This thesis describes the findings of an experimental and numerical investigation into the mechanical behaviour of a soft clay. Chapter 1 outlines the need for continued research on soft clay behaviour. The aims and objectives of the work are set out, as well as the structure of the research. The mechanical behaviour of soft clays has been reported by a large number of researchers and the findings of thee are considered in Chapter 2. A review of Bothkennar clay is given. Chapter 3 sets out the details of two constitutive models, whose purpose is to model the efforts of plastic anisotropy and destructuration. In Chapter 4 the experimental apparatus and procedures used in the testing programme are described. The majority of testing was carried out in triaxial cells and the arrangement of this equipment is detailed. In Chapter 5, the main experimental programme is described. Chapter 6 contains results from triaxial tests on vertically oriented samples. These tests include multi-stage stress path tests and standard drained shearing tests to failure. Triaxial tests on horizontally oriented samples are presented in Chapter 7. These provide a more generalized assessment of the role of anisotropy and destructuration, aided by the use of local strain measuring devices. In Chapter 8 tests results on both vertical and horizontal samples are compared with simulations from the constitutive model S-CLAY1 in order to assess how well this constitutive model predicts the stress-strain response of the soil, with particular respect to plastic anisotropy. Chapter 9 is concerned with the constitutive modelling of both plastic anisotropy and destructuration. In Chapter 10, it is concluded that the constitutive models introduced in Chapter 4 have provided considerably improved predictions of the mechanical behaviour of natural soft clays when compared to commonly used modelling techniques.

The key parameters representing the small strain response of geomaterials are the very small strain shear modulus, Gmax, shear modulus degradation G/Gmax, and damping ratio. These are also important parameters in the design of foundations where only small deformation takes place. A review of the literature suggests that shear modulus and damping ratio at small strain are significantly influenced by void ratio and mean effective stress. They are also influenced by other parameters such as confinement time, anisotropy, number of loading cycles, and OCR, etc. However, there was little evidence in the literature showing the influence of particle characteristics on the small strain response of geomaterials. In this research the effects of some particle characteristics such as particle size and particle shape on the small strain response of soils are investigated. Granular materials with different particle shapes, namely glass Ballotini with various diameters, Leighton Buzzard sand fraction B and E, glass glitter, glass nugget, as well as mixtures of Leighton Buzzard sand fraction B and 0.1 mm mica, are tested using a fixed-free resonant column apparatus (RCA). The test results suggest that particle shape significantly influences the small strain response of geomaterials. Both particle form and particle roundness have correlations with the values of Gmax normalised by a void ratio function, F(e). Normalised Gmax increases with increasing sphericity and roundness of the particle. At the same void ratio, the stress exponent, n, elastic threshold strain, e, and shear modulus degradation, G/Gmax, for granular materials decrease with an increase in sphericity and roundness. Material damping ratio also increases with increasing sphericity and roundness. Particle size was also found to significantly influence the small strain response of glass Ballotini. At the same void ratio and effective stress, Gmax increases with an increase in particle diameter. Elastic threshold strain, e, and G/Gmax also increase with an increase in particle diameter. In addition, stress exponent, n, and material damping ratio decrease with increasing particle diameter. It can be concluded that fine soils are more susceptible to an increment in shear strain and effective stress than coarse soils. The addition of a small proportion of 0.1 mm mica to Leighton Buzzard sand fraction B (LBSB) considerably reduces Gmax, even though the void ratios of the mixtures are lower than those of the sand alone. The stiffness reduction of the mixtures of LBSB and 0.1 mm mica can be attributed to the effects of both platy particle shape and fine particle size of mica. The effects of particle characteristics on the small strain response of geomaterials can be explained using the proposed porous discontinuous-solid model. A dry soil elementis assumed to consist of two phases, namely the pore and the discontinuous solid, where the stiffness of discontinuities is represented by a shear wave velocity through the contact network, Vcontact, which is a function of particle characteristics. Particle size and particle shape create both macro effects (e.g. effect of void ratio) and micro effects (at the contact level) on the small strain response of granular materials. The model postulates that an increase in void ratio will increase travel length, and hence decrease Vs. The model suggests that the macro effect of void ratio on Gmax can be normalised using the theoretical (universal) void ratio function, F(e) = (1 + e)−3, which can be applied for both clays and sands with various void ratio range. And by doing so, the micro effects of particle shape and particle size can be observed and taken into account using a particle characteristic coefficient, Cp, which increases with increasing particle diameter, sphericity, and roundness. In addition, during testing relatively stiff specimens using the RCA, equipment compliance was observed, leading to an significant underestimation of the natural frequencies of the specimens. In order to identify the source of compliance and evaluate the influence of equipment compliance on the measured data using the RCA, several finite element (ABAQUS) models were developed. The numerical analysis results suggest that the stiffness of the drive mechanism, the mass and/or fixity of the test base, and calibration bar design significantly affect the test results. To correct for the effects of system compliance, a new model termed two spring model is developed. The model key parameters i.e. stiffness of the equipment, Kequipment, and mass polar moment of inertia of the drive mechanism, I0, can be calibrated through testing of a series of aluminium calibration bars.

