Stochastic modelling of slopes and foundations on heterogeneous unsaturated soils

Le, Thi Minh Hue (2012) Stochastic modelling of slopes and foundations on heterogeneous unsaturated soils. PhD thesis, University of Glasgow.

Due to Embargo and/or Third Party Copyright restrictions, this thesis is not available in this service.
Printed Thesis Information:


Many geotechnical engineering structures, including embankments, slopes and foundations involve unsaturated soils, either because they are constructed of compacted fills which are unsaturated after placement or because they include a zone of unsaturated natural soils above the water table. Moreover, both compacted fills and natural soils are inherently heterogeneous, which leads to an important but complex class of problems whose analysis involves consideration of heterogeneous unsaturated soil. Both partial saturation and material heterogeneity significantly influence the overall behaviour of the soil domain and have been widely acknowledged as particularly important factors to be accounted for during geotechnical analysis. Research on these two areas has however evolved almost independently, leaving a gap in the understanding of problems relating to heterogeneous unsaturated soils. The aim of this thesis is to contribute to bridging this gap.

Three boundary value problems, namely seepage through a flood defence embankment, stability of a slope and differential settlement (rotation) and bearing capacity of a rigid foundation, were investigated with reference to unsaturated soils which have spatially varying material properties. The finite element method was combined with random field theory to stochastically analyse each problem by employing unsaturated soil constitutive models covering both hydraulic and mechanical behaviour and randomly varying, but spatially correlated, soil properties. Stochastic modelling was undertaken by means of Monte Carlo simulations which involved a large number of finite element analyses, each with a different random property configuration. The collective responses were statistically characterised by their mean, standard deviation and probability distribution function, which provide a powerful probabilistic tool for risk and reliability assessment.

The thesis demonstrates that the simultaneous existence of heterogeneity and partial saturation can significantly modify the soil response to a change of hydraulic boundary conditions (with respect to the homogeneous case or where the effects of partial saturation are neglected) and, in many cases, can exacerbate the risk of unsatisfactory performance of slopes and foundations. Notably, heterogeneity of soil porosity causes variability of both permeability and degree of saturation in unsaturated conditions, which can lead to a counter-intuitive tendency of water preferably following a path through low porosity regions, instead of high porosity regions as in the saturated case. This is because the influence of porosity on the degree of saturation at a given suction can counteract the influence of porosity on the intrinsic permeability. Moreover, if the influence of porosity on both intrinsic permeability and degree of saturation is taken into account, it is possible to obtain rather unexpected results such as for example the reduction of the mean and variation of steady state flow across randomly heterogeneous embankments compared with the case in which porosity only affects permeability.

The thesis also shows that wetting events (e.g. rainfall) can cause, during transient state, a significant reduction in the stability of an unsaturated slope but can also lead to an increase in the range of variation of the factor of safety if the slope has spatially varying porosity. Consequently, the probability of failure of the slope tends to be significantly higher during transient state than at a final steady state when hydrostatic conditions have been reinstated. An important finding from the study of the heterogeneous unsaturated slope is that rainfall infiltration results not only in a decrease in factor of safety but also in a decrease in size of the sliding mass. Therefore, the risk assessment of a heterogeneous unsaturated slope taking into account both factor of safety and size of the sliding mass can lead to different conclusions compared to the conventional case when only the factor of safety is considered.

The study of a rigid foundation resting on a heterogeneous unsaturated soil has found that the settlements induced by rainfall infiltration depend on the nature of the spatially varying material property. A rigid symmetrical foundation lying on soils with heterogeneous preconsolidation stress rotates (i.e. experiences differential settlement between the foundation ends) even though the wetting front migrates into the soil domain symmetrically about the centreline of the foundation. Conversely, the same foundation lying on soils with non-uniform porosity rotates because of the asymmetrical migration of the wetting front into the ground, which is caused by the variability of permeability and degree of saturation associated with porosity heterogeneity. Importantly, the thesis emphasizes that the rotation of the foundation cannot be detected if partial saturation and spatial variability are not simultaneously taken into account.

Item Type: Thesis (PhD)
Qualification Level: Doctoral
Additional Information: Due to copyright restrictions the full text of this thesis cannot be made available online. Access to the printed version is available.
Keywords: random, slope, foundation, heterogeneous, unsaturated, probability of failure
Subjects: T Technology > TA Engineering (General). Civil engineering (General)
Colleges/Schools: College of Science and Engineering > School of Engineering > Infrastructure and Environment
Supervisor's Name: Gallipoli, Dr. Domenico
Date of Award: 2012
Depositing User: Ms Thi Minh Hue Le
Unique ID: glathesis:2012-3097
Copyright: Copyright of this thesis is held by the author.
Date Deposited: 30 Jan 2012
Last Modified: 10 Dec 2012 14:03

Actions (login required)

View Item View Item