Empirical Methods for the Prediction of Response Spectra

Abstract

Three empirical methods of estimating response spectra are reviewed. It is shown that these methods produce comparable results, provided that the comparisons are made in a consistent manner. Although the traditional method of scaling amplification factors by peak ground motions is fundamentally reasonable, there exist some significant discrepancies between scaled and other types of spectra. The basic reason for this is that the amplification factors commonly used in the engineering community are biased toward those of a large magnitude earthquake recorded on a soil site. The dependence of spectral shape on magnitude and site conditions can be accounted for by independent estimates of peak ground acceleration, velocity and displacement to scale an unbiased set of amplification factors derived for particular site conditions. Other methods of estimating spectra account for the dependence of spectral shape on magnitude and site conditions in a more direct manner. In order to reduce the discrepancies between scaled and other types of spectra, it is recommended that a new set of amplification factors be estimated using a suite of response spectra due to earthquakes whose magnitudes are evenly distributed in the range 5.0 ≤ M ≤ 7.5 and which have been recorded on sites having similar soil conditions.