654

UHF partial discharge

detection system for GIS:
Application guide for
sensitivity verification

Working Group
D1.25

April 2016
UHF PARTIAL DISCHARGE
DETECTION SYSTEM FOR GIS:
APPLICATION GUIDE FOR
SENSITIVITY VERIFICATION
WG D1.25

Members

U. Schichler, Convenor (AT), W. Koltunowicz, Secretary (AT),

D. Gautschi (CH), A. Girodet (FR), H.Hama (JP), K. Juhre (DE), J. Lopez‐Roldan (AU),
S. Neuhold (CH), C. Neumann (DE), S. Okabe (JP), J. Pearson (UK), R. Pietsch (DE),
U. Riechert (CH), S. Tenbohlen (DE)

Copyright © 2016

“All rights to this Technical Brochure are retained by CIGRE. It is strictly prohibited to reproduce or provide this publication
in any form or by any means to any third party. Only CIGRE Collective Members companies are allowed to store their copy
on their internal intranet or other company network provided access is restricted to their own employees. No part of this
publication may be reproduced or utilized without permission from CIGRE”.

Disclaimer notice

“CIGRE gives no warranty or assurance about the contents of this publication, nor does it accept any responsibility, as to the
accuracy or exhaustiveness of the information. All implied warranties and conditions are excluded to the maximum extent
permitted by law”.

ISBN : 978-2-85873-357-6
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

UHF Partial Discharge Detection

System for GIS: Application Guide
for Sensitivity Verification
Table of Contents

Executive Summary .............................................................................................................................................. 3

1 Introduction ...................................................................................................................................................... 4
2 Sensitivity verification Step 1: laboratory test ......................................................................................... 5
2.1 Background and general aspects ....................................................................................................... 5
2.2 Test setup ................................................................................................................................................. 6
2.3 UHF measuring device .......................................................................................................................... 8
2.4 Pulse generator ...................................................................................................................................... 9
2.5 Determination of artificial pulse magnitude ..................................................................................... 9
3 Sensitivity verification Step 2: on-site test .............................................................................................. 12
3.1 General aspects .................................................................................................................................. 12
3.2 Arrangement of the sensors: principle, details and important aspects .................................... 12
3.2.1 Sensor locations according to ELECTRA Report.................................................................. 12
3.2.2 Alternative method for the location of sensors .................................................................. 14
3.3 Influence of switching devices ........................................................................................................... 16
3.4 Test equipment .................................................................................................................................... 17
3.5 Execution of the on-site sensitivity verification .............................................................................. 17
3.6 Criteria to pass the on-site test ........................................................................................................ 17
4 Conclusion ..................................................................................................................................................... 18
References .......................................................................................................................................................... 20
Annexes
Annex 1: Fundamental PD signal propagation characteristics................................................................. 22
Annex 2: Sensors ............................................................................................................................................... 25
Annex 3: Distance between sensors .............................................................................................................. 29
Annex 4: Type of PD defects ......................................................................................................................... 30
Annex 5: Mixed technology switchgear ....................................................................................................... 33
Annex 6: Vintage GIS ...................................................................................................................................... 34

Page 2
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

EXECUTIVE SUMMARY
Gas-insulated switchgear (GIS) have been in operation for more than 45 years and it shows a high level of
reliability. However, the return of experience indicates that some of the in-service failures are related to defects
in the insulation system. Many of these defects can be detected by partial discharge (PD) diagnostics. The ultra-
high frequency (UHF) measurement method, which was introduced in the late 1980’s for PD detection, is used
worldwide by GIS manufacturers during routine testing in factory, during commissioning on-site and by utilities
for continuous monitoring in service. The UHF method is less sensitive to noise, so easier to handle in comparison
with the conventional method according to IEC 60270.

An Electra Report published in 1999 by CIGRE Task Force 15/33.03.05 describes the two-step procedure for
the sensitivity verification of the UHF system in a very general way. After 15 years of its application, it became
obvious that a more detailed description of the sensitivity verification procedure is necessary to avoid technical
misinterpretations.

This Technical Brochure collects the available experience on sensitivity verification and describes its practical
applications for GIS. A detailed description of the two-step procedure is given and supported by examples.
Guidelines will help manufacturers and users for the effective application of the UHF method for PD detection on
GIS.

In Chapter 2, the sensitivity verification (Step 1) is described. The aim is to determine in the laboratory an artificial
PD pulse magnitude equivalent to 5 pC of apparent charge of a defined defect, which will be applied later on-
site during Step 2. The compact test set-up for single-phase GIS is defined (Chapter 2.1), as well as for a
complete bay of three-phase GIS (Chapter 2.2). The examples of PD defect types to be used in Step 1 are given
in Chapter 2.2 and in Annex 4.

The components of the measuring chain like e.g. UHF sensor (Annex 2), PD acquisition unit (Chapter 2.3) and pulse
generator (Chapter 2.4) are described. The determination of the artificial pulse magnitude is shown with
particular attention to different methods of comparison of UHF spectra from the real PD defect with the spectra
of artificial calibration pulse (Chapter 2.5).

In Chapter 3, the sensitivity verification (Step 2) to be performed on-site is described. The aim of this step is to
verify that the installed sensors and the UHF measurement or monitoring system have sufficient sensitivity to detect
signals, equivalent to those from a specific type of PD defect, within any compartment of the GIS being checked.
At the same time, the correct functioning of the sensors themselves and the measurement chain is also tested. The
principle of the Step 2 procedure and sensor arrangement is described in Chapter 3.2. The typical and
alternative methods for the location of sensors leading to the reduction in the number of sensors, by limiting the
overall detection sensitivity for all GIS compartments to the required detection sensitivity of 5 pC, are also
proposed (Chapter 3.2). The influence of the position of the GIS switching devices on the attenuation of the UHF
signal is shown in Chapter 3.3. The criteria to pass the UHF sensitivity verification test are described in Chapter
3.6 and Chapter 4 presents the conclusions.

Page 3
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

1 INTRODUCTION
Gas-insulated switchgear (GIS) have been in operation for more than 45 years and they have shown a high level
of reliability with very low failure rates. This is the result of quality assurance during the development and
manufacturing process as well as during installation and commissioning. However, the return of experience shows
that some of the in-service failures are related to defects in the insulation system. Many of these defects can be
detected by partial discharge (PD) diagnostics.

Nowadays the UHF method, which was introduced in the late 1980’s for PD detection, is used worldwide by GIS
manufacturers in the factory, during commissioning and by utilities for monitoring in service based on positive
return of experience from the field. The UHF method is less sensitive to noise and easier to handle in comparison
with the conventional method according to IEC 60270. For the UHF method a calibration to “apparent charge in
pC” is not possible due to the complex PD pulse propagation characteristics. The PD detection sensitivity depends
on various parameters like the distance between the PD defect (UHF signal source) and the sensor.

In 1999 a report was published by CIGRE Task Force 15/33.03.05 which described the sensitivity verification
on GIS applicable to the UHF method. The proposed two-step procedure ensures that defects causing an
apparent charge of 5 pC or greater can be detected by the UHF method [1]. Manufacturers and users gained
a lot of experience since the proposed sensitivity verification method was applied for GIS commissioning and PD
monitoring in service. The previous challenges according to PD identification and risk assessment based on PD
diagnostics were discussed in the last years by CIGRE WG D1.03 (TF 09) and the published report provides
valuable information to facilitate the application of UHF method [2]. Today the on-site testing of GIS with lightning
impulse voltage is nearly completely replaced by the sensitive PD measurement using the UHF method. However,
it is necessary to keep in mind that the UHF method for PD detection is still a complex technique and by now not
standardized.

The ELECTRA report from CIGRE Task Force 15/33.03.05 describes the two-step procedure for sensitivity
verification in a very general way. During the last years it became obvious that a more detailed description of
the sensitivity verification is necessary to avoid technical misinterpretations on the proposed two-step procedure
and the required measures at the laboratory test (Step 1) and the on-site test (Step 2).

This Technical Brochure collects the available experience on sensitivity verification and describes practical
applications of the sensitivity verification for GIS. A detailed description of the two-step procedure is given and
supported by examples. Guidelines will help manufacturers and users in the effective application of the UHF
method for PD detection on GIS.

