Characterization of site-specific soil spatial variability and evaluation of its impacts on geotechnical performance

Spatial variability involving geological and geotechnical properties are often the cause of unexpected geotechnical performance. Under practical situations, the design engineers often possess a small number of sampled points with known values of parameters. In other words, the associated uncertainties or random fields are 'conditioned' according to the distribution of known sampled points. This paper presents a holistic framework that incorporates systematic characterization of geotechnical investigation data, in order to derive site-specific spatial correlation structures. Based on these results, conditioned random fields can be generated, with the simulated geological profiles and/or geotechnical properties passing through the known data points. The simulated profiles are then utilized for probabilistic analyses of piled foundation settlements. The framework aims to maximize the value of geotechnical investigation data, so that uncertainties and risks of arising from soil spatial variability can be properly quantified and managed.