Hydraulic model based sensor placement, predicting sensor failures, and localizing background leaks in water distribution systems

This research addresses critical challenges in water distribution systems, focusing on leakage management, optimal sensor placement, and predictive maintenance of sensors. Leakage is a significant issue for water utilities, with background leaks being particularly difficult to detect due to large energy loss perturbations when compared to large leaks in water networks. The limited number of on-field sensors, driven by cost constraints, further complicates localization efforts. To tackle this, a novel hydraulic model-based iterative clustering technique is developed to effectively localize background leaks and isolate leaky sections of the network. Additionally, the study proposes a robust and resilient graph-based solution for optimal sensor placement, ensuring that limited financial resources are utilized efficiently to support digital water solutions such as calibration, leak detection and also daily operational needs. This approach enhances the ability of water utilities to monitor and manage their networks effectively. Finally, recognizing the importance of maintaining sensor functionality, a knowledge-based, data-driven solution is developed to predict sensor failures. This predictive maintenance strategy enables water utilities to proactively address anomalies, optimize maintenance schedules, and improve planning, ultimately ensuring the reliability and efficiency of water distribution systems. Together, the developed methodologies provide a comprehensive framework for enhancing the resilience and sustainability of water networks.