Virtual Material Testing - A Computational Tool for the Prediction of Macroscale Properties based on Homogenization

Research Project, 2011 – 2013

The proposed project concerns the development of a computational strategy, coined *virtual material testing*, for predicting the macroscopic mechanical properties such as elasticity, plasticity and damage. Essential features are computational homogenization on Representative Volume Elements (RVE) and the statistical sampling of RVE-realizations. The tool will be applied to a few model materials with distinct (and different) mesoscale features: Duplex stainless steel (DSS) and nodular cast-iron. In particular, the purpose is to produce *virtual experimental data* for upscaling, i.e. for calibration of phenomenological constitutive models relevant to the chosen model materials, or it will be used as part of *nested* meso-macroscale computations (FE^2-technology). The following main tasks will be carried out: (1) Develop the *virtual material testing* strategy for estimating bounds on effective properties with given confidence. (2) Develop a strategy for dimensional reduction from 3D subscale modeling to 2D (or 1D) macroscale modeling. (3) Compute macroscale yield and subsequent loading surfaces in the macroscale stress space. (4) Simulate Manson-Coffin and Wöhler relations pertinent to Low-Cycle-Fatigue (LCF). (5) Calibrate the subscale constitutive models from macroscale tests, whereby experimental data from the literature and from (current and previous) partners involved in experimental research will be used.