Investigation of optimal sensor placement based on contaminant backtracking in water distribution systems

Abstract:

Nondominated Sorting Genetic Algorithm II was used to find the Pareto front between minimum overlap of possible detection times of two events and the best probability of detection. This methodology was applied to an example network for optimizing sensor placement in water distribution systems. The solutions obtained were then compared to the ones optimized by considering probability of detection and time to detection. The proposed method performs better than the benchmark method in detecting a contamination event and identifying its possible source.