CIGRE-633 2023 CIGRE Canada Conference & Exhibition

Vancouver, BC, Sept.25 – 28, 2023

UHF PD Monitoring Data Enables Condition-Based Maintenance of the GIS

BHARAT NANDULA
QUALITROL DMS
Canada

SUMMARY

The condition assessment of high voltage gas insulated switchgear (GIS) has become
increasingly important in recent years, as economic factors drive users to extend the service
life of equipment and optimise maintenance practices to prevent unexpected outages. This is
especially true in the case of aging GIS, where defects in the insulation system can lead to in-
service failures.
To address these issues, improved quality assurance measures must be implemented during
manufacturing and on-site assembly, and more efficient diagnostic checks and testing
procedures must be adopted. UHF partial discharge measurements are a well-established
technique for assessing the condition of GIS by measuring the electromagnetic signals
generated by partial discharges. This provides the capability to detect incipient defects at an
early stage, which can aid maintenance planning and avoid potential breakdowns during
commissioning and operation.
UHF partial discharge diagnosis plays a vital role in providing condition assessments of GIS
during commissioning and operation, supporting decision-makers in making informed
choices. Continuous monitoring or periodic PD measurements can provide valuable
information on the condition of the GIS, and implementation of these methods, addressing the
reliability of PD diagnosis results and their influence on a conditional assessment and risk
assessment of substation equipment, is the ultimate goal.
To achieve these objectives, a system of approach must be established to obtain sensitive PD
measurements to detect critical defects. Once a PD signal is detected, steps must be taken to
obtain information about the type and location of the defect inside the GIS, together with
other factors such as laboratory measurements, manufacturer experience, design aspects, and
the trend of the PD activity. This information can help estimate the failure probability and
support risk assessment.
The implementation of PD diagnosis systems and operational procedures is evolving in many
organisations and asset owners, and in some cases, not fully developed. The data provided by
diagnostic and monitoring equipment can be difficult to interpret for operators and
maintenance crews, requiring experts to assess it. Moreover, condition monitoring is
sometimes implemented without defining supporting structures, leading to inefficiencies.
Therefore, it is essential to establish logical and simplified assessment procedures and
“monitoring-friendly” organisational structures to facilitate successful implementation of
condition-based maintenance and monitoring. This paper aims to provide guidance and
strategies, with clear stages involved, to obtain sensitive partial discharge data to support
users during risk assessment. Furthermore, it provides practical examples of detected partial

[email protected]
discharges from actual on-site experience, emphasising the importance of how to handle
detected PD to prevent failures and optimise maintenance practices.
One crucial aspect of successful PD diagnosis and monitoring is the use of appropriate
measurement equipment and techniques. For instance, the sensitivity of the measuring
equipment must be adequate to detect PD signals reliably. The electromagnetic signals
generated by partial discharges are typically in the ultra-high-frequency (UHF) range,
specialised UHF sensors and measurement techniques are required. The use of broadband
sensors can be beneficial in detecting PD.
Another critical factor is the interpretation of PD data, which requires specialised knowledge
and experience. The recorded data must be analysed to determine the type and location of
defects, the trend of PD activity, and other critical factors. This information can aid in
estimating the probability of failure and help prioritise maintenance activities.
To facilitate the interpretation of PD data, organisational structures must be established,
including the definition of roles and responsibilities of the personnel involved in PD diagnosis
and monitoring. The use of standardised PD measurement procedures and protocols can also
aid in the interpretation of data and support consistency in decision-making.

KEYWORDS

GIS-Gas Insulated Switchgear

UHF- Ultra High Frequency
PD- Partial discharge
EMI- Electromagnetic interference

[email protected]
The UHF PD Diagnostic Technique

Dielectric breakdowns in service, caused by various types of discharges such as free particles,
protrusions, insulation defects, and similar conditions in a GIS, are typically preceded by partial
discharges for many months before sufficient damage occurs to cause complete failure. In some cases,
failure may be imminent. The size of the partial discharge itself does not indicate the risk of failure,
and the cause must be identified before an accurate prognosis can be made. Therefore, the detection of
partial discharges (PD) using the UHF technique is becoming increasingly important in gas-insulated
substations (GIS) because manufacturers and utilities recognize its suitability as a sensitive method for
monitoring insulation integrity.

