Evolutionary network flow models for obtaining operation rules in multi-reservoir water systems

Abstract

Obtaining operation rules (OR) for multi-reservoir water systems through optimization and simulation processes has been an intensely studied topic. However, an innovative approach for the integration of two approaches – network flow simulation models and evolutionary multi-objective optimization (EMO) – is proposed for obtaining the operation rules for integrated water resource management (IWRM). This paper demonstrates a methodology based on the coupling of an EMO algorithm (NSGA-II or Non-dominated Sorting Genetic Algorithm) with an existing water resources allocation simulation network flow model (SIMGES). The implementation is made for a real case study, the Mijares River basin (Spain) which is characterized by severe drought events, a very traditional water rights system and its historical implementation of the conjunctive use of surface and ground water. The established operation rules aim to minimize the maximum deficit in the short term without compromising the maximum deficits in the long term. This research proves the utility of the proposed methodology by coupling NSGA-II and SIMGES to find the optimal reservoir operation rules in multi-reservoir water systems.