It’s tempting to think that water scarcity is mainly a problem of the developing world. After all, the number of people lacking access to clean water in Africa alone is predicted to rise to more than 300 million by 2050 – an almost fourfold increase from 2016.
But high income and industrialized countries are not immune. The effects of water scarcity are becoming increasingly noticeable there, too. The WRI’s water stress ranking lists six Western European states among the 47 countries facing high or extremely high water stress. Worldwide, we face a 1% rise in water demand every year, and the UN warns that “half the world’s population is already experiencing severe water scarcity at least one month a year.”
Yet almost a third of all drinking water, which is equal to the amount of all fresh water withdrawn annually in France, Germany, Italy, Spain and the UK combined, is lost in distribution every year because of outdated infrastructure.
That’s why we need to manage our water more efficiently.
It’s not going to be easy. In many parts of the world the water infrastructure – fresh and wastewater piping in particular – is aging, but it’s complex and costly to replace. London, for instance, famously still relies on Victorian- and Edwardian-era pipes for large parts of its 10,000-mile-long network. At current replacement rates, pipes in European countries that are put in today would have to last for the next 200 years.
We need new ideas, and that’s where artificial intelligence (AI) comes in.
Pinpointing leaks with AI
Digital technologies have not yet been widely adopted by the water sector, but two real-life examples show us how AI can already have a profound impact on the industry.
Swedish water company VA SYD supplies drinking water to more than half a million customers in Malmö, Sweden’s third-largest city, and the country’s Southwest Skåne region. The company used to lose about a tenth of its water. Although this was already much better than most water providers (and although Sweden is not a dry country), VA SYD was determined to do more to further reduce leakages across the network it operates, which consists of 5,000 kilometers of pipelines.
The company rolled out SIWA Leak Finder, which is an AI technology that analyzes water flow data to detect leaks as small as 0.25 liters and can reduce water losses from pipe leaks by up to 50 percent This focus on the smallest leaks might seem counterintuitive, but for most suppliers, water escaping through tiny gaps and cracks causes much bigger losses than the old pipe bursting. This is also true for VA SYD. By finding and fixing those small leaks, VA SYD reduced non-revenue water in the drinking water network from 10% down to 8%. “We really want to beat that. There are municipalities that achieve much lower figures, and we want to be able to measure up to the best,” says Simon Granath, Development Engineer at VA SYD. This AI-powered project has been recognized as a best practice in Sweden, leading other municipalities to adopt the same solution for water distribution.
Use a bucket – not a hose
If you visit the website of Yorkshire Water, a supplier in Northern England, you’ll be greeted by the words: “Use a bucket not a hose, so our water flows.” The slogan is testament that even in the UK, with its reputation for rainy weather, water is a precious resource.
It turns out that excessive rain can be part of the problem, but so can AI be part of the solution.
Yorkshire Water, like many other suppliers, purifies sewage from households and runoff water in dedicated plants before re-releasing it back into nature. When it rains heavily, these plants purposely release excess water and sewage to prevent flooding in public areas. But debris can block the release system. These blockages are hard to predict and are often only identified when the release valves fail and flooding occurs.
So Yorkshire Water installed a blockage-predicting AI model. Trained on data from thousands of sensors in the network, it detects unexpected flows and potential blockages before it’s too late. “The data has allowed us to identify problems with our network quickly, giving our teams the opportunity to attend to them before pollution occurs”, says Heather Sheffield, manager of operational planning and technology at Yorkshire Water. The AI identifies 90% of potential issues, which makes it three times more effective than traditional methods.
AI is thirsty – is that a fatal flaw?
AI might increase resource efficiency in the water sector, but there’s a problem: it also uses immense amounts of water, mostly to cool servers in huge data centers.
It’s fair to question whether the benefits outweigh this. My colleague Pina Schlombs, Sustainability Lead Siemens Digital Industries Software, agrees that both sides of the coin need to be considered. But her overall assessment is positive: “With a holistic perspective, we can see the sustainability benefits of the strategic use of AI outweighing the invested resources to train and run it. It's important to note that the outlook of AI computing efficiency is also increasing drastically, thanks to innovation already underway targeting LLM efficiency, how the infrastructure architecture is designed, the hardware the models run on, and more.”
The examples of VA SYD and Yorkshire Water show how AI can help to cut water loss and significantly reduce flooding risk without the need for costly and disruptive replacements of entire fresh and wastewater pipe networks.
So when it’s used, the impact of AI on water management can be profound, and its role is likely to continue to grow. If water companies start to make the most of the self-learning nature of AI-powered tools, the technology could even become an overarching monitor of the entire water cycle. For AI to be effective in this context, it must be industrial-grade—reliable, secure, and trustworthy; engineered to meet the rigorous standards of demanding environments.
Are you ready for the next wave of AI-powered water? To push forward this evolution at pace and scale, there needs to be an ecosystem of partners. The open digital business platform Siemens Xcelerator is at the forefront of this transformation, offering AI-driven applications to a broad ecosystem easier, faster and at scale.
