Nonlinear Site Amplification Models And Used In Ground-Motion Prediction Equations

Ground-motion prediction equations (GMPEs) plays an important role in seismic safety evaluation,earthquake hazard analysis and determining quantitative fortification standards for seismic design. Nonlinear site model is an important part of ground-motion prediction equations (GMPEs) and can be constructed in a number of ways:with the best way being to derive directly from strong-motion records. However, the number of strong-motion records in subduction zone area, for example in Japan strong-motion records that contain the effect of strong nonlinear soil response is still too small to derive nonlinear site terms. As an alternative we attempt to use 1-D numerical modeling to derive nonlinear site amplification ratios. The rock site strong-motion records used in this thesis were from different earthquake categories in Japan and the PEER dataset. Those records had a wide range of earthquake magnitudes, source distances and peak ground accelerations. The 1-D models were constructed based on the shear-wave velocity profiles from Kik-net strong-motion stations in Japan and have a wide range of site periods, soil depth and impedance ratios. The nonlinear site models derived from 1-D analysis is for GMPEs using site class as the site parameter. To overcome the limitations of the 1-D site models, we designed a method to adjust the 1-D model so that it can be used in a GMPE. Based on the above mentioned ideas, the following work has been carried out:(1) Collect a moderately large number of strong-motion records from rock sites obtained in Japan and other countries in the Pacific Earthquake Engineering Research (PEER) database, the dataset has a wide range of earthquake magnitudes, source distances and peak ground accelerations.(2) We used the measured shear-wave velocity profiles to compute site period, and then divided the site into site classes according to their site periods. For each soil layer, a nonlinear model was selected based on not only the depth and the shear-wave velocity of the layer but also a random selection within a given range. Then we used SHAKE to carry out equivalent linear analyses and calculated the site amplification ratios.(3) Next we fitted an empirical nonlinear site amplification model to the computed amplification ratios. Based on existing models used by the others, we carried out step-by step residual analyses, to determine the model parameters. The final amplification model accounts for the effect of impedance ratios, magnitude and source distance of the rock site records.(4) Based on modelling and the parameters selected in the last step, we constructed a random effects model to derive model parameters. During the regression analyses, we tested each parameter using the increase in the maximum log-likelihood and the ratios between the median and the standard deviation for each coefficient. When the ratio between the median value for a parameter and its standard deviation is less than 2 it is likely that this term is not statistically significant. We started testing the term that had the smallest median value/standard deviation ratio. If this term is not statistically significant, the maximum log-likelihood is reduced by less than 1.0.when this term is deleted from the model. We establish one set of models the site classes we used.(5) We also designed a method to adjust the 1-D model so that it can be used in a GMPE accounting for the fact that a 1-D model is an overly simplistic assumption for many real strong-motion recording stations in many parts of the world. The model adjusted can be incorporated into a GMPE as nonlinear site terms.