Abstract: To obtain multi-objective optimal operation schemes of reservoirs under the condition of uncertain inflow runoff, this study examines the Three Gorges reservoir with a combined storage capacity for flood control and power generation. The forecast error of reservoir inflow process is quantified and the principle of benefit maximization by coupling output and cost isolines in microeconomics is applied. Then, the marginal substitution rate is used in multi-objective reservoir optimization and decision to seek the best point balancing flood control with power generation. The results show that this model can directly describe the mutual feedback transformation relationship between flood control benefit and power generation benefit under different inflow scenarios, and it also obtains the optimal reservoir operation stage for the early flood period when the inflow forecasting error is considered, theoretically and practically helping make full use of reservoir benefits while ensuring flood control safety.

Key words: forecast error, reservoir, multi-objective optimal operation, scheme decision, marginal substitution rate, best balance point, mutual feedback relationship