The effects of base motion variability and soil heterogeneity on lateral spreading of mildly sloping ground

Abstract

Centrifuge modeling has been used to observe some key characteristics of liquefiable soils during seismic motions. If carefully conducted, the results of centrifuge tests can be used for validation of constitutive models and numerical modeling tools. However, a thorough evaluation of numerical models requires knowledge of the soil properties and the uncertainties associated with these properties. Moreover, the base excitations achieved in centrifuge tests often vary from the target base excitation, and a fair evaluation of the quality of blind prediction of a centrifuge test requires an account of the uncertainties associated with the achieved base motion. This paper presents an analysis of the effects of the inherent variability present in the soil density and base motion on the lateral spreading of mildly sloping ground. The analysis combines fully-coupled non-linear finite element modeling with Monte Carlo simulation. The stochastic analysis is based on the variabilities observed in the density of the soil specimens and in the achieved base motions of the centrifuge tests conducted for the Liquefaction Experiments and Analysis Projects (LEAP): LEAP-GWU-2015 and LEAP-UCD-2017. The two-surface plasticity model for sand is used to model the soil response. The model is calibrated against element tests performed on Ottawa-F65 sand to determine its liquefaction strength. The finite element model is pre-validated through deterministic simulation of the centrifuge experiments conducted during the LEAP project. The sources of variability are identified and their magnitudes are quantified based on the results obtained from the LEAP centrifuge experiments. First, the effects of the soil spatial variability are presented. Then, the effects of the base motion variability are discussed. Finally the combined effects of the variability in the soil density and base motion are evaluated. The results obtained from the stochastic analysis are compared with the variability in the soil response observed in the centrifuge experiments. The results obtained from this study show that by carefully modeling the different sources of variability, the stochastic analysis was able to model the observed variability in the lateral displacement of the centrifuge experiments. The results obtained confirm the observation that the lateral displacement of liquefiable soil is more sensitive to the base excitation variability than the variability in the soil density.