Towards Quantitative Crystal Plasticity Model Validation Using Experimental In-plane Deformation Maps

Background Mechanistically based constitutive models incorporate several key parameters that describe the macroscopic response and microstructure evolution of a polycrystalline metal alloy under external loading. To date, careful calibration of these parameters has for the most part relied on characterization methods that lack spatial resolution. Objective With the intent of facilitating future model development, this work investigates feasibility of using Digital Image Correlation (DIC) to characterize realistic strain fields in polycrystalline metals. Methods First, crystal plasticity based simulations are used to obtain displacement and strain fields in a polycrystalline aluminum alloy with average grain size ~ 200 µm under uniaxial tension up to 10% macroscopic strain. Second, random speckle patterns are generated on a metal alloy surface and their optical images are acquired. The images are numerically deformed according to the displacement field predicted by the model. DIC strain maps are obtained using standard methods with subset diameter within 15–50% of the grain size and are compared against the applied fields. Results Overall, the results demonstrate that DIC is capable of obtaining sufficiently accurate strain fields to validate or challenge simulations. At various applied strain magnitudes, DIC strain maps measured from a high-quality pattern are able to reproduce over 95% of the simulation strain field with relative errors less than 10%, and over 60% of the simulation strain field with relative errors less than 5%. Achieving this level accuracy relies on the proper DIC analysis settings. Conclusions When the deformation is locally homogeneous, the DIC accuracy increases as the subset window size increases. In contrast, in regions where strain within each subset is highly localized, DIC accuracy increases as the subset window decreases. Suggestions for improving standard DIC algorithms to enable quantitative model-data comparison are discussed.