Real-Time River Flood Control under Historical and Future Climatic Conditions: Flanders Case Study

Abstract

Model predictive control (MPC) has shown to be an efficient technique for real-time flood control. The evaluation of the control performance is, however, typically restricted to a limited set of flood events. In this paper, the control performance is evaluated for a long-term time series of 116 years of meteorological data as well as after climate scenarios. Such an evaluation became feasible thanks to the use of a computationally efficient MPC approach based on a fast conceptual river model and an adapted genetic algorithm. The uncertainties related to the river model and the rainfall forecasts were accounted for. The influence of these uncertainties on the MPC control performance was, however, found to be limited after applying data assimilation. Comparing the proposed MPC approach to a standard programmable logic control (PLC)-based regulation shows that – despite the presence of uncertainties – MPC outperforms the PLC-based approach because it strongly reduces the incurred damage cost, the flood risk, and the frequency of flooding. This is still the case after considering the climate scenarios.