Le comportement mécanique des alliages métalliques industriels, notamment ceux utilisés dans le domaine de l’aéronautique, est contrôlé par la présence de différents types de précipités et par la nucléation et propagation de défauts cristallins tels que les dislocations. La compréhension du comportement de ces matériaux nécessite des modèles continus afin d’accéder à l’échelle macroscopique. Cependant, même aujourd’hui, les théories conventionnelles de la plasticité utilisent des variables mésoscopique et des équations d’évolution qui ne reposent pas sur la notion de transport de dislocations. En conséquence, ces théories sont basées sur des lois phénoménologiques qu’il est nécessaire de calibrer pour chaque matériau et chaque application. Il est donc souhaitable d’établir le lien entre les échelles micro et macro afin de générer une théorie continue de la plasticité déduite analytiquement des équations fondamentales de la dynamique des dislocations. L’objet de cette thèse est précisément de contribuer à l’élaboration d’une telle théorie. La première étape a consisté à établir rigoureusement la procédure de changement d’échelle dans une situation simplifiée. Nous avons alors abouti à un système d’équations de transport hyperboliques sur des densités de dislocations contrôlées par des contraintes locales de friction et de backstress qui émergent du changement d’échelle. Nous avons ensuite développé une procédure numérique pour calculer ces termes et analyser leur comportement. Finalement, nous avons développé un schéma numérique efficace pour intégrer les équations de transport ainsi qu’un schéma spectral multi-grille pour résoudre l’équilibre élastique associé à un champ de déformation propre quelconque dans un milieu élastiquement anisotrope et inhomogène. / The mechanical behavior of industrial metallic alloys, in particular those used in the aerospace industry, is controlled by the existence of several types of precipitates and by the nucleation and propagation of crystalline defects such as dis- locations. The understanding of this behavior requires continuous models to access the macroscopic scale. However, even today, conventional plasticity theories use mesoscopic variables and evolution equations that are not based on the transport of dislocations. Therefore, these theories are based on phenomenological laws that must be calibrated for each material, or, for each specific applications. It is therefore highly desirable to make link between the micro and macro scales, in order to derive a continuous theory of plasticity from the fundamental equations of the dislocation dynamics. The aim of this thesis is precisely to contribute the elaboration of such a theory. The first step has consisted to rigorously establish a coarse graining procedure in a simplified situation. We have then obtained a set of hyperbolic transport equations on dislocation densities, controlled by a local friction stress and a local back-stress that emerge from the scale change. We have then developed a numerical procedure to compute these local terms and analyze their behavior. Finally, we have developed an efficient numerical scheme to integrate the transport equations as well as a multigrid spectral scheme to solve elastic equilibrium associated to an arbitrary eigenstrain in an elastically heterogeneous and anisotropic medium.

Owing to continuous miniaturization, many modern high-technology applications such as medical and optical devices, thermal barrier coatings, electronics, micro- and nano-electro mechanical systems (MEMS and NEMS), gems industry and semiconductors increasingly use components with sizes down to a few micrometers and even smaller. Understanding their deformation mechanisms and assessing their mechanical performance help to achieve new insights or design new material systems with superior properties through controlled microstructure at the appropriate scales. However, a fundamental understanding of mechanical response in surface-dominated structures, different than their bulk behaviours, is still elusive. In this thesis, the size effect in a single-crystal Ti alloy (Ti15V3Cr3Al3Sn) is investigated. To achieve this, nanoindentation and micropillar (with a square cross-section) compression tests were carried out in collaboration with Swiss Federal Laboratories for Materials Testing and Research (EMPA), Switzerland. Three-dimensional finite element models of compression and indentation with an implicit time-integration scheme incorporating a strain-gradient crystal-plasticity (SGCP) theory were developed to accurately represent deformation of the studied body-centered cubic metallic material. An appropriate hardening model was implemented to account for strain-hardening of the active slip systems, determined experimentally. The optimized set of parameters characterizing the deformation behaviour of Ti alloy was obtained based on a direct comparison of simulations and the experiments. An enhanced model based on the SGCP theory (EMSGCP), accounting for an initial microstructure of samples in terms of different types of dislocations (statistically stored and geometrically necessary dislocations), was suggested and used in the numerical analysis. This meso-scale continuum theory bridges the gap between the discrete-dislocation dynamics theory, where simulations are performed at strain rates several orders of magnitude higher than those in experiments, and the classical continuum-plasticity theory, which cannot explain the dependence of mechanical response on a specimen s size since there is no length scale in its constitutive description. A case study was performed using a cylindrical pillar to examine, on the one hand, accuracy of the proposed EMSGCP theory and, on the other hand, its universality for different pillar geometries. An extensive numerical study of the size effect in micron-size pillars was also implemented. On the other hand, an anisotropic character of surface topographies around indents along different crystallographic orientations of single crystals obtained in numerical simulations was compared to experimental findings. The size effect in nano-indentation was studied numerically. The differences in the observed hardness values for various indenter types were investigated using the developed EMSGCP theory.