CATSIZE

From the industrial point of view as well as the scientific one, multiple scales models and simulation tools are needed to provide a good prediction of the mechanical behavior of metallic alloys from the knowledge of the microstructure (volume fractions, sizes of constituents, crystallographic orientations etc). The stakes of the present project in terms of applications are (i) for ARCELOR: to optimize the microstructures of steels and to develop new ones able to give the desired functional properties targeting alloy design (fine grained steels, optimized second phase etc), and, (ii) for EDF: to study the behavior and the fracture properties of steels used in nuclear plants (in the continuity of the PERFECT project). The difficulties arise from the fact that the resolution is truly multiscale. Until now, different approaches were developed to study and solve theoretically as well as numerically the aforementioned problem. They are of three types: Discrete Dislocation Dynamics, Generalized Continuum Mechanics and Micro-Macro scale transitions. Nevertheless, each of them considered alone is not able to respond to the highlighted objectives. That's the reason why, this project demands an efficient and concerted collaboration between the different investigators of the three recalled modeling types. Thus, it appears to be scientifically relevant and useful for industrial partners to propose a new approach which enables to optimize the accuracy of the predictions (e.g. size effect) and an operational manipulability in a concern to apply the methodology to complex industrial cases. In this framework, both LPMM and ARCELOR have recently initiated a first approach adapted to the objectives and aiming to respond to the current needs. This approach is based on an original micromechanical model which accounts for the presence of geometrically necessary dislocations (GND) due to lattice incompatibility in the grain boundary region. A « physical » internal length linked to strong strain gradients is naturally introduced. First results well reproduced a grain size effect on strain hardening. This first approach lies in an hypothesis of mean fields within phases; this one will be denoted Mean Field Approaches with Internal Length(s) Scale(s) ('ACMLI' in French). Even though this methodology seems to be promising, it needs some steps of validation of the different assumptions and predictions. It is then judicious to use in parallel finer but more time consuming approaches such as Discrete Dislocation Dynamics (DDD) developed at SIMAP (GPM2) and 3D polycrystals based on Generalized Continuum Mechanics performed at MINES Paris. This comparative validation will ensure the relevancy of the approach developed in this project at the different scales.