A comparison between FEM predictions and DIC results of crack tip displacement field in AA2024-T3 CT specimens

• DIC results are compared with FEM predictions in CT specimens with a thickness of 1 mm. • The elastic behaviour is well predicted, which indicates a good modelling of geometry and Young modulus. • The numerical CTODp is below experimental values, even for the plane stress state. • The change of isotropic hardening model from Swift to Voce increased the plastic CTOD. • The material models must be validated with DIC measurements in thin specimens. In the current paper, fatigue crack growth is modelled using a node-release strategy based on crack tip plastic deformation. The accuracy of material elastic–plastic modeling is crucial for the quality of the numerical predictions. Stress–strain loops obtained in low cycle fatigue (LCF) were used to fit material constants for 2024-T3 aluminium alloy, which is the procedure usually followed. The material model was validated with Digital Image Correlation (DIC), with results obtained at different distances behind the crack tip in a thin CT specimen. Significant differences were found between FEM predictions and experimental results for plastic COD, which were attributed to an inadequate elastic–plastic model. In fact, the LCF tests always have small strain ranges, much smaller than the observed at the crack tip. The change of the isotropic hardening law from Swift to Voce improved the predicted plastic COD values. The final conclusion is that the fitting of the material parameters using LCF results must be validated with DIC-FEM comparisons for the case of thin specimens.