Scenario Reduction for Long-term Hydropower Scheduling using Shape-based Block Decomposition

This paper proposes a framework consisting of a forepart precursor and the Scenario Fan Problem (SFP) for long-term hydropower scheduling, using shape-based feature extraction. The domain feature of the forepart precursor is scenario decomposition and scenario selection. Then, the selected scenarios will be the input of SFP, based on the feature of the new inflow scenario. Besides, we compared the performance of 5 clustering methods to generate the boundary variation on a hydro-thermal test case, using in-sample and out-of-sample analysis. Our results show that the proposed framework can significantly reduce the computational time of SFP by around 90%. The results also reveal that the shape-based method performs better regarding the boundary variation generation of the reservoir levels. In addition, this study demonstrates the possibility of performing long-term hydropower scheduling at a disaggregated level with detailed topology information and a higher time resolution. This provides more opportunities to deal with the penetration of Renewable Energy Sources (RES) in long-term hydropower scheduling and take into account more uncertainties in other variable renewable energy sources.