This will continue from the previous SYNAPS 1 project — the installation of the current sensors on LV feeders with known faults to determine their location. This work will focus on optimising the network calibration procedure and tuning the location technology to improve accuracy of the fault information and minimise the time required to generate it.
Objectives
Basic Fault Location
This will continue from the previous SYNAPS 1 project — the installation of the current sensors on LV feeders with known faults to determine their location. This work will focus on optimising the network calibration procedure and tuning the location technology to improve accuracy of the fault information and minimise the time required to generate it.
Once a fault is identified the equipment will be moved between feeder locations at regular intervals after liaison with UKPN. This will allow continued collection of fault data in multiple cable type environments. Fault location information will be communicated to asset teams for investigation and validation.
DNO Operational IT System Connection Specification
Development of a specification for an interface between the SYNAPS cloud server and DNO operational IT systems to enable DNO operational staff to be informed in real time when LV faults are detected on specific feeders and to provide fault location information when this is available. The target system will be determined through mutual discussion between the NIA partners and could include iHost or similar platform.
Advanced Fault Location
This will be based on around ten sets of the next generation SYNAPS sensors. These will be divided between SSEN and UKPN and the installation of the sensors will be on LV feeders with known faults and/or fault history. In addition to increasing the number of data collection sites, this work package will also aim to test and validate the technology in support of the following use cases:
- Low voltage networks that run in parallel known in the industry as fully meshed networks
- Long Term Fault Evolution. Investigation of the network disturbances noted in SYNAPS 1 project. to determine which are early faults, to follow their evolution into permanent faults, to seek to determine at what stage they can be located and to ascertain the scope for predicting the time before they become critical.
- Investigate if faults can be identified as either cable or joint faults.
Prototype DNO Operational IT System Connection
Based on the “DNO Operational IT System Connection specification” developed above, implementation and testing of a working interface between the SYNAPS system and the DNO Operational IT system.
Business as Usual Demonstration
A demonstration of the final TRL8 commercial prototype will be carried out on the distribution network with the intent to demonstrate the solution in a network environment and to evaluate the data connectivity.
Learnings
Outcomes
Success Criteria
SYNAPS 2 system locates faults on an LV network with a precision better than existing technologies.
The SYNAPS 2 NIA project has made significant improvements since the SYNAPS 1 NIA Project and has demonstrated that the SYNAPS solution consistently detects, classifies, and locates events (pre-fault events) before other technologies such as reclosers and temporary fuses. COPPsystem MCUs between December 2023 and March 2024 detected 7,382 events, and classified 475 as fault events.
The event data is detected by LV Monitor units (Chameleon MCU) connected to LV circuits in the SSEN network and captured as waveform data. This data is sent to and stored on servers at the Lucy GridKey data centre and is available to, and shared with, SSEN engineers. In the data centre the events are classified as fault events, load events or unclassified events. Fault events are analysed by the AI deep learning algorithm, the location on the network is determined and the SSEN engineers are notified.
Fault accuracy improvements using calibration process.
Evidence has demonstrated that calibration assists in fault location analysis.
The requirement for calibration is minimised where the DNO has online information available about their electrical network. This has been demonstrated with GE's Smallworld Electric Office (EO) and will also be possible with the Common Information Model (CIM):
- The Common Information Model (CIM) is an electric transmission and distribution standard developed by the electric power industry. It aims to allow application software to exchange information about an electrical network. It has been officially adopted by the International Electrotechnical Commission (IEC).
- GE's Smallworld Electric Office (EO) helps utilities model network connectivity, design and build workflow management, and orchestrate infrastructure asset management challenges — throughout the entire network asset lifecycle. EO provides business applications for geospatial network modelling and design based on a proven geospatial network-based model.
The introduction of Machine Learning All Feeders (MLAF) will reduce the callibration requirement . MLAF will replace Machine Learning Single Feeder (MLSF). MLSF which requires calibration to be carried out on each individual feeder where no online electrical network data (EO, CIM etc.) is available. With the introduction of MLAF no, or minimal, calibration is required at a network level, as long as online electrical network data is available and accurate. MLAF is being trialled for both single-ended and double-ended fault location.
Improve the speed with which the location can be determined.
Speed of location has improved from three weeks in SYNAPS 1 to less than 1 day in SYNAPS 2 during 2022. The model now provides instantaneous fault locations (subject to successful validation of models).
Validate the technology on fully meshed networks.
The technology has been installed and validated on meshed networks at UKPN Technology validated in radial networks, validation on meshed networks in progress. Different cable types and network architectures covered. Enhanced calibration being investigated to improve fault locations on all types of networks.
Specify and demonstrate a basic interface between the SYNAPS 2 system and DNO IT systems.
Within the NIA project an interface was successfully designed and implemented as a proof of concept.
Improve deployment procedures and expose more operational staff to the technology.
Thirteen (3 carried forward from Synaps 1 project, 5 for SSEN and 5 for UK Power Networks) sites have been installed with Synaps 2 Chameleon MCUs, this involved calibration and commissioning processes. An additional ten locations have been installed with pre-production prototype multi-port COPPsystems. DNO personnel, Lucy Gridkey or Lucy Electric service personnel and third party (Freedom Group) personnel have been exposed to the technology.
Demonstrate installation on an operational meshed LV network.
