Market View: Can AI help us cope with increasing water demand?

Mark Deighton, head of Insight Driven Transformation at Capgemini UK, weighs up the options.

Global temperatures are up 1.8 degrees since 1980 and 55% of the world’s population now live in urban areas, which is set to increase to 68% by 2050. In fact UK city dwelling is up 3% in the last 7 years.

Overall demand for water in the UK is up by almost a third over the last 50 years, and if none of that was a heady enough combination for the water industry, now we’ve experienced the longest dry spell in 40 years. 

All this combines to form a perfect storm for water demand. It is therefore no surprise that a recent report from the National Infrastructure Commission suggested that in the next 30 years there’s roughly a 1 in 4 chance that water supply will be cut off for extended periods due to severe drought.

But such water supply restrictions could cost London’s economy £330m every day, according to Thames Water, which is obviously a figure that would be reflected in cities across the UK. And as a result the long-term need for resilient and affordable water supply is increasingly under the spotlight.

What is being done?

Seven million people in the North-West of England were given a hosepipe ban on 5 August 2018. But this is a short-term, reactive and largely cosmetic measure, as domestic consumption represents less than 5% of total water use.

Instead, a sharp focus on distribution leakage as opposed to domestic consumers might provide the answer.

A 2018 Environment Agency report said three billion litres of water a day are lost through leakage – that’s around 30% of all water produced.  In response, Ofwat  set targets for a 15% reduction in leakage by 2020, so this is now one of the top priorities for water companies.

Tech solutions

Modern technologies and innovations are being explored to deal with this challenge, including acoustic listening devices, drones, satellite images, and intelligent demand forecasting using weather and consumption patterns. These have all been employed to help detect and locate leaks faster, but most require significant capital investment and years of planning.

Acoustic sensors have proven highly effective and can now be deployed without breaking ground. Equipped with 3G and other over-the-air communication capabilities they enable reactive leak awareness and localisation. However, a useful range of around 300 metres creates high costs due to the volume of sensors required.

Research by UKWIR also concluded that acoustic sensors struggle to detect leaks on plastic pipes; in noisy environments, such as city centres; noise from smaller leaks masking larger leaks; locating leaks in large diameter pipes; and locating leaks in areas with pipes using varied materials.

AI to the rescue?

But the rise of machine learning (or AI) offers a lower cost alternative and has proven highly effective in detecting and locating priority leaks; typically delivering over 80% accuracy.

Using telemetry data from existing pressure and flow sensors across underground water networks, complex algorithms – assisted by AI – can identify and locate leaks as they occur rather than the norm of weeks or months later.

By truncating the leak life-cycle, leaks can be fixed faster, they can require less intervention effort, cause less damage and inconvenience, and leak less water. The outcome is more water available to consumers and reduced operating costs for water companies. Is AI the silver bullet we’ve all been looking for?

Ofwat’s view

Ofwat has already identified that the best solution to sustain household consumption without a reduction in the level of utility or quality of our water is to leverage technology and service innovation.

And as water becomes an increasingly precious resource, the time to pursue this strategy is now.