Abstract: This study develops a multi-objective optimal drawdown operation model of multiple reservoirs by considering the demands of the river’s natural flow patterns and using three objective functions－the maximum hydropower output, maximum water supply reliability, and minimum flow alteration. This model is solved using the fast nondominated sorting genetic algorithm-II (NSGA-II), and applied in a case study of multiple reservoirs in the Xiangjiang River basin: the Dongjiang, Shuangpai, and Centianhe reservoirs. The results show that relative to conventional operation in the wet, normal and dry year cases, the maximum hydropower output scheme raises the power generation by 15.71 million kW?h (11.2%), 11.67 million kW?h (10.9%), and 8.11 million kW?h (10.0%), respectively; The minimum flow alteration scheme reduces river flow deviation by 1.56 (61.2%), 0.33 (6.3%), and 0.89 (13.6%), respectively; The maximum water supply reliability scheme promotes water supply satisfaction by 2.1%, 2.6%, and 3.4%, respectively. Our new model proves applicable and effective and would be useful for the drawdown operation of multi-reservoir systems in the Yangtze River basin.

Key words: drawdown operation, multi-objective optimization, fast nondominated sorting genetic algorithm-II, decision analysis, Xiangjiang River basin