Pdf

Using artificial intelligence to deliver advanced water network management

Our Systems Thinking approach is one of our competitive advantages, as outlined in Our competitive advantage, and we continue to increase our capabilities as part of this approach. One area that demonstrates this is in managing leakage.

We need to operate and maintain our water network to reduce leakage and to reliably deliver excellent quality water at the appropriate pressure to our customers' taps. 

In the past this has often relied upon customers informing us of an issue before we've carried out emergency repairs and restored service.

Due to advances in network monitoring technology, coupled with artificial intelligence, we're now able to provide proactive, and often predictive, management and maintenance of our water network.  We're leading the way in the UK water industry with our use of artificial intelligence to improve customer service and operational efficiency.

We supply water to around seven million customers, and to do this effectively the water distribution network is divided into approximately 3,000 District Metered Areas (DMAs). In collaboration with a leading university, we've developed an artificial intelligence system known as Event Recognition in the Water Network (ERWAN).

ERWAN applies Systems Thinking

ERWAN uses advanced analytics to learn the typical patterns of the system from our network of sensors, identifying the 'normal' system signature for each DMA so that it can recognise any deviation to this signature and generate an immediate alert. It applies Systems Thinking to determine the likely root cause of an alert, such as a faulty valve or water main leak/burst, and understanding adapts automatically over time. Traditional analysis often focuses on individual items, whereas Systems Thinking also looks at how these items are connected and interact.

The application of ERWAN has resulted in multiple benefits including the avoidance or reduction in issues such as poor water pressure, no water, or poor water quality, thereby improving our service to customers. It has also reduced asset maintenance costs by informing the need for maintenance prior to asset failure, and avoiding unneeded maintenance visits. Operational costs are also reduced as it enables problems to be dealt with proactively which is much less expensive than dealing with asset and service failures.

The use of ERWAN has contributed to the three per cent reduction in leakage and 29 per cent reduction in water network incidents between 2011 and 2017.

Expanding our use of artificial intelligence

Artificial intelligence and machine learning methods provide the opportunity for us to operate and maintain our asset base at a lower totex than was previously possible, while being able to minimise levels of customer disruption and improve the service we offer.

Following the success of ERWAN, we're looking to apply similar artificial intelligence methods in other areas of the business. We have a number of collaboration projects with universities and specialist analytics companies to capitalise on the opportunities further as we move into the next five-year investment period (2020–25).