The Drinking Water Inspectorate is expected to legislate the accelerated removal of lead pipes from water networks in the coming years. Approximately 95% of the UK’s drinking water network is phosphate-dosed, which reduces the leaching of lead into drinking water, and there are high levels of uncertainty about the location of lead service pipes.
Using machine learning and known pipe and water quality samples, predictions can be made across whole water networks to target properties and optimise lead pipe replacement programmes. The predictive model also allows us to systematically review each property against the available data and to respond quickly to new rules and data assumptions. This enables water companies to identify those most at risk to lead exposure and to identify opportunities for programme efficiency, where multiple pipes can be replaced at one time.
Our lead service pipe identification tool was originally developed for a US municipality, where we used the predictive model to prioritise our efforts to find and replace lead pipes after it was estimated that 64,000 – 84,000 lead service line replacements were needed. We are currently working with South West Water on a pilot project to develop and test the model in the UK.