Quantifying the contributions of multi-source uncertainties in dynamic soil-pile-nuclear power plant structure response via probability density evolution method

Soil–pile–nuclear power plant (SPNPP) structures on non-bedrock sites exhibit complex seismic behaviors due to inherent uncertainties in ground motion, geotechnical properties, and structural materials. However, comprehensive uncertainty quantification remains limited. This work employs the Probability Density Evolution Method (PDEM), integrated with the Change of Probability Measure and Fréchet-derivative-based sensitivity analysis, to systematically assess the stochastic dynamic response of SPNPP structures. Taking the HPR1000 nuclear power plant as an engineering case study, a refined three-dimensional (3D) finite element model of the SPNPP system is established. The effects of 25 uncertain parameters, covering three uncertain sources (ground motion, geotechnical materials, and structural materials), are systematically analyzed. Based on 900 dynamic time-history simulations, the relative influence of each parameter is quantified. Results indicate that when all uncertainty sources are considered simultaneously, the coefficient of variation (COV) of the SPNPP structural response increases significantly. Ground motion uncertainty contributes most substantially to the overall structural response variation. For the top drift ratio response of the containment structure, uncertainties in soil shear wave velocity and structural elastic modulus both play key roles. Global sensitivity analysis (GSA) results for the 25 uncertain parameters are also presented. This study provides a practical framework for uncertainty quantification and performance-based optimization in seismic design of nuclear power structures on non-bedrock sites.