Author(s): Ashley Zhang; Zheng Yi Wu; Alvin Chew; Fred Cao
Linked Author(s): Hui Zhang, Alvin Chew
Keywords: Smart water grids optimization sensor placement water distribution networks
Abstract: Collecting field data is essential to manage and operate water distribution systems, especially for smart water grid management and analytics in real time. Pressure data can be used not only for monitoring but also for anomaly detection and localization, hydraulic model calibration, etc. To maximize the value of pressure data and meanwhile save cost, it is critical to optimize how many and where to place pressure sensors in the network. This work presents an improved method, which has been implemented as a software tool, to optimize pressure sensor placement based on the nodal pressure-drop sensitivity to emulated leaks. Built on the previous research and development, the method and software tool have been improved to enable users to optimize the pressure sensor placement for (1) the selected zones in a large system; (2) better coverage on leakage-prone areas such as the aging pipes and large water mains; and (3) additional sensors on top of existing sensors in the system. Each leakage event is generated by Monte Carlo sampling on junction nodes with user-defined ranges of number of leaks and emitter coefficients. Hydraulic simulation is performed for each event to produce a database with candidate sensor junctions and their individual detection status. A high-performance optimization framework is utilized to maximize leakage detection coverage. The optimization is speeded by parallel computing via Message Passing Interface (MPI) to efficiently handle large hydraulic models and large amounts of sensors. The effectiveness of the method is demonstrated with a case study on a large network.
Year: 2025