Page 4
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

2 SENSITIVITY VERIFICATION STEP 1: LABORATORY TEST

2.1 Background and general aspects
The signals in the UHF frequency range can be detected by means of dedicated sensors [3]. Because of the
distortion of UHF signals described, a certain number of sensors have to be installed in a GIS in order to cover
it overall with sufficient sensitivity to detect PD defects. Due to the complexity of the resonance patterns, the
magnitude of the detected signals depends strongly on the PD location relative to the sensor and to a minor
degree on the orientation of the defect and the sensor installed on the GIS compartment (Annex 1). A detailed
high frequency model would be required to enable the formulation of an overall transfer function between PD
defect (emitter) and sensor (receiver). Nowadays a numerical calculation seems to be not possible because the
precise location and orientation of the defect are generally not known, and moreover different defect types
generate different Radio frequency (RF) spectra. Creating an accurate RF model of a GIS would already pose
a formidable challenge, requiring highly sophisticated RF modelling software and entry of every internal
construction detail of the GIS. However, such a model would represent only a perfect version of the GIS, mostly
due to the skin effect, which determines the penetration depth of RF signals on conductors. For aluminium at 100
MHz, the skin depth is already on the order of 10 µm. Since a GIS is built to transport kiloamperes and kilovolts
at typical power frequencies - many orders away from the magnitude of the PD signal in the UHF frequency
range - the inner contact design and assemblies simply do not meet the precision and repeatability at µm-scales
to guarantee a reliable RF signal transfer. Thus even if a high quality RF model were to be created, both trying
to verify its validity and assuming its accuracy for predicting actual transfer functions would be virtually
impossible.
When attempting to compare the magnitude of PD using the UHF method versus the conventional method
according to IEC 60270, the actual charge transferred at the defect results in the well-known apparent charge
measured via a coupling capacitor. However, PD signals measured with the UHF method depend on the following
factors:
1. The type of defect
2. The location of the defect within the GIS compartment
3. Propagation effects including reflection, dispersion, interference and attenuation [4, 5]
4. The position of the sensor relative to the defect and the signal transmission path [6, 7, 8]
5. The characteristics of the sensor, i. e. its own transfer function
6. The characteristics of the complete chain of the measurement system (from sensors to data acquisition
and display)

The above mentioned factors are well-known and have been widely documented in the literature. The position
dependence of UHF PD signal strength is easy to demonstrate in a laboratory. The situation described clearly
indicates that a charge calibration of the UHF method is not possible for GIS installations [1, 9]. However, when
carrying out PD measurements, e.g. for the purpose of commissioning tests or when implementing PD monitoring
systems utilizing the UHF technique, it is highly desirable to be able to verify that those systems are functioning
at required level of sensitivity. Therefore, a method of verifying this detection sensitivity is proposed in this report,
based on correlating the UHF signal to the apparent charge measured in an IEC 60270 set-up using a real PD
defect. Using this comparison, a two-step procedure can be used to establish that a UHF measuring system is
functioning and able to detect defects with an apparent charge of e.g. 5 pC [1].
In general the UHF method has proved to be at least as sensitive in detecting most of the defects as the
conventional method according to IEC 60270, and this is mainly due to the GIS enclosure functioning as a Faraday
cage, thus helping to screen out external electromagnetic interference (EMI) disturbances and thus enable a low
background noise level [10]. The sensitivity of the UHF measuring system can be influenced by using suitable
sensors, amplifiers, filters, and associated signal-processing equipment (Annex 2). Tests in laboratories and on-
site have shown that critical defects - and even other defects - may be detected [11].

Page 5
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

2.2 Test set-up

The laboratory test must be performed in order to determine the magnitude of an artificial PD pulse, which will
be applied later on-site during Step 2 of sensitivity verification, by comparison to a real defect measured
according to IEC 60270.
The laboratory test is carried out as follows: A real defect is placed close to the UHF sensor C1 of a laboratory
set-up as shown in Figure 1a. The defect will start to discharge when the applied voltage is high enough. When
the apparent charge of the related PD signal, measured according to IEC 60270, reaches the threshold of e.g.
5 pC, the value of the UHF signal related to the signal intensity (e.g. pulse energy, pulse magnitude) is measured
at sensor C2. This UHF signal magnitude A (signal A) will be used for comparison in the next step of the laboratory
test.

The artificial pulses - as described below - are injected into sensor C1 as indicated in Figure 1b. The UHF signal
is again acquired at sensor C2, as during the preceding step. The resulting UHF signal magnitude B (signal B) is
to be compared with the magnitude A from the preceding HV measurements. The amplitude of the artificial pulse
has to be varied until the magnitude of the measured UHF signal B is equivalent to the magnitude of the UHF
signal A within an accepted tolerance of ± 20 % [1].
The lowest possible attenuation (shortest distance) between two sensors is preferable [1]. If not possible, the
sensitivity verification Step 1 could be performed, without any disadvantages, using a complete GIS bay
(Annex 3).

Figure 1: Laboratory set-up for the high voltage measurements (a) and low-voltage measurements (b)
during sensitivity verification Step 1 [1]

Figure 2 shows a typical test setup which is used in the laboratory [12, 13]. It consists of two GIS compartments
in which sensors are installed. The compartments are divided by an insulator and connected to a high voltage
transformer. The test setup is equipped with a coupling capacitor and a conventional PD measuring system
according to IEC 60270. The complete test setup does have a background noise level below 2 pC. Sensor C1 is
only used for the injection of artificial voltage pulses from a pulse generator (PG). The UHF PD measurement
system which is used in the laboratory test is always connected to sensor C2.

Page 6
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

Sensor C 2
Sensor C1

HV Connection

Defect location

Figure 2: Typical test set-up for sensitivity verification Step 1 of single-phase GIS [13]

The defect has to be placed inside the compartment that is as near to sensor C1 as possible. Often a moving
particle is used, e.g. with a length of 3 - 5 mm and a diameter of 1 mm. It is normally placed on the enclosure
at a location near to sensor C1. Instead of a moving particle, protrusions on the high voltage conductor, simulated
by a needle with a length of 5 - 10 mm and a tip radius of about 0.5 mm that could be positioned preferably
right underneath the sensor (Figure 3), are used [13].

Position Sensor C1

Needle

Figure 3: Needle placed inside the GIS compartment to simulate a protrusion on the inner conductor [13]

On three-phase encapsulated GIS the sensitivity verification Step 1 can be performed as well on small
arrangements like described in Figure 2. Nevertheless, due to their limited size and the less homogenous
transmission line characteristic of these GIS the sensitivity verification might be performed on a complete bay as
presented in Figure 4.

Figure 4: Test set-up for sensitivity verification Step 1 on a three-phase encapsulated 145 kV GIS
(Courtesy of GE Grid (Switzerland) GmbH)

Page 7
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

Just as in single-phase arrangements the sensor C1 and the defect have to be placed as near as possible to
each other. Investigations have shown that the distance between the sensors C1 and C2 has a negligible effect
as long as it is in the range of some meters (Annex 3, [14]).

2.3 UHF measuring device

UHF signals can be detected in the time domain or the frequency domain. The results obtained in the time domain
can be characterized by the magnitude of the UHF signals. Measurements in the frequency domain result in
spectra which show the amplitudes of the various frequency resonances stimulated by the PD pulses.
There is no recommendation for any specific acquisition and evaluation method, but it is mandatory that the same
method should be used throughout for the sensitivity verification. Therefore any meaningful method may be used.

Narrow-band systems (Figure 5) examine a part of the frequency range for PD signals (e.g. spectrum analyzer).
This has the advantages of being able to provide high rejection of ambient interference signals and improved
signal-to-noise ratio. In addition to these advantages, the examination of the frequency spectrum itself often
reveals important information about the PD defect type and location [4].

Figure 5: Narrow-band spectrum and PD system block diagram [4]

Wide-band systems (Figure 6) amplify a broadband frequency spectrum (e.g. 300 - 1500 MHz) and feed the
signal directly into a detector. The output is shaped and sent directly to a display device (oscilloscope) or to
an A/D converter, which might be part of PD measurement or monitoring systems.

Figure 6: Wide-band spectrum and PD system block diagram [4]

2.4 Pulse generator

Page 8
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

Depending on the means by which the UHF signal is recorded and compared, the magnitude for the artificial
pulses may vary slightly [15]. These variations are not significant in view of the accuracy which is inherent in the
present procedure. The measuring cables should be suitable for frequencies in the applied UHF range.

The exact shape of the artificial pulse is not critical. However, the following parameters are of importance:
• The rise time of the artificial pulse determines the upper frequency limit of its output spectrum and thus that
of the signal emitted at the sensor. Therefore, the rise time of the artificial pulse must be appropriate that such
frequencies across the whole measurement bandwidth are excited. As different bandwidths can be used for the
detection of the UHF signals, there is no need for a detailed definition, however the rise time of the artificial
pulses must not exceed 0.5 ns.
• The magnitude of the pulse voltage (chapter 2.5).
• The time between consecutive pulses must be greater than the longest duration of the resonating UHF signals
observed at the sensors in order to avoid superposition of multiple pulses. Usually any repetition rate less than
100 kHz can be used, e.g. 50 Hz or 60 Hz.

