Oana Cazacu

In this paper, we present a deep generative framework based on Conditional Variational Autoencoders (CVAE) developed for inverse design of crystallographic textures. Unlike forward prediction approaches that map microstructures to macroscopic properties, this work addresses the challenging inverse problem: given desired Lankford coefficients for three orientations in the plane of the sheet, e.g. r0, r45, r90, generate the crystallographic textures that match the targeted macroscopic plastic anisotropy. To this end, we developed a new methodology for generating an extensive training dataset consisting of spherical harmonic representations of various textures and associated Lankford coefficients, calculated using a crystal plasticity model. The developed CVAE, called TextureNet, learns a probabilistic mapping from the target r0, r45, r90 to crystallographic textures. Once trained, TextureNet is able to generate several potential textures. The capabilities of the developed computationally efficient framework for inverse materials design are illustrated for several test cases. The results demonstrate that the generated textures reproduce the target anisotropy.

Micro-mechanical study of damage evolution in isotropic metallic materials KNAAK Karl, REVIL-BAUDARD Benoit, CAZACU Oana [button link="https://www.mrforum.c

This paper presents a new polycrystalline model and Lagrangian computational framework for describing the large-scale thermo-mechanical response of energetic materials under dynamic loadings. In our multi-scale computational polycrystalline framework, at the grain level, the elastic response is modeled using an anisotropic Hooke's law, while the plastic behavior is described with a recently developed quadratic anisotropic single-crystal model that accounts for the intrinsic symmetries associated with the lattice of the constituent crystals. The orientation, plastic strains, and stresses in the individual grains are continuously updated, so the predicted macroscopic scale response takes into account the evolution of the thermo-mechanical state at the meso-scale. First, we illustrate the polycrystalline model capabilities by simulating the response of a pentaerythritol tetranitrate (PETN) polycrystalline high energetic system when subjected to dynamic compression. It is shown that there are strong differences in temperature and stresses between the constituent grains, depending on their relative orientation with respect to the wave direction. Moreover, it is shown that the rise in temperature in certain grains may be well in excess of the macroscopic value. We also present 3D finite element simulations of the impact of a penetrator made of a high-strength steel containing the same polycrystalline PETN system. Insights into the complex interactions between the energetic system and the metallic casing material are provided. Furthermore, it is shown that if the crystallinity is neglected, the predicted temperature rise and the extent of the zone of maximum heating in the energetic system during the impact event differ noticeably from those obtained with our polycrystalline model, which accounts for the crystallinity of the PETN material and the anisotropy in the plastic flow of its constituent crystals.

For realistic simulations the use of improved constitutive models for the description of the mechanical behaviour and accurate prediction of formability are essential. Key is the development and use of well-posed yield functions which are applicable to any type of loadings and a correct and efficient implementation in the FE framework. For general three-dimensional (3-D) loadings,

Given the current trends of globalization and active competition, the reduction of the lead time in transfer of new alloys and design is critical. Virtual testing and design using finite element (FE) analyses may contribute to this reduction. Indeed, FE analysis has been applied extensively to compare design options, understand the influence of process conditions on both

The 6000-series aluminum alloys are generally used as structural materials in building structures and automotive parts where the engineering design requires the use of a light weight material. The credibility of any finite element analysis conducted for these materials depends critically on the constitutive models used to describe the plastic deformation and fracture of these