Advances in extraneous water detection of urban sewer networks: From conventional methods to data-driven approaches

Extraneous water intrusion significantly undermines the performance of urban sewer systems by elevating treatment costs and increasing environmental risks of sewer overflow. This study proposes a systematic inflow and infiltration (I/I) detection framework, emphasizing the evolution from conventional inspection and modeling techniques toward intelligent, data-driven solutions. While traditional methods such as physical detection and chemical mass balance (CMB) provide foundational insights, they are limited by labor intensity, low scalability, and detection lag. In contrast, emerging approaches centered on digital twins, machine learning (ML), and edge-cloud architectures enable real-time monitoring, predictive analytics, and dynamic decision-making. A smart water system combining Internet of Things (IoT) sensing, geographic information system (GIS)-based assessment, and artificial intelligence (AI)-driven anomaly detection was demonstrated in a 63.5 km² sewer catchment in Guangzhou, China. On-site investigations verified the identified I/I sources, showing a 90% accuracy rate for the model-inferred sources. The results highlight key advancements in detection accuracy, system automation, and cost-effectiveness. This work offers a roadmap for advancing innovative drainage system detection and practicing sustainable urban water management.