The above mentioned parameters of the pulse generator have to be documented by test certificates or
measurements during the laboratory test. The same type of pulse generator must be used for the on-site test
(sensitivity verification Step 2) to ensure that the tests are comparable. It is recommended to use the same type
of pulse generator at the laboratory tests (sensitivity verification Step 1) and on-site (sensitivity verification Step
2).
Measurements have been carried out at several laboratories using different GIS designs and different methods
for generating the artificial pulses. The available results indicate that the pulse magnitude of the artificial pulses
typically lie in the range from 5 - 20 V depending on pulse shape, sensor design and type of GIS.

Note: Export control regulations for high-speed pulse generators must be considered in EU, US, and JP.

2.5 Determination of artificial pulse magnitude

The methods presented here are not a calibration. However, the aim is to find the best possible match between
a real PD defect and the artificial pulse magnitude in order to verify the sensitivity of the measurement system
during the on-site sensitivity verification (Step 2). A tolerance of ± 20% is acceptable for the determination of
the artificial pulse magnitude.

A) Determination of artificial pulse magnitude in frequency domain (using a spectrum analyzer)

Figure 7 shows the measured amplitude spectrum of a 5 pC PD defect compared with the noise spectrum, i.e.
without applied high voltage [12, 16, 17]. The different lines or peaks in the noise signal are caused by radio
and TV stations, mobile radio transmitters and from a nearby airport radar [16]. This is not a result of a poorly
shielded measurement arrangement, but rather these interfering signals are directly coupled into the GIS test-
setup, e.g. via the flanges or in the present case through the bushings. Figure 8 shows that the spectrum of the
applied signal at sensor C1 is best aligned with that of the real 5 pC PD using a pulse with a magnitude of 2 V.

Page 9
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

Figure 7: Amplitude spectra - noise signal without applied high voltage and
signal of a 5 pC PD - measured at sensor C2

Figure 8: Comparison of the amplitude spectrum measured at sensor C2 for a real 5 pC PD

with the amplitude spectrum of an artificial pulse with a magnitude of 2 V

A comparison with the spectra of different voltage pulses can be done by visual comparison or with the aid of
statistical tools. The choice of an extracted characteristic based on spectrum power is one of the most relevant
parameters [18] compared to average amplitude (AA). Average amplitude (AA), maximum measured power
(MP) and the average power (AP) applied for a moving particle are listed in Table 1.

Table 1: Statistical values of frequency spectra calculated in case of a particle defect

showing 5 pC apparent charge and different pulse magnitudes from a pulse generator

MP AP AA
[dBm] [dBm] [dBµV]
Moving particle -25.7 -50.9 53.4
10 V -25.8 -51.2 50.9
20 V -19.5 -44.8 56.5

In this example a voltage pulse magnitude of 10 V fits well with the signal extracted from the particle. A visual
comparison confirms that this method can be used for such type of defect.
Typically, a comparison of the entire frequency spectrum is used for the determination of the required artificial
impulse amplitude and narrow-band measurements are used in addition because of higher sensitivity. In the case
where a narrow-band measurement system with fixed frequency band is used without comparing the entire
spectrum, special attention has to be drawn on the sensitivity verification. Narrow-band measurements have the
disadvantage that the results do not only depend on the PD defect signal but also on the measuring frequency.
This can turn into an advantage (meaning an even lower susceptibility to external noise) if the measuring

Page 10
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

frequencies are carefully chosen [13]. That has been done for the current setup with a narrow-band UHF PD
measuring system. Table 2 contains the measured equivalent voltage pulses. There is a good correlation in case
of the measurements made at 740 MHz and it can be seen that the results obtained at the two other frequencies
show significant deviations.

Table 2: Equivalent voltage pulses of both artificial defects of 5 pC measured

with a narrow band system at different frequencies

Moving
Frequency Protrusion
particle
540 MHz 2-5V 1-2V
740 MHz 2-5V about 10 V
1240 MHz 1-2V 10 - 20 V

B) Determination of artificial pulse magnitude in time domain (using PRPD pattern)

For wide-band peak detection systems like PD monitoring systems a comparison can be made using PRPD pattern
(Figure 9). In the given example an artificial pulse magnitude of 20 V is equivalent to 5 pC caused by a moving
particle.

Figure 9: Determination of artificial pulse magnitude by using PRPD pattern for a PD defect showing e.g.
an apparent charge of 5 pC from moving particle (left) and artificial pulses with different magnitudes (right)

Page 11
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

3 SENSITIVITY VERIFICATION STEP 2: ON-SITE TEST

3.1 General aspects
The second step of the CIGRE sensitivity verification is carried out on-site on the installed GIS (same type of GIS
as used in the laboratory test) in order to verify that the installed sensors and the UHF measurement or monitoring
system has sufficient sensitivity to detect signals, equivalent to those from a specific type of PD defect, within any
compartment of the GIS being checked. In addition and at the same time, the correct functioning of the sensors
themselves and the measurement chain is tested.
It is understood that the same type of sensors and the same pulse generator must be used during the on-site
sensitivity verification as employed for the laboratory test (Step 1). Step 2 should be carried out with the same
measuring system which was used at Step 1 or with a system with similar or better detection sensitivity.
In general the on-site sensitivity verification is fulfilled if the injected artificial pulse can be measured at the
adjacent sensors as described in [1].

The number of sensors required to be installed in the GIS depends on the PD defect and the required detection
sensitivity. In addition to the sensor arrangement according to [1] an alternative proposal is presented for sensor
arrangements which are strongly limited to the required detection sensitivity.

3.2 Arrangement of the sensors: principle and important aspects

3.2.1 Sensor locations according to ELECTRA Report
The positioning of the sensors within the GIS should be done in such a way that a PD defect (e.g. a moving
particle showing an apparent charge of 5 pC) can be detected in any compartment of the GIS by at least two
sensors [1].
The number and location of the sensors will determine the sensitivity of the UHF PD measurement. Many
parameters impact the attenuation observed between two sensors, for instance diameter changes of the enclosure
or geometrical shapes of gas compartments [11]. It is not possible to establish fixed rules describing the best
arrangement of the sensors.

Principle: The locations of the UHF PD-sensors should be chosen in a way that a predefined sensitivity is
guaranteed for the entire GIS. Figure 10 shows a GIS section with three sensor locations. In the laboratory test
(Step 1), the pulse generator magnitude has been determined, e.g. 10 V.

Figure 10: Principle of CIGRE sensitivity verification Step 2

At the GIS erected on site, the pulse generator is connected to one sensor (II) and the UHF PD measuring system
is connected to the neighbouring sensors (I) and (III). When the predetermined pulse magnitude injected from the
pulse generator (e.g. 10 V signal as above) can be measured at the adjacent sensors, the sensitivity check is
fulfilled for both GIS sections. If this is valid for all sensor configurations it is considered that the combination of

Page 12
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

sensor locations and measurement system is sufficient to measure the entire GIS with a predefined sensitivity (e.g.
5 pC of a moving particle).

In detail: Consider a single phase encapsulated GIS with five UHF PD-sensors located at the positions A, B, C, D
and E on each phase according to Figure 11 [19]. The cable termination A and the transformer bushing E
correspond to the endpoints of the GIS.

Figure 11: 420 kV GIS with five sensor locations: A - E (section view (left) and schematic (right))

An illustration of the measuring sequence is presented in Figure 12a for different injection and measuring points
on the GIS. To simplify this representation, the attenuation between two sensors is considered to be linear (see
Annex 1 for details). Two cases of defect location will be considered.
Case A: A 5 pC defect is located in the middle of the section B-C. From Figure 12b it can be deduced that the
defect can be detect by sensor B and sensor C but not by sensors A and D.

Case B: A 5 pC defect is located close to sensor C. From Figure 12c it can be deduced that the defect can be
detected by UHF PD-sensor B, C, D and E.

Figure 12a: Simplified UHF signal attenuation for different injection and measurement locations

Page 13
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

UHF
signal
[dBm]A B C D E
Level measured Level measured
in B in C
[m]
X:
Distance along GIS

Figure 12b: Simplified UHF measuring level for a PD defect location in the middle of the section B-C

UHF
signal
[dBm] B C D E
Level measured in C
Level measured in D
Level measured in B
[m]
A X:
Distance along GIS

Figure 12c: Simplified UHF measuring level for a 5 pC equivalent defect close to C

The method proposed in [1] for on-site sensitivity verification may lead to the installation of more sensors than
effectively necessary to detect a 5 pC PD defect. Whatever the position of the defect, it can be detected by at
least two sensors. This also means that the detection sensitivity with such locations and number of sensors is much
better than 5 pC.

3.2.2 Alternative method for the location of sensors

In this chapter an alternative method is shown which allows a reduction in the number of sensors by limiting the
overall detection sensitivity for all GIS compartments to the required detection sensitivity of e.g. 5 pC [19].
Considering Figure 13a, only the sensors at locations B and D are required to fulfil the requirement of 5 pC
detection level in the sections A-B, B-D and D-E.

