Influence of Soil Spatial Variability on Embankment Deformation Reliability Analysis Using the Random Finite Element Method

Resumo

When dealing with practical engineering problems, the uncertainties associated with soil are a major limiting factor in terms of defining the subsurface characterization and the input parameters used for design considerations in computational analysis. These uncertainties are mostly associated with the fact that soil is a natural material, formed by different geological processes that entail inherent variability. The increasing use o CPTu for site investigations allows geotechnical engineers to obtain reliable data to describe the spatial variability of soil. In this study, a geotechnical engineering situation is presented, in which the deformation of an embankment section is analyzed. On-site investigations are provided by CPT profile data, which are used to estimate soil
parameters using well-known correlations from the literature. The inherent spatial variabilities in soil geotechnical properties along the profile are described using the decomposition method by a smoothly varying trend function and residuals components. Then, the random finite element method (RFEM) is applied to compare the results with homogeneous layer probabilistic analysis. The results obtained show that consideration of spatial variability can lead to almost identical average displacement values compared to homogeneous layer probabilistic analysis, but with tighter distribution of the findings.