Multi-Objective Optimal Operation of a Large Deep Reservoir during Storage Period Considering the Outflow-Temperature Demand based on NSGA-II

Abstract Thermal stratification occurs in deep reservoirs and causes an outflow temperature that is significantly different from the inflow temperature, which is harmful to the ecology of downstream reaches. Optimal operation of reservoirs using the multi-objective evolutionary algorithm (MOEA), which focuses on power generation, flood control, agricultural irrigation, navigation, etc., has been performed. However, the outflow temperature has not been considered. Focusing on this deficiency, we considered Sanbanxi Reservoir as a case study and used the non-dominated sorting genetic algorithm II (NSGA-II) by combining it with the calibrated 1-D General Lake Model. The trade-off relationships among three objectives, namely, total power generation (TPG), root-mean-square deviation between the inflow and outflow temperatures (RMSDtem) and weighted aggregate eco-index (EI) based on indicators of hydrologic alteration (IHAs) parameters, were estimated as Pareto frontiers during the storage period from March 15 to July 15. The results demonstrated the following. 1) Through reservoir re-operation, RMSDtem could be reduced by 1.60 °C from 5.99 to 4.39 °C at the expense of reducing TPG from 7.43 × 108 to 6.56 × 108 kW·h (–11.8%) during a normal year. 2) The optimized operating schemes were extracted as Pareto frontiers, and the slope of the outflow discharge curve (SODC) could be a useful parameter for determining the operating scheme in practice. A large SODC yielded high TPG, a small SODC resulted in low RMSDtem, and a medium SODC yielded minimum EI. The medium SODCs ranged from –0.1 to 0.4 m3/(s·d), –0.4 to 0.6 m3/(s·d), and –0.6 to 1.2 m3/(s·d) for dry, normal, and wet years, respectively. 3) As the yearly inflow increased, the absolute value of the SODC for the best individuals for a single objective became larger. The range of TPG and RMSDtem in the Pareto frontier expanded, and that of EI remained basically unchanged. The results of the Sanbanxi Reservoir can be applied to other reservoirs to support reservoir re-operation, especially for the deep reservoirs in southwest China that suffer from outflow-temperature problems.