Predictive control of flow rates and concentrations in sewage transport and treatment systems

We design a predictive flow rate and concentration controller for wastewater transport and treatment networks. It manages flow rates to avoid overflows during times of high flow, and maximizes treatment efficiency when the system is within capacity limits. The underlying optimization is nonlinear due to the microbial growth kinetics and bilinear mass flows. Using a second-order cone relaxation of the microbial growth constraints and the alternating direction method of multipliers, we break down the problem into second-order cone and quadratic programs. This allows us to solve the problem at large scales in real-time. In a case study based on the wastewater transport and treatment system in the City of Paris, our controller outperforms the conventional flowrate-based controller by removing 13.7% more pollutant mass while treating the same amount of wastewater.