Whilst the use of AI, automation, and data science is typically thought of as being the preserve of industries like telecoms, energy, or precision manufacturing, an example that might not immediately spring to mind is ensuring that water always flows from your taps. However, this is exactly what those of us at Anglian Water are now doing as part of the Safe Smart Systems project, in partnership with Jacobs, Affinity Water, South West Water, Portsmouth Water, Airbus Protect, Microsoft, The University of Sheffield, Imperial College London, Information Junction, Skanska, BIM4Water and an international cohort of supporting partners. The project will help inform a framework for the UK’s national digital twin.
The impetus for this comes from needing to improve visibility of the water delivery infrastructure. Just as logistics businesses are looking to make the delivery of goods more efficient and avoid disruptions, so too is the water industry looking to upgrade its networks so that we can more quickly identify leaks and other issues, remediate them, and eventually get to a position where the network can be self-healing so that leaks do not have an impact on the water supply.
From pipe dreams to reality
When it comes to water distribution, the two critical parameters are flow and pressure. We are somewhat fortunate in that water hydraulics are predictable and so can be modelled, making irregularities easily identifiable. The challenge is achieving the granularity necessary across what is effectively a highly distributed food-grade production facility to create an effective digital twin and corresponding AI decision engine.
One of the most important pieces of the puzzle is visibility at the customer’s property. The challenge, of course, lies in how often we receive readings from customers. You might be asked for a meter reading every three months, or in some cases six, and sometimes those readings might be late or forgotten about entirely. In any case, it is a world away from the visibility needed. To overcome this, smart metering, which is still fairly nascent in the industry, has and continues to be rolled out at scale to help piece together the flow of water from the reservoir to the customer in a more immediate way.
Obviously, understanding the amount of water that leaves the reservoir and how much of that gets to customers is one thing, but to identify and solve problems quickly we are need greater visibility along the network of pipes that supply water across the region. Pressure monitors are key to this. Making this as accurate as possible has meant deploying a higher density of pressure monitors than one might normally, using five per districted area instead of one, but doing so has allowed for quick and exact triangulation of data so that we can zero in on where a leak might be happening.
Once an issue is identified, the ultimate goal is for the system to automatically isolate the problem pipe and reroute the flow of water so that it is not wasted. Whilst pump control is not necessarily new, where Safe Smart Systems looks to take a new approach is to place more of the decision making for pressurising the system at the edge, ultimately placing greater trust in automation, instead of setting the system up on a modelled basis, so that it can react more quickly. The other key element is creating more integration between different districts. By using bi-directional pressure controls at the boundaries of these districts, the aim is to get away from the siloed, zonal design that has plagued the industry to create greater resiliency by allowing supply from one to flow to another when needed.
Rivers of data to insight
At the heart of making the whole system work is the data upon which decisions are made. This means integrating highly varied data sets both from across the networks of pipes, pumps, and sensors that we operate and from elsewhere, such as weather data and critical suppliers that we rely upon, such as the energy network. From here we can start to build a comprehensive model that can account, not just for leaks, but can also assess the likelihood and impact of other scenarios, such as what would happen if heat stress knocked out part of the power network. On top of this, we are also looking at integrating customer-based alarming into the system, so that as a problem is reported to customer service or is reported online, it can be fed into the model.
Data quality and the interoperability of live insights is a major challenge given that some of the infrastructure is decades old, from a time when the asset record was drawn out by hand. However, it is absolutely necessary for creating the feedback loops through automated data flows that enable real time decisions about water supply that can drive faster actions.
A future on solid ground
For water companies, there is a clear need to reduce leakage and we see Safe Smart Systems as being a key part of our achieving this. Realising a level of autonomous operation that effectively allows the network to become self-healing will have a significant impact not just for customers, both in terms of reliability of service and lower water bills, but also on the environment; fewer leaks mean a reduced demand on the freshwater supply. This is really important at a time when population growth, data centres, and the production of hydrogen as a renewable energy source put more pressure on that supply from multiple angles. Developing a self-healing network at this scale has never been done before, but Safe Smart Systems should become a critical building block to achieving this goal, enabling others to follow suit.
These learnings will not only apply to the water industry. As part of a national UK digital framework initiative originally developed by the Centre for Digital Built Britain, it will also inform digital twins across numerous sectors. The partnership between Anglian Water and the Department of Business, Energy & Industrial Strategy is helping us understand how various sectors, such as construction and infrastructure, can create a digital approach to better design, build, operate, and integrate the built environment. The lessons being learnt using IoT and AI can be instructional for all of us as we look to build a more effective and resilient built environment in the UK and beyond, today and for the future.
Fionn Boyle is the Strategic Innovation Lead at Anglian Water, where he leads the Shop Window innovation strategy and is responsible for working with business in delivery of key strategic innovation initiatives.
