Risk Analysis for Reservoir Real-Time Optimal Operation Using the Scenario Tree-Based Stochastic Optimization Method

The inherent uncertainty of inflow forecasts hinders the reservoir real-time optimal operation. This paper proposes a risk analysis model for reservoir real-time optimal operation using the scenario tree-based stochastic optimization method. We quantify the probability distribution of inflow forecast uncertainty by developing the relationship between two forecast accuracy metrics and the standard deviation of relative forecast error. An inflow scenario tree is generated via Monte Carlo simulation to represent the uncertain inflow forecasts. We establish a scenario tree-based stochastic optimization model to explicitly incorporate inflow forecast uncertainty into the stochastic optimization process. We develop a risk analysis model based on the principle of maximum entropy (POME) to evaluate the uncertainty propagation process from flood forecasts to optimal operation. We apply the proposed methodology to a flood control system in the Daduhe River Basin, China. In addition, numerical experiments are carried out to investigate the effect of two different forecast accuracy metrics and different forecast accuracy levels on reservoir optimal flood control operation as well as risk analysis. The results indicate that the proposed methods can provide decision-makers with valuable risk information for guiding reservoir real-time optimal operation and enable risk-informed decisions to be made with higher reliabilities