Crystal plasticity parameter identification via statistical relevant micropillar compression

Abstract

In order to predict ductile damage initiation at the microstructure level, especially for multi-phase materials, it is essential to have high-fidelity crystal plasticity parameters. They need to accurately represent the evolving phase contrast, which implies that the initial phase contrast and the individual strain hardening of the phases has to be mapped precisely. This paper presents a methodology for calibrating the parameters of a phenomenological crystal plasticity model for a DP800 steel based on the critical resolved shear stress from in situ micropillar compression tests taken out of macroscopic tensile tests at various prestrain levels. Furthermore, the influence of mechanical size effects was incorporated through the inclusion of statistical relevant micropillar compression tests of varying prestrains and dimensions. The data were used to calibrate a model, which successfully predicted the homogenized macroscopic stress–strain curve from uniaxial tensile tests with a mean absolute error of only $20.7\pm 7.7\mathrm{MPa}$ and a mean absolute percentage error of 3.3%. Furthermore, it was shown that the influence of the strain hardening of the martensite can be neglected under certain conditions, especially when predicting the homogenized stress response for low strains. This result demonstrates the importance of high-fidelity parameter calibration for damage prediction, when compared to a synthetic parameter set, which leads to a different stress and strain partitioning for the same homogenized stress–strain curve.