An Evolutionary Algorithm For Deriving Optimal Operating Policy Under Uncertainties For Tehran Multi-Reservoir System

Projecting future inflows under climate change and rapidly growing population has large uncertainty and requires serious attention for proper utilization of limited water resources. existing algorithms can only optimize the operation policy for a specified scenario (e.g., drought, wet, or normal year; decreased or increased demands) and when established, the system would face serious operational difficulty if the expected scenario does not occur. on the other hand, most of water resources systems involve more than three objectives and demand proper techniques to handle computational complexities in so-called many-objective problems. this paper aims at providing a many-objective optimization algorithm using social choice (sc) and melody search (mes) algorithms that is able to efficiently derive general system operation rules suitable for all possible future scenarios. in other words, the proposed algorithm overcomes uncertainties in the occurrence of future scenarios and works optimally regardless of future conditions; whether it be variable streamflows and/or increased water demands. to evaluate the performance of the proposed algorithm, a system consisting of five reservoirs in the tehran region with four objective functions. it is shown that in all cases the general multi-scenario rule derived by the proposed method performs as good as each of the operation rules derived for every specific scenario assuming the occurrence of that scenario. moreover, the proposed many-objective algorithm is able to handle as many objectives as needed without any computational burden and/or algorithm complexity.