A Combined Physical and Virtual Digital Image Correlation Technique for Calibration and Validation of Elasto-Plastic-Damage Models

The calibration and validation of elasto-plastic-damage models are commonly performed using simple uniaxial or more complex multi-axial setups. While, the uniaxial tests fail to incorporate the influence of triaxiality, the multi-axial methods present challenges related to complexity of setup and specimen design.

This study proposes a simple methodology combining Digital Image Correlation (DIC) and the Finite Element Method (FEM) to parameterize triaxiality-dependent elasto-plastic-damage models using specimens with non-uniform geometries and uniaxial loading.

Uniaxially loaded specimens with a hole or notch were utilized to obtain triaxiality-dependent inhomogeneous response and the resulting strain fields were captured using DIC. The strain fields and load-displacement response were then reproduced using FEM to extract the parameters of the Gurson-Tvergaard-Needleman (GTN) model. A Virtual-DIC method was also developed to introduce noise in the FEM response arising from speckling.

Comparisons between DIC, FEM and Virtual-DIC clearly demarcated the influence of speckles on variations in local strain fields obtained from DIC at low applied strains. For certain cases, Virtual-DIC showed a closer agreement with DIC than FEM indicating that incorporating the variations introduced by speckles may provide more accurate calibration of the model. The approach accurately captured the evolution of triaxiality at the process zone of crack tip, thus validating the suitability of the model and its parameters.

Overall, the results suggest that the proposed approach can be used to calibrate and validate elasto-plastic-damage models accurately.