A Novel Stochastic Approach for Optimization of Diversion System Dimension by Considering Hydrological and Hydraulic Uncertainties

This study proposes a new stochastic approach for optimizing diversion system design and its construction schedule by considering different hydrological and hydraulic uncertainties sources. For this purpose, a multi-objective optimization-simulation model was developed to evaluate the failure of a diversion system to flood. Two objective functions, the expected flood damage (EFD) and cost-benefit (CB) index of a diversion system, are optimized in this study using a non-dominated sorting genetic algorithm II (NSGA-II). The approach is tested for four different compositions of uncertainties (Base Case, Case1, Case2, and Case3) to estimate their impacts based on distance index (D) and the boxplot. Finally, finance constraints are evaluated based on the construction period of the project. The Karun-4 dam, located in Iran, is considered as the case study. The obtained results demonstrate that the hydrological uncertainty with $${D}_{case2}^{basecase}=21.335$$ and $${IQR}_{basecase}=2.1M$$ has the highest effect on the Pareto optimal front and the hydraulic uncertainty of downstream cofferdam with $${D}_{case3}^{basecase}=5.789$$ and $${IQR}_{case2}=1.8M$$ has the lowest effect on the Pareto optimal front. The best value of the CB index is related to the base case (66.42%) using the pseudo weight factor. The study indicates that the total investment of the water diversion system is lower than the consultant's plan by 20.23%, 18.33%, 17.28%, and 18.81% when the different components of uncertainty are considered. An implementation period of 6-year and 11-year is the best option for no financial constraints and financial constraints, respectively. The stochastic simulation-optimization approach proposed in the present study provides decision-makers reliable insight into planning dam construction.