Flood routing model for water resources optimization models based on a network flow

Real-time operations of multireservoir models should contain a flow routing procedure to predict the impacts of the observed and/or predicted inflows hydrographs on the downstream parts of the river system. Solving separately the hydraulic and operational constraints is only possible when the water resources management model is based on an evolutionary algorithm. Conceptual models for channel routing like Muskingum model can be imbedded within the mathematical formulation of the optimization algorithm when it is linear. Nevertheless, no channel routing model has been developed for algorithms based on network flow approach, whereas they are commonly used to manage operations of multi-reservoir. The purpose of this work is to propose a special network structure which allows the consideration of river routing. The routing model is considered as a network flow and the parameters are calibrated using a genetic algorithm. Simplicity and accuracy are the main reasons for this choice, which are relevant characteristics to be adopted in optimization models. The flood routing model is validated on the Wilson event, and the results obtained are compared to the Muskingum’s model results. A real example application on a reach of 40 km in the south west of France provided excellent results.