Multi-Objective VLGA Model for Flood Control and Water Supply Management in a Cascade System of Reservoirs

Developing a tradeoff between long-term objectives of reservoir operation such as water supply and short-term objectives like flood control and power generation is one of the challenging issues in reservoir operation management. In most of the monthly reservoir operation optimization models, flood control objective is only considered as a constraint for flood control storage and the probable flood damages are not usually taken into account. This can be a more important issue in cascade reservoirs, which provide more flexibility in achieving the long and short-term objectives. In this paper, a new approach for incorporating both flood control and water supply objectives in a monthly reservoir operation optimization model is presented. The optimization model is developed using Varying Chromosome Length Genetic Algorithm (VLGA). K-Nearest Neighbor (K-NN) is also used as a method for selecting the initial solutions considering similarities between reservoir inflows and storages in different water years. To incorporate the expected value of the flood damage in each month, peak flow frequency analysis has been carried out and the flood damages associated with each flood have been estimated by linking a hydraulic flood routing model with GIS tools for considering the land-use information in the flood plain. Damages due to deficit in supplying agricultural water demands are also calculated based on the functions of crop yield responses to deficit irrigation. The multi-objective optimization model is formulated to minimize the expected flood and agricultural water deficit losses, maximize reliability and resiliency, and minimize total deficit and surplus in supplying water demands. The developed model is applied to the cascade system of the Dez and Bakhtiari Reservoirs in southwest of Iran. The results of multi-objective model demonstrate efficiency of the model in finding pareto optimal solutions with more reliability and resiliency and less economic losses when compared with the single objective VLGA model.