Optimal Pressure Sensor Placement in Water Distribution Systems via coupling Geostatistical Tools with Genetic Algorithm

This paper intends to address the optimization problem and the solution process for design of pressure sensor placement to make the estimation of the pressure field more accurate and efficient in the WDS. In this work, the major contribution will be to define and quantify an objective function and ways to optimize it. This entails both the accuracy of the mean pressure estimation in a direct way and the economic cost of the data collection implicitly if not explicitly. To solve this problem, a well-known geostatistical variance-based reduction method so called variance of residuals computed via block ordinary kriging is successfully used. The proposed methodology is implemented on a literature test case of Anytown and a real-life system so called C-Town WDSs. The results show that for both systems, a geostatistical tool coupled with a random search optimizer could clearly be an appropriate and efficient approach to build the decaying function of block variance of residuals versus the number of sensors which can also help determining the minimum number of sensors necessary to satisfy the desired accuracy and their optimal placement.

کلیدواژه ها:

Water distribution systems ، Sensor placement ، Variogram ، Block ordinary kriging.