Swift hydraulic models for real-time control applications in sewer networks

Abstract

Real-time control (RTC) is an important tool for safe and cost-effective operations of sewer systems by, for example, reducing sewer overflow or enhancing sulfide mitigation. Due to the long transport time of sewage and the inherent dynamics in sewage flow rates, model-predictive control is often needed, which requires the prediction of sewage hydraulic characteristics across the network. The full hydraulic models are often unsuitable for such purposes due to their high computational demands, which are not affordable as the models need to be called for numerous times in each optimisation step. In this study, two swift, data-driven hydraulic models are developed to predict sewage flow rates in gravity sewers receiving feeds from rising main(s) and gravity main(s), respectively. The models are shown to be able to predict both the sewage flow rate and the cross-sectional flow area in high fidelities with solutions of Saint-Venant Equations, but reduce the computational time by up to four orders of magnitude. The swift hydraulic models were then integrated into an RTC strategy for NaOH dosing in a simulated real-life sewer network, and achieved cost-effective control of sulfide. These models could potentially be used for other sewer RTC applications.

Keywords: Chemical dosing; Data-driven model; Gravity sewer; Hydraulic models; Real-time control; Sewer network.