Sensor location in water distribution networks to detect contamination events - A multiobjective approach based on NSGA-II

Abstract

The distribution network is the most exposed part of water supply systems due to the large number and geographical dispersion of derivation nodes and access points. Therefore, a reliable monitoring and surveillance system based on a sensor network is necessary to timely detect contamination events. The sensor location problem in water distribution networks to detect (accidental or intentional) contamination events has been tackled by optimization approaches aimed to determine the best location for a set of sensors, thus allowing the management entity to detect those events in a short period of time and be able to minimize their impact on the population served. This paper presents a multiobjective evolutionary approach to determine the location of sensors in a water distribution network to detect a contamination event and minimize its potential consequences according to multiple, incommensurate and conflicting evaluation aspects of the merits of each solution. The objective functions are the expected time of detection, the expected population affected prior to detection, the expected consumption of contaminated water prior to detection, and the detection likelihood. A set of nondominated solutions representing the Pareto front is obtained, which have been validated with known solutions for the case studies. Further, this information enables to exploit tradeoffs and identify good compromise solutions according to the decision maker's preferences.