Search

Blockage, which arises from accumulated materials like sand, stones, sediment, uncontrolled growth of tree roots within pipes, and the inadvertent partial closure of in-line valves, poses a pervasive challenge in the operational sphere of water distribution networks (WDNs). This results in significant head loss in the pipes, leading to disruptions in the water supply. Identifying such blockages conventionally requires specialized human resources and is often encumbered by cost and time challenges. This paper introduces a method for precisely determining the location and magnitude of blockages within WDNs. This methodology is firmly grounded in a calibration-optimization framework, harnessing a specially crafted hybrid optimization algorithm known as PSOHS, which combines the strengths of particle swarm optimization (PSO) and harmony search (HS) algorithms. Detailed comparisons are drawn between simulated pressure data (derived from the hydraulic network model) and corresponding field data at multiple points to minimize disparities, thereby facilitating the precise identification of potential blockages in a WDN. The proposed method undergoes rigorous evaluation across two benchmark networks (modified Poulakis and Hanoi) under diverse scenarios as well as on an actual WDN. Results from over 60,000 simulated blockage scenarios underscore the method's remarkable proficiency, achieving blockage localization accuracy of 99.9% and magnitude estimation with a maximum absolute error (MAE) of less than 3%. Particularly noteworthy is the superior performance of the PSOHS algorithm as proposed, surpassing the PSO algorithm by enhancing blockage localization accuracy by a minimum of 10.5% and reducing MAE by 3%. Further, adopting the proposed algorithm over the HS algorithm yields improvements of 27.5% and 3.5%, respectively. Applying the proposed method in an actual network effectively pinpointed its blockage. Consequently, this algorithm presents a valuable tool for WDN operators, augmenting their capacity for effective network management. [ABSTRACT FROM AUTHOR]