Reliability assessment of real-time hybrid simulation with model updating

Laboratory experiments play a critical role in earthquake engineering research. Hybrid simulation provides a viable technique to assess structural performance through component tests. One challenge exists for current practice of hybrid simulation when a complex structure has more critical components than those could be accommodated in laboratories. Hybrid simulation with model updating has been developed to updating the model parameters for analytical substructures based on the observed behavior of similar parts within experimental substructures. Hybrid simulation with model updating thus has great potential to be extended to real-time hybrid simulation to account for rate-dependent behavior within structures beyond existing laboratory capacity in terms of space and equipment. It however also raises concern on how to quantify the cumulative effect of modeling errors in analytical substructures throughout the experiments. This thesis investigates on-line model updating and its corresponding modeling error for several types of lateral force resisting systems. To investigate the effect of modeling error the thesis adopts a previously proposed tool that has been demonstrated to be highly effective for detecting the presence of modeling error and thereby enables future reliability assessment of hybrid simulation results when actual structural response is not available for immediate comparison