Potential game-changer: finding imminent power line faults before they happen14 July 2020
Following successful trials in Australia, the USA and China, Melbourne-based IND Technology is tasked with commercialising the new EFD (early fault detection) preventive maintenance system for power lines developed by RMIT University after signing a commercialisation agreement with the university.
The potentially game-changing EFD system remotely detects and locates the radio frequency (RF) signals emitted by incipient power line asset faults, ie, situations which could develop into blackouts or bushfires if left unremedied. The
RF signals can be caused by abnormal electricity leakage, high levels of corona discharge into the air and internal micro-arcing inside electrical equipment.
The new EFD system is a combination of both hardware and software elements, with an architecture comprising wireless sensors, edge computing and an innovative IoT platform.
The RF signals produced by degraded equipment can radiate outwards through the air and be detected by mobile vehicle or helicopter patrols to identify sections of the powerline for more detailed investigation. On-site surveys using ultrasonic acoustic sensing and corona cameras can then pinpoint specific components that are arcing, tracking or leaking current.
This approach has severe limitations: it is inherently a point-in-time check and offers no guarantee of detection of faults yet to emerge; emissions from degraded components are very intermittent and may not be present when a patrol or survey is performed; patrol identification of areas for on-site investigation may not be sufficiently accurate; and, radiated emissions may be masked by other sources of RF interference such as industrial or agricultural machinery or nearby radio transmitters.
The other transmission channel for radio frequency emissions from failing components is along the powerline itself. Powerline wires can act like wave-guides enabling long-range transmission of weak RF signals. A network of regularly-spaced fixed sensors can continuously ‘listen’ for RF signals carried along powerline wires. The GPS satellite network offers very accurate time-synchronisation for RF signal detection, so detections from multiple sensors can triangulate the location of the RF signal source. Because mobile surveys are not involved, this monitoring can be automatic and continuous (24/365) and the location algorithm inherently discriminates against RF interference from extraneous sources that are not powerline related. The EFD technology applies this approach.
Data streams from EFD sensor units typically located four to five kilometres apart are sent over a 4G/3G communication network to a secure cloud server where complex algorithms correlate and analyse data from multiple sensors. The cloud server database populates a web portal with rich data visualisation. Network operators can thereby identify emerging issues and develop preventive maintenance strategy and take operational decisions in response to detected asset deterioration or damage and incipient or actual faults.
The EFD system works independently of the mains-frequency operation of the power system and finds emerging low-current faults just as well as it finds high-current ones.
Associate RMIT University professor Alan Wong, who led development of the technology at the university and is now CEO of IND Technology, said: “The most exciting part is the technology’s success in identifying faults that are about to happen through deterioration before they even happen – which solves the problem of having to do network repair and maintenance in response to damage already done and enables more proactive and cost-effective management of electricity network assets.”
Wong describes the system as being unlike anything else on the market due to its patented sensing method and data processing algorithm, which can identify the precise location of expected faults down to a 10 m section of a power line many kilometres long.
“This level of performance means electrical asset inspection every few years will soon be a thing of the past. With the EFD system, the network owners can now monitor every network asset, every second, 24/7 including during extreme weather when asset failures are likely to first appear,” he said.
EFD’s fixed hardware is placed at intervals of several kilometres apart and ‘listens’ for emerging issues.
It uses the GPS satellite network, offering very accurate time-synchronisation for signal detection and pin-point location of an emerging- fault signal source. The software features of EFD include continuous data streaming, big data analysis, cloud-based information management, machine learning, real-time data analytics, system alerts and 24/7 device connectivity.
The EFD system can guide preventive maintenance, including vegetation management, replacement of failing and damaged equipment, and pre-emptive asset replacement programmes, while in conditions of high fire or network supply risk, the system can detect rapidly emerging issues.
To date the EFD system has been proven to successfully detect the following items remotely, proactively, in real-time and down to 10 m: imminent pole fires; deteriorated and broken conductors; cracked and dirty insulators; touches and close approaches from vegetation in primary and secondary circuits; mechanical failures of pole top structures; animal-related activity; conductor clash or slap; conductor with bullet damage; fuse malfunction; lightning damage; deteriorated underground and overhead cables; and internal equipment insulation breakdown in assets such as pole top transformers.
As well as being voltage agnostic, able to find incipient faults in the high voltage primary network (at transmission and distribution voltages) as well as the low voltage or secondary system, it can also be applied to an overhead system and an underground network.
To date, EFD has been implemented in the Australian single-conductor network, 3-wire systems as well as 4-wire systems.
In North America trials the system was installed in a 12 kV network and detected and located many issues, including vegetation contacts, conductor damage, insulator tracking and conductor clashes, to a high accuracy. More systems are to be installed shortly in California and elsewhere in the USA. The technology was one of 17 finalists selected from over 130 applicants to participate in EPRI’s Incubatenergy Labs Challenge pitch day, April 2020 – postponed to early 2021 due to Covid-19.
In Hong Kong, a pilot system has been installed on the 25 kV railway network, with trials projected to continue until the end of 2020 and positive results expected.
The largest, and most impressive, trial to date was that carried out as part of the Victorian government funded Powerline Bushfire Safety Program, from late 2017 to mid-2019.
