Author(s): Duan Chen; Xinlong Deng
Linked Author(s): Duan Chen
Keywords: Reservoir Operation Inflow Ensemble Forecast Optimization Under Uncertainty
Abstract: Ensemble streamflow forecast (ESF) has provided more flexible information for reservoir operation than single streamflow forecast whereas the embedded uncertainty has not been fully considered. How to quantify the uncertainty of ESF and how to consider its propagation and control remains two critical research questions in the optimization of reservoir operation under ESF uncertainty. This study proposed an optimization model framework by integrating robust optimization concept and Karhunen-Loeve (KL) expansion method with attempt to provide a new routine of ESF uncertainty quantification and control in reservoir operation. First, the so-called uncertain set in the concept of robust optimization (RO) is used to describe the ESF uncertainty, rather than a predefined probability distribution that are often assumed in the routine of Stochastic Optimization. Then the KL expansion method is employed to quantify the uncertainty by preserving the statistical structure of the ESF with the combination of eigen values and eigen functions. Finally, the KL expansion is integrated in the routine of a multi-objective optimization model and provide the uncertainty set for each iteration of the optimization. The optimization model is then executed to find the the optimal reservoir outflow which maximize the operational objective but with the least variance in facing of multiple possible inflow scenarios of the ESF. The effectiveness of the proposed model is demonstrated by a case study of reservoir operation in Liuxihe basin, China and compared with a benchmark experiment that used extensive Monte Carlo simulation. The results showed that the proposed model provided an alternative for optimization of reservoir operation under ESF uncertainty with a straightforward framework and high computational tractability.
Year: 2025