Exploring Data-Driven Prediction and Risk Assessment of Pipe Failures in Water Distribution Networks

Licentiatavhandling, 2026

Access to safe drinking water is essential for public health and for the functioning of modern societies. Water distribution networks (WDNs) play a vital role in ensuring the reliable delivery of drinking water to consumers. However, ageing WDNs, increasing pipe failure rates, and constrained renewal budgets are placing growing pressure on water utilities and driving a shift from reactive maintenance towards more proactive asset management. Data-driven methods can support such proactive management by providing outputs that are meaningful to water utilities. This thesis focuses on the development and evaluation of practical data-driven methods for pipe failure prediction and risk-based prioritization in WDNs. Using data obtained from water utilities in Sweden, the thesis investigates how data-driven methods can be used to support utility decision-making. This is done by estimating the likelihood of pipe failures and integrating failure likelihood with the consequences of pipe failures to generate outputs that are useful for maintenance planning, renewal decisions, and risk reduction. In addition, the thesis explores how such methods can be used under real-world deployment conditions by predicting future pipe failures and comparing the predictions with failures observed in the following years.