Figure 13a: Simplified UHF signal attenuation according to sensor injection point

Page 14
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

A 5 pC defect in the section B-C will be detected by sensor B whereas one in section C-D is detected by sensor
D. A defect in the sections A-B or D-E are respectively detected by sensor B or D. Only two sensors at the correct
position instead of five sensors are necessary to fulfil the sensitivity requirement. A defect at position C should
be detected either by sensor B or sensor D (Figure 13b). The position of sensors B and D has to be validated
during the on-site sensitivity check. Due to the long distance separating both sensors, the artificial pulse injected
on one sensor cannot be detected on the second one. Nevertheless, different procedures are available to perform
the on-site sensitivity verification for this alternative arrangement of sensors.

UHF
signal
[dBm]A B C D E
Level measured Level measured for
Level measured for a defect in C a defect in C
for a defect in A
[m]
X:
Distance along GIS

Figure 13b: Simplified UHF measuring level for a defect close to location C

Procedure 1: The same sensor (built-in type expected) as those installed on the GIS is installed at the intermediate
position (position C) on one phase and used only for pulse injection. The three phases of a single-phase
encapsulated GIS can be considered similar from the design and consequently attenuation point of view. This is
based on having the identical sensor location on the compartments of the three phases and a limited variation of
length between two sensors which will not modify the attenuation significantly. It is however recommended to
equip the longest phase of the GIS section.

Procedure 2: An external UHF PD-sensor is applied at the intermediate position (position C) and used only for
pulse injection. In this case the determination of pulse generator magnitude should also be done in the laboratory
test (Step 1) for the external sensor. Consequently two voltage levels might be determined corresponding to the
injection by the internal and the external sensor.

Procedure 3: The same GIS bay sections are generally provided for different substations and only the connection
to the transformers and bushings may vary. In this case the on-site sensitivity verification can be omitted on GIS
sections assuming that the sensitivity verification has been passed with a complete set of sensors and the same
sensor locations and the same distance in between is respected for both GIS.

However, for the alternative arrangement of sensors the correct functioning of the sensors themselves and the
measurement chain can be tested at any suitable pulse generator magnitude.

It is necessary to mention that a reduced number of internal sensors (causing only one sensor to detect the PD
signal) can reduce the possibility of PD defect location by means of electrical time-of-flight measurement, since
additional external sensors cannot be applied in all GIS designs.

Page 15
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

3.3 Position of switching devices

From a general point of view, the GIS can be considered as having a coaxial wave guide geometry with its
internal HV conductor and grounded enclosure. Disconnectors and circuit breakers respect this geometry only in
the closed position, where the gap between contacts is open this causes a discontinuity which acts to prevent the
TEM mode from propagating from one side to the other. The waves propagating in the TEM mode will be
partially reflected and partially converted to TE and TM modes, thus modifying the frequency content of the
traveling wave. Therefore, the position of switching devices influences the attenuation of the UHF signals and has
to be taken into account for the on-site sensitivity verification.
Measurements performed by a PD monitoring system on a three-phase encapsulated GIS bay illustrate the
described effect. The sensor used for pulse injection is located on the upper busbar of the GIS bay and the sensor
location for signal detection is on the cable termination. The combination of switching device positions (OPEN or
CLOSE) significantly influences the resulting amplitude of the measured UHF signal (Figure 14). In this example
the measured UHF signal can vary from 58% to 100%.

detection:
injection: sensor at
sensor at Q1 Q2 Q0 Q9 cable
busbar termination
C O C C 71%
O C C C 100%
pulse
generator C O O C 65%
O C O C 84%
O O O O 58% 3-phase encapsulated 145 kV GIS

Figure 14: Influence of the switching device positions on the measured UHF signals (PRPD)

The same type of measurements have been performed in the frequency domain on a single-phase encapsulated
245 kV GIS [20]. The frequency spectrum shown in Figure 15, from 300 MHz to 1200 MHz, represents the
maximum amplitudes of the spectra. When the circuit-breaker is switched from the CLOSE to OPEN position, some
frequency resonances disappear in the upper part of the spectrum (800 to 1200 MHz), and some are dominant
in the lower frequency band (less than 800 MHz).

measured spectra test set-up

Signal
Injection

Measurement

1-phase encapsulated 245 kV GIS

Figure 15: Influence of the circuit-breaker switching position (CLOSE/OPEN) on the measured UHF signal

Page 16
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

It can be concluded that the magnitude of the measured UHF signals depends on the switching position of the
different switching devices (disconnector switches, circuit-breaker) located between two sensors. Measurements
in frequency domain show that the change in UHF signal does not correspond to a complete attenuation on the
spectrum. Instead, there is a complete modification of the propagation modes which prevents defining a unique
attenuation factor on a per-component basis. Therefore, the sensitivity verification can be made using any
reasonable position of the switching devices. It is recommended that on-site sensitivity verification should be
carried out for the most frequently used configuration in service.

3.4 Test equipment

Sensors
The same type of sensors must be used during the laboratory test (Step 1) and for the on-site sensitivity
verification (Step 2). The length and type of any cables used to connect the sensors to the PD measuring or
monitoring equipment should be the same at Step 1 as in the final configuration used for the monitoring
measurements. When preamplifiers are connected directly to the sensors, e.g. for on-site tests, reasonable and
convenient lengths of the measurement cables may be applied.
Pulse generator
As shown in [21] the rise time of the pulse calibrator can influence the resulting received UHF signal level and
spectrum shape (i.e. higher cut-off frequency). Consequently, the same type of pulse generator as used in Step 1
shall be used also for Step 2.
Measuring equipment
UHF measurements can be performed using narrow band or broadband systems. The detection level and the
minimum noise depend on the type of equipment. Step 2 of the proposed sensitivity verification should be carried
out with the same measuring system which was used at Step 1 or with a system with similar or better sensitivity.

3.5 Execution of the on-site sensitivity verification

For each of the different GIS sections the following steps should be performed and repeated for all sensor
combinations (sensor and adjacent sensor) installed on the GIS:
a. The pulse generator is connected to one UHF PD-sensor.
b. The UHF measurement device is connected to the adjacent sensor of the investigated GIS section.
c. The first measurement is made with the pulse generator switched off (in order to make a measurement of the
background noise level). This has to be done for about 1 min in order to take account of stochastic interference
signals occurring on site.
d. The second measurement is made with the pulse generator switched on and set to the voltage magnitude
determined by the CIGRE sensitivity verification Step 1, performed in the laboratory.
e. The magnitude of the measured UHF signal should be checked and documented (chapter 3.6)

It should be noted that the on-site sensitivity verification can be applied during commissioning as well as during
regular operation of a GIS.

3.6 Criteria to pass the on-site test

To pass the on-site sensitivity verification, when the artificial pulse is injected, the magnitude of the measured
UHF signal should be clearly above the noise level (PRPD display) or it can be considered as sufficient when
some resonance frequencies clearly appear above the background noise spectrum (spectrum analyzer).

Page 17
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

4 CONCLUSION
The return of experience from GIS indicates that some of the in-service failures are related to defects in the
insulation system. Many of these defects can be detected by UHF PD diagnostics. An Electra Report published in
1999 describes the two-step procedure for the sensitivity verification of the UHF system. This Technical Brochure
collects the available experience on sensitivity verification from the last 15 years and describes its practical
applications for GIS. Table 3 summarizes the established guidelines and recommendations which will help
manufacturers and users in the effective application of the UHF method for PD detection on GIS.

