One of the critical aspects of the geotechnical earthquake engineering is to evaluate the response of soil grounds under earthquake loading. Generally, three types of uncertainties are contributed to the seismic site response analysis — ground motion variability, site property variability and uncertainty associated with the constitutive soil models and numerical methods used for the analysis. This study focuses on quantifying the effects of ground motion variability and site property variability on the seismic ground response analysis. An equivalent-linear approach and a modified version considering frequency-dependent modulus and damping are used to perform the seismic site response analysis....[ Read more ]

One of the critical aspects of the geotechnical earthquake engineering is to evaluate the response of soil grounds under earthquake loading. Generally, three types of uncertainties are contributed to the seismic site response analysis — ground motion variability, site property variability and uncertainty associated with the constitutive soil models and numerical methods used for the analysis. This study focuses on quantifying the effects of ground motion variability and site property variability on the seismic ground response analysis. An equivalent-linear approach and a modified version considering frequency-dependent modulus and damping are used to perform the seismic site response analysis.

Two types of intrinsic variability in ground motion intensity measures are considered in the site response analysis. The first type is the variability from a scenario earthquake, given the earthquake magnitude, rupture distance etc, and is termed as “unconditional” variability. The second type is the variability of the ground motion intensity measures conditioned on a specified response spectrum value at a specified spectral period, which is termed as “conditional” variability. In this study, a new ground motion selection scheme is implemented to select ground motions that preserve the conditional and unconditional variability of the spectral acceleration (Sa). Due to the complexity of ground motion time histories, different ground motion intensity measures can only represent certain aspects of ground motion characteristics. Thus the ground motion selection scheme is further extended to simultaneously capture the joint unconditional and conditional variability of multiple intensity measures, specifically the spectral accelerations and cumulative absolute velocity (CAV). Soil property variability is also an important source of uncertainties in seismic ground analysis since soil properties vary spatially across the site and variations in soil properties can change the surface response considerably. Two statistical models were implemented to simulate the site property variability in terms of shear velocity profiles, as well as the shear moduli and damping ratios of the soils.

The influences of the ground motion variability and soil property variability on site response are examined by Monte Carlo simulations. Ground motion variability is found to be the most important controlling factor compared with the soil property variability. A new site-specific response analysis approach is therefore proposed to evaluate the site response by considering the ground motion variability at a specified seismic hazard level. In this approach, seismic hazard deaggregation is performed to identify the controlling earthquake scenarios. The ground motions are selected and modified to preserve the conditional variability of the scenario earthquakes conditioned on the Uniform Hazard Spectrum (UHS) at different conditioning periods. Seismic site response analyses are performed on two sites located in California, representing a stiff soil site and a soft soil site, respectively. Compared with the traditional method that selects ground motions to tightly fit the target UHS curve at all periods, the new approach is more consistent with the probabilistic seismic hazard analysis (PSHA) framework. The study also shows that the effects of spectral accelerations and CAVs of the input motions are not coupled in the site response analysis using the equivalent linear analysis method.