UPDATED: United Utilities successfully uses FIDO AI to negate false positives, eliminate human error and halve leak run time.
United Utilities has one of the largest fixed logger estates in the world with 67,000 devices generating 600 new leak alarms every day.
With no way to identify false positives, each alert had to be analysed manually in order to schedule appropriate follow up by a technician.
It calculated that this task was taking the equivalent of seven full time analysts a day, diverting their skills from other, more complex, leakage tasks such as pressure management.
There was also a significant margin of error, with many non-leaks scheduled for a logger follow-up (LFU).
Having a tough regulatory leakage target to meet, the company wanted more value from its substantial logger investment. This meant reducing the time from alert to LFU and making better use of limited skilled resource by reducing the time people needed to spend managing alerts and responding to false positives.
The FIDO AI Solution
We deployed FIDO AI through API with the client’s existing logger estate portal in August 2020.
It analysed all of United Utilities acoustic logger alarms in real time and deliver instant leak/no leak decisions and leak size. All the results were available before the start of each working day.
FIDO also created a benchmark data set by evaluating tens of thousands of historic logger files dating back to the start of 2020.
All the key leak process data was brought together into a single place where it could be used as a reference point for tracking accurate performance improvement.
Using FIDO AI for rapid, accurate and timely daily analysis of logger files completely removed the need for manual analysis, freeing up the equivalent of seven skilled analysts.
The results proved that 72% of leak alarms generated by the company’s loggers were false positives. This meant that using FIDO AI reduced the likelihood of sending engineers to a non-leak.
FIDO’s leak-sizing identified the six per cent of leaks larger than 2.5m3 per hour. This allowed United Utilities to prioritise its large leaks and then batch smaller leaks for greater efficiency.
Comparison of pre- and post-FIDO performance over the 12 months to July 2021 showed that by using FIDO AI to identify and prioritise large leaks United Utilities had increased its daily leak finding rate by more than a third (+34.9%) and almost halved (-49%) its overall leak run time.
Additionally, the number of LFU visits per day rose by 19.3% and the time from alarm to completed LFU dropped by 39.4%.
“FIDO AI gives me the benefit as a leakage manager to instantly assess all the alarms on our current acoustic logger fleet. We can then plan those priority leaks ensuring that we can efficiently use our leakage technicians to locate leaks on our network, whilst removing a large number of abortive visits.”Paul Parr, Leakage Technical Manager
Logger alarm files analysed per month
Overall leak run time almost halved
Rise in number of leaks found per day