Stochastic modeling and sensitivity analysis for seismic response in bridge structures

The seismic performance of bridge structures is governed by high-dimensional uncertainties stemming from ground motion variability and structural parameter uncertainties.This paper presents a stochastic modeling framework for forward uncertainty quantification (UQ) and global sensitivity analysis (GSA) of bridge seismic responses.The proposed approach integrates physics-based dimensionality reduction with multivariate conditional distribution modeling to efficiently propagate uncertainties through nonlinear response history analyses (NLRHA).The framework's performance is demonstrated through a case study on the Auburn Ravine bridge model.Results indicate that the stochastic simulator achieves high accuracy while reducing computational cost compared to conventional methods.Variance-based sensitivity analysis reveals that ground motion uncertainties account for more than 60% of the response variability.The framework offers a practical and computationally efficient tool for seismic UQ and sensitivity analysis of bridge structures.