Yorkshire Water has improved its leak detection with an artificial intelligence trial that has learnt to differentiate sounds from acoustic loggers to reduce false alarms by 60 per cent.

The project with Siemens and Artesia analysed data from nearly 40,000 loggers across Yorkshire’s clean water network to listen for evidence of leaks that should trigger a manual investigation.

These can be prone to false positive alarms when a background noise is mistaken for a leak. However, the audio analysis with advanced AI has improved the process by learning to distinguish between leaks and other sounds.

Innovation project manager at Yorkshire Sam Bright said the technology is key to the company hitting its leakage target of 15 per cent reduction in AMP7. If successful, Bright said the trial will let the company target its activities more effectively.

Following Yorkshire’s appeal to the Competition and Markets Authority (CMA) it, along with Anglian, Bristol and Northumbrian, was permitted an additional spending for leakage but the reduction target remained as set by Ofwat at PR19.

The CMA increased totex spending by 3 per cent and provided enhancement spending of £75 million for company specific schemes.

Elsewhere, the company has deployed remote actuators to remotely manage water flow centrally from a control room to respond more quickly to pipe bursts and interruptions.

Its smart meter rollout reportedly reduced leakage by more than 90,000 litres daily by pinpointing bursts within the smart network and identifying customer-side leakage for repair.

Cutting leakage, together with lowering per capita consumption, has been identified as key to ensuring resilience of water supplies in the face of population growth and increasingly changeable weather patterns due to climate change.