Table 3: Guidelines and recommendations for sensitivity verification

No Topic Guidelines and Recommendations Reference

Sensitivity Verification Step 1: Laboratory
1 Aim Determination of an artificial PD pulse magnitude equivalent to 5 pC Chapter 2.1
of apparent charge of a defined defect, which will be applied later
on-site during Step 2 of sensitivity verification on the same type of GIS.
2 Test setup - Compact test setup for single-phase GIS: Figure 1 and Figure 2 Chapter 2.2
- Complete bay for three-phase GIS: Figure 4
3 PD defect and Often a moving particle is used, e.g. with a length of 3 - 5 mm and a Chapter 2.2,
detection diameter of 1 mm. It is placed on the enclosure at a location near to the Annex 4
sensitivity sensor which is used for pulse injection: Figure 1, sensor C1. Instead of
a moving particle sometimes protrusions on the high voltage conductor,
simulated by a needle with a length of 5 - 10 mm and a tip radius of
about 0.5 mm (Figure 3), are also used. The required detection
sensitivity usually corresponds to an apparent charge of 5 pC according
to IEC 60270.
4 Sensor Internal and external sensors can be applied with suitable frequency Annex 2
response. Comparison and optimization of sensors is possible by using
different methods (GTEM cell, cone arrangement etc.)
5 PD measuring UHF signals can be detected in the time domain or in the frequency Chapter 2.3
device domain by narrow or wide band systems (Figure 5 and 6): oscilloscopes,
spectrum analysers, PD instruments, PD monitoring systems with or
without amplifiers and filters.
6 Pulse generator Any pulse generator with variable output and able to generate Chapter 2.4
artificial pulses of:
- rise time ≤ 0.5 ns
- variable magnitude (see chapter 2.5)
- pulse repetition rate: less than 100 kHz, e.g. 50 Hz or 60 Hz
7 Determination of The method presented here is not a calibration. However, the aim is to Chapter 2.5
artificial pulse find the best possible match between the real PD defect and the
magnitude artificial pulse in order to verify the sensitivity of the UHF measurement.
A comparison with the spectra of different voltage pulses can be done
by visual comparison or with the aid of statistical tools. A comparison
could also be made using PRPD pattern (Figure 9).
A tolerance of ± 20% is acceptable for the determination of the
artificial pulse magnitude.
8 Result Test report showing magnitude of artificial pulse and documentation of
test setup and PD measuring equipment

Page 18
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

Sensitivity Verification Step 2: On-site

9 Aim Verify that the installed sensors and the UHF measurement or monitoring Chapter 3.1
system has sufficient sensitivity to detect signals, equivalent to those
from a specific type of PD defect, within any compartment of the GIS
being checked. In addition and at the same time, the correct functioning
of the sensors themselves and the measurement chain is tested.
10 Principle of The pulse generator is connected to one sensor and the UHF PD Chapter 3.2
on-site sensitivity measuring system is connected to the adjacent sensors (Figure 10). If the
verification signal of the pulse generator (magnitude according to Step 1) can be
measured at the adjacent sensors, the sensitivity check is fulfilled for the
GIS sections. If this is valid for all sensor configurations it is considered
that the combination of sensor locations and measurement system is
sufficient to measure the entire GIS with a predefined sensitivity (e.g. 5
pC of a moving particle).
11 Arrangement of - Sensor location should be done in such a way that a PD defect (e.g. Chapter 3.2
sensors a moving particle showing an apparent charge of 5 pC) can be
detected in any compartment of the GIS by at least two sensors.
- Alternative method for the location of sensors realizes a reduction in
the number of sensors by limiting the overall detection sensitivity for
all GIS compartments to the required detection sensitivity of e.g. 5 pC
(PD defect detection by at least one sensor).
12 Position of The magnitude of the measured UHF signals depend on the switching Chapter 3.3
switching devices position of the different devices (disconnectors, circuit-breakers) located
between two sensors.
Therefore, the sensitivity verification should be made using a
reasonable position of the switching devices. It is recommended that on-
site sensitivity verification should be carried out for the most frequently
used configuration in service.
13 Test equipment Step 2 of the proposed sensitivity verification should be carried out with Chapter 3.4
the same measuring system which was used at Step 1 or with a system
with similar or better detection sensitivity.
The same type of sensors and the same pulse generator must be used
during the laboratory test (Step 1) and for the on-site sensitivity
verification (Step 2).
14 Criteria to pass When the artificial pulse is injected, the magnitude of the measured Chapter 3.6
the test UHF signal should be clearly above the noise level (PRPD display) or it
can be considered as sufficient when some resonance frequencies
clearly appear above the background noise spectrum (spectrum
analyzer).
15 Result Test report showing verification results and documentation of PD
measuring equipment.

Page 19
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

REFERENCES
[1] CIGRE Joint Task Force 15/33.03.05: Partial Discharge Detection System for GIS: Sensitivity Verification
for the UHF Method and the Acoustic Method. ELECTRA, No. 183, pp. 75 - 87, 1999
[2] CIGRE Working Group D1.03 (Task Force 09): Risk Assessment on Defects in GIS Based on PD Diagnostics.
CIGRE Technical Brochure No. 525, 2013
[3] Hampton, Meats, Pryor, Watson-Jones: The Application of Partial Discharge Measurements to GIS.
International Symposium on Gas Insulated Substations, Toronto, Canada, pp. 313 - 321, 1985
[4] Doi, Muto, Fuji, Kamei: Frequency Spectrum of Various Partial Discharges in GIS. International Symposium
on Electrical Insulation Materials, Toyohashi, Japan, 1998
[5] Reid, Judd, Stewart, Fouracre: Frequency Distribution of RF Energy from PD Sources and its Application in
Combined RF and IEC 60270 Measurements. Conference on Electrical Insulation and Dielectric Phenomena,
Kansas City, USA, 2006
[6] Masayuki, Ohtsuka, Ueta, Okabe, Hoshino, Maruyama: Influence of Insulating Spacer Type on Propagation
Properties of PD-induced Electromagnetic Wave in GIS. IEEE Transactions on Dielectrics and Electrical
Insulation, Vol. 17, No. 5, 2010
[7] Masayuki, Ohtsuka, Teshima, Okabe, Kaneko: Examination of Electromagnetic Mode Propagation
Characteristics in Straight and L-Section GIS Model using FD-TD Analysis. IEEE Transactions on Dielectrics
and Electrical Insulation, Vol. 14, No. 6, 2007
[8] Okabe, Kaneko, Yoshimura, Muto, Nishida, Kamei: Partial Discharge Diagnosis Method using Electro-
magnetic Wave Mode Transformation in Gas Insulated Switchgear, IEEE Transactions on Dielectrics and
Electrical Insulation, Vol. 14, No. 3, 2007
[9] Sellars, MacGregor, Farish: Calibrating the UHF Technique of Partial Discharge Detection using a PD
Simulator. IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 2, No. 1, 1995
[10] CIGRE WG 15.03: Diagnostic Methods for GIS Insulating Systems. CIGRE Report 15/23-01, Paris, 1992
[11] Behrmann, Neuhold, Pietsch: Results of UHF measurements in a 220 kV GIS Substation during on-site
Commissioning Tests. 10th Int. Symposium on High Voltage Engineering, Montreal, Canada, 1997
[12] Riechert: Gas-insulated Switchgear, Diagnostics & Monitoring - Present & Future. International Conference
on Condition Monitoring and Diagnosis, User Oriented Workshop, Seoul, Korea, 2014
[13] Tröger, Riechert: Influence of Different Parameters on Sensitivity Verification for UHF PD Measurement.
16th International Symposium on High Voltage Engineering, Paper B-33, Cape Town, South Africa, 2009
[14] Tröger, Riechert, Burow, Tenbohlen: Sensitivity Evaluation of Different Types of PD-Sensors for UHF-PD-
Measurements. International Conference on Condition Monitoring and Diagnosis, Paper P1-49, Tokyo,
Japan, 2010
[15] Albiez, Leijon: PD-Measurement in GIS with Electric Field Sensor and Acoustic Sensor. 7th International
Symposium on High Voltage Engineering, Dresden, Germany, 1991
[16] Riechert, Linn, Winkler, Pietsch: Reasonable Application of UHF-Partial Discharge Measurements in
Development, Production and Service of Gas Insulated Switchgear (GIS). CIGRE SC 15 Symposium “Gas
Insulated Systems”, Dubai, United Arabic Emirates, 2001
[17] Hoek, Riechert, Strehl, Tenbohlen, Feser: A New Procedure for Partial Discharge Location in Gas-insulated
Switchgear in Frequency Domain. 14th International Symposium on High Voltage Engineering, Paper G-
005, Beijing, China, 2005
[18] Harscoet, Taillebois, Prieur, Girodet: Application of the UHF Method for Partial Discharge Measurement
to Gas Insulated Substation Monitoring. CIGRE Report 15-303, Paris, France, 2000
[19] Schichler, Reuter, Gorablenkow: Partial Discharge Diagnostics on GIS using UHF and Acoustic Method. 16th
International Symposium on High Voltage Engineering, Paper D-9, Cape Town, South Africa, 2009
[20] Girodet, Fifi, Gautschi, Luna, Lebreton: Improvement of Defect Detection by Measurement of the UHF
Signal Transmission inside Single Phase and Three Phase Encapsulated GIS and Power Transformers. CIGRE
Report D1-308, Paris, France, 2012

