Real-time model predictive and rule-based control with green infrastructures to reduce combined sewer overflows

Abstract

The impact of integrating large-scale distribution of green infrastructures (GIs) with different real-time control strategies on combined sewer overflows (CSOs) is assessed for the southern area of the City of Montreal's combined sewer system (Canada). An iterative process involving a synthetic design rainfall event and model predictive control (MPC) of the sewer system is developed to distribute GIs according to cost-efficiency and spatial analysis criteria. The distributed GIs are alternatively integrated with static, rule-based control (RBC) and MPC, for which model simulations are performed for a two-month period. The performance of the three strategies is compared in terms of CSO volume and frequency reductions, fulfillment of the outfall environmental priorities and transfer of runoff capture to CSO volume reduction. A gradual increase in GI implementation levels and an alternative scenario of GIs distribution are also considered to assess the performance of the two real-time control (RTC) strategies. By comparing the scenarios where GIs are uniformly distributed with those where no GIs are implemented and omitting the most extreme rainfall event, average CSO volume reduction is about 65%, 82% and 92%, respectively, for static control, RBC and MPC. Moreover, the scenario integrating GIs with MPC is the only one permitting to avoid almost all CSO events and the fulfilment of the outfall environmental priorities. GIs efficiency performance (the transferability between global runoff capture and CSO volume reduction) is also the highest under MPC, even when considering varying GI implementation levels and spatial distribution schemes.

Keywords: Low impact development; Model predictive control; Real-time control; Rule-based control; Source control measure; Urban drainage modelling.