No description available.

No description available.

Unsaturated soils are present in a wide range of geotechnical applications such as slopes, highways or earth dams, in addition to many other geo-environmental applications such as underground disposal of radioactive waste or landfills. Despite a significant number of improvements in understanding unsaturated soil behaviour over the last five decades, there are still several aspects to be addressed and to be better understood, including in the areas of constitutive modelling, development of experimental procedures and improvement of numerical techniques. This work concentrates on two of these aspects: constitutive modelling of unsaturated soils accounting for the coupling between mechanical and water retention response; and associated numerical and computational aspects employed to solve engineering problems involving unsaturated soil mechanics. Wheeler et al. (2003) proposed a coupled elasto-plastic constitutive model for the mechanical and water retention behaviour of unsaturated soils and this model constitutes the basis of the work developed within this thesis. For isotropic stress states this coupled model employs as stress state variables mean Bishop’s stress p* and modified suction s*. In the model, plastic volumetric strains occur when the stress state reaches a LC yield curve and plastic changes of degree of saturation Sr occur when the stress state reaches a SD or SI yield curve. Coupled movements of the yield curves represent the influence of plastic changes of degree of saturation on mechanical behaviour and the influence of plastic volumetric strains on water retention behaviour. According to this constitutive model, during many types of loading or wetting under isotropic stress states the soil state will ultimately arrive at the corner between the LC and SD yield curves. Analysis of the model suggests that such states should fall on a unique planar normal compression surface in v:lnp*:lns* space and also on a second unique planar surface in Sr:lnp*:lns* space. The experimental results of Sivakumar (1993) from constant suction isotropic loading of compacted speswhite kaolin are presented in these spaces, and are shown to lie on planar surfaces, as predicted by the model. Presenting experimental normal compression results in this way gives an ideal method for determining the values of the key soil parameters k1, k2, λs and κs in the model. A simple extension of the Wheeler et al. (2003) model to the stress conditions of the triaxial test, by including the role of deviator stress q, has been proposed by Lloret et al. ii (2008b) and Raveendiraraj (2009). According to this extended model, critical states will commonly occur at the intersection of the LC yield surface and the SD yield surface, and such critical states should fall on a second pair of unique planar surfaces in the v:lnp*:lns* and Sr:lnp*:lns* spaces. The experimental critical state data of Sivakumar (1993) have been plotted in these spaces and the results obtained are presented and discussed. The new constitutive model has been extended to general 3D stress conditions and generalised stress-strain relationships required for numerical integration of the model are presented. Furthermore, 3D extended functions to identify the different elasto-plastic mechanisms of the model are proposed. A partial validation of this extended model is also presented using the experimental results of Sivakumar (1993). Two strain-driver algorithms for the integration of the generalised model are presented including first and second order error schemes. The complete formulation of these explicit algorithms is described and the computed results from both integration schemes is compared and discussed. Finally, the generalised fully coupled constitutive model has been implemented within the finite element program CODE_BRIGHT (Olivella et al., 1996) to solve boundary value problems involving unsaturated soil. The performance of the new implemented model is analysed and discussed by considering application to a boundary value problem involving wetting of a cylindrical soil specimen.

No description available.

The National Grid Transco Company sponsored this project in order to promote the understanding of NAPL migration through b-horizon soils and retarding effects upon non aqueous species migration. Soil structure and texture was also studied using conservative (Bromide) and non-conservative (Phosphate) tracers. Experimental data was produced using a laboratory ½ metre scale automated lysimeter designed and constn1cted at Plymouth. The tracers were compared before oil injection, to calibrate differences in soil texture, and after oil injection to detect any changes in the flow patterns caused by the oil injection. It was found that the Crediton, Sollom and Conway soils respectively offered least resistance to the tracers with the non-conservative tracer behaving much more unpredictably than the conservative tracer. After oil injection it could be seen that the oil had heavily retarded the ability of the tracers to migrate from the injection site. This retardation was identified as analogous to perturbations of the soil structure. Statistical analysis of the data showed that the experiments were all internally self consistent and visible patterns could be seen in the corrected data caused by inclusion of oil in the injection site. Methods of dispersal for the oil and tracer are suggested in the concluding chapter with references to the work of previous authors. Development of a hazard assessment framework was facilitated by the simulation of soil structures using a pedo transfer function developed at the National Soils Resource Institute. To allow the modelling of soils the Pore-Cor software had an annealed simplex algorithm integrated into the data inversion engine to allow the simulation of 3-D soil structures using 2-D data from pedo transfer functions or experimentally derived water retention curves. An extensive sensitivity analysis upon the model highlighted limitations, due to the data set the current pedo transfer function is based upon. It was suggested that inclusion of choices of different pedo transfer functions could be used to overcome this problem. A suitable framework was derived for the identification of priority soils using a validated computer model. Experimental data was compared to the simulated data in order to try and develop an understanding of practical upscaling of the data. The use of the "Scaleway" method is discussed in the concluding Chapter.

No description available.

No description available.

No description available.

No description available.

No description available.