Page 20
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

[21] Okabe, Ueta, Kaneko, Ito, Nishida, Kamei: A New Verification Method of the UHF PD Detection Technique.
16th International Symposium on High Voltage Engineering, Cape Town, South Africa, 2009
[22] Reid, Judd: Ultra-wide Bandwidth Measurement of Partial Discharge Current Pulses in SF6. Journal of
Physics D: Applied Physics, Vol. 45, No. 16, 2012
[23] Hoeck, Riechert, Strehl, Feser, Tenbohlen: New Procedures for Partial Discharge Localization in Gas-
Insulated Switchgears in Frequency and Time Domain. 15th International Symposium on High Voltage
Engineering, Ljubljana, Slovenia, 2007
[24] Kaneko, Okabe, Yoshimura, Muto, Nishida, Kamei: Partial Discharge Diagnosis Method Using
Electromagnetic Wave Mode Transformation in Actual GIS Structure. IEEE Transactions on Dielectrics and
Electrical Insulation, Vol. 15, No. 5, 2008
[25] Park, Goo, Yoon, Hong, Kang: Measurement of Ultra-high Frequency (UHF) Partial Discharge Sensor
Sensitivity and Partial Discharge (PD) Signal Losses in the 800 kV Gas-insulated Substation (GIS). 13th
International Symposium on High Voltage Engineering, Delft, Netherlands, 2003
[26] Kurrer, Feser: Attenuation Measurements of Ultra-High-Frequency Partial Discharge Signals in
Gasinsulated Substations. 10th Int. Symposium on High Voltage Engineering, Montreal, Canada, 1997
[27] Neuhold: Abnahme- und Diagnoseprüfungen von GIS vor Ort - Essenzen und Trends. ETG/FKH-Fachtagung
„Trends bei Hochspannungs-Schaltanlagen“, Baden, Switzerland, 2013
[28] Okubo, Yoshida, Takahashi, Hoshino, Hikita, Miyazaki: Partial Discharge Measurement in a Long Distance
SF6 Gas Insulated Transmission Line (GIL). IEEE Transactions on Power Delivery, Vol. 13, No. 3, 1998
[29] Schoeffner, Boeck, Graf, Diessner: Attenuation of UHF-signals in GIL. 12th International Symposium on High
Voltage Engineering, Bangalore, India, 2001
[30] Riechert, Tröger, Schraudolph, Bräunlich, Neuhold: PD Diagnostics of Gas-insulated Switchgear - Sensitivity
Verification. ETG-Fachbericht 119 “Diagnostik elektrischer Betriebsmittel“, pp. 477 - 482, VDE Verlag,
2009
[31] Hanai: Relation between the Conventional PD Measurement and the UHF Measurement in GIS. CIGRE SC
D1 Session, Contribution to Question 2.13, Paris, 2008
[32] Endo, Hama, Matsumoto, Hironaka: Innovation of GIS Insulation Monitoring Techniques and Application to
Remote Monitoring System. CIGRE Report 15-103, Paris, France, 2002
[33] Putro, Nishigouchi, Khayam, Suwarno, Kozako, Hikita, Urano, Min: Influence of Spacer Aperture Size on
PD-induced Electromagnetic Wave measured with UHF External Sensor in 66 kV GIS Model. International
Conference on Condition Monitoring and Diagnosis, Bali, Indonesia, 2012
[34] Gautschi, Bertholet: Calibration of UHF Sensors for GIS: Comparison of different Methods and Testing of
a Calibration System based on a Conical Antenna. International Conference on High Voltage Engineering
and Application, New Orleans, USA, 2010
[35] Lopez-Roldan, Blundell, Irwin, Charlson: Partial Discharge Diagnostics for Mixed-Technology Switchgear
(MTS) in Outdoor Substations. IEEE Electrical Insulation Magazine. Vol. 29, No. 3, 2013
[36] Lopez-Roldan, Blundell, Allan, Scott, Saha: Insulation Monitoring of Hybrid Switchgear. CIGRE Report A3-
201, Paris, France, 2008
[37] Neuhold, Heizmann, Bräunlich, Koechli, Riechert, Dehne: Experiences with UHF PD Detection in GIS using
External Capacitive Sensors on Windows and Disk-insulators. 15th International Symposium on High
Voltage Engineering, Paper T7-480, Ljubljana, Slovenia, 2007
[38] Albiez: Teilentladungsmessung an SF6-isolierten Schaltanlagen, PhD Thesis, ETH Zurich, Switzerland, 1992
[39] Neumann, Krampe, Feger, Feser, Knapp, Breuer, Rees: PD Measurements on GIS of Different Designs by
Non-conventional UHF Sensors. CIGRE Report 15-305, Paris, France, 2000
[40] CIGRE Joint Working Group 33/23.12: Insulation Co-Ordination of GIS: Return of Experience, On Site
Tests and Diagnostic Techniques. ELECTRA, No. 176, pp. 66 - 97, 1998

Page 21
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

Annex 1: Fundamental PD signal propagation characteristics

Partial discharges (PD) in GIS are caused by defects of the insulating system. The resulting discharge currents
have rise-times which are known to be much less than one hundred picoseconds [22]. The defects, e.g. moving
particles, protrusions, floating components etc. cause electromagnetic transients whose frequency spectra exceed
2 GHz. The resulting signals propagate within the coaxial structure of a GIS not only in the basic mode (TEM00)
but also in many higher order modes (TEmn, TMmn). The higher order modes propagate only above their cut-off
frequencies (fc). In Figure A1.1, the cut-off frequencies of the first waveguide modes are shown for three different
compartment diameters, respectively different types of GIS [23].
Reflections occur at the numerous discontinuities in the arrangement and lead to the formation of multiple standing
waves of varying frequencies. In addition, there are coupling effects between the modes which also influence the
field patterns. Due to the finite conductivity of the metallic conductors, coatings, the skin effect, dielectric losses
(e.g. spacers) and other effects, the propagating signals undergo both loss in amplitude and dispersion. The total
reduction in signal strength and signal distortion (from the original fast pulse) is highly frequency dependent and
occurs mainly at the numerous boundaries and discontinuities present in a typical GIS. The result is a complex
resonance pattern of electromagnetic waves within each compartment. Each compartment can be thought of as
a complex composite of high frequency resonator, high-Q filter function (including deep notch filtering), and non-
ideal impedance. A real GIS is composed of many of these units chained together.

Figure A1.1: Cut-off frequencies (fc) within a GIS for 300 kV, 362 kV and 550 kV [23]

Calculations performed on L-shape and T-shape GIS sections show, that TE mode can be preserved or changed
into TEM mode, or that discontinuities such as shields or insulating supports also change the distribution of the TE
mode into a TE and TEM mode [24].
Based on measurements, some values of attenuation have been estimated [25, 26]. They can vary according to
the frequency and depends on the GIS design. Two examples with different signal damping behaviour are shown
in Figures A1.2 and A1.3.
In Figure A1.2 an example of the frequency dependent attenuation characteristics along the busbar of a single-
phase encapsulated 220 kV GIS is shown [27]. The busbar of this type of GIS and this configuration shows quite
low signal damping. The pulse generator signal used for carrying out the on-site sensitivity verification can even
be identified at the sensor 14 bays further away (at 495 MHz). It can be seen that the signal-to-noise ratio is
higher for the frequencies below 1 GHz compared to the frequencies above 1 GHz. Furthermore, with increasing
distance from the artificial pulse signal injection point, the frequency content tends to concentrate on specific
resonance frequencies with decreasing bandwidth.

Page 22
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

Figure A1.2: UHF signal attenuation characteristics along 220 kV GIS busbar [27]

In Figure A1.3 an example of a strong damping effect in a 220 kV single-phase encapsulated GIS disconnector
module is shown [11]. Although the disconnector module is below 1 m in length, the frequency content of a UHF
signal passing through this element decreases significantly. This kind of abrupt low-pass filtering for discrete
frequencies of the signal spectra can also be observed at GIS configurations where the enclosure undergoes a
significant change in diameter.
The signal attenuation differs according to the propagation mode (TEM, TE and TM) and is frequency dependent
[28, 29]. The lower attenuation measured on straight GIS portions compared to other components can be
explained by the reduced number of reflections and dispersion of the signal. Due to the complex PD signal
propagation characteristics it is not possible just to sum up attenuation of individual GIS modules and arrive at
an accurate estimate of the attenuation between two sensors in a GIS configuration. Furthermore, since the exact
location of a possible PD defect is unknown, technically demanding calculations of propagation path coefficients
would need to be done for every possible combination of PD defect and sensor location and is therefore
technically not feasible with today’s available technologies.

From a general point of view and based on return of experience, the distance between two sensors typically
falls in the range of a few meters to approx. 20 m. This length can be extended up to more than 100 m in the
case of GIL or GIB.

Page 23
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

Figure A1.3: UHF signal attenuation of a disconnector switch module [11]

Page 24
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

ANNEX 2: Sensors
A) Typical sensors used for UHF PD measurement
In most types of GIS, the UHF energy is concentrated between 100 MHz and 2 GHz. The sensor's frequency
response depends on its size, shape and the connection method used. Most sensors are themselves resonant
structures at UHF frequencies and this can be used to advantage. Typical sensors are shown in Figure A2.1.