The EFD system was trialled on SWER (single wire earth return) networks operated by AusNet Services and Powercor in rural Victoria to allow these network owner/operators to gain experience of the technology and assess its value as a fire-risk mitigation tool.
The system successfully identified a number of real powerline bushfire risks in high fire consequence areas and demonstrated how EFD data could potentially transform network asset management. Experience gained during the trial also resulted in some improvements to the technology.
Three fire-risk situations were detected early by the EFD technology during the Victoria trial:
- Broken steel strand on a powerline. This case (at Ross Creek) received attention from local media. It was exactly the same pre-fault situation as that implicated in the disastrous Kilmore-East/Kinglake fire on Black Saturday (7 February 2009) which caused huge loss of life and property.The broken strand was above a grassy paddock in rolling hills south of Ballarat, ie, the fire consequence of a fallen powerline in such worst-case conditions could be high.
- Arcing low-voltage service line. The failure of this privately-owned overhead low-voltage service line was of a type that has led to fires in the past. Regulatory changes in 2009 require such lines to be progressively replaced with underground cable to prevent fires. This fault was detected only half a km from the Ross Creek broken steel strand. Again, fire consequence in such worst-case conditions could be considerable.
- Deteriorating substation transformer. This transformer problem was at an unoccupied house next to an extensive heavily wooded area north-east of Beechworth, ie, with very high fire consequence. Signals received by the nearest EFD sensor were becoming more intense at the same time as the weather was deteriorating towards extreme fire risk levels. Replacement of the transformer appeared to fix the immediate problem.
The three EFD-detected situations had clear significance to fire-risk. It was concluded that on the evidence from the trial, EFD technology offers potential fire-risk benefits sufficient to be of material significance to Victoria if it were to be applied more widely across the state’s rural SWER networks.
The EFD technology also detected and continuously monitored a small number of network assets that exhibited intermittently high levels of signals indicating abnormal electricity discharge. Among 924 poles (311 of which had substation transformers mounted on them), three poles appeared to have assets that were actively deteriorating.
Two of the three sources of abnormal signals were located by the EFD technology to an accuracy of ten metres on inter-sensor network paths. They both proved to be at substation poles. The third was located at the end of a spur line beyond the inter-sensor network path and again, was considered to be most likely to be located at the termination substation pole.
The EFD SWER trial confirmed that EFD technology had sensitivity, signal-to-noise ratio and location accuracy that exceeded all stakeholder expectations. The first sign of EFD technology’s sensitivity was when it was realised that the EFD systems were detecting and locating the impacts of individual raindrops on the powerline conductor during heavy storms. Deep investigations into the EFD data revealed background noise levels were generally at least one order of magnitude less than the smallest signals from network assets. At higher signal strengths the signal-to-noise ratio was around one million to one. Detailed investigations also demonstrated that signal sources were located to an accuracy of plus or minus ten metres, which was the same order of accuracy as pole location records in network owners’ databases.
The EFD SWER Trial demonstrated the performance of the EFD system exceeded any level that could reasonably be required for network operation and asset management.
SWER powerlines and fires
Powerlines can cause catastrophic bushfires and have done so in Victoria, causing some of the worst fires on Black Saturday in 2009.
There are three ways powerline faults start fires:
- High voltage electric arcs near vegetation, usually dry grass (wire down).
- Vegetation conducting high voltage current (tree touching wire or wire dropped into bush).
- Incandescent particles landing in dry grass (when live wires clash).
The most catastrophic fire on Black Saturday (the Kilmore-East/Kinglake fire), started when an SWER powerline fell to the ground and ignited dry grass. Two other major fires that day (the Horsham and Coleraine fires) were started by SWER powerlines that came loose and contacted nearby vegetation.
Since Black Saturday, the government of Victoria has sponsored a major programme of research to better understand how powerlines start fires and to find ways to cut powerline fire-risk. This research has shaped a multi-faceted programme of work to improve powerline bushfire safety. Action on SWER powerlines has included selective undergrounding and use of covered conductor, and the installation of more than 2000 high-sensitivity, fast-acting, remotely-settable automatic circuit reclosers (ACRs) to more quickly detect and isolate faulted SWER powerlines when fire-risk is high.
ACRs can operate fast enough to prevent fires from SWER powerline faults provided the fault causes sufficient increase in powerline current to be quickly detected. However, ACRs cannot detect low-current faults because the fault current cannot be distinguished from normal variations in customer load current. Unfortunately, low- current faults can still cause fires. The detection of low-current faults on SWER powerlines is one of the last remaining gaps in the powerline bushfire safety tool kit. As yet, there is no reliable method to detect low-current SWER faults quickly enough to prevent a fire so there is a high focus on improved network O&M to prevent such faults.
This led to the Victorian government’s sponsorship of the EFD SWER trial reported here.
The trial revealed that EFD technology works well on real networks and can deliver material fire-risk reduction benefits.
Existing business processes in electricity distribution companies are designed around the concept of fast, effective response to network faults after they occur (and have resulted in supply outages or fires). For maximum benefits from EFD technology, new business processes must respond to warnings of imminent faults that have not yet happened. This is potentially a profound change.
Further reading: EFD SWER trial, final report, 23 June 2019