Efficient management of a sewer system includes the control of the conveyed wastewater quality to adequately operate treatment plants and protect the receiving water bodies. Moreover, these systems are vulnerable to either accidental spills or intentional unauthorized discharges. To properly manage them, a limited number of sensors could be placed at different locations to monitor the water quality. In this paper, multiobjective and single-objective optimization procedures to optimally locate sensors in sewer systems are proposed, tested, and compared. The multiobjective procedures include objective functions related to information theory (IT procedure), detection time and reliability (DR procedure), and a combination of them (IT_DR procedure). The single-objective procedures include a greedy-based objective function (GR procedure) and a merged objective function (DR_IT_GR procedure). The procedures show a similar performance when applied on a small network, whereas in a real system, the results show that (1) the IT-based method can be effectively used as a filtering technique(2) the DR_IT_GR procedure outperforms the other multiobjective onesand (3) the GR procedure is very efficient in finding the Pareto extreme solutions.

A core problem associated with the water quality monitoring in the sewer system is the optimal placement of a limited number of monitoring sites. A methodology is provided for optimally design water quality monitoring stations in sewer networks. The methodology is based on information theory, formulated as a multi-objective optimization problem and solved using NSGA-II. Computer code is written to estimate two entropy quantities, namely Joint Entropy, a measure of information content, and Total Correlation, a measure of redundancy, which are maximized and minimized, respectively. The test on a real sewer network suggests the effectiveness of the proposed methodology.