Power cables are important assets in electric utility transmission networks for the reliable operation of power systems. To prevent catastrophic failures in underground cables and terminations, it is important to know how to leverage the various types of available sensors that can detect partial discharge (PD) and locate the source so that appropriate action can be taken to avoid in-service failures. This article answers questions about how technology can help identify partial discharge in underground cables without expensive shutdown time.
Power cable systems consist of the cables themselves and their accessories — joints and terminations. In recent years, the use of high-voltage (HV) power cables has increased due to more densely populated urban areas and new power cables being installed in old networks.
New synthetic polymer materials have boosted the birth of extruded XLPE power cables, and the use of XLPE insulation in HV cables is increasing due to its advantages: low dielectric losses, suitability for high operating temperatures, and relatively easy and low-cost manufacturing.
Let’s address some key questions that arise around the topic of PD testing.
Q: How can you ensure the reliability and availability of cables that experience transients, lightning, or switching surges and fault voltages?
Electrical diagnostic tests have played a role, though this often involves expensive shutdowns and other cost implications. Most dielectric failures in HV XLPE cables are associated with defects in joints and terminations that develop over the lifetime of a cable system. To detect such changes at an early stage, detailed information on insulation condition is necessary. As an alternative to traditional diagnostic methods, this information can be achieved by PD monitoring during the operation of the equipment.
The most effective tool to detect local damage, defects, and/or localized aging processes in extruded cable systems is measuring PD. On-line PD measurements are most suitable for detecting problems in cable terminations and joints. Harmful levels of PD can be detected well before a breakdown.
With continuous measurement, reliable estimations about insulation condition can be made.
Possible applications for online PD measurements are:
- Quality assurance of installed cables
- Continuous cable monitoring
- Location of problematic joints and terminations
- Prioritizing cable replacements
Q: What are the advantages of online PD monitoring over offline PD measurement?
Using PD technology has proven to be an excellent method to identify problems in insulation. IEC 60270, High-Voltage Test Techniques — Partial Discharge Measurements, details the most adaptive and conventional method of measuring PD. However, this method is most suitable for offline diagnostics in environments with less electromagnetic interference. Since aging of the insulation of in-service HV components is ongoing, on-site PD testing and diagnosis have attracted increased interest for use in condition monitoring. Since conventional PD measuring systems used in a controlled factory environment are not generally suitable for on-site application, specialized PD detection and measurement methods have been introduced.
PD signals can be detected by unconventional PD measuring methods and systems that use the physical characteristics and properties of the PD processes. These methods have gained significant popularity as they can be applied while equipment is operating. For example, unconventional PD coupling methods based on inductive or capacitive sensors have led to increased sensitivity at the accessories (joints) compared to conventional PD detection at the cable end. One promising approach is the use of inductive PD sensors at terminations and cross-bonding the links of long HV cable systems.
Q: What type of sensors or sensing techniques are used in online PD testing?
To detect PD activity online, non-intrusive sensors must be used. For online PD detection, high-frequency current transformers (HFCTs) are used to detect current pulses from PD in the cables and switchgear. Transient earth voltage sensors (TEVs) are used to detect electromagnetic radiation from PD activity from terminations and switchgear. By using a combination of sensors, sensitivity to various types of PD can be obtained, and the measurements from different sensors can be correlated to aid in the diagnosis. A combination of HFCTs and TEV (Fig. 1) makes it possible to separate the signals that are coming from the switchgear panel, terminations, and cables.
HFCT
The most suitable sensors for underground cable PD measurement are inductive-type HFCTs. These sensors can be installed when the cables are in use (i.e., energized). Typically, HFCTs are mounted on the ground straps of the MV or HV cable at the connection boxes. To design and select appropriate HFCTs for these applications, the lower and higher cut-off frequencies, polarity, and saturation characteristics must be considered.
A lower cut-off frequency is an essential parameter for HFCTs. It’s a compromise between the ability to detect dispersed signals from long cables and tolerating noise pickup from mains. Most of the HFCTs on the market for lower cut-off frequencies operate in a range from 50 kHz to 150 kHz. For special applications where no dispersed signals need to be captured, much higher-frequency ranges can be used with the benefit of obtaining lower noise levels.
