Robustness of crystal plasticity parameters validation with Digital Image Correlation for fatigue modeling

Abstract

Advances in computational power has made microstructure-sensitive modeling more accessible in industry. For example, crystal plasticity models enable the understanding of the microstructure influence on fatigue crack initiation and propagation. However, these models require extended parameterizations with multiple coefficients, making it not only time consuming, but often inaccessible to engineers. Although multiple methods have approached parameters estimation by correlating models with macroscopic experimental measurements from a monotic test, macroscale calibrations carry significant local uncertainties. In this sense, Digital Image Correlation (DIC) can offer mesoscale validation by providing a full-field measurement of strain that can be compared to the simulation. The main difficulty of this approach is the strain accuracy and the signal-to-noise ratio. This work employs DIC to study parameter sensitivity and adequate parameterization strategies. We explore the strain difference after a model parameter variation using virtual experiments with different loading scenarios. A simple criterion comparing strain sensitivity to the model parameters and the measured DIC noise is proposed to assess the calibration robustness. Different case studies highlight the difficulty of validation under cyclic paths owing to a low signal-to-noise ratio. This is the main limitation for calibrations of models that aim to capture the cyclic behavior. We conclude with further recommendations to use DIC for model parameterization.