Seismic collapse risk assessment of buildings : effects ofintensity measure selection and computational approach

Life safety and collapse prevention have always been primary goals of earthquake engineering. Although the collapse risk of structures has not been explicitly quantified until recently, interest in doing so has risen significantly. A primary reason for this is the advent of performance-based earthquake engineering, which considers uncertainties in the seismic hazard and structural response and seeks to engineer structures that will achieve a desired level of performance in terms of expected monetary losses, downtime and casualties. Collapse risk assessment of a structure involves combining information about the seismic hazard at the site with the behavior of the structure. Typically an intensity measure (IM) is used to describe the level of ground motion shaking and quantify the seismic hazard. The behavior of the structure is then characterized through nonlinear response history analysis by subjecting the structure to many different ground motions at various intensity levels. A collapse fragility curve, which describes how the probability of collapse increases as a function of the ground motion intensity, is constructed based on analysis results and combined with the seismic hazard curve to compute the mean annual frequency of collapse. This dissertation focuses on evaluating the effects of IM selection and computational approach on the computed collapse risk, and contributions are made in the following areas: (1) quantifying the uncertainty in the collapse risk estimate due to the number of ground motions used in structural analysis; (2) characterizing the ability of different IMs to efficiently predict structural collapse and provide reliable estimates of the collapse risk; (3) explaining why certain IMs perform better than others with respect to the previous point; and (4) because the spectral acceleration averaged over a period range has been identified as a promising IM for collapse risk assessment, providing recommendations on the period ranges that maximize the efficiency of this IM as a function of structural properties.