Optimization of Sensor Selection and Placement for State Estimation in a Water Distribution Systems

Abstract Submission: Accurate and comprehensive information on water distribution systems (WDS) is essential for their effective operation and monitoring. Pressure and flow are key state variables in WDS management, often simulated using software like EPANET. These models typically assume perfect knowledge of system parameters, which is rarely the case in practice. To address this challenge, state estimation techniques can be employed to integrate information from hydraulic models with sensor measurements, providing a more accurate representation of real-time system dynamics. However, installing sensors throughout an entire WDS is impractical due to financial and logistical constraints. Therefore, it is essential to optimize the type, number, and placement of sensors to maximize their effectiveness under constrained conditions. We propose an approach that integrates greedy optimization with the Kalman Filter for state estimation, selecting the optimal placement of sensors, including both pressure sensors and flowmeters, to achieve satisfactory estimation accuracy. We evaluate the effectiveness of combining pressure sensors and flow sensors, comparing their performance to that of using a single sensor type. The analysis is demonstrated on a WDS serving a small remote community in Alaska, where we assess state estimation performance across various sensor configurations. This study provides guidance on strategic sensor selection and placement, offering insights for improving the observability and operational efficiency of WDSs.

Learning Objectives/Expected Outcome (Optional) :