Figure A2.1: Examples of sensors [30]

Internal sensors are normally mounted at a recess in the enclosure. In this region, the radial component of electric
field is predominant. The intensity of this electric field is therefore the primary factor affecting the signal level
that can be obtained from the sensor. Internal sensors must be fitted to the GIS during manufacturing or retrofitted
during maintenance, because degassing of the GIS chambers is necessary. These sensors often take the form of
a metal disc insulated from the GIS enclosure by a dielectric material. The measurement connection is made by
a coaxial connector that is usually connected to the centre of the disc. According to the geometric requirements
of switchgear, the appropriate sensor type has to be chosen (Figure A2.2).

Figure A2.2: Types of internal sensors [31]

Externally mounted sensors (e.g. at an inspection window or barrier insulator) will be affected by the field
patterns in the structure on which they are mounted. In this case, the mounting arrangement should be considered

Page 25
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

as part of the sensor. External sensors are fitted to an aperture in the chamber wall, such as an inspection window
or exposed barrier edge. These sensors are suitable for installation on vintage GIS that cannot be retrofitted
with internal sensors or for periodic PD testing of GIS for which a permanently installed sensor is not needed.
The UHF signal is attenuated by impedance discontinuities at the surfaces of the barrier and window materials,
and usually the gap between the chamber flanges or the port on which the window is mounted acts as a high-
pass filter. However, unless the window or barrier on which it is mounted is too small, external sensors can still
reach sufficient sensitivities. Because they are more exposed than internal sensors to interference signals, shielding
of the window or barrier might be necessary.
The diameter of a window sensor, the thickness of the glass and the window recess are factors that influence
the sensitivity. The sensitivity of a barrier sensor with respect to the dimensions of the insulator aperture is
displayed in Figure A2.3. Up to a certain point, increasing the length of the insulator aperture gives no
significant increase of the UHF signal. It should be noted that the sensitivity of an internal sensor in general
cannot be achieved with a window or barrier sensor.

Figure A2.3: Electromagnetic wave (EMW) intensity of a barrier sensor as function of the length of insulator
aperture [33]

In Figure A2.4 a horn antenna type is shown mounted between two metallic flanges [32]. The bottom of the horn
antenna, where the antenna is connected with the enclosure, builds a polarization plane which allows
detection of signals from a PD event. Changing the axis direction of the horn antenna by turning, also changes
the sensitivity. A maximum sensitivity is achieved if the polarization plane lies on the x-axis of the enclosure
[32].

Figure A2.4: Installation of a horn antenna between flanges and PD signals detected
at different orientations of the polarization plane [32]

B) Sensor characterization and comparison

Page 26
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

The sensor performance and its frequency response can be determined using different methods:
1. Sensor installed in the switchgear: by the use of the CIGRE sensitivity verification procedure.
2. Sensor installed inside a GIS component: by using special conical injection and termination elements (tapers).
3. Sensor installed on a stand-alone, specific calibration cell.

Each of the above characterization methods has advantages and shortcomings. The approach 1 is the most
favourable method because it takes into account the layout of the switchgear and the surrounding of the sensor.
Unfortunately in special cases this approach cannot be used. This can happen for example when vintage GIS are
retrofitted with UHF sensors or if third party sensors (e.g. external sensors) are used with unknown characteristics.
In these cases it might not be possible to repeat the CIGRE sensitivity check Step 1 which means that the artificial
pulse magnitude might not be determined. It is therefore needed that the sensor performance is characterised in
another way.
In Figure A2.5 an example arrangement according to approach 2 is presented [37]. By the comparison of the
frequency responses of the known sensor (U1) and the external sensors (sensor I1 or I2), the characteristic of the
sensor I1 or I2 can be determined. The same approach can be used by replacing the internal sensor U1 with a
different type of sensor with unknown characteristic. Via the comparison of the different spectra or the measuring
results it is possible to characterise the unknown sensor regarding its sensitivity. This can be done using narrow
band or wideband measuring equipment.

Figure A2.5: Double cone arrangement for comparison of internal and external sensors
(I1, I2: external sensors applied on casting apertures, U1: internal sensor) [37]

To further reduce the complexity of the test setup the comparison of different sensors can be done as well in
dedicated calibration cells (approach 3, [34]). Figure A2.6 shows two possible calibration cells. According to
[34] the type of calibration cell has to be carefully chosen.

Figure A2.6: GTEM calibration cell (left) and conical calibration cell with installed UHF sensor (right) [34]

All of the presented approaches can be used to optimise and to compare sensors. While the approaches 1 and
2 takes the real surrounding of the switchgear into account they are the most demanding in terms of complexity

Page 27
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

because a fully equipped GIS or GIS component is needed. Approach 3 is less complex but does not take the
surrounding of the GIS into account.

If the sensitivity check Step 1 has been performed with a sensor of well-known characteristic then the same
artificial pulse can be applied for comparison measurements between a new sensor and the sensor with known
characteristic. The comparison measurements can be made with GIS components or with dedicated calibration
cells.

Page 28
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

ANNEX 3: Distance between sensors

The influence of the distance between sensors was investigated using real GIS compartments. In the described
experiments the distance between the sensor for the voltage injection and the sensor for the measurement was
as short as possible. The sensors are placed in adjacent GIS compartments. However, in GIS installations the
sensors are placed throughout the GIS according to rules based on experience and the topology of each
individual GIS installation. The distance between the sensors could reach more than 20 m.
Tests were performed on different test set-ups for the sensitivity verification Step 1 in order to determine the
influence of the test set-up on the resulting artificial pulse magnitude. Figure A3.1 shows the different test set-ups
[14].
In test set-up A, a disconnector module (DES) and a fast-acting earthing switch module (FAES) were installed
between the two UHF sensors. In test set-up B the sensors were directly installed in adjacent compartments. Sensor
1 was always used for the injection of voltage pulses. The UHF PD measurement system was always connected
to sensor 2.

Figure A3.1: Test set-up A (top) and B (bottom) for sensitivity verification Step 1 [14]

Table A3 shows a comparison of the results for both test set-ups based on a 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 [14].

Table A3: Equivalent artificial pulse magnitude using different test setups for sensitivity verification Step 1

Test setup A Test setup B

artificial pulse
10 V 10 V
magnitude

Page 29
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

ANNEX 4: Type of PD defect

Insulation imperfections in GIS may cause dielectric breakdown during commissioning or in service. Many of these
insulation imperfections produce PD activity before a complete breakdown finally occurs. Therefore, it has been
a goal for decades to detect these PD signals [2]. As a result, PD measurements are an important tool for testing
the dielectric integrity of a GIS during both the design and production stages. The following paragraphs describe
some of the typical insulation imperfections encountered in GIS along with their probability of occurrence (Figure
A4.1).

Remark: If a detection sensitivity for critical defects like particles on insulation or protrusions on the HV conductor
is required (to detect critical defects sensitive to lightning impulse voltages or to very fast transient
voltages from switching operations [40]) the sensitivity verification Step 1 has to be done with the critical
defect size of this specific defect type. As a consequence, the derived pulse magnitude for the specific
defect type has to be applied for the on-site sensitivity verification (Step 2).

Figure A4.1: Typical defects in GIS [16]

A) Moving Particles
While every attempt is made to prevent contamination during manufacture and assembly, sometimes tiny particles
find their way into the GIS enclosure. Conducting particles located on the inner surface of a GIS enclosure are
charged by the electric field of the applied high voltage. The induced charge and the electric field result in a
force on the particle. At the moment this exceeds the gravitational force it is termed the lifting force since the
particle is lifted off from the enclosure surface. The distance the particle flies depends on its size (i.e. length,
width, diameter, mass), the charge induced on it, and the frequency, amplitude, polarity and phase of the applied
voltage. These parameters result in a complex array of forces being exerted on the particle. It may simply
levitate on the surface of the enclosure or, more often, it will begin moving up and down.
Depending on the local electrical field distribution a particle may remain hopping in one location or it might move
around. It may reach a low field region, where gravitational forces are greater than electrical forces, with the
result that it will stop hopping. It may reach a high stress region, and if conditions permit, it may actually reach
the centre conductor. In this case particles can initiate a gas breakdown. Sometimes when this occurs, the energy
of the flashover destroys the particle (which is usually very small). As the particle bounces (or ‚hops‘), small sparks
between the particle and the electrodes (enclosure and high-voltage conductor) occur. The electrical discharges
and the mechanical impact (as the particle hops) generate electric and acoustic signals that can be detected using
various methods. Extensive tests are performed in the factory and during commissioning so that particles rarely
ever lead to in-service faults.

