REMOTE LEAK LOCATION – FIXED CORRELATING NETWORKS
- 1. PPS-GEN-044-4.0 Smart Water Utilities 2019 Reducing Water Leakage Across the Network Roger Ironmonger, Managing Director Remote Leak Location - Fixed Correlating Networks
- 2. UK water leakage situation The Environment, Food and Rural Affairs – select committee report The target for water companies to reduce leakage by 15% by 2025 is not ambitious enough – the water industry should aim to reduce leakage by 50% by 2040 National Infrastructure Commission – advises UK government A twin-track approach of improving infrastructure and reducing leakage…. to ensure the network is fit to meet the future pressures from climate change and increased population – otherwise we’ll see Cape Town style droughts in the future Anglian Water CEO – Peter Simpson Under its five year plan AW is investing £240m to cut leaks by 22% across it’s 40,000km of pipes…… innovative solutions are needed in order to establish long term, resilient water supplies
- 3. The technology High sensitivity hydrophone sensor - detects leak noise inside water pipe Correlation processing – detects more leaks (vs conventional leak listening) Very accurate, daily time synchronisation - for accurate leak position + repeatability Simple underground installation – no above ground equipment required Daily audio data transmission and ‘cloud-based’ correlation - reduces leak run-time - separates leakage from customer use
- 4. On-line correlation and data delivery Leak + logger positions on Google Maps Historical data available to aid leakage management Audio file saved - listening for confirmation of leak noise Alarm notifications via email Street-view* to aid visualisation * Courtesy of Google Maps
- 5. Adaptive Filter used 531 Hz – 816 HzNo leak using standard filters The technology - automatic filtering Live Adaptive Filter operates in the frequency domain applying over 100 different filter frequency ranges to the correlation data Difficult leaks found – can be missed when correlation with conventional filtering!
- 6. Proof of concept – Louth 2826m correlation to a comm pipe leak 1899m of PVC main (33 mains) 375m of CI main (10 mains) 552m of HDPE main (13 mains) 56 Different Mains Materials 3002m correlation to a burst main 367m of AC main (8 mains) 1313m of CI main (31 mains) 2m of MDPE (1 main) 1298m of PVC main (33 mains) 21m of HDPE main (3 mains) 76 Different Mains
- 7. Proof of concept Bespoke installation chambers Historic minimums surpassed – generally high quantity of smaller leaks, that were either previously undetectable In this example 16 hydrophone loggers covered the area but 655 conventional noise loggers would be required DMA ref KPIPHLMA minimum halved (l/s)
- 8. Rural leak location Leak located over 4.6km between loggers resulting in a leak being located within 5 metres
- 9. Leak positions “Unlike more widely used methods, this new technology can be used on plastic pipes, which make up 60% of our waters mains. This is a real game- changer in how we tackle leakage. It will revolutionise what we do in terms of leakage.” Andy Smith, Regional Optimisation Manager, Anglian Water. Anglian Water have reduced leakage by 20% since 2010 and committed to a further 22% reduction by 2025 Technology comparisons performed over mains length of 40km of mixed pipe materials So far 1325 leaks found from a potential of 1600 - a conversion rate of 83% Conventional noise logger approach required 800 devices at cost of £200,000 compared with 58 Primayer devices costing £69,600 In areas where work completed a 1.4 million litres/day reduction – a 35% shift in those areas Now have installation of 3500 Enigma3hyQ underway with a further 1000 to follow Demonstratable ROI
- 10. Meeting the challenge – leakage management
- 11. At December 2018 the NRW is 24%; target is to reduce to 5% by 2025 Approximately 600 Enigma3m plus 50 Enigma3hyQ installed in 5 DMA’s In Bandar Putra B DMA 295 devices installed on metal + polyethylene pipes After 3 months 115 leaks located with NNF reduced from 30.99 l/s to 20.08 l/s Total water saved in one DMA is 705 m3/day giving total cost saving of >£3000/month Water Supply Company Johore (SAJ), Malaysia Leakage Confident Factor (LCF): Red = 11 Nos Yellow = 59 Nos Blue = 225 Nos
- 12. Angers, France 792 correlating noise loggers permanently installed in the city center of Angers in 12 areas Florimond NAULLEAU, Operating Officer: “In 2018, the team found 50 underground leaks equivalent to 30 cubic meters / hour, or 720 m3 of water saved per day! ". (City consumes 50 million liters of water per day)
- 13. Mass deployment = huge data analytics Over 21,000 Enigma3 correlating loggers now installed - over 5000 of these with hydrophone sensors Customers with large deployments have a very large volume of data Often several leak markers can point to one point-of- interest Need easy way to review data!
- 14. Data analytics Automatic optimisation of filter selection Multiple points of interest (pointing to the same position) are clustered together and most confident is highlighted Ability to categorise these points Multi-day correlations are automatically triggered and used to calculate an overall confidence factor Leak suppression – suppress automatic leak markers in a specific location for a number of days (until they should be fixed) Manual analysis Automatic analysis
- 15. Summary Table
- 16. Allows a shift towards proactive leakage – potentially locating leaks before customer knows about them Enhanced analytics – multi-day and multi-logger ‘Street-view’ helps plan repair work Accurate remote pinpointing such that site visit correlation not required - could we train repair crew to confirm leak position with ground microphone hence saving operational costs? Cost effective remote leak location - no above ground equipment - less sensor points - data analytics Operational benefits
- 17. Artificial Neural Network learns the typical range of behaviour for a signal The output of the ANN are the bounds of anticipated behaviour Actual telemetry data outside of the bounds for a sustained period causes an alarm Sizes of events are estimated Data analytics - future…..self-learning event detection
- 18. 2.73 l/s Burst occurs at 3:15 a.m. 20-Jan Verified burst repair at 6:30 p.m. 20-Jan The future…..example bust identified and repair verified
- 19. A = AWARENESS L = LOCATION R = REPAIR Water loss: m3/day 75 50 25 Leak Type Water Lost Daily loss 1.1 days Catastrophic Pipe Burst Mains: 82.5 m3 75 m3 YesVisible Leak? A L R 16 days Reported Leak Communication pipe: 400 m3 25 m3 Yes Active Leak Detection Supply pipe: 3650 m3 20 m3 182.5 days A L R No THE IDEAL Supply pipe: 80 m3 20 m3 4 days No Example showing that the smaller, non-visible leaks account for the majority of water loss. Most of these run, not just for 182 days as per the smallest leak example here, but go undetected for years Courtesy of Bristol Water Reduced leak run-time – increased network resilience
- 20. Thank you Innovation for leakage reduction…