The discharge always occurs across very small gaps in high-pressure SF6, resulting in extremely rapid
breakdown times, typically in the sub-nanosecond range. The fast-rising PD pulse contains frequencies
that extend well above 1 GHz, exciting various modes of resonance in the GIS chambers. These
resonances have low amplitudes and can be detected by couplers built into the GIS or external sensors
placed at dielectric openings on the GIS. These openings provide sensitive means of detecting and
identifying the PD. It is worth noting that electromagnetic energy emitted from the PD signal is
directly coupled into the GIS chamber, causing it to resonate. These electromagnetic signals produced
by PD are in the UHF frequency range, specifically between 300 MHz and 3000 MHz
The UHF detection of PD in GIS involves three main stages, as depicted in Figure 2: a) excitation of
the UHF signal by a PD, b) propagation of UHF signals within the GIS, and c) capturing the UHF
signals from the GIS using a coupler. This technique utilizes a coupling sensor to extract the UHF
signals that are excited within the GIS by discharges of PD current occurring at a defect location. UHF
PD sensors can be internal to the GIS and installed during the manufacturing process. These internal
sensors are mechanically robust, with proven records of service reliability. In GIS units without
internal UHF PD couplers/sensors, it is often possible to retrofit specially designed external sensors at
dielectric exposed barriers, apertures on the barrier, and inspection windows. While internal sensors
are less susceptible to electromagnetic interferences and very sensitive, external sensors are easier to
retrofit. To mitigate electromagnetic interferences, EMI shielding belts can be utilized.
Overall, the UHF PD diagnostic technique is a valuable method for monitoring the insulation integrity
of GIS and detecting early signs of partial discharges. By identifying and addressing these issues in a
timely manner, the risk of equipment failure can be minimized, ensuring the reliable operation of gas-
insulated substations.

a) b)

Figure 1: a) Internal type PD sensor, b) External type barrier coupler

Ensuring the frequency response of the sensors is verified before their installation on the Gas Insulated
Switchgear (GIS) is essential to demonstrate their optimal sensitivity tailored to the GIS requirements.

3
The frequency response of the sensors determines their ability to detect and capture Ultra High-
Frequency (UHF) signals generated by partial discharges accurately. Since PD signals typically
operate within the UHF range, specialized UHF sensors are required to capture these electromagnetic
emissions effectively.
Before deploying the sensors on the GIS, it is crucial to subject them to rigorous testing and
verification. This process involves analyzing their response to different frequencies, especially within
the UHF range. By conducting thorough frequency response testing, any potential limitations or
deviations in the sensors' sensitivity can be identified and addressed, ensuring they are optimized for
detecting PD signals specific to the GIS's operating conditions.
Moreover, by verifying the frequency response, the sensors' performance can be aligned with the GIS's
unique characteristics, such as its size, configuration, and insulation system. This alignment ensures
that the sensors are well-suited to the GIS environment, enhancing their sensitivity to early-stage PD
detection and enabling accurate condition monitoring.
In summary, pre-installation verification of the sensors' frequency response is a critical step in
ensuring the effectiveness and reliability of the PD monitoring system. It guarantees that the sensors
are tailored to the GIS's specific UHF PD detection requirements, enabling early identification of
potential defects and supporting proactive maintenance strategies to prevent unexpected failures.

1. Implementation of UHF PD for sensitive measurements

Detection of Partial discharge in UHF has been adopted by manufacturers for routine and
onsite commissioning test and utilities for continues monitoring in Gas insulated substation
(GIS) applications because as in nature less sensitive to EMI and ease of handling compared
to conventional methods and importantly signals generated by PD can be detected by nearest
sensors regardless of the total size of GIS. UHF PD method is not suitable for measure
apparent charge like IEC 60270 due to its complex pulse propagation. However, it is possible
and proved in many cases that is sensitive to detect defects causing apparent charge of 5pC or
greater can be detected by UHF method. Cigre W.G-D 1.25 clearly stated two-step procedure
in technical brochure that ensures apparent charge of 5pC can be detected with UHF method.
The conventional method for detecting PD used coupling capacitor which is placed in parallel
to test object and discharge signals are measured across an external impedance and this
discharge expressed in apparent charge. This method has advantage as able to calibrate the
system. Yet sensitivity dependents on the ratio of the coupling capacitance to capacitance of
the test object, hence technique e is not suitable to monitor large section of a GIS. Also, this is
very prone to EMI and showed difficulties during onsite commissioning and not suitable for
online continues monitoring. In GIS standard practice and permissible level for partial
discharge level shall not exceed 5pC at the test voltage defined as per standards. Conventional
method expresses PD in apparent charge. However, PD signals measured in UHF depend on
the PD source, location of the defect, propagation effects and coupling factors and complete
measurement chain as shown in figure 2.

