Probabilistic seismic demand analysis: Spectrum matching and design

The current state and trend of Earthquake Engineering practice calls for better understanding of structural nonlinear response due to strong ground motion excitation. The present study deals with issues on such structural-systems understanding, in light of recent developments in Probabilistic Seismic Demand Analysis (PSDA), which is a practical interface between the elastic and inelastic representations of structural behavior and ground motion hazard, in a probabilistic framework.; One part of this study deals with requirements of PSDA in terms of structural response representation: efficient and accurate estimations of response. Ground motion response spectral shape is studied and is recognized as having an important contribution to structural displacement demands, especially for structures that are forced into large nonlinear displacements. Methods based on the use of particular spectral shapes for providing accurate and efficient estimates of response are studied.; In particular, the use of Spectrum Matched Records (also known as Spectrum Compatible Records) for the prediction of M-R (magnitude-distance) scenario-based displacement demands is discussed. It was found that the use of Spectrum Matched accelerograms, for the estimation of these nonlinear demands associated to a scenario event, can lead (albeit very efficiently) to obtaining unconservative results. An important cause of this bias is identified.; The second part of this study “dis-assembles” the components of PSDA for structural design purposes. The final product of PSDA is a probabilistic representation of a structure's response given its seismic environment. A methodology is proposed by assigning a desired exceedence probability to a particular response level of interest (e.g., collapse) in order to try to estimate the design-level ground motion to consider for design. The proposed method is tested and presented with an example, for which a 3-D model of a steel offshore platform is considered. Some issues that arise from considering the traditionally uni-dimensional PSDA and spectrum matching (one horizontal component of ground motion) on a three-dimensional system (two horizontal components of response) are discussed as well.