[Truncated abstract] This research has studied the penetration and extraction resistance profiles of different types of penetrometers in soft clay. The penetrometers of interest include the cone, T–bar, ball and plate. Effects of the surface roughness and aspect ratio of the T–bar penetrometer on its resistance have also been investigated. Undrained shear strength, Su, profiles derived from the penetration tests are compared with the shear strengths measured from field vane shear tests and laboratory (triaxial and simple shear) tests. Both in situ and centrifuge model penetration tests were undertaken for the research. In addition, ‘undisturbed’? tube samples were retrieved from both the field and the centrifuge strongbox samples (after completion of the centrifuge tests) for laboratory testing. The in situ testing was carried out in Western Australia, at the Burswood site near Perth, with tests including cone, T–bar, ball and plate penetrometer tests, and vane shear tests. Interestingly, the T–bar, ball and plate (‘full-flow’) penetrometers showed a narrow band of resistance profiles both during penetration and extraction, with a range of around 15 % between the highest and lowest profiles and standard deviation of 15 %. However, the cone penetrometer gave similar resistance at shallow depths but increasingly higher penetration resistance at depths greater than 7 m – a phenomenon that is also common in offshore results. During extraction, the cone penetrometer gave a higher resistance profile than the full–flow penetrometers for much of the depth of interest. The Su profile measured directly from the vane shear tests falls within the Su profiles derived from the penetration resistances of the full–flow penetrometers, using a single bearing factor, N = 10.5 (the value originally suggested in the literature for a T–bar penetration test). Again, the cone penetrometer demonstrated diverging results, requiring two separate values for the cone factor, Nkt (10.5 initially increasing to 13 for depths below 10 m) in order to give Su similar to the vane shear tests. This highlights the possible variability of the cone factor with depth. Cyclic penetration and extraction tests were performed at specific depths for each fullflow penetrometer. These tests comprised displacement cycles of ±0.5 m about the relevant depth, recording the penetration and extraction resistances over five full cycles. The results may be used to derive the remoulded strength and sensitivity of the soil. Laboratory tests such as triaxial and simple shear tests were performed on ‘undisturbed’ tube samples retrieved from the same site to evaluate the in situ shear strengths in the laboratory. However, the resulting Su data were rather scattered, much of which may be attributed to variable sample quality due to the presence of frequent shell fragments and occasional silt lenses within the test samples. In general, N factors for the full–low penetrometers, back–calculated using Su values measured from the simple shear tests, fell mainly in a range between 9.7 and 12.8 (between 10.4 and 12.2 for the standard size T–bar (250 mm x 40 mm))

Dans la pratique actuelle du génie géotechnique, les essais de pénétration tels que les CPT, SPT, Panda sont largement utilisés pour caractériser mécaniquement les sols, au travers notamment d’une caractéristique de rupture appelée résistance de pointe. Par ailleurs, les dernières évolutions technologiques apportées aux essais de pénétration dynamique (Panda 3) permettent d’obtenir pendant chaque impact une courbe charge–enfoncement donnant la charge en pointe en fonction de l’enfoncement à partir de la mesure et du découplage des ondes générées durant l’essai. L’exploitation de cette nouvelle courbe fournit des informations non seulement sur la résistance de pointe dynamique mais également sur des paramètres mécaniques complémentaires mis en jeu pendant l’enfoncement de la pointe. L’objectif de cette thèse est de développer un modèle numérique en 2D capable de reproduire les signaux pénétrométriques obtenus expérimentalement par essais de type statique ou dynamique. Ce modèle est basé sur la méthode des éléments discrets avec une loi de contact linéaire simple. Une fois le modèle validé, une étude paramétrique a été réalisée en jouant essentiellement sur les modes d’application de la sollicitation (vitesse d’impact ou de pénétration), la granulométrie du matériau ainsi que l’arrangement granulaire (variation de la densité). Outre l’influence de ces paramètres sur les signaux pénétrométriques et la résistance de pointe mesurée, une attention particulière est portée sur l’analyse micromécanique : dissipation d’énergie dans le milieu, évolution des chaines de force, orientations des contacts. Cette analyse nécessite de développer des outils numériques spécifiques afin de mieux comprendre le mécanisme de l’enfoncement et tenter d’expliquer la réponse mécanique macroscopique obtenue. L’effet de la vitesse n’influence significativement que sur les essais de pénétration statiques et dynamiques en régime d’écoulement dense. A vitesse d’enfoncement comparable, il n’y a aucune différence significative au niveau microscopique entre les deux modes de sollicitation statique et dynamique. En ce qui concerne l’influence des caractéristiques du matériau, les résultats obtenus par le modèle numérique conforment aux celui réel lors que le frottement entre particules ou la compacité du milieu varie. Concernant la granulométrie, la variation de la courbe charge-enfoncement et la force de pointe dynamique augmente lorsque le diamètre moyen augmente. / In the field of in situ mechanical characterization of soils, penetration tests are commonly used. Penetration tests measure the properties of soils in the domain of large deformations. The tip resistances, deduced from pile driving theory, can be measured either in dynamic conditions (q d ) either in static conditions (q c ). Recently, the measurement technique in dynamic conditions has been improved and it is now possible to record the whole response of the soil during one impact in terms of tip force and penetration distance. The exploitation of this new curve provides information not only on dynamic tip resistance but also on additional mechanical parameters involved during the driving of the tip. The objective of this work is to develop a numerical model in 2D able to reproduce the penetrometric record obtained experimentally by static or dynamic penetration tests. This model is based on the discrete element method with a simple linear contact model. After the validation of the model, a parametric study was performed essentially on the loading type (static or dynamic), the penetration rate, the particle size of the granular material and the arrangement (density variation). Besides the influence of these parameters on the penetrometer signals and the tip resistance, a particular attention was focused on micromechanical analysis: energy dissipation in the medium, force chain evolution, contact orientation. This analysis requires the development of specific numerical tools to better understand the penetration mechanism and try to explain the macroscopic mechanical response obtained. The penetration rate influences significantly only in the dense flow regime on the static and dynamic penetration tests. There is no significant microscopic difference between static and dynamic penetration tests with similar penetration rates. Regarding the influence of the characteristics of the material, the numerical results obtained conform to the real results when the particle friction or the compactness of the medium varies. Concerning the particle size, the dynamic signal variation and the dynamic tip force increases when the average particle diameter increases.