Optimization of sensor locations for contaminant detection in water distribution networks

A method for the optimization of sensor locations in water distribution networks is presented with respect to effective and efficient detection of contaminations. The optimization problem is formulated as a twin-objective minimization problem with the objectives being the sensor cost and the risk of contamination. Unlike past approaches, the risk of contamination is explicitly evaluated as the product of the non-detection probability of an intrusion by a given set of sensors and the consequence of that failure (expressed as effected population). An Importance-based Sampling Method is presented and used to effectively determine the relative importance of contamination events, thus reducing the overall computation time. The above problem is solved by using the Non-dominated Sorting Genetic Algorithm II (NSGA-II). The methodology is tested on a case study involving the water distribution system of Almelo (Netherlands) and the potential intrusion of E. coli bacteria. The results obtained show that the algorithm is capable of efficiently solving the above problem. The estimated Pareto front suggests that a reasonable level of contaminant protection can be achieved using a small number of strategically located sensors.