Improving Burst Detection Performance of Small-Scale Areas in Water Distribution Networks with State Estimation Methods

Pipe bursts caused by corrosion, ground movement, or pressure surges can lead to severe water loss and structural damage, while intelligent detection methods help mitigate these risks by enabling more efficient repairs. The evolution of monitoring technologies has facilitated more densely distributed flow measurement, allowing for management in smaller areas, and posing new challenges for existing burst detection methods. A smaller number of customers will magnify the impact of unpredictable random water usage, which further leads to fluctuations in measurements and causes false alarms. On the other hand, pipes serving smaller areas exhibit relatively lower burst flow rates and are less destructive within a short time, creating a detectable window for differentiating true bursts from noise. This study proposes a burst state estimation (BSE) framework that strategically balances detection timeliness and false alarm rates by integrating the density-based spatial clustering of applications with noise (DBSCAN) algorithm with the Kalman filter state estimation method. Through case studies of three real-world small-scale areas with burst repair records, the proposed BSE method shows superior performance by achieving a 79%–90% reduction in false alarms at equivalent detection rates compared to traditional methods that are simply based on anomaly detection, demonstrating the effectiveness and practical feasibility of the method.