Model Predictive Control for Stormwater Reservoirs: Investigating Effects of Climate Change and Urbanization

Today's stormwater systems are not designed for the adverse effects of climate change and deep urbanization. These effects can increase the magnitude of floods and cause significant damages and fatalities. A typical solution for flood control in urban areas is the implementation of stormwater reservoirs. These reservoirs typically require space which can be very costly in densely populated cities. In contrast to this investment-heavy solution, retrofitting current systems with advanced flow sensors and controllers (i.e., orifice valves and variable speed pumps) can provide alternatives to flood management by allowing optimal, real-time control. In this paper, we investigate the performance of controlled and uncontrolled stormwater reservoirs against combined Scenarios of urbanization and climate change. To this end, we develop a novel state-space, dynamic model representing watersheds, reservoirs, and one-dimensional channels hydrodynamic processes. We also develop a model predictive control (MPC) algorithm to optimally determine valve operations in stormwater reservoirs, by focusing on decreasing the flood effects in downstream channels. Results reveal that MPC yields significantly better flood control, compared to passive, model-unaware control strategies. The results illustrate the need for more model-driven control algorithms for climate change-aware control of stormwater systems.