A two-step procedure proposed by CIGRE can be used to establish that the UHF PD
measuring system is functioning and able to detect defects within 5pC. UHF method has
proved sensitive to detect defect as the conventional method, this is due to GIS enclosure acts
as a faraday cage to screen out electromagnetic disturbances. PD produces current pulses and
consequently it produces EM signals in GIS. These pulses rise less than few hundred Pico
seconds and width in the nanoseconds. Elector magnetic emission produced by these current
pulses in GIS in the range of ultra-high frequency and these fields can be detected by electrical
sensing device which placed in GIS so called UHF couplers as described in section 1. It is an
important step to perform during implementation of the UHF measurements to prove that it is
sensitive detect PD anywhere in the GIS. Other essential factor to be considered for sensitive
PD measurements is to make sure best signal to noise ratio. PD measurements strongly depend
on the ratio of PD signal to background interferences caused by external interference. Signals
from radar or mobile telephones operating in the frequency range between 300 MHz to 2 GHz

4
 are the main interference for UHF PD measurements. In the case of broad-band PD detection
measurements the interferences signals can be tackled with filtering by band selections or
appropriate filters.

Figure 2. propagation of the signal

PD measurements would be performed on various stages of the asset life cycle. It is important
to perform the PD measurements after installation of the GIS on-site once it’s fully assembled
to detect critical defects that may occur during production, shipment, or assembly at on site. It
should be checked prior to measurements that UHF sensors and measurements system is
sensitive enough to detect the sensitivity of at least 5pC or equivalent by following two step
procedure recommended by CIGRE technical brochure 654 as described above.

PD measurements carried out on existing substation and newly installed GIS to improve the
reliability of the equipment. Usually, they perform periodic measurements or continuous on-
line monitoring to capture the PD data. Having continuous online monitoring right from the
installation of the GIS have inherent advantage that ability to assess condition better compared
to periodic checks, since PD can be intermittent. Measurements must be performed in a way
that PD intensity is not only one relevant parameter to estimate the probability of dielectric
failure.
Following factors must be considered to have sensitive PD measurements.

- The type and location of the defect

- Signal propagation characteristic which depending on GIS size and configuration.
- Sensor location and sensitivity
- Measurements chain characteristics (from sensor to data acquisition unit).
- Measurements instruments characteristics

2. Role of UHF PD in Condition monitoring and Condition based maintenance.

Gas insulated switchgear (GIS) technology is widely utilized to improve supply reliability
with minimal space compared to AIS. Although it is designed to be maintenance free, GIS do
deteriorate and can be damaged because of various operating conditions. Failure of the GIS
can occur while it is in service before the scheduled maintenance. Moreover, the replacement
of an old model GIS is complicated and requires significant effort and cost. Prime concern of
the utilities is the condition assessment and lifetime estimation of the GIS and other assets in

5
 T&D grid to guarantee the reliable electricity. Traditionally, maintenance of the GIS is
conducted according to pre-determined interval recommended by the manufacturer and
guidelines by standards known as preventive maintenance or time-based maintenance. This
type of practice is not optimal due to excess maintenance and outage cost. Hence, it has
shifted across world by many utilities shifting towards condition-based approach with
determining the actual condition of the GIS and its components to properly plan the
maintenance task. To know the actual condition of GIS components, condition monitoring and
diagnostic techniques plays key role to detect abnormal conditions. Studies and experience
show that majority of the failures are resulted from the dielectric failures. These failures can
be detected while GIS is in service using UHF PD monitoring. It is important to have logical
implementation of principals of Partial discharge alarm handling and health assessment of
GIS. As detailed in section 2, it is important stage to consider listed parameters to have
sensitive PD measurement data from either periodic measurements or online monitoring.
Condition monitoring is merely just not meant to look for a condition at a certain moment. It
would be more beneficial to have the change of a condition from one point in time to another.

