Comparing 1D-3C et 1D-1C nonlinear dynamic responses of deep and shallow japanese sites, considering various assumptions
Résumé
In this paper, we present the results obtained with the CyberQuake computation code for two Japanese sites selected from the KiK-net and PARI accelerometric networks, which were proposed within the PRENOLIN international benchmark. The deep Kushiro (KiK-net KSRH10) and the shallow Sendai (PARI) sites were chosen as they are very close to a 1D geometry (horizontal layers). During the PRENOLIN benchmark, various 1D-1C computing assumptions with vertical incident waves were considered. Here, we present further works comparing results of 1D-1C / 1D-3C nonlinear transient dynamic computations, considering effective-stress / total-stress approaches, as well as vertical / oblique incident input motions, in order to review the usual main assumptions in 1D nonlinear site effects analyses, depending on site conditions. For the selected shallow site, the 1D-1C analysis with vertical incidence seems to be a better option. No clear trend is found in this case with respect to preferring the effective-stress or the total-stress model. The effective-stress analysis provides however better results in terms of predicted time-history ground motions at the site. For the deep site, the new computations performed in this study confirm that 1D-3C effective-stress analysis do improve the predictions. However, the discrepancy between the predictions and the recordings is still remaining since PRENOLIN, and cannot be explained by the new assumptions tested in this study. For both sites, no significant effect is observed, when considering oblique incidence for propagating waves. Computations with vertical incidence lead sometimes to better results.
Origine : Fichiers produits par l'(les) auteur(s)
Loading...