Advanced Seismic Risk Assessment of California Box-Girder Bridges Using Emerging Modeling Techniques and Innovative Risk Models

Seismic fragility models depict the structural failure probability under earthquakes and play an essential role in planning mitigation strategies for, and prioritizing emergency response after, a natural hazard. This dissertation concentrates on developing a new generation of seismic fragility models for select concrete box-girder bridges in California in terms of advanced numerical bridge models, comprehensive bridge component capacity models, and robust seismic risk analysis methodologies. The dissertation first introduces emerging modeling techniques that can improve the fidelity of numerical models. Most importantly, an abutment backwall fracture model is proposed to eliminate an enormous error due to excessive lateral supports from abutment foundations in conventional abutment models. In the aspect of capacity models, seven damage states for columns are established based on a newly developed column dataset with 198 laboratory tests. Next, appropriate geometrical and material uncertainties are identified and applied in the finite element bridge models. Furthermore, to ensure that the 352 virtual bridge realizations meet the design criteria in California, three sampling techniques are proposed to correlate different uncertainties. After acquiring seismic response demands of bridge components, several methods of establishing a probabilistic seismic demand model (PSDM), relating structural seismic demand and ground motion intensity measurement, are examined. A new method called modified multiple adaptive regression splines (M-MARS) is proposed to construct the PSDM. Following is the development of four-level fragility models, from low-level component fragilities to high-level system fragilities. Ultimately, conclusions are made based on the research findings and comparisons of results through a developed bridge grouping method.Ph.D