Figure 3: Flow chart for the implementation of the effective CBM

However, data provided from the diagnostic and monitoring equipment are often difficult to interpret
by the operators and maintenance crews. It’s required experts to assess the data. Furthermore,
Condition monitoring most of the time is implemented in the organisation without defining proper
organisation structure in place.
CIGRE published brochure 525 (1) in 2013, outlining a structured risk assessment approach for
detected partial discharges (PD) within gas insulated switchgear (GIS). In this discussion, we will
explore an alternative approach based on logical decision chains, focusing primarily on online GIS PD
monitoring. When a PD alarm is triggered, a crucial decision must be made on how to proceed. An
example of the organisational flow chart for effective CBM implementation is illustrated in Figure 3.
The main challenge lies in the decision maker's confidence in their assessment. This confidence can be
influenced by the level of knowledge and the reliability of the monitoring system. If the user lacks
confidence, it may lead to underutilization or even disabling of the condition monitoring system to
avoid discomfort. The PDM (Partial Discharge Monitoring) alarm handling organization structure is a

6
crucial component of an effective Condition-Based Maintenance (CBM) program for high voltage gas
insulated switchgear (GIS). Properly organizing and defining roles within the PDM alarm handling
team ensures efficient response to alarms and timely decision-making to address potential issues.
Below in Figure 4 is suggested PDM alarm handling organization structure.

Figure 4: Organisation Procedure to Handle PD Alarms

The procedures outlined below are primarily focused on how to handle PD alarms generated by online
monitoring systems. However, in some cases, these procedures can also be applied to PD alarms
generated by periodic measurements.
In other words, the following steps and methods are designed with the main purpose of addressing PD
alarms from continuous online monitoring systems. However, they can still be adapted and utilized to
some extent for PD alarms detected through periodic measurements. The applicability may vary
depending on the specific context and capabilities of the monitoring systems in use. Condition
monitoring is not solely aimed at assessing the current condition of the equipment but rather at
identifying changes in condition over time. Threshold levels are employed to detect such changes or
evolving conditions. For instance, a PD alarm threshold, established based on historical data, will
indicate a shift in condition once the alarm is triggered. Changes in condition may result from various
factors, such as the issue of new Electromagnetic interference e.g., new transmission antennas or
frequencies for walkie-talkies), change in external PD sources (e.g., increased pollution on bushing

7
surfaces leading to heightened external discharge activities or external corona), or the development of
new or modified internal discharges due to incipient faults. Figure 5 illustrates the diverse potential
causes for a new PD alarm. Modern PD monitoring systems should utilize automated noise gating or
pattern recognition techniques to avoid generating unnecessary alarms or notifications. However, it is
crucial to ensure that the user is promptly informed when external PDs or disturbances are evolving,
potentially concealing, or masking internal discharges. In all three scenarios, appropriate actions must
be taken to address the identified conditions.

PD Alarm or Signal

Internal PD External PD Disturbances

Figure 5: Possible scenarios of the PD alarm

Disturbances
In the context of this discussion, disturbance refer to events not correlated with the phase of the bus
bar voltage. These events can raise from various sources, such as newly installed transmitting antennas
or device operating in the UHF range (e.g., mobile telephone antennas, radar systems, or other
transmitting applications). potentially concealing or masking internal discharges. In all three
scenarios, appropriate actions must be taken to address the identified conditions. When displayed on a
PD monitoring system with phase-related PD pattern displays, these disturbances appear as increased
background noise across the entire voltage cycle. On the other hand, certain typical pattern can be
identified in real-life PD displays using pattern recognition techniques (examples shows in figure 6).
As these patterns are typically associated with specific frequencies in the UHF Frequency domain,
once detected, they can be effectively eliminated by implementation UHF filtering tailor made to the
frequency involved.