Technology installed on meshed networks at UK Power Networks, six sites for 28 months.
Identify the fault as either cable (main cable or spur) and joint faults
RINA’s reports provide evidence that SYNAPS can identify faults on the main cable and associated spurs that breech off the main cables.
Investigate long term fault evolution on faulty cables (Asset Management.)
Long term fault evolution at Trojan Way and Oakwood Close identified faults before a catastrophic breakdown had occurred. The early warning delivered by Synaps could help when determining the cause of a cable system fault which may otherwise be lost if a failure had occurred. This can also be used to inform future asset management planning.
Demonstration of the BAU commercial prototype on DNO network.
In preparation for BAU readiness single feeder prototypes for COPPsystem MCU have been demonstrated at SSEN and UK Power Networks. These have been built to a commercial standard and can be procured by DNO’s.
Fault Locations Identified and excavated.
There have been 13 fault locations modelled. Out of the fault locations modelled 11 have been successfully excavated and validated as the cause of the fault. These have been validated at RINA or by DNO Engineers. The site locations are:
SSEN (radial):
- Trojan Way 2 locations excavated & verified (RINA report 2022-0117)
- Oakwood 3 locations excavated & verified (RINA 2023-0005)
- Penn Hill location excavated & verified by SSEN.
- Rose Street Aberdeen 2 locations excavated & verified by SSEN.
- Rose Street location excavated & verified by SSEN.
- Sheldrake location excavated & verified by SSEN.
- Knowle Village excavated & verified by SSEN.
UKPN (radial):
- Pelham 1 location verified by UKPN, and cable rerouted due to tree root damage.
- McRae Lane 1 location identified and verified by UKPN. Excavation & repair carried out at this location due to permanent fault.
UKPN (meshed/interconnected)
- Vincent Square location excavated & verified by UKPN. Initially the location given by Lucy Gridkey was about 25m out, using voltage only fault localisation, but correctly identified the spur that the fault was on. Subsequently voltage and current localisation to was applied to the original event data and the exact location was identified.
- Baker Street location excavated & verified (RINA Report pending)
- Fenchurch Street/Leadenhall Street location excavated & verified (RINA Report pending)
Lessons Learnt
Operational Issues
Calibration Unit Failure
An issue was encountered with the operation of the original bespoke calibration unit when completing the SSEN installation1 (Trojan Way). The calibration unit was returned to the manufacturer who modified the unit to offer a higher rating and extra protection that stops network fuses from tripping out. The key learning from this is that calibration work should have been tested on the PNDC test network before using on the DNO Network.
The crocodile clip ends of the leads used with the calibration device need to be redesigned to avoid “sparking” and build-up of debris on the surface after repeated use.
Uninterruptible power source for sensors
Fuse failure detection on a heavy fault on UKPN caused a fuse operation, but unfortunately the auxiliary power socket at the substation in this case, was downstream of the fuse, so the sending end sensor experienced a loss of power at the same moment. This issue has now been solved and is an operational learning outcome for future installations. As a result of the protected mains source operational issue the installation procedure was improved to ensure an uninterruptible mains source is used, where available.
Fault Location Confirmation
Using complementary equipment such as Megger ‘Fault Sniffer 2’ non-intrusive gas tester, ‘Thermal imaging Observation techniques for Underground Cable Networks (TOUCAN)’ and EA Technology ‘ALVIN’ Reclose can confirm SYNAPS 2 pre-fault location improved asset management and cable health analysis. Identifying an accurate pre-fault location can be used to focus on early event identification impacting the performance and the reduction in location time after fuse/Alvin reclose operation. ALVIN Reclosers, Kelvatek WEEZAPS/Bidoyngs and VisNets from EA tech can provide useful supplementary information for pre-faults.
Early Location Identification based on Non-Intrusive Gas Testing
The DNOs in the project have carried out analysis of early fault confirmation using fault locations provided by Lucy’s SYNAPS solution. This was carried out using Megger Fault Sniffer 2 non-intrusive gas tester to confirm fault locations as early as possible in the detection process. There is a need to classify notifications to indicate those which are more urgent than others.
Proactive Control Room
SSEN have created a Project Control Room team to learn how to model and interpret fault locations. This team also dispatch to fault locations, to enable operational staff to investigate and validate fault locations. Without this Team it would not have been possible to initiate investigations.
Third Party Validation
To validate that Synaps can detect faults at earlier stages than other types of fault monitoring equipment, RINA Technologies was used to examine joints that had not had any visible breakdown. Cable and joints were removed from the SSEN Oakwood substation based on Synaps fault location and low fault activity. This has proved to be a valuable exercise; without RINA Technologies the Project would not have been able to validate the fault on this installation, thus third-party validation is an important lesson learned when using new technologies like this on the operational network.
Schneider Board Connection (Holdenhurst)
Connection to the Schneider distribution board at Holdenhurst could not be achieved with the available voltage clamps and/or Rogowski coils. Further investigation is required to identify a suitable solution.
Fuse Failure Detection on UK Power Networks Meshed Network
When installing the sensors at the receiving end of the meshed circuit, it was immediately apparent from the lack of current on one phase that the network fuse had already ruptured and so was only fed from the sending end. The key learning from this is the installation on this part of the DNO Network identified abnormal network condition.