Coordinated Decentralization-Based Optimization of Disinfectant Dosing in Large-Scale Water Distribution Networks

Abstract

Disinfectants are typically used in water distribution networks (WDNs) to maintain the microbiological quality of potable water throughout the network. Disinfectant residual levels in such large distribution networks need to be maintained within the prescribed bounds to address two major problems: (1) preventing microbial regrowth, and (2) minimizing harmful disinfection by-product (DBP) formation resulting from high levels of the disinfectant itself. These requirements pose a two-point constrained control problem of disinfectant residual levels. Another important aspect of consideration is that such WDNs exhibit spatial and temporal variations in water quality at different points in the network. Therefore, conventional systems engineering tools such as modeling, optimization, and control need to be adapted to accommodate these variations for water quality management in distribution networks. In this paper, we propose a novel zone control based approach to address the two-point quality control problem. Further, to combat the spatial complexity, we combine the zone control approach with a decentralized optimization strategy, namely, the model coordination method, which has been widely proposed to solve large-scale optimization problems. The resulting novel formulation is proposed here to address the aforementioned problems in operational optimization of booster chlorination water distribution systems by using a realistic multispecies model developed from a previous study. We demonstrate the efficacy of this approach on a real-scale distribution network of Cherry Hill Brushy Plains (CH/BP) (EPANET Example network 2).