8
Figure 6: Examples of Fluorescent faulty light signal PRPD and Singel cycle display

External PD
In this paper, external PD refers to discharges related to the phase of the bus bar voltage, appearing in
phase-resolved PD pattern displays similar to internal PD. These external discharges can arise from
various sources, such as discharges associated with SF6-filled silicone composite bushings, old-style
ignition devices in defective neon lights, or other external sparking entering the GIS through "UHF
windows," like insulation disks or windows.
Since these external discharges exhibit patterns resembling real PD (e.g., bus bar or chamber corona
resulting from discharges on the bushings), PD pattern recognition techniques may classify them as
real PD, triggering PD alarms. Example of the external PD which resembles real PD shown in Figure
7. To address this issue, two main techniques are utilized.
For single-phase encapsulated designs, modern PD systems offer the option of "coincident filtering." If
the same PD impulses appear in multiple phases, they will be gated out and not recorded by the PD
system. However, this method is not applicable to three-phase encapsulated GIS.
In the case of three-phase encapsulated GIS, where external discharges may couple in a single phase
only, "noise antennas" are employed. These antennas are placed near the discharge source to capture
external discharge impulses. When detected by an external antenna and the affected internal UHF
sensor, the discharge impulse is gated out, like coincident filtering, and not recorded by the PD
system.
Both methods have proven to be highly effective in preventing external PD from triggering false
alarms. However, it is essential to use coincident filtering and noise gating with external antennas only
when necessary. Incorrect settings can lead to the system being blinded throughout the measurement
time. If any of these settings are changed, a sensitivity verification for the related UHF PD sensors
must be repeated to ensure that the system can still detect internal discharges. If the intensity of
external discharges is too high to apply gating effectively (resulting in the inability to detect internal
discharges), other actions should be considered, such as cleaning the bushings or repairing faulty
lights, ultimately removing the cause of the external PD.

9
Figure 7: External signal, single cycle display and PRPD

Internal PD

In the event of an Internal PD alarm, three crucial questions must be answered:

1. Is the recorded pattern phase-related to the bus bar voltage?
2. Does the recorded pattern appear in the same manner (but 120 degrees shifted) in one or all
the other phases (in the case of a single-phase encapsulated GIS)?
3. Does the recorded pattern appear in the same manner at the noise antenna?
If the answer to question one is 'No,' it indicates a new disturbance source has emerged.
If the answer to question one is 'Yes,' and either question two or three is 'Yes,' it suggests a new
external discharge has occurred. The handling of both situations has been described above.
If the answer to question one is 'Yes,' and both other questions are 'No,' it indicates a new internal PD
has surfaced, which requires a risk assessment.

To assess the risk, it is crucial not to overreact but instead read the signs of urgency. The following
questions need to be answered:
4. What is the rate of change in terms of amplitude and count rate?
5. What is the type of defect?
For question four, historical data should be used to understand how long it took for the discharge to
develop to its current stage. This provides an indication of the required response time to a specific
alarm. A slow development suggests a slower reaction time (perhaps waiting until the next outage),
while a rapid development implies an urgent response. Generally, if a discharge develops over months,
it will take months to lead to failure, whereas a discharge developing over hours might cause GIS
failure within hours.
Question five can be answered today by automated pattern recognition algorithms integrated into
modern PD systems. Different types of defects carry varying levels of risk. For instance, a void type of
defect usually requires a long development time before becoming dangerous, while a particle,
continuously active after switching operations, could signal the need for a breaker overhaul. Floating
type defects generally have high discharge amplitude and may not be dangerous over an extended
period with stable discharges but can indicate a fault when amplitude and/or count rate change rapidly.
The answers to questions four and five must always be considered in their context.

10
 PRPD display

Single Cycle display

Figure 8: Provides example where proper health assessments were applied to detect particles leading
to the avoidance of breakdowns. PRPD display and Single cycle display.

