Simulation-Optimization Framework for the Optimal Flood Mitigation Operation of Multi-Reservoir System

A novel multi-objective simulation-optimization (S-O) framework for flood mitigation in a river-reservoir system with multiple reservoirs and control points is proposed. Different combinations of objectives regarding flood mitigation in one or more of the reservoirs and/or one or more of the control points can be specified within the framework, namely, (1) maximizing the flood releases from the reservoirs to make space for the subsequent floods; (2) minimize the excess cumulative flow over the channel capacity at the control points; and (3) maximize the flood peak mitigation at the control points, with an option to constrain the upper limit of one or more of the end-of-horizon reservoir elevations. Thus, the S-O formulation aims to achieve simultaneous mitigation of the peak as well as the volume of flood at the control points, while increasing the reserve storage spaces at the different reservoirs for moderating the subsequent flood, which are conflicting objectives. The other constraints are concerned with the physical and the operational constraints of the river-reservoir system. The main feature is that the S-O framework arrives at the optimal operating strategy for the multi-reservoir flood mitigation operation without the use of any penalty or weighting functions in the formulation. The involvement of more than one control point in the flood damage mitigation makes the decision process complex due to their spatial locations and their respective priorities for flood mitigation, based on land use and development. The evolutionary search based non-dominated sorting genetic algorithm-II (NSGA-II) has been used as the solution technique. The objective functions proposed serve as surrogates for flood damage functions at the control points. Also, compromising solutions are obtained considering the conflicting objectives formulated.