Page 30
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

B) Particle on insulation
Particles lying on the surface of insulators produce PD signals of low magnitude (both electrically and acoustically)
but are critical to transient voltages. Since particles on insulators are difficult to detect, the number of horizontal
spacers in GIS is kept to a minimum to lower the probability of particles resting on insulator surfaces.

C) Protrusions
A protrusion might cause a local electric field enhancement. If the field stress on the protrusion tip exceeds a
critical value a corona discharge starts. Note that the corona inception does not necessarily cause an immediate
gas breakdown due to the effect of corona stabilization (sharp protrusion). The discharge produces both a
displacement current and electromagnetic waves, which can be detected by electrical PD measurement
techniques. Further details of the theory of this PD activity are reported in [2, 3, 10]. As with particles, an acoustic
shock wave is produced as a result of local heating of the gas near the tip of the protrusion. Fixed protrusions
are rare in GIS due to both the specific design and the proprietary fabrication methods used. In addition, they
can be detected easily during routine testing in the factory.

D) Floating Electrode
A conducting object which is not galvanically connected to either the enclosure or the inner conductor will acquire
charge and an undetermined floating potential. As the AC potential changes, the object will charge and discharge
accordingly. The potential difference between the floating component and the adjacent conductor may be
bridged by tiny sparks. Although the PD signals produced are often of high amplitude, such floating defects are
not necessarily harmful. In rare cases this type of discharge process will cause surface charges to accumulate on
an adjacent insulator, resulting in a field inhomogeneity and higher probability for surface breakdowns. In other
cases such defects result in erosion of material. One type of floating defect is formed when metallic components
inside the GIS with no (or poor) electrical contact to the electrodes, e.g. loose shields, cause emissions. Sometimes
the charge/discharge cycle of these defects is modulated by mechanical vibrations in the GIS i.e. at twice the
power frequency, as the object makes and breaks electrical contact. The probability of having floating parts in
a GIS depends on design and assembly procedures. With correct designs and assembly, these faults are rare.

E) Voids
Voids or cracks in spacers or delamination at the surface boundaries of cast-in electrodes are usually filled with
a low-pressure gas mixture. Depending on geometry and / or surface roughness, a local field enhancement may
occur which reaches the critical value for the gas. If free electrons are present a discharge process is initiated
causing a displacement current between the insulator terminals or sparking between electrode and insulator. Both
processes can be generally detected via electrical PD measurement. Discharge processes in epoxy spacers may
also lead to treeing and eventual breakdown. The probability of getting ‘dangerous’ voids is influenced by the
shape of the spacer (or bushing), the material, the production process and the field stress level. Voids and
delaminations can be virtually eliminated by appropriate design of these components and optimized production
processes. Factory PD testing filters out spacers containing voids and so virtually eliminates their appearing in the
field.

Further information on critical PD defects and necessary detection sensitivity are described in [2, 40]. Table A4
shows the relation between artificial voltage pulse magnitudes from a pulse generator to the magnitude of a
protrusion and a moving particle showing an apparent charge of 5 pC. The artificial pulse magnitudes differ
from 2 V for a protrusion to 10 V for a moving particle. Such a difference may have an impact on the number
of sensors at the GIS. However, nowadays it is common practice to focus on moving particles as PD defect for
sensitivity verification.

Page 31
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

Table A4: Artificial voltage pulse magnitudes in reference to magnitudes of PD defects

showing 5 pC apparent charge during sensitivity verification Step 1 using a peak detection system

Artificial pulse Moving

Protrusion
magnitude particle
1V 86.6% 72.8%
2V 96.7% 80.4%
5V 110.0% 92.4%
10 V --- 100.7%

Page 32
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

ANNEX 5: Mixed technology switchgear (MTS)

The term “Mixed Technology” refers to a combination of both traditional air insulated switchgear (AIS) and
newer GIS technologies. The MTS module is a self-contained switching module suitable for use in outdoor
substations. It utilises existing proven GIS components and uses a conventional air insulated busbar to connect the
various modules [35]. The hybrid module itself is an assembly of a GIS type circuit-breaker, disconnector switch,
earthing switch and instrument transformers with SF6/air bushings at either end to permit connection to an air-
insulated bus on one side and a circuit e.g. overhead line or transformer, on the other side. In essence, the
assembly or module, forms a complete switching bay.

Figure A5.1: (a) PD Diagnostic equipment installed in a car; (b) UHF method applied
to 132 kV MTS module; (c, d) external window sensor placed on the inspection window [36].

In many MTS module installations there is just one point to apply an external window sensor (Figure A5.1).
Therefore, the on-site sensitivity verification as used for UHF PD detection in GIS cannot be applied. Experiences
show that a single sensor per 132 kV MTS module is sufficient to cover the complete unit and the application of
additional sensors does not provide any gain in terms of increase in the sensitivity of the UHF PD detection method
[36]. The required detection sensitivity of e.g. 5 pC for the PD measurement system is assumed by the small
dimensions of such MTS modules and can be proven during special agreed laboratory tests with an additional
sensor [36].

Note: Noise rejection measures have to be applied if the background noise level hampers sensitive PD
measurements.

Page 33
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

ANNEX 6: Vintage GIS

Although UHF PD detection is a common on-site insulation diagnostic technique for GIS today, elder GIS are
often equipped with sensors for which Step 1 of the sensitivity verification was not done in the laboratory. In
some cases, a GIS internal shielding of e.g. an earthing switch shows enough bandwidth for a sensitive UHF PD
measurement, but with unknown pulse generator magnitude [30]. Some elder GIS do not feature internal sensors
but have some dielectric apertures in the enclosure large enough for the application of external sensors [37, 39].
Trying to perform an UHF PD measurement on such kind of installation, the question of detection sensitivity and
sensor frequency response arises. The following two examples are given for the evaluation of the sensitivity and
an example for the determination of the frequency-characteristics of a non-conventional sensor.

A) Sensitivity check by using the capacitance of sensors

In the early period (up to 1995) of the application of VHF and UHF PD measurement in a moderate frequency
range around 100 MHz the detection sensitivity of capacitive sensors were verified by a method related to the
PD calibration according to IEC 60270 [38]. A defined charge Q was transferred by the sensor capacitance C1
(capacitance of the sensor electrode to the HV conductor) of one sensor in order to simulate a PD pulse and to
verify the sensitivity of the other sensors [37, 38]. The transferred charge Q was produced by a defined step
voltage U, applied to the exciting sensor capacitance C1 (Q = C1 . U).
In Figure A6.1 the relation by how much the amplitude of a reference step voltage determined by the CIGRÉ
proposal differs from the initial defined charge injection to the HV conductor (Uc = Qc/ C1) is shown [30].

Figure A6.1: Comparison of pulse generator magnitudes -

CIGRE sensitivity verification Step 1 versus calculation from C1 for different sensor capacitances [30]

It can be concluded that for sensor capacitances C1 around 1 pF a correlation exists between the artificial pulse
magnitude determined by sensitivity verification Step 1 and the determination on the basis of the sensor
capacitance C1. However, considerable deviations could be observed for other values. When trying to achieve
a simple estimation of the detection sensitivity of internal sensors on vintage GIS, the measurement of the sensor
capacity C1 could give useful information under specific circumstances [30].

B) Sensitivity check by using the GIS apertures and external sensors

For GIS not equipped with internal sensors, different types of external sensors have been designed [37]. The
sensitivity and the usable bandwidth are strongly dependent on the type of sensor and the kind and size of
apertures in the GIS. Examples of apertures at vintage GIS are shown in Figure A6.2.

Page 34
 UHF Partial Discharge Detection System for GIS: Application Guide for Sensitivity Verification

Figure A6.2: Examples of GIS apertures like inspection window (left) and casting aperture of
insulating material at metallic barrier flange (right) [37]

In the following, an example of sensitivity verification with external sensors at a vintage GIS installation is shown.
The sensitivity verification Step 1 in the laboratory was carried out using a real 5 pC PD defect (needle on the
HV electrode). These tests were carried out in a GIS test setup according to Figure A6.3 with original spare parts
of the vintage GIS to be investigated [37]. The use of the same type of GIS parts is important since differences
in the internal geometry influence the frequency spectrum of the signal and therefore the outcome of the
measurements. Two external sensors were applied to casting apertures of the GIS.

Figure A6.3: Sensitivity verification laboratory test setup with spare parts from vintage GIS
(left side: measuring external sensor; right side: external sensor used for pulse injection) [37]

At the laboratory test, a pulse generator magnitude of 6.5 V was determined to be equivalent for a 5 pC needle
defect. It is important to prevent any unwanted external signals coupling in between the external PD sensors and
the flange. It is often necessary to employ appropriate electromagnetic sealing between the GIS enclosure and
the external sensor, e.g. specialized EMC gasketing material composed of conductive elastomer material or wire
mesh. A measurement to check the sealing effectiveness is recommended.

Page 35