS ( Figure 1). The PD sensor consists of a small modular
unit which fits into earthing switch flanges of GIS components
and is designed to pick up the very fast transient radio
frequency pulses produced by PD.
The sensor consists of metal flange plate which house an inner
sensing element supported in a coaxial arrangement by
insulating material. The components are designed so as to
minimize reflections and other unwanted effects, thus
improving sensitivity in the UHF frequency range. The high-
voltage capacitance is between the inner conductor of the GIS
and the sensing element, while the stray capacitance is
between the sensing element and the GIS enclosure.
Measurement equipment is connected to the sensor via a
standard type N-connector.
Figure 1 Photo of UHF PD sensor
Conference Proceedings of CMD2010
The PD sensor is available for all types of ABB GIS from
170 kV to 1100 kV and was developed to have a flat
frequency response up to and beyond 2 GHz which has been
verified by laboratory experiments [2].
In addition to the purpose-built PD sensor, it is possible to use
other components inside the GIS for sensing UHF PD,
although since these are not optimized for this function [3].
This includes the earthing switch field shields. Earthing switch
field shields are provided in all combined disconnector /
earthing switches (DES) and fast-acting earthing switches
(FAES), as shown in Figure 2.
Figure 2 Drawing of the earthing switch shield in the disconnector
/earthing switch (DES, above) and of the earthing switch shield
in the fast-acting earthing switch (FAES, below): 1 Earthing
switch shield, 2 N-connector, 3 Moving contact, 4 Insulation, 5
GIS enclosure
Originally, these earthing shields have not been designed for
the sole purpose of PD measurement but also for zero
potential measurement. Because of the increasing importance
of PD diagnostic and monitoring also the field shields were
optimized to realize a good signal-to-noise performance and
sensitivity. Based on physical limitations it is not possible to
reach the same flat response over the whole frequency range
compared to the PD sensor. Figure 3 shows the frequency
respond of the different sensors. Both the DES and the FAES
sensor have a widely flat frequency response, with narrow
band sections with lower sensitivity between 1.1 GHz and
1.5 GHz.
Nevertheless, both the DES and the FAES sensor are suitable
for on-site PD measurement during commissioning tests,
service checks and for continuous PD monitoring and allow
sensitive measurements in the UHF frequency range. The
frequency characteristic of the sensors has to be considered
during sensitivity verification and for the PD measurement,
especially if narrow-band measuring systems are used.
-80
-60
-40
-20
0
00.511.52
Frequency [GHz]
Amplitude [dBm]
DES
FAES
Figure 3 Sensor frequency response
III. SENSETIVITY CHECK
An apparent “disadvantage” of UHF measurements is that the
UHF signal cannot be clearly correlated with the apparent
charge of the PD source. In other words, a calibration
according to IEC 60270 is not possible because of physical
and measurement reasons. It is possible, however to perform a
sensitivity verification according to report of the CIGRÉ Task
Force 15/33.03.05 [1]. As a result, the GIS operator can also
check the sensitivity of the UHF sensors and the measuring
system according to the CIGRÉ sensitivity verification
procedure on-site.
A. Comparison of Different Sensors
Figure 4 shows the test setup used for the experiments. It
consists of 4 adjacent GIS compartments in which different
sensors are installed. The compartments are divided by
partition or support insulators. They are connected with a high
voltage transformer that is equipped with a conventional PD
measuring system. The complete test setup has a noise level
below 2 pC. PD Sensor 2 is only used for the injection of
voltage pulses. Different UHF PD measurement systems
which are used in the experiments are always connected to PD
sensor 1 or to the shields.
In order to compare the characteristic properties of a voltage
impulse with the properties of typical defects in a GIS system,
an artificial defect was placed inside the compartment which is
near to PD sensor 2. As artificial defect hopping metal
particles with a length of 3 mm - 5 mm and a diameter of
1 mm were used. They were placed on the encapsulation and
the paint was removed in that area in order to ensure that the
particles are electrically charged if a high voltage is applied to
the system.
High voltage was applied to the system and was adjusted to a
level where the conventionally measured PD amplitude of
5 pC was reached. UHF PD measurement systems which were
Conference Proceedings of CMD2010
connected to the other three sensors measured the PD signal at
the same time. The corresponding voltage impulse was
determined by comparison of the measured UHF PD signals
with the recorded signals of the voltage impulses which were
injected in sensor 2. A commercially available “GIS
calibrator” which was able to produce voltage steps with
different amplitudes was used as source for the pulses. The
origins of the PD signals that are caused by the defect and by
the injected voltage pulse were very close to each other, so
that the influence of the GIS geometry on the signal
transmission was identical and can be neglected.
