Ductile fracture criteria assessment on AA2024-T3 aluminum alloys sheet using digital image correlation technique and analytical calculation

The accuracy of predicting fracture behaviors in AA2024-T3 sheets was investigated using ductile fracture criteria, specifically Lou-Huh and modified Mohr-Coulomb (MMC), while considering the anisotropic yield criterion to assist the experimental DIC technique and numerical simulation. First, the relationship pairs of stress triaxialities and equivalent plastic strains were established in the form of fracture loci (FLs) based on Lou-Huh and MMC in combination with Hill’48. Second, experimental and numerical simulations on Nakajima specimens were conducted under uniaxial, plane strain, and biaxial stress states to evaluate the accuracy of these fracture criteria predictions. Consequently, the FLs derived from the investigated fracture criteria were incorporated into the finite element (FE) models using the element-deletion technique in ABAQUS. The strain distribution histories during the forming tests were monitored using DIC techniques and compared with the results from the FE simulations. Finally, the accuracy of the fracture criteria predictions was presented in terms of the predicted punch force-displacement curves, as well as strain and stress based on fracture forming limit curves (FFLC, FFLSC), strain and stress paths, and fracture locations on the specimens, all of which corresponded well with experimental results. Additionally, the predicted drawing depths and equivalent plastic strains at fracture were accurately established. The predicted drawing depth from the Lou-Huh uniaxial stress state specimen showed a slight error of 0.45%, while the least accurate prediction from the MMC biaxial loading specimen resulted in an error of 3.00%. Moreover, the simulated equivalent plastic strains based on MMC for all specimens exhibited smaller errors compared to those based on Lou-Huh. Lastly, the simulated fracture locations and shapes using the Lou-Huh model aligned more closely with the experimental results than those predicted by the MMC model.