Deriving mixed reservoir operating rules for flood control based on weighted non-dominated sorting genetic algorithm II

The optimization of the mixed reservoir system, consisting of parallel and cascade reservoirs, is a complex issue due to its nonlinear storage curves, complex hydraulic connection and coordination. This paper aims to derive optimal operating rules for the flood control of large-scale mixed reservoir using a new algorithm known as weighted non-dominated sorting genetic algorithm II (WNSGA II), which could efficiently locate the Pareto front in the non-dominated region due to directed searching with the weighted crowding distance. A case study was performed in the Xijiang river basin in China, where there are eight flood control reservoirs and five flood control sections in four tributaries and the mainstream. Three scenarios, including conventional operating rules (COR), single-objective piecewise linear operating rules derived by genetic algorithm (PLO-GA), and multi-objective piecewise linear operating rules derived by WNSGA II (PLO-WNSGA II), are compared, and the resilience of three reservoir operating rules under inflow uncertainty is investigated as well. The results indicate that: (1) WNSGA II can locate non-dominated solutions more efficiently and provide better Pareto front than non-dominated sorting genetic algorithm II (NSGA II) due to the weighted crowding distance; (2) PLO-WNSGA II outperforms COR and PLO-GA for the control of basin-wide floods; and (3) PLO-WNSGA II is more resilient under inflow uncertainty. Therefore, PLO-WNSGA II is effective and efficient for the control of basin-wide floods in a mixed reservoir system.