ABSTRACT
This paper presents an algorithm to select records from a ground motion database for multiple sites under unconditional or conditional earthquake scenarios. Major steps in the selection algorithm are to construct the target distribution (i.e., mean and covariance matrix) of spectral accelerations (SAs) at multiple sites by incorporating the ground-motion spatial correlation, randomly simulate suites of response spectra given the “target”, and select a final suite of recorded ground motions with response spectra individually matching the simulated spectra. Ground-motion selection examples are provided to validate the performance of the algorithm developed; it is demonstrated that the selected multiple-site ground motions realistically capture the target distribution (i.e., mean, covariance, and spatial correlation) of response spectra for both unconditional and conditional scenarios. The computational expense of the algorithm is examined to be generally acceptable. The algorithm presented can be used for selecting ground motions, as inputs to conduct dynamic analysis of spatially distributed buildings.
Acknowledgments
The authors greatly thank two anonymous reviewers for their helpful comments to improve this manuscript.