Understanding the relationships between microstructures and properties of materials is a key to developing new materials with more suitable qualities or employing the appropriate materials in special uses. In the present world of material research, the main focus is on microstructural control to cost-effectively enhance properties and meet performance specifications. This present work is directed towards improving the fundamental understanding of the microscale deformation mechanisms and mechanical behavior of metallic alloys, particularly focusing on face centered cubic (FCC) structured metals through a unique computational methodology called three-dimensional dislocation dynamics (3D-DD). In these simulations, the equations of motion for dislocations are mathematically solved to determine the evolution and interaction of dislocations. Microstructure details and stress-strain curves are a direct observation in the simulation and can be used to validate experimental results. The effect of initial dislocation microstructure on the yield strength has been studied. It has been shown that dislocation density based crystal plasticity formulations only work when dislocation densities/numbers are sufficiently large so that a statistically accurate description of the microstructure can be obtainable. The evolution of the flow stress for grain sizes ranging from 0.5 to 10 µm under uniaxial tension was simulated using an improvised model by integrating dislocation pile-up mechanism at grain boundaries has been performed. This study showed that for a same initial dislocation density, the Hall–Petch relationship holds well at small grain sizes (0.5–2 µm), beyond which the yield strength remains constant as the grain size increases.

L’acier 316L(N) est le matériau de référence pour les structures du circuit primaire des réacteurs nucléaires de quatrième génération, en raison de leur résistance mécanique à la température de fonctionnement, de l’ordre de 550°C. La thèse a permis de développer un modèle polycristallin, capable de prédire le comportement de ces aciers, basé sur la description du glissement viscoplastique des dislocation à haute température, de mise en œuvre simple et avec l’identification d’un nombre de paramètres matériau limité. La démarche de modélisation a été progressive. Lors de la première étape, nous avons proposé et validé une loi d'homogénéisation élasto-viscoplastique à champs moyens, grâce à de nombreux calculs par éléments finis, en considérant des durcissements plastique et des viscosités cristallines. Ensuite, un modèle de viscoplasticité cristalline, reposant sur les lois d’évolution des densités de différents types de dislocations, a été implémenté et les prédictions ont été validés en considérant un très grand nombre de résultats expérimentaux à faible. Le modèle a ensuite été enrichi afin de prendre en compte les mécanismes physiques supplémentaires observés à température élevée, comme la montée des dislocations, le vieillissement dynamique et l’apparition d’une structure de dislocation très hétérogène. Le modèle proposé nécessite uniquement l’ajustement de trois paramètres par identification inverse, utilisant seulement des essais de traction monotone avec saut de vitesse. Les prédictions du comportement mécanique en chargement uniaxial et cyclique sont également en bon accord avec les mesures expérimentales aux températures élevées. / The 316L(N) stainless steels is the reference material for the primary circuit structures of fourth-generation nuclear reactors. This alloy present high mechanical resistance at the operation temperature range of these reactors, of the order of 550 °C. This PhD allowed to develop a polycrystalline model based on the description of the viscoplastic dislocation slip at high temperatures, with straightforward implementation and with identification of a limited number of material parameters. The modeling process was progressive. In a first step, we proposed and validated a mean-field elastic-viscoplastic homogenization law, in comparison to numerous finite element calculations, considering crystalline plastic hardening and crystalline viscosity. Then, a model of crystalline viscoplasticity, based on the evolution laws of the different dislocations densities was implemented and the predictions were validated considering a very large number of experimental results at low temperature. The model was then enhanced to take into account the additional physical mechanisms observed at high temperature, such as dislocation climb, dynamic strain aging and the appearance of a very heterogeneous dislocation structure. The proposed model requires the adjustment of only three parameters by inverse identification, using only monotonic tensile tests at different strain rates. The mechanical behavior predictions in uniaxial and cyclic loading are also in good agreement with experimental measurements at high temperature.