Real-Time Multistep Prediction of Sewer Flow for Online Chemical Dosing Control

Abstract

Chemical dosing is the most common strategy for sulfide control in sewers. Recent research has shown that online control of chemical dosing can significantly reduce dosing costs, while achieving better control performance. One of the bottlenecks of online control is the prediction of sewage retention time in sewers, governed by future sewage flows. This study developed a methodology for real-time future flow prediction in sewers based on autoregressive moving average (ARMA) models and multistep iterative prediction. This methodology was validated with flow data collected from two pumping stations with different flow characteristics and different wet-well storage capacities. The results showed that the proposed methodology was capable of predicting future flow rates with good accuracy under different weather conditions. Online control of chemical dosing with real-time sewer flow prediction was tested through a simulation study. Results showed that future flow prediction improved sulfide control and significantly reduced chemical dosage.