Managing Water: How Digital Twins Reveal Hidden Leaks

^{Image above: A composite visualization of a wastewater treatment plant, taken from the BIM model created by WSP}

One of the biggest challenges for anyone working with water is that it is often out of sight underground, whether in a natural aquifer or a city’s supply network. Now, though, new ways of collecting and combining information promise to shed light on what’s really going on.

“Better use of data can help make the invisible visible,” explains Anna Dahlman Petri, a senior water consultant at WSP in Stockholm. A lack of useable information often means utility companies are in the dark when it comes to understanding their own networks, she says, creating a variety of problems including — notoriously — the vexed issue of leaking pipes and why water utilities do not fix more of them.

“Fixing leaks is difficult and expensive, not least because you might not know exactly where the leak is. And then if you do find it and fix it, the resulting pressure change can often cause another leak to spring up somewhere else in the system.” If a repair is not handled intelligently, says Dahlman Petri, the interruptions to supply, and the cost and inconvenience of digging up roads, simply repeat themselves in an expensive game of whack-a-mole.

To remedy this, WSP is helping water companies create “digital twins”: computer models of a supply network that integrate data such as water levels, pipe capacities and water pressures. Water companies already collect data, but this is often done manually, and it is stored separately and used for disparate purposes. “Some are already overwhelmed with more data than they know how to use,” says Dahlman Petri. “So we are asked to help them establish the level of detail they need, and how to combine data usefully. A digital twin can help them do that.”

To remedy this, WSP is helping water companies create “digital twins”: computer models of a supply network that integrate data such as water levels, pipe capacities and water pressures. Water companies already collect data, but this is often done manually, and it is stored separately and used for disparate purposes. “Some are already overwhelmed with more data than they know how to use,” says Dahlman Petri. “So we are asked to help them establish the level of detail they need, and how to combine data usefully. A digital twin can help them do that.”

For example, a twin can help with fixing leaks by narrowing down the area where water is exiting the system: “So you might only have to shut off supply to three houses instead of 300. And the effect of the repair on local pressure can be modelled and accurate adjustments made so that the chances of more leaks as a knock-on effect is minimized.” They can also inform the planning and design process for new infrastructure: “A twin can simulate how it will impact the existing system, and also help you to decide what capacity is needed so you do not over- or under-invest.”

Neither does it need to be complete to be useful, she adds. “Even a simple twin gives you a picture of what is happening underground. You can add more granular detail over time, and the more detail you have, the clearer the picture becomes.”

Integrating different data streams into one model could enable more effective, collaborative relationships too, says Louise Elliott, WSP’s UK strategy director for water, similar to the way that building information modelling (BIM) has brought construction teams together. “Water companies quite often have plenty of data, but they tend to have different systems to manage asset operation, to maintain assets, to understand performance and risk, and yet another to optimize design improvements. In the UK, regional water companies often struggle to cooperate efficiently with each other. Yet they need to, partly because rivers and catchment areas and supply needs don’t respect their geographical areas of control. A virtual reality-type representation of their system helps them to understand what they have to offer and what they need from each other. And, being visual, it helps to crystallize issues and to clearly articulate what is going on for all those involved in the industry.”

This is just the start. As ways of collecting and handling data become more sophisticated, so the potential of data to maximize efficiency is growing. Artificial intelligence, for example, can usefully incorporate digital feedback from a well-monitored system.

“Machine-learning algorithms and neural network strategies are now enabling us to do something much more profound with managing resources,” says Seattle-based Nicole DeNovio, leader of WSP’s global groundwater practice. “We use machine learning to help manage some very complex subsurface pumping patterns. If you have several wells providing groundwater, they may be of different depths and tap into different aquifers with different recharge capacities. In the natural environment, you have so much complexity. So AI is helping us to identify the best way to balance water injection and extraction, and establish what maximum sustainable yield really is in the face of climate change.”

Beyond capacity and delivery, data is also essential for utilities seeking to influence how water is used — the consumer side of the equation. “Perhaps you do something to ‘nudge’ consumers to behave in a certain way,” says Dahlman Petri. “If consumption is unmetered, or if the meter is only read once a year, then how do you know whether your nudge has been successful? But if you have detailed, real-time information on where water is going, then you have the vital feedback to quickly and accurately assess whether your intervention is working, and whether it is delivering the expected benefits.”