A probabilistic assessment of resolution in the SASW test and its impact on the prediction of ground vibrations

Abstract

SUMMARY The Spectral Analysis of Surface Waves (SASW) method is a technique for the identification of the thickness, dynamic shear modulus, and damping ratio of shallow soil layers. The method consists of an in situ experiment to determine the dispersion curve of the soil and the solution of an inverse problem where the corresponding soil profile is identified. The SASW method has been used to investigate pavement systems, to assess the quality of ground improvement, to determine the thickness of waste deposits, and to identify the dynamic soil properties for the prediction of ground vibrations. In this paper, the focus is on the last application. The information on the dynamic soil properties provided by the dispersion curve is limited. The dispersion curve is insensitive to variations of the soil properties on a small spatial scale and at a large depth. As a result, the solution of the inverse problem in the SASW method is non-unique and hence uncertain. The prediction of ground vibrations is therefore based on a soil model with uncertain properties. In this study, a Bayesian approach is followed to solve the inverse problem in the SASW method. A prior stochastic soil model is first formulated using the information that is available before the SASW test is performed. A Markov chain Monte Carlo method is used to transform the prior model into a posterior stochastic soil model that accounts for the SASW test results. Finally, the prediction of ground vibrations is addressed. The posterior soil model is used to assess the robustness of the predicted vibrations, accounting for the uncertainty on the results of the SASW test. As an example, the free field vibrations due to a hammer impact on a concrete foundation are considered. More complicated problems, such as the prediction of road and rail traffic induced vibrations, can be addressed in a similar way.