Selection of Ground Motions for Performance-Based Seismic Assessment of Structures

Abstract

It is well known that one of the most significant sources of uncertainty and variability in seismic demand prediction arises from ground motion selection. The task of selecting appropriate ground motions can become a formidable given the limited database of earthquake records that satisfy the required site parameters. Moreover, from a practical consideration, it is necessary to limit the number of ground motions used in the evaluation process while at the same time minimizing the dispersion in the demand estimation. This presentation will examine the effectiveness of three ground-motion selection schemes: (i) magnitude scaling; (ii) spectrum matching; and (iii) conditional mean spectra. Findings from comprehensive nonlinear time history simulations indicate that while spectral matching is slightly more effective in reducing dispersion compared to scaling, it may modify ground motion content which can alter the location of peak demands. The conditional mean spectrum is less effective in reducing dispersion in demands when amplitude scaling is used