Comparative Study of Uncoupled Ductile-Fracture Models on Fracture Prediction of Structural Steels under Monotonic Loading

Since the middle of the last century, numerous fracture models have been reported in the literature. These models, whether proposed on the basis of theoretical derivation, experimental observation, or a combination of the two, are found to vary greatly from each other in terms of both the forms and complexities in mathematics. Generally speaking, more fracture parameters are supposed to provide a better description of fracture initiation and propagation, which, however, would also create huge obstacles for their application in practice. Instead of proposing a totally new model, it seems more practical to perform a comparative study on the flexibility and operability of existing models for fracture prediction of structural steels. Specifically, the ductile fracture characteristics of Q355 structural steel were first studied experimentally with six types of notched specimens pulled up to failure under monotonic loading. Corresponding numerical simulations were also performed using an