Bayesian nonlinear finite element model updating of a full-scale bridge column: insights into calibration of inherent damping model

This study presents a Bayesian framework for calibrating a nonlinear finite element (FE) model of a full-scale reinforced concrete (RC) bridge column tested on the Large High-Performance Outdoor Shake Table (LHPOST) at University of California, San Diego (UCSD).Using experimental input-output data from three selected earthquake ground motions (GMs) of increasing intensity, material, damping, and noise parameters are calibrated via the Transitional Markov Chain Monte Carlo (TMCMC) method.The calibrated FE model yields improved predictive accuracy, with postprocessed effective modal damping ratios extracted from the calibrated FE model exhibiting strong response amplitude dependence both within and across seismic excitations. INTRODUCTIONDigital twin-based approaches for structural health monitoring (SHM) and damage prognosis (DP) are emerging as powerful framework for intelligent infrastructure management by integrating physics-based FE models with measurement data.Bayesian updating of nonlinear FE models using input-output measurement data allows iterative calibration that accounts for noise, parameter uncertainty, and model form error, supporting probabilistic assessment and improved predictive reliability over a structure's lifespan [1].Nonlinear FE models have advanced significantly, capturing key seismic response features, including stiffness degradation, concrete cracking, plastic hinging, and collapse mechanisms under strong ground motions (GM) [2].However, predictive accuracy hinges on properly calibrating uncertain parameters such as material properties, boundary conditions, and damping.Bayesian calibration is increasingly adopted for its ability to incorporate prior knowledge, manage measurement noise, and quantify posterior uncertainty [3].Among various modeling uncertainties, inherent damping-representing energy dissipation mechanisms not explicitly modeled through material nonlinearity-remains particularly