Improving the characterization of seismic hazard for performance-based earthquake engineering design

The research presented in this dissertation addresses several topics pertaining to the probabilistic characterization of earthquake ground motion and its application to performance-based earthquake engineering (PBEE). Hazard characterization for PBEE applications consists of (1) the development of hazard curves describing the likelihood of observing various levels of ground motion intensity measures (IMs) within a given time period and (2) the selection of ground motion records compatible with the IM hazard for use in structural simulations.;Geotechnical ground response analyses (GRA) are performed for eight sites to define site-specific amplification factors (AFs) calculated for multiple input motions. Those AFs modify the median and standard deviation of IM that would be expected given a scenario earthquake event (magnitude and distance). Alternative estimates of standard deviation that are empirically-based and that are derived for a theoretical idealization are considered. The modified median and standard deviation are then used inside of the hazard integral. The impact on PSHA results of site-specific versus generic site-factors is shown to be large for soft soil sites for PGA and 1.0 sec spectral acceleration and modest for stiff soil sites. Those differences are caused by a combination of the different median and standard deviation and by a modification in the sources controlling the hazard towards larger magnitude events on more distant faults for which rare strong realizations of ground motions are achieved.;Once the hazard is properly characterized, it is possible to select ground motion records for the structural analyses. Because the record-to-record variability generally drives the dispersion of structural response, this step is critical. The problem of record selection is explored as part of the PEER benchmark project and is continued with the ongoing research within the Ground Motion Selection and Modification (GMSM) program. The GMSM program first developed a methodology to evaluate different methods of GMSM and preliminary findings show that when the nonlinear properties of the structural response are taken into account within a method, the structural response predictions tend to be more accurate and present less dispersion that when records are selected solely on seismological parameters.