Determining dynamic water level control boundaries for a multi-reservoir system during flood seasons with considering channel storage

The dynamic control of reservoir water level during flood seasons (WLFS) allows a trade-off between flood control and water resources conservation. In this process, the WLFS dynamic control boundaries are key parameters. The channel flood routing affects the dynamic control operations. However, flood routing is seldom considered in multi-reservoir aggregation (a popular method for tackling the curse of dimensionality in multi-reservoir issues). To address this knowledge gap, this study conceptualises channel flood routing as a virtual channel storage reservoir (CSR) based on Muskingum flood routing in the aggregation method. The dynamic control boundaries for the multi-reservoir WLFS are then derived using the following aggregation-decomposition method: (a) the aggregation reservoir is created by considering channel storages, and the flood risk is quantified using pre-discharge rules; and (b) the aggregated reservoir storages are decomposed by maximising hydropower generation with constraints. The Xiluodu, Xiangjiaba and Three Gorges reservoir system on the Yangtze River, China, were selected for a case study. The annual power generation is 175.40 billion kWh when using static operation rules. With dynamic operation, the output would be 177.05 billion kWh if CSRs were not considered and 177.53 billion kWh if CSRs were considered. Thus, the proposed method can efficiently determine the dynamic control boundaries for the multi-reservoir WLFS.