CCWI2017: F19 'Multi-object approach for WSN Partitioning in the framework of Pressure Driven Analysis'

This paper proposes a novel methodology for WSN (water supply network) partitioning, made up of two serial steps and aimed at investigating leakage reduction benefits as a dual-use value of water network partitioning (WNP). Step 1 makes use of a spectral clustering algorithm to define the optimal layout of the districts, exploiting the properties of the connectivity matrix and giving a mathematical elegance to the arduous problem of optimal cluster definition. Taking the partitioning results of step 1 as granted, Step 2 deals with closure of isolation valves and installation of flow meters in the boundary pipes between either district. In the context of Step 2, a bi-objective optimization, aimed at maximizing the daily water volume supplied to the WSN users and at minimizing the leakage, is performed, in which network behavior is tested in PDA (pressure driven analysis). The applicability of the methodology is shown in a real case study.