Key points are not available for this paper at this time.
A severe obstacle for the routine use of crystal plasticity models is the effort associated with determining their constitutive parameters. Obtaining these parameters usually requires time-consuming micromechanical tests that allow probing of individual grains. In this study, a novel, computationally efficient, and fully automated approach is introduced which allows the identification of constitutive parameters from macroscopic tests. The approach presented here uses the response surface methodology together with a genetic algorithm to determine an optimal set of parameters. It is especially suited for complex models with a large number of parameters. The proposed approach also helps to develop a quantitative and thorough understanding of the relative influence of the different constitutive parameters and their interactions. Such general insights into parameter relations in complex models can be used to improve constitutive laws and reduce redundancy in parameter sets. The merits of the methodology are demonstrated on the examples of a dislocation-density-based crystal plasticity model for bcc steel, a phenomenological crystal plasticity model for fcc copper, and a phenomenological crystal plasticity model incorporating twinning deformation for hcp magnesium. The approach proposed is, however, model-independent and can be also used to identify parameters of, for instance, fatigue, creep and damage models. The method has been implemented into the Düsseldorf Advanced Material Simulation Kit (DAMASK) and is available as free and open-source software. The capability of translating complex material response into a micromechanical digital twin is an essential precondition for the ongoing digitalization of material property prediction, quality control of semi-finished parts, material response in manufacturing and the long-term behavior of products and materials when in service.
Sedighiani et al. (Fri,) studied this question.