Diversion risk prediction for pumped storage power plants under extreme rainfall: a case study

Construction diversion risk is critical for the safety of pumped storage power plants (PSPPs), particularly under extreme rainfall when the upper and lower reservoirs are constructed simultaneously. Conventional risk assessments often neglect spatial dependence between adjacent reservoirs and may therefore underestimate the overall system risk. Using the Qiannan PSPP in China as a case study, 60 years of precipitation and runoff data were analysed. A Gumbel-Hougaard copula was selected to model the dependence structure between extreme rainfall at the upper reservoir and flood volume at the lower reservoir. Monte Carlo simulation was then employed to generate correlated construction floods and diversion risk was quantified for individual structures and the overall system. The results revealed significant risk superposition at the system level, where the integrated diversion risk substantially exceeds the sum of individual structural risks. Accounting for spatial dependence amplified the lower reservoir’s risk by up to 25.93%, demonstrating that univariate approaches assuming hydrological independence systematically underestimate true system risk. This case study application framework provides a practical and replicable tool for construction diversion design and flood risk management in rainfall-sensitive mountainous PSPPs.