Effect of probabilistic variation in soil properties and profile of site response

Abstract

Site response is a function of the soil profile, and the probable distribution of the soil profile has a significant effect on the seismic site response. In the present study, the influence of random variations in soil characterizations on the site response is investigated using different probabilistic distributions. The important characteristics of the local soil, corresponding to the layering, the shear wave velocity (Vs), the decrease in the nonlinear modulus, and the damping (MRD) curves, are considered when carrying out these random variations. Stochastic processes are generated by using different distribution models and keeping in mind the effect of the coefficients of the variations. In this research, a proposed procedure is developed and coded to perform the variations in soil characterizations. The coding of this new procedure is based on the original SHAKE91 framework. However, instead of using the fixed soil properties and profile, the uncertainties of the MRD curves, the layer thickness, and Vs are generated as the input data. This analysis shows that the use of median Vs, obtained from all the possible inputs under the different stochastic processes, yields good agreements with the baseline profile. Modelling the variabilities in the layering and the Vs profile is seen to have a slight effect on the performance of the site response. Additionally, the results of these analyses indicate that the variabilities in nonlinear soil properties have a significant impact on the median surface response spectrum and the amplification spectrum of the surface motions.