Abstract
Proper characterisations of soil properties and their variations are pivotal to the geotechnical design process. Although multiple in situ soil tests are routinely specified and performed in geotechnical investigation programmes, the information they provide regarding the spatial correlations of soil properties are often not fully utilised. This paper presents a holistic framework to characterise the three-dimensional, anisotropic, spatial variability of soil properties, using results of in situ tests such as standard penetration tests or vane shear tests. The restricted maximum likelihood method is implemented with an anisotropic covariance model, leading to improved predictive capabilities compared to conventional approaches, and allows quantification of the uncertainties on soil properties at unsampled locations, represented as distributions of prediction variance across the entire subsurface three-dimensional domain. The magnitudes of prediction variance at different locations can be used to provide guidance on the necessities and locations of additional soil sampling. They can also provide key input parameters for random field models in site-specific probabilistic analyses of geotechnical projects. The proposed approach is applied to the study of two project sites in Hong Kong, where it is shown that the three-dimensional spatial correlation features may be interpreted together with the geological settings at the site.
Original language | English |
---|---|
Pages (from-to) | 805-819 |
Number of pages | 15 |
Journal | Geotechnique |
Volume | 68 |
Issue number | 9 |
DOIs | |
Publication status | Published - 1 Sep 2018 |
Keywords
- In situ testing
- Site investigation
- Statistical analysis
ASJC Scopus subject areas
- Geotechnical Engineering and Engineering Geology
- Earth and Planetary Sciences (miscellaneous)