Spatial variability of geological profiles influences the performance of geotechnical structure, and such variability can be modelled by the random field theory, which simulates the geological profile as a set of spatially-correlated random variables. Uncertainties in certain zones of the subsoil profile would pose a higher risk to the overall geotechnical performance, and a proper assessment of risks contributed from different zones will provide insights into the geotechnical investigation and subsequent design processes. One possible tool of risk quantification is the Sobol index, which is a sensitivity index used to quantify the influence of each random variable on the performance variability. This paper illustrates the evaluation of the Sobol index coupled with advanced probabilistic tools such as the polynomial chaos expansion and Latin hypercube sampling with dependence. The proposed methodology is demonstrated using a shallow foundation case with spatially-varying Young's modulus, cohesion and friction angle. The most important zones are identified as having the maximum Sobol index values. The presented package aims to integrate the latest probabilistic tools in order to systematically evaluate the risks associated with foundation performance due to spatial uncertainties of geotechnical parameters.
ASJC Scopus subject areas
- Civil and Structural Engineering
- Building and Construction
- Geotechnical Engineering and Engineering Geology