Optimizing Short-Term Operation of a Multireservoir System during Transition of Objectives and Constraints

Abstract

The goals of reservoir operation change periodically during a hydrological year due to varying considerations on human and ecological interests in different seasons. This requires adapting operation of reservoir in a short time frame (transition period) by shifting objectives and/or constraints of the reservoir. Determining operational strategies that can achieve optimal output on all the shifting objectives in the transition period is a major challenge of reservoir operation, in particular when it involves multiple reservoirs. This study took ten reservoirs on the Columbia River as a case and presented an optimization model that consider multiple shifting objectives and constraints for the multi-reservoir system during a two-week period. For the first week, the objectives of the reservoir system are fish migration and hydropower revenue. The objectives reduce to only maximizing hydropower revenue in the second week. The varied purposes of reservoir operation were combined in the proposed model as a multi-objective optimization problem, in which the different objectives and constraints were applied only to their corresponding period. An additional objective was considered in the model, which aims to meet corresponding historical water surface elevation (WSE) of the reservoirs at the end of the optimization period. A hybrid optimization method combining the Non-dominated Sorting Genetic Algorithm-II with local search method was used as a numerically efficient optimizer. The study showed that the proposed model can successfully achieve Paretooptimal solutions for the reservoir operation in the transition period of objectives and constraints. Most of the Pareto-optimal solutions were found superior on both fish migration and hydropower revenue objective, compared to the historical operation. The optimization result implied that satisfying fish migration requirement would decrease hydropower revenue. However, hydropower revenue can be largely increased by allowing a small deviation of the historical WSE at the end of the optimization period. It demonstrated that the proposed optimization model offered a more flexible context for decision making by fully investigating interaction between different objectives of the reservoir operation in the transition period.