Figure 4 Picture of the experimental layout of indoor test facility (test
set-up A)
The amplitude of the artificial defect was measured by a
spectrum analyzer with custom made preamplifiers. A full
frequency sweep in the range from 300 MHz – 2 GHz was
performed. The measured spectrum consists of the maximum
amplitudes which have been determined for each frequency
during 60 s of measurement time. Moreover the average
spectrum was also recorded. A comparison with the spectra of
different voltage pulses can be done directly or with the aid of
statistical tools like the measured power (MP) and the average
power (AP) in the frequency spectrum, the maximum
amplitude (MA) or the averaged area per data point (AR) [4].
A direct comparison of the spectra measured at the PD sensor
shows that the amplitude of a corresponding voltage pulse is
between 9 V and 11 V in case of a hopping particle. This
result was also confirmed by measurements with a second
commercial peak detection system. In case of the earthing
shields the required corresponding voltage pulse is between
12 V and 13 V and therefore a little higher compared to the
PD sensor.
The corresponding statistical values for all sensors are listed in
TABLE I and TABLE II. TABLE I shows the results
evaluated from the average spectrum, whereas TABLE II
shows the values evaluated from the max-hold spectrum. The
results based on the statistical values are basically in
conformity with the direct comparison of the measured
spectrum, especially for the average spectra (MP and AP). Of
course, there are differences between the evaluation methods.
Depending on the frequency response of the sensor it is
important to choose the best frequency range to avoid a wrong
interpretation. For the FAES sensor a frequency range from
600 MHz to 2 GHz gives a better result. Moreover, prominent
peaks in the measured spectra influence the calculation results.
TABLE I STATISTICAL VALUES OF SPECTRA MEASURED AND
CALCULATED FOR THE DIFFERENT SENSORS (AVERAGE SPECTRUM)
Parameter
Voltage
Pulse /
Particle MP [dBm] AP [dBm] MA [dBm] AR
PD Sensor 1
Hopping
Particle 34 5.9 18 1.6
5 V 25 -4.1 15 0.1
7.5 V 29 1.1 19 0.8
12.5 V 32 4.2 23 1.1
FAES (600 MHz – 2 GHz)
Hopping
Particle 26 (22) -3 (-6) 11 (9) 0.23 (0.22)
5 V 27 (16) -1.8 (-12) 19 (3) 0.05 (0.05)
7.5 V 29 (17) 0.4 (-10) 21 (3) 0.1 (0.06)
12.5 V 31 (22) 3.7 (-7) 25 (10) 0.5 (0.25)
DES
Hopping
Particle 32 3.6 19 0.7
5 V 28 -0.4 19 0.1
7.5 V 29 0.8 14 0.3
12.5 V 33 4.5 20 0.7
TABLE II STATISTICAL VALUES OF SPECTRA MEASURED AND
CALCULATED FOR THE DIFFERENT SENSORS (MAXIMUM SPECTRUM)
Parameter
Voltage
Pulse /
Particle MP [dBm] AP [dBm] MA [dBm] AR
PD Sensor 1
Hopping
Particle 36 8.5 24 1.9
5 V 34 5.3 24 0.6
7.5 V 36 8.8 26 1.4
12.5 V 39 12 28 2.1
FAES (600 MHz – 2 GHz)
Hopping
Particle 34 (32) 5.4 (3.9) 23 (19) 0.6 (0.5)
5 V 35 (25) 6.8 (-3) 26 (15) 0.6 (0.2)
7.5 V 36 (28) 8.9 (-0.2) 27 (22) 0.7 (0.25)
12.5 V 39 (32) 11.6 (4) 29 (23) 1.4 (0.6)
DES
Hopping
Particle 42 14 27 3
5 V 37 8.4 23 0.9
7.5 V 39 11.4 25 1.9
12.5 V 42 15 27 3.6
Conference Proceedings of CMD2010
B. Influence of the Test Set-up
In laboratory experiments, the distance of the sensor for the
voltage injection and the sensor for the measurement is as
short as possible. The sensors are placed in adjacent GIS
compartments. In GIS installations the sensors are placed
throughout the GIS according to rules based on experience and
the topology of each individual installation. The distance
between the sensors could reach more than 20 m. In order to
determine the influence of the test set-up the results presented
in chapter 3A were compared to the results using a typical
laboratory test set-up for the first step of the sensitivity
verification [5]. Figure 5 shows the test set-up which was used
for the second test series. It consists of two adjacent GIS
compartments. Sensor 1 was only used for the injection of
voltage pulses. The UHF PD measurement system was always
connected to sensor 2.
