Water scarcity is becoming a significant environmental and social concern, with experts warning that the crisis should be treated with the same urgency as climate change - which is of course inextricably linked. With this in mind, the UK’s water services regulation authority, Ofwat, has awarded a range of partners £7.5 million to work on a project called Safe Smart Systems, with the aim of using artificial intelligence and mathematical optimization to improve long-term operational resilience of water systems across the UK.
The project was made possible because of the British Government’s Innovation Fund and is being led by Anglian Water, which supplies 4.3 million people and is the largest water and sewerage company in England and Wales geographically (covering 20% of the land area). It operates 143 water treatment works, supplies over 1 billion liters of water every day to 2.5 million households and 110,000 businesses.
Alongside Anglian Water, Safe Smart Systems includes a range of partners that range from the Centre for Digital Built Britain, to Imperial College of Science, Microsoft, Unity Software and Airbus Defence and Space (amongst others).
diginomica got the chance to speak with Fionn Boyle, who is the strategic innovation lead at Anglian Water, who said the thinking is to ensure that the Safe Smart Systems project remains as agile as possible - where the outcomes are being driven by true research and exploratory methods. Boyle said:
All the other projects that we saw being developed or being funded, they were quite well defined. They were like: ‘we're gonna work with these partners, we're gonna deliver this exact output’.
Whereas when it came to Safe Smart Systems, we wanted to leave space to really try and push the boundaries of what's possible.
Anglian Water is a leader in managing water leakages in the UK - in that it is the expert in how to do this effectively. Every day it has 300 people finding leaks, it has the best record for detecting and tackling leaks in the UK per kilometer of pipe and in 2020 it reduced leakage to its lowest ever levels for the ninth year in a row. However, this experience has proven to Anglian Water that is needs a different approach. Boyle explained:
Because of that position, we’d got to the point where we know that managing leakage, or managing any problem on its own, isn't the right way to go about it. Because you only drive one outcome.
So we settled on a systems-based approach, which is really focused on enhancing resilience for the long term. And trying to cover multiple benefits at once.
As such, we put forward this project that is true agile delivery, which for a water company might be a bit alien. That’s because typically we like waterfall projects, we build capital assets, that's what we know, that's what we've always done. We’re recognizing more so now that that isn't the right way to do it.
Boyle said that Anglian Water wanted to work with partners that are key leaders in their respective areas. The project was set up to understand: where is the best in class? Even if they are siloed solutions, where does that exist? And then to know whether or not water companies can really use AI and automation to manage their systems. The key is that this isn’t for Anglian Water itself, but is a knowledge sharing exercise that aims to benefit the rest of the UK - or even the globe.
The key drivers for this attempt to better manage systems end-to-end using data and automation is, as highlighted above, because climate change and population growth are driving water scarcity in the UK, as well as across the globe. Boyle said:
In the East of England, we have three of the fastest growing cities outside of London. And we have below the national average rainfall. You feel like it rains everywhere in England, but it doesn’t. And it's a very real fact - if we did nothing in the East of England, part of our region would run out of water within the next 20 years.
That’s not a long time. So we need to think differently when it comes to managing that. And the way in which we see that future coming just through building new assets, spending money on concrete and steel, it's about how do we manage the assets that we already have better?
How do we make them last longer and how do we do different things with that to build resilience into the system?
This sort of project doesn’t come without its challenges, however. Understandably, much of the sewage and water networks in the UK are old, the assets (pipes) are often over 80 years old, for example. In addition to this, the records of these assets are sometimes not the best - which is difficult when you’re thinking about collecting data for analysis. Boyle added:
From a data perspective, because we're a legacy industry as well, we have issues with data quality, we have issues with latency from some of our operational technology.
We knew we were going to have to solve those issues, to be able to get the feedback loops going, when thinking about machine learning and AI.
That's where the bulk of our work to date has actually been: building those things in. It's not the sexy stuff, but it's the stuff that's most important. It’s the stuff that has the longest lasting impact, rather than just that shiny AI decision engine that sits on top.
In addition to this, in order to remain truly agile, Anglian Water also knew that it would have to change how it organizes itself within teh company. Boyle said:
Actually what we're seeing now is that the learning that we're getting out of Safe Smart Systems is fundamentally changing and forcing us to rethink how we structure ourselves. And not just in terms of hierarchy, but capability-wise: what are the skills that we need to have in the organization and where we're going to get those from?
Then you add into that that at least 30% of our workforce is likely to retire within the next five to 10 years. How do we upskill the young people? How do we bring in new people and where we're gonna get those from?
Currently, Smart Safe Systems is in what Boyle calls the ‘concept and feasibility’ phase - or minimum viable product stage. It’s building digital solutions, building things for feedback loops, and is providing better tracing through the system.
The first MVP it's been working on is around anomaly detection, which focuses on trying to work out if there is a signal that says there’s a burst water main or a leak. If it gets an alert, is that true or is there an issue with the sensor? On some of the other use cases being worked through, according to Boyle, include:
Enabling routine proactive maintenance and moving away from time based maintenance; we're looking at digitalization of asset records for asset performance; better ways of collating operational asset data for data capture; operational response, localizing network faults automatically and doing root cause analysis for why something has gone wrong or why it could go wrong.
Then we move into the theme of dynamic networks, which is about being able to balance a supply abstraction from the environment through to the customers tap; better management practices; and the configuration of the network automatically; and the idea of having a self-healing system so that if something changes we don't need to interact with it, the system will understand what's happened and automatically reconfigure itself as the best possible condition.
All of Smart Safe System’s sensors are third party, However, the project has also been building its own digital solutions in-house. Boyle said that this isn’t something it’s focused on doing, but argued that finding piecemeal solutions in the market where it could pick best of breed isn’t easy. He said that typically commercial software providers want you to buy a whole suite, which isn’t what’s needed for this smart utilities project. Boyle said:
That's an evolution I think the water industry and our supply chain need to go through in really understanding how you create the space for plug and play solutions that give you besting best in class, rather than having to take quite a vanilla approach and choosing one that can offer lots of different things, but never quite leading.
Understanding what’s achievable
Boyle said that Safe Smart Systems is as much about finding out what the industry can’t do, as it is about finding out what it can do. It’s looking to push the boundaries, but also seeking to understand where it will fall over. This is something Boyle said it already has experienced, which has helped to better its progress. He said;
We will fall over. In fact, we have fallen over - but we picked ourselves back up again on a number of things, particularly around data quality and data structures.
But AI won't be the answer to everything, right away. If we're talking about full system automation, I think we're a very long way away from being able to do end-to-end. When we talk about AI, people go straight to ChatGPT, and all of those visions of complete automation.
I think, for us, it's the right flavor of advanced analytics that we need to provide the best service. So we're always trying to weigh those things up. You can't have a silver plated or gold plated tap everywhere. So you need to think about what's the best cost for the customer and value for the environment.