After identifying the type of discharge and the rate of change, the next question is:
6. Where is the discharge located?
This question is addressed by conducting an on-site measurement known as Time-of-Flight. If a
discharge can be detected on two couplers, the time difference of arrival of the PD impulse on the two
sensors is measured using a fast oscilloscope (5 to 10GS/s) with at least 2 channels. The time
difference allows the calculation of the discharge location between the sensors or determines if the PD
source is located outside the area covered by the two sensors.
Knowing the type of discharge and its location can already provide insight into the part of the GIS
causing the PD. By adding the rate of change of amplitude and count rate, the estimated time left for
repair can be determined. For instance, a constant (by amplitude and count rate) void type discharge
has low potential to develop immediately into a fault, whereas a constant corona type discharge
(resulting from protrusion on a conductor or encapsulation) is sensitive to overvoltage caused by
breaker or disconnector operations, posing an immediate risk of a serious fault during these events.
Once an internal discharge is detected, it is of utmost importance to consider environmental
conditions. Factors such as transformer tap position changes, load variations, or temperature
fluctuations can influence the discharge behavior. Understanding the GIS's history or the same type of
GIS and determining if the discharge is triggered by a switching operation are also vital.

The use of the UHF PD monitoring for assessing the condition of the high voltage gas insulated
switchgear is not new, it is well established technique that has been successfully used in various
locations worldwide, especially in Middle east and Asia pacific and some parts of the Europe. Above
mentioned guidance and strategies based on existing best practices and practical examples of partial
discharges detected at the site and data collected from the monitoring system. It emphasises the

11
importance of systematic implementation and clear organisation structures to ensure the effectiveness
of UHF PD monitoring.

Conclusion
In conclusion, the assessment of partial discharges (PD) in high voltage gas insulated switchgear (GIS)
has become increasingly critical as users aim to extend equipment service life and optimize
maintenance practices to prevent unexpected outages. Utilizing UHF PD measurements in condition
monitoring during commissioning and operation is a well-established technique to detect incipient
defects at an early stage, enabling informed decision-making for maintenance planning and risk
assessment.
To implement successful condition-based maintenance and monitoring, it is essential to establish a
logical and simplified assessment procedure with a "monitoring-friendly" organizational structure. The
commitment of the asset owner and the involvement of the operation team are vital in ensuring the
effectiveness of the PD monitoring program.
A structured risk assessment approach, as described in CIGRE Brochure 525, should guide the
handling of PD alarms. Smart application of pattern recognition techniques, coincident filtering, and
noise gating with external antennas help prevent unnecessary alarms triggered by external
disturbances. Proper interpretation of PD data, considering historical trends and defect type, allows for
appropriate risk assessment and timely maintenance decisions.
This paper provides, strategies for sensitive PD data collection, the successful implementation of these
strategies may involve overcoming various challenges related to equipment, expertise, data
management, and environmental factors (EMI interference). Each site circumstances and GIS
configuration may influence the nature of these challenges. Therefore, a well-planned and organized
approach to implementation is crucial for reaping the benefits of UHF PD monitoring. There are
opportunities for future research to address the limitations of false alarms for e.g., external
disturbances which appears to be more like PD like signals and can contribute to more effective and
sustainable condition monitoring of the GIS through UHF PD measurements.
In summary, effective condition monitoring of GIS through UHF PD measurements requires a
systematic approach involving committed asset owners, well-defined roles, and reliable data
acquisition systems. By accurately interpreting PD alarms, distinguishing internal and external
discharges, and assessing risks based on type, location, and rate of change, organizations can optimize
maintenance practices, prevent failures, and ensure the reliable operation of gas-insulated substations.
Applying these strategies can lead to increased equipment reliability, reduced downtime, and
improved asset management overall.

BIBLIOGRAPHY

[1] CIGRE Brochure 525- Risk Assessment on Defect in GIS based on PD diagnostics.
[2] CIGRE Brochure 654- UHF Partial discharge detection system for GIS: application guide for
sensitivity verification.
[3] J.S. Pearson, B.F. Hampton, and A.G. Sellars, ‘’A continuous UHF Monitor for Gas-Insulated
Substations’’, IEEE transactions on Electrical Insulation Vol. 26 No. 3, June 1991
[4] CIGRE WG 15-03: Effects of Particles on GIS Insulation and Evaluation of Relevant Diagnostic
Tools, CIGRE, 1994, Report 15-103
[5] B. Hampton, R.J. Meats: Diagnostic Measurements at UHF in gas-insulated Substations, IEE
Proceedings, Vol. 135, No. 2, 1988, pp. 137 - 144

12