Impact of objective function selection on optimal placement of sensors in water distribution networks

Abstract

This paper investigates how the choice of the effectiveness-related objective function affects the results of the optimal placement of water quality sensors in water distribution networks (WDNs). The methodology adopted is based on bi-objective optimization, with the number of installed sensors as first objective function to be minimized. As for the choice of the second objective function, representative of the effectiveness of the monitoring system to react to a set of potential contamination events, two variants of optimization were considered: variant 1 - event detection likelihood to be maximized; variant 2 - average contaminated population to be minimized. The analysis of the results of the optimizations, which were re-evaluated in terms of four different system effectiveness indicators, proved that neither optimization variant is numerically superior. The choice of the objective function also impacts the physical placement of the sensors, with locations at maximum distance from the source(s) and more scattered over the layout for the two variants respectively.