Edge Hill University and SME Water Limited KTP 24_25 R1

Water companies lose millions of litres of water every day through leaks they cannot see in time to stop them. This project builds a machine learning platform that automatically captures, analyses, and summarises transient pressure data from water pipes—the sudden spikes and drops that often signal a developing leak before it becomes a burst. Currently, engineers must manually sift through vast streams of pressure readings, a slow process that misses many early warning signs. The platform will replace that manual work with rapid, automated detection and reporting. If it succeeds, water companies could spot leaks hours or days earlier than they do now, reducing water loss, cutting repair costs, and minimising disruption to roads and homes. The work is applied, not fundamental science: it takes existing machine learning techniques and adapts them to a specific, messy real-world data problem. The direct impact is on infrastructure maintenance, not on daily life in a visible way—but every avoided burst means fewer road closures, less wasted water, and lower bills for customers.