Full Program »

Blind identification of nonlinear civil structures using sparse seismic response data

Several approaches for damage identification (DID) of civil structures have been proposed and studied in the literature for post-earthquake assessment of structural safety. Since the structural response during damage-inducing events is far from linear elastic, it is crucial to develop reliable, robust, and accurate DID methods able to capture the complex nonlinear behavior exhibited by structural systems when they are subjected to intense excitations. In many practical applications, the actual input excitations acting on the structure of interest are difficult to be accurately and precisely measured (e.g., due to soil-structure-interaction effects) or the input excitations may be only partially measured (e.g., extended civil structures such as long-span bridges with sensors in only some supports). Such erroneous and/or incomplete input components, if not treated correctly, introduce some errors in the estimation of the modeling parameters involved in the model updating procedure. In this paper a Bayesian nonlinear finite element (FE) model updating methodology for DID of civil structures when the inputs are unknown, i.e. only limited output measurements are available, is proposed. The proposed methodology allow estimating unknown time-invariant modeling parameters of mechanics-based nonlinear FE model of the structure and the unknown input excitation using spatially sparse structural response measurements recorded during a damage-inducing earthquake event. A tri-dimensional 4-story 2-by-1 bay steel frame building subjected to unknown bidirectional seismic base excitation is used to validate the proposed approach.

Author(s):

Rodrigo Astroza    
University of California, San Diego
United States

Hamed Ebrahimian    
University of California, San Diego
United States

Joel P. Conte    
University of California, San Diego
United States