Abstract:
Reliability estimation is one of the key ingredients of performance evaluation and risk assessment strategies in structural and earthquake engineering. Residual life estimation is of great importance since design principles are constructed on a framework of performance-based engineering. Conventional methods of reliability estimation are mainly based on theoretical content without experimental information. System identification provides additional experimental information to update or verify analytical models by dynamic response measurements. Therefore, integration of system identification and reliability estimation could improve the accuracy of reliability estimation with rational and state-of-the-art information. In this study, a reliability estimation methodology, which incorporates vibration-based system identification, is presented. A two-span, three-bent reinforced concrete bridge structure is exposed to a series of white noise and earthquake excitations, and damage progress is documented. Two procedures are followed to modify base finite element model at different damage states. Updated procedure involves system identification and finite element model updating, whereas non-updated procedure is based on post-event structural parameter prediction using nonlinear time history analysis. For each damage state, updated and non-updated base models are used to estimate reliabilities by fragility curves. Fragility curves, considering peak ground acceleration as random variable, are estimated by a series of nonlinear time history analyses under various input ground motions. After all, the distinction between updated and non-updated fragility curves, and the effect of structural damage on fragility curves is disclosed. The results have proven that the proposed methodology, which utilizes system identification for reliability estimation, significantly changes estimation results and is essential for making accurate estimates of residual life.
