Back to results list
Show full item record
Please use this identifier to cite or link to this item:
|Title:||Characterisation of soil spatial variability and applications in ground exploration||Authors:||Liu, Wenfei||Degree:||Ph.D.||Issue Date:||2018||Abstract:||Proper characterisations of soil properties and their variations are pivotal to the geotechnical design process. While spatial variability in geological profiles and geotechnical properties is one of the primary sources of uncertainties encountered in the design and construction of geotechnical engineering projects, there has been limited study on the site-specific characterisation of these features, and how they may be included in the engineering decision-making process. This thesis proposes a new integrated framework that incorporates the restricted maximum likelihood (REML) method with data transform and rigorous residuals diagnostics to ensure that the available geotechnical data is best utilised. Meanwhile, the procedures ensure that fundamental statistical assumptions are satisfied, thereby enhancing the reliability of residual analysis. The framework also consists of a rational detrending process to determine the optimal polynomial order, and allows detection of outliers in the dataset, both of which have not been considered in previous attempts to characterise spatial variability. The proposed approach is illustrated through investigations on spatial correlation features of geological profiles at two sites, and the existence of geological faults is found to significantly affect these features, as indicated by the reduced scales of fluctuation and spatial dependence, which correspond to increased uncertainty in areas intersected by faults. In order to characterise spatial correlations in soil properties, the framework is extended to reveal the three-dimensional anisotropic spatial variability in soils that exhibit significant heterogeneity. This is illustrated through results of in situ tests including standard penetration tests and vane shear tests from two project sites, where the spatial correlation features may be interpreted together with the geological settings at the sites.
Improvements in prediction capability of the proposed model are illustrated through comparisons with conventional geotechnical approaches. The framework also allows quantification of the uncertainties on soil properties at unsampled locations, represented as distributions of prediction variance across the entire three-dimensional subsurface domain. The magnitudes of prediction variance at different locations can be used to provide guidance on the necessities and locations of additional soil sampling. Considering these uncertainties, an approach is then presented to quantify the cost-effectiveness of ground exploration and its implications to the budget in civil engineering projects. Meanwhile, the optimal sampling strategy can be determined through spatial tessellation techniques including generation of Voronoi diagrams and two- and three-dimensional Delaunay triangulation techniques. Through the proposed approach, this thesis promotes better utilisation of geotechnical information and rational assessments of project risks associated with their variability, which may lead to improved planning and resource allocation of the projects.
|Subjects:||Hong Kong Polytechnic University -- Dissertations
|Pages:||xxx, 272 pages : color illustrations|
|Appears in Collections:||Thesis|
View full-text via https://theses.lib.polyu.edu.hk/handle/200/9465
Citations as of Oct 1, 2023
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.