Clustering-Based Maintenance Strategy of Isolation Valves in Water Distribution Networks

This paper presents a novel method to identify strategic valves of Water Distribution Networks (WDNs) to be prioritized for maintenance, as their guaranteed operability mitigates the impact of potential failures of other valves, by keeping an overall high level of WDN reliability. Notably, after clustering the dual segment/valve topology of the WDN, the novel method identifies the valves on the boundaries of the clusters as the strategic ones to be maintained always operable. The clustering algorithm utilizes a modularity formulation to minimize the number of boundary valves while maximizing the uniformity of demand across clusters. The methodology is applied to a real-world case study, and its performance is compared with a benchmark scenario, involving no valve maintenance, and with an engineering judgment-based maintenance approach targeting only valves on the boundaries of segments with large diameter pipes. Given the uncertainty on the location of failing valves and on the failure probability, various random multiple valve failure scenarios at different failure rates are generated to evaluate the maintenance strategy. The results show that the maintenance strategy is capable of defining a small subset of strategic valves to significantly reduce the impact of valve failures on WDN reliability, while outperforming the engineering judgment-based maintenance approach.