The main purpose of this thesis is the development of a new constitutive soil model emphasising the use of thermodynamic principles. This new approach to plasticity modelling, termed ‘hyperplasticity’, was first developed by Collins and Houlsby (1997) and Houlsby and Puzrin (2000). This idea has been further extended to continuous hyperplasticity in which smooth transitions between elastic and plastic behaviour can be modelled (Puzrin and Houlsby, 2001b). Applying hyperplasticity to this research, a kinematic hardening model specified by means of two scalar functionals is used to accommodate the effect of stress history on stiffness. A rate-dependent calculation for an approximation of the incremental stress-strain response is introduced. The model developed in the research is named ‘kinematic hardening modified Cam-clay (KHMCC) model’ and requires eight parameters (plus an extra parameter for rate-dependent analysis). Triaxial test results from the Asian Institute of Technology (AIT) and cyclic undrained triaxial data from Chulalongkorn University are employed to establish the soil parameters for the new model. The model is initially developed in terms of triaxial stress-strain parameters for the purpose of comparison with the experimental data on Bangkok clay. The model is expressed in FORTRAN code for implementation into the OXFEM finite element program. Two examples of real geotechnical projects in Bangkok (a road embankment and tunnelling in soft ground) are analysed under plane strain conditions. Comparisons of the numerical analysis results with field data are made. In addition, factors affecting the results of the analysis such as stress history and K0, are investigated.

Orientador : Alberto Pio Fiori / Co-orientadores : Naldy Emerson Canali, Paulo Cesar Soares / Dissertação (mestrado) - Universidade Federal do Paraná / Resumo: O presente trabalho consiste na aplicação de um modelo matemático acoplado à utilização de técnicas de mapeamento temático em ambiente SIG, objetivando desenvolver uma metodologia voltada ao estudo dos fatores de segurança na estabilidade de encostas. Este modelo matemático, estruturado na distribuição das forças atuantes em uma vertente infinita, leva em conta a influência da vegetação sobre as encostas, e os índices físicos e as propriedades mecânicas de cada unidade de solos. Os índices físicos e as propriedades mecânicas dos solos, bem como o papel da vegetação na estabilidade das encostas, permitiram a quantificação dos riscos de escorregamento para cada unidade de solo analisada. Como um primeiro resultado, apresenta-se a criação de um banco de dados digital, contendo informações preliminares das características do terreno, através da confecção de cartas temáticas tais como: topográfica, geológica, declividade e de uso e ocupação do solo dos anos de 1953, 1986 e 1999. Todas essas cartas foram transferidas para meio digital, permitindo as suas interpolações em ambiente SIG. Para os cálculos dos fatores de segurança das vertentes foram considerados como atributos: a inclinação das encostas; os índices físicos e as propriedades mecânicas dos solos, com a presença de água (solo saturado); a influência da cobertura vegetal (peso das árvores e sistema solo-raiz) e a força do vento atuando nas copas das árvores. As propriedades físicas e mecânicas dos solos (ângulo de atrito interno e coesão), foram determinadas pelo método de retroanálise desenvolvido por Lopes (1995). O método envolveu inicialmente a foto interpretação de cicatrizes de escorregamentos, para a elaboração de perfis topográficos, que determinaram os pares de altura/ângulo de taludes limites. Como resultado final, apresenta-se uma metodologia que possibilita o mapeamento dos diferentes graus de riscos ao escorregamento, a partir da interpolação em geoprocessamento das cartas temáticas com os dados obtidos pelas equações de fator de segurança. / Abstract: The purpose of this dissertation is to develop a methodology concerned with the study of safety factors on slope stability through the application of a mathematical model combined with the utilisation of techniques of thematic mapping within GIS. Such mathematical model, structured on the distribution of operating forces on an infinite slope, takes into account the influence of vegetation cover over slopes, as well as the physical index and mechanical features of each soil unit. The physical index and the mechanical features of the soils, as well as the role played by vegetation cover on slope stability, allowed the quantification of the risks of landslides for each soil unit analysed. As a first result, was created a bank of digital data comprising preliminar information on thematic charts such as: topographical, geological, geomorphologic, declivity, land use and land occupation for 1953, 1986 and 1999. All these charts were digitalised in order to make possible its interpolations within GIS. The safety factors were calculated based upon attributes such as: inclination of the slopes, physical index and mechanical features of the soils, existence of water (saturated soil), influences of vegetation cover (trees weight and system soil-root) and wind speed on the top of the tress. The mechanical and physical properties (angle of internal friction and cohesion) were estimated by the method of retroanalysis developed by Lopes (1995). Such method initially involved photo interpretation of landslide scars so as to elaborate topographical profiles, which determined the pairs of height/angle of slopes-limit. The final achievement is shown to have created a methodology that allows the mapping of different degrees of slope failure risks, by using the interpolation of thematic charts within geoprocessing, with the data obtained by means of equations of safety factor.