Output-only Finite Element Model Updating for Post-Earthquake Monitoring

Abstract

After any seismic event, it is critical to rapidly assess the damage condition of bridges. This process could be facilitated using digital technologies to eliminate the barriers of manual inspections. The post-event response of a bridge can be simulated using the bridgeﾒs prior finite element (FE) model and recorded ground motions at nearby stations. But, the simulated responses are unreliable due to the uncertainties in the structural and soil properties, as well as input motion at the foundation level. These uncertainties can be reduced through integration of measurement data and the prior FE model using model updating techniques. This study presents a Bayesian model updating technique that jointly updates the FE model of bridge, including the soil-structure effects, and foundation input motions using the measured structural responses. After verification of this technique in a numerically simulated environment, real seismic data are used in a series of case studies. The agreement between the measurements and FE-predicted responses shows the applicability of this technique in a real-world setting. Moreover, comparison between estimated foundation input motions and free-field motions reveals valuable information regarding the soil-structure interaction effects at the bridge site. The updated model is the bridge digital twin and is used for virtual sensing. Using the bridge digital twin, the bridgeﾒs local-level responses can be monitored to identify the location/severity of damage.