Abstract:
This thesis implements various scalar and vector methods of probabilistic seismic risk assessment. Methods implemented differ by type of scaling and selection of ground motion records as well as by primary scalar intensity measure (IM) used as ground motion intensity metric. Two different ground motion scaling and selection methods are used, namely stripe scaling and cloud scaling. Conceptual differences between stripe scaling and cloud scaling as well as the differences they produce in risk assessment results are presented and discussed. Different scalar ground motion intensity measures used are fundamental period spectral acceleration (SAT(T1)) and two distinct weighted average spectral accelerations. Weighted average spectral acceleration is a novel intensity measure developed and presented within this thesis. It is shown to be a good predictor of structural response at high intensity levels where structure undergoes intense yielding. Weighted average spectral acceleration enables analyst to set weights to contributions of spectral accelerations at different periods according to their expected importance to structural response. Conditional spectra for ground motion selection are developed for all three considered scalar IMs. Multivariate normal distributions are presented as useful mathematical tools and they are used for various purposes throughout this thesis (conditional spectra, vector IM hazard, structural response modeling, etc.). Maximum likelihood estimation (MLE) based method is also developed for estimation of parameters used for modeling structural response using structural analyses data.