Higher cut-off frequencies — the maximum frequency at which an HFCT needs to respond — depend on the application. A high-frequency HFCT cut-off is generally 10 MHz or higher based on the application. The HFCT should be clearly marked in terms of its polarity relationship (e.g., an arrow pointing toward the grounding point of the cable or earthing strap). The orientation doesn’t seem to matter, but if more than one HFCT is used in a test and the polarity of the signals needs to be compared, a clear and common definition is needed.
TEV
Another type of sensor widely used in medium-voltage (MV) switchgear employs the transient earth voltage (TEV) phenomenon, which has been more widely exploited for condition monitoring and asset management of MV switchgear. Transient earth voltage sensors make use of the skin effect to measure electromagnetic radiation due to internal partial discharge. This is an attractive sensing option because measurements are inherently safe and can be made without any physical intrusion or modification to the switchgear. The benefits of TEV measurement derive from the ability to install sensors non-intrusively on in-service equipment.
Q: What are the possible approaches for monitoring PD?
Two approaches can be used to perform online PD testing on cables: periodic monitoring and continuous monitoring. Periodic monitoring is reasonably immediate to deploy, with tests between a few minutes and a few hours per circuit.
Continuous online monitoring provides 24/7 monitoring and the ability to do trending. By trending this summary data, changes in PD activity during the monitoring session can be observed. For example, increases in PD magnitude indicate the defect is getting larger, and increases in PD count indicate defects are discharging more rapidly. When the activity meets pre-set event criteria, discharge magnitude levels, and discharge rate, the system can generate alarms.
Q: What is the greatest challenge in performing online PD measurement?
The greatest challenge in online PD measurement is distinguishing between actual PD signals and noise from various sources such as electromagnetic (EM) interferences, adjacent circuits, and corona when cables are connected to overhead lines. Several advanced software techniques and hardware solutions (Fig. 2) are useful for discriminating between actual PDs and noise interferences.
In the following case study, capturing 50 non-consecutive cycles and all pulses in the time domain — while at the same time capturing the phase-resolved PD data — showed the data in different dimensions.
Case study
Periodic online monitoring was conducted on 3-phase, 268-m, single-core 33kV cables terminated to switchgear. The remote ends were terminated to a transformer. The PD monitoring system used HFCTs, TEV, and a portable measuring instrument. PD recording was performed at an MV metal-clad switchgear cable termination box. HFCTs were clamped around individual earth shields of power cables, and TEV sensors were attached to the inner walls of the switchgear cable termination enclosure.
Fifty non-consecutive cycles were recorded. With the help of software and algorithms like artificial intelligence (AI) and pulse shape analysis, the data was divided into PD and non-PD categories, as shown in Fig. 3 and Fig. 4.
The investigation further considered data including phase activity, pulse shapes, and phase-resolved PD compared with TEV and HFCT graphs. Note that phase graphs were phase consistent, and pulse-shape graphs confirmed the PD source from Phase C.
The next task is to determine the location of the partial discharges. Comparing the HFCT and TEV data indicated the same activity from both sensors. As this software and monitoring device can capture individual pulses, we further looked at data in the time domain. The same activity was observed in pulse-shape analysis graphs from the HFCT, which was installed on phase C, and the TEV sensor, which was placed on the panel.
Observing Phase C and TEV activity in the time domain (Fig. 5), it was clear that the PD activity was located at the termination end. You can see that HFCT and TEV activity was occurring at the same time at a high magnitude.
When the terminations were opened a few months later, it was clear that there was sparking at the termination, which could eventually break down the bushing connected to the cable (see Photos).
Conclusion
Online PD measurement is an important technique for use on power cables and switchgear during routine inspection and after installation to assess the condition of equipment and prevent catastrophic failures. Implementing online periodic monitoring is an effective solution to avoid catastrophic failures. The case study explained how important it is to have time-domain measurements. Time-domain characteristics can be used to classify the signals, and the method of separation based on time characteristics and AI allows noise pulses and PD to be isolated.
