Cyclic behaviour modeling of geotechnical structures including regularization methods

The main technical objective of this PhD thesis deals with the development of a soil behavior numerical tool. It should be robust, efficient and adapted to model the mechanical behavior of geotechnical structures (e.g. embankment dam) under the worst loading scenarios such earthquakes. In the finite element method, the quality results is directly linked to the soil constitutive model, the integration scheme and the numerical resolution. In this PhD Thesis, the ECP elastoplastic soil model is introduced in Code-Aster through an implicit scheme. An implicit scheme ensures to respect the theoretical formulation of the model. The ECP constitutive model is one of the models available in the literature to represent the behaviour of different kinds of soils under cyclic loadings and it is used since the 80's by hydraulic engineers at EDF. It is expressed in terms of effective stresses and infinitesimal strains. The developments are validated for laboratory tests in a large scale of loading paths. On the other hand, the study of a sand embankment was performed and compared to the results obtained with the finite element software GEFDyn developed at ECP. The ECP model is based on a non-associated flow rule and it is able to reproduce a softening behavior. When shear bands occur in the structure, these properties lead to a pathological sensitivity of the results depending on the mesh size. Therefore, a regularization technique has to be used to circumvent this problem and to obtain objective results with respect to the mesh. The second gradient of dilation model is thus chosen to be coupled to the ECP model and in this way, to ensure a spatiotemporal independence of results. However, this mesh independence still evolves in a potential domain of solutions, when instability occurs. The simulations of drained biaxial tests on laboratory samples show a dependence of shear bands thickness in regard to the initial stress state and material properties. An analytical problem of a dilatant shear band is used to extract the key factors. These theoretical solutions are compared and validated to numerical responses, which are in good accordance. A bearing capacity problem was also solved to demonstrate the potential of the method. The conclusion of this work establishes the main role of initial stress state over the shear band thickness in the context of the second gradient of dilation model and the models based on Cam-Clay approach. (author)