A multi-objective optimization method based on NSGA-III for water quality sensor placement with the aim of reducing potential contamination of important nodes. Issue 1 (14th July 2021)

• Record Type: Journal Article
• Authors: Jafari, Hamideh; Nazif, Sara; Rajaee, Taher
• Is Part Of: Water Supply. Volume 22:Issue 1(2022)
• Page Start: 928
• Page End: 944
• Publication Date: 2021-07-14
• Subjects: contaminant detection -- contamination of important nodes -- critical damage -- NSGA-III -- water quality sensor placement strategy
• DOI: 10.2166/ws.2021.222
• Languages: English
• ISSNs: 1606-9749

Abstract

Abstract: One way to mitigate the risk of consumption of contaminated water in water distribution networks is optimal placement of the quality sensors. A considerable challenge in this respect is the significance of contamination at a junction. Beside the population affected and the volume of the contaminated water consumed, importance of each junction is a parameter that must be taken into account in placing the sensors. This parameter directly concerns the service provided by each junction as well as the sensitivity and social consequences of junction contamination. The present study defines a new objective function for minimizing the effect of junction contamination with respect to its importance. Using a robust approach, this study applied the NSGA-III algorithm to solve a 5-objective problem. The algorithm was tested on a hypothetical network and a benchmark network and the Pareto response was selected for each scenario based on the slope of the different points. The proposed method suggested 12, 12, and 11 sensors for the three scenarios in the hypothetical network. The results show that sensor placement by this method yielded good performance in comparison with the other solutions presented in a benchmark network. HIGHLIGHTS: A NSGA-III algorithm for five-objective sensor placement based on the robust optimization models is proposed. By employing two typical water distribution networks, the benefits of the proposed algorithm are illustrated and some critical factors are also addressed. A new objective function describes the impact of contamination of important nodes. Uncertainty at the point of entry and the entry time of the contamination are considered.