Let AI guide leakage investment, not assumptions


Home > News > Let AI guide leakage investment, not assumptions.

AHEAD of her opening case study at this year’s Global Water AI Innovation and Data Usage Congress from April 27 to 28, FIDO CEO Victoria Edwards discusses the need to replace assumptions with accuracy by letting AI guide leakage investment decisions.

I don’t like the term non-revenue water. It makes an abstract concept out of something we should all lose sleep over. It’s the amount of water utilities take from the environment, treat to drinking standard, but never bill for because it goes missing somewhere along the line.

There are a few reasons for this, but the sizeable and most problematic one is the simplest to understand – leakage. Recent research suggests cutting NRW by one third would yield the same amount of water it takes to supply 800 million people. You wouldn’t catch most manufacturers taking that sort of hit on their bottom line.

Leaks are hard for humans to find

It’s not that utilities aren’t taking it seriously. It’s just that, despite all the advances in network technologies, too much current operational and capital investment in leakage is still based on guesswork.

Only five per cent of leaks show above ground making them very hard to find and fix. Trusted technologies like the listening stick have been digitised into sophisticated acoustic sensors and correlators which detect and record the sounds of leaks.

But until now, it has still depended on human beings to interpret whether a sound is a leak or not, actually pinpoint it out on site and then, based on best guesses, decide where to plan investment for the best results.

Avoid investment based on assumptions

Working with many utility clients, I’ve been astonished to find out just how much business strategy is built on assumptions. Where do they think the water’s being lost? Is it customer side? Or is it at the point of production?

Estimations are made on number of leaks they’ll find, how big they’ll be, how long it will take to find and stop them, and how much water that will save. Finally, they’ll take an educated decision on which parts of the network are most in need of maintenance.

Some of these assumptions lead to huge capital investment programmes to build extra assets to replace water that is being lost.

Remove human fallibility

Until now, managers have largely been playing the hand they’ve been dealt. Investing in technologies to give them more certainty of data on which to base those decisions, but accepting that, ultimately, big strategic decisions come down to them, experience and a bit of luck.

The advent of AI into this space has been a game changer. Not only does AI analyse acoustic and kinetic sensor data in a tiny fraction of the time it takes an analyst, it’s massively more accurate. With FIDO AI you also add the ability to accurately assess a leak’s size before you even put an engineer in the van.

You remove human fallibility from a key part of the process. Set up automated scheduling too and, hey presto, FIDO is uploading alarm files at scale, sifting out the genuine leaks and sending your available engineers straight to the ones that are losing you the most water.

Improve operational performance

Bye bye assumptions. By helping one client prioritise the six percent of its hundreds of daily leak alarms which were genuine large leaks, FIDO cut average leak run time from more than 18 days to three.

That equates to a huge saving over the year just in NRW. Add in additional savings due to fewer wasted visits, dry digs and night-time soundings and the leakage savings are almost double.

So, FIDO gets you to your biggest leaks early. In terms of performance and day-to-day resource management it’s hugely important. But because FIDO assigns each leak a unique ID, to track the whole process from first file to final fix, each leak suddenly becomes an asset.

Let AI guide leakage investment

FIDO is starting to use these multiple new assets with data from other systems – such as historical records on material, volume of water lost, repair, usage and individual engineer activities – to build a comprehensive picture of network leakage performance.

FIDO can do this because it is totally hardware, platform and material agnostic. Its intelligence accuracy is built on verified data gathered from some of the world’s most mature networks, and proved through physical interactions with real events.

We call this concept Leak Central, using FIDO’s algorithm to analyse explicit and implicit data to generate reliable outcomes on which to base strategic business decisions with confidence. All of which means you let data-led AI, rather than assumptions, guide your leakage investment.

FIDO is lead sponsor at this year’s Global AI & Data Usage Innovation Congress on April 27 and 28, 2021. Click below for details. You can also take advantage of our 25% discount code or complete a short survey on attitudes to AI in the water sector for the chance of a free pass.