Probabilistic soil model for seismic risk assessment based on SDMT results

Abstract

A geotechnical site investigation has been conducted in the southern breakwater basin of the Port of Barcelona, for which hundreds of in-situ and lab tests have been performed. Among these tests, this study focuses on the evaluation of triple points. That is, at the prospected locations, there are results from seismic dilatometer Marchetti test (SDMT), cone penetration tests (CPTu), and
laboratory tests based on soil samples. By using this information, the probabilistic distribution of the dynamic and geometrical properties of the soil profiles can be properly characterized. In this way, eleven closely spaced boreholes have been used to characterise the statistical properties of the input variables. Then, it has been performed a probabilistic generation of one-thousand soil profiles, which are statistically compatible with the data provided by the eleven aforementioned boreholes. Then, it has been analysed how the elastic properties of the generated soil profiles evolve once seismic waves have passed through them. To do so, a large set of ground motion recorded in hard soils have been employed. The main objective has been to obtain mathematical arrangements exhibiting highefficiency to predict parameters that can be used to characterize the nonlinear dynamic response
of the soil. This has been achieved by employing bivariate and multivariate analyses in the log-log space. Results show that the soil response can be properly parametrized if considering intensity measures extracted from the ground motions acting at the bedrock level and specific soil parameters. Finally, fragility curves and surfaces have been derived for risk assessment purposes.