SITE-SPECIFIC GROUND MOTION PREDICTION EQUATIONS ON THE BASIS OF INSTRUMENTALLY VERIFIED SUBSOIL CLASSES OF STRONG-MOTION RECORDING SITES AND THEIR APPLICATION TO NPP SITES IN GERMANY

摘要

A new set of Site Specific Ground Motion Prediction Equations (SGMPEs) is presented in this paper. They are based on strong-motion recordings from sites with instrumentally verified subsoil conditions in California, Turkey and at the NPP target sites. The approach follows analogy considerations by assuming that in the case of similarities in the transfer characteristics also similar ground motion recordings have to be expected. For this purpose, a cluster analysis is performed, taking H/V spectra from the NPP site measurements as search and evaluation criterion. By a scoring procedure different site measurement points MP_i (inside and outside the NPP area) can be considered. Each measurement is providing a target function (H/V spectra) which is the basis for a ranking of the best fitting strong-motion recording sites. The target function represents a statistically evaluated, several days covering time window of instrumental data where the H/V-spectra are determined in a sequence of equal time steps. The scatter due to day time and working phases is smoothed out by calculating different fractiles of the H/V spectra. Having identified the best fitting recording stations, the available data is combined within a special data group standing for the most likely subsoil class of the NPP-Site. Basic steps of the whole concept are illustrated for two NPP-Sites in Germany being classified by quite different subsoil classes. The procedure is open for sensitivity and parameter studies, and offers the advantage of a "self-growing" database and the adoption of modified single-station approach.