Multi-objective simulation-optimization for reservoir quality-quantity management with selective withdrawal and model Fusion

In this paper, a multi-objective simulation-optimization model for reservoir selective withdrawal considering different releases scenario and model Fusion method is developed, NSGA-II model, CE-QUAL-W2 model and validated model Fusion method are developed to determine the quality of reservoir releases from outlets. To generate possible reservoir releases as well as automatic running of calibrated CE-QUAL-W2 model for different possible scenarios, two codes are developed in MATLAB® software, separately. The main objectives function of this study are minimizing difference between reservoir water releases and different demands of stakeholders, maximizing Iran Water Quality Index of outlet water and the reservoir storage. Based on this methodology, the input-output data sets of simulation model are saved and utilized for training and validating the MLP, GRNN and RBF Meta models. The Ordered Weighted Average is used for fused the meta-models and for determining the model weighted is used Orness method. Then, the validated Meta model is linked to NSGA-II model to determine the objectives. For obtaining an appropriate solution considering different involved stakeholders on the pareto front, Nash theory and different social choice rules method are used. At the end, the fallback bargaining theory is used to choose the best reservoir selective withdrawal solution.