Time-variant seismic fragility analysis framework for arch dams considering material deterioration and stochastic ground motions

To address the high computational cost associated with time-variant seismic reliability analysis of large-scale structures, this study proposes an efficient probabilistic framework for lifecycle seismic fragility assessment of arch dams considering material deterioration and stochastic ground motions. A velocity divergence–based generalized adaptive probability density evolution method (VD-GAPDEM) is employed to evaluate the probabilistic evolution of nonlinear dynamic systems with time-dependent material properties. To improve computational efficiency, a representative point selection strategy based on the generalized F-discrepancy measure and iso-probability mapping is developed, together with an efficient numerical scheme combining smoothed-kernel estimation and second-order Runge–Kutta integration. The initial probabilistic state for lifecycle analysis is established through seismic-response-induced probability evolution, enabling consistent coupling between seismic loading and long-term material degradation. The proposed framework is validated using a deteriorating beam example, demonstrating high accuracy with significantly reduced computational effort compared with Monte Carlo simulation (MCS) and the probability density evolution method (PDEM). It is further applied to the Baihetan Arch Dam to perform lifecycle seismic fragility analysis incorporating spatio-temporal variability of material properties and stochastic ground motions. The results indicate that material deterioration and spatial variability have significant effects on dynamic responses and seismic fragility, highlighting the importance of considering time-dependent material behavior in the seismic safety evaluation of arch dams.