Figure 5 Picture of the experimental layout of indoor test facility (test
set-up B)
TABLE III shows a comparison of the results for both test set-
ups, based on a direct comparison of the measured frequency
spectra. It could be concluded, that the effect of the test set-up
and therefore, the effect of the distance between the sensors
for the sensitivity verification could be neglected.
TABLE III EQUIVALENT VOLTAGE PULSE FOR HOPPING PARTICLES
Test set-up A Test set-up B
Required voltage pulse 9 V – 11 V 10 V
IV. OTHER TESTS
Beside the PD diagnostic and monitoring function, it is
important to know, that the sensors as well as the connected
measuring system is able to withstand the system
requirements. When switching capacitive currents, the focus is
on VFTO (very fast transient overvoltages) in particular. In
addition to the requirements for the disconnector itself, the
overvoltages that occur represent a major challenge in relation
to the EMC of the measurement and control equipment [6].
As the sensor may be struck by VFTO during earth switch /
disconnector operations, tests were carried out to verify that
the protection mechanisms of the continuous PD monitoring
systems are sufficient to prevent damage. Two system
immunity tests were carried out: disconnector switching
according to Annex F of IEC 62271-102 and closing operation
of the earthing switch with a trapped charge voltage of 1 pu at
the busbar section. The PD system was checked before, during
and after each test for correct operation. The system operated
correctly at all times, no failures or damage occurred during
either test. The tests have shown that the PD system is suitably
protected. For the FAES it is required to prove the short-
circuit making performance as well as the Short-time
withstand current and peak withstand current according to
IEC 62271-102. The PD measuring system was connected to
the sensor during these tests. Also in that case, the system
operated correctly at all times. Concluding, the shield
electrodes used for PD measurement are suitable for UHF PD
measurements and continuous monitoring.
V. CONCLUSIONS
A partial discharge sensor is available for all types of ABB
GIS from 170 kV to 1100 kV and was developed to have a flat
frequency response up to and beyond 2 GHz which has been
verified by laboratory experiments.
In addition to the PD sensor, it is possible to use other
components inside the GIS for sensing UHF PD. This includes
the earthing switch field shields. Earthing switch field shields
are provided in all combined disconnector / earthing switches
and fast-acting earthing switches.
During broadband measurements with the spectrum analyzer,
the spectrum of the maximal and average PD amplitudes was
recorded for a frequency span of 300 MHz – 2 GHz. A
comparison with the measurement of the PD sensors results
lead to the similar result if the equivalent voltage pulse was
determined by direct comparison of the spectra.
A comparison of statistical values that were derived from the
measured spectra did not lead to the same result in any case.
Depending on the frequency response of the sensor it is
important to choose the best frequency range to avoid a wrong
interpretation. The presented frequency response and the
sensitivity check of these sensors prove the suitability for PD
diagnostic and monitoring in almost the same manner
compared to the PD sensors.
REFERENCES
[1] CIGRÉ, Joint Task Force 15/33/03.05: Partial Discharge Detection
System for GIS: Sensitivity Verification for the UHF Method and the
Acoustic Method, Électra, No. 183, April 1999, pp. 75 – 87
[2] Riechert, U.; Tröger, A.; Schraudolph, M.; Bräunlich, R.; Neuhold, S:
PD Diagnostics of Gas-Insulated Switchgear - Sensitivity Verification,
Internationaler ETG-Kongress 2009, 27.-28. Oktober 2009, Düsseldorf,
2009, ETG-Fachbericht 119, 2009, S. 477-482,
[3] Hampton, B.F.; Meats, R.J.: Diagnostic measurements at UHF in gas
insulated substations, IEE Proc., Vol. 135, No. 2, 1988, pp. 137 – 144
[4] Meijer, S.: Partial Discharge Diagnosis of High-Voltage Gas-Insulated
Systems, Dissertation University of Delft, 2001
[5] Tröger, A.; Riechert, U.: Influence of Different Parameters on
Sensitivity Verification for UHF PD Measurement, 16th ISH 2009,
Cape Town, South Africa, 24 to 28 August 2009, conference
proceedings, paper B-33, pp. 521-524
[6] Riechert, U.; Krüsi, U.; Sologuren, D.: Very Fast Transient
Overvoltages during Switching of Bus-Charging Currents by 1100 kV
Disconnector, CIGRÉ Report A3-107, 43rd CIGRÉ Session, August 22-
August 27, 2010, Palais des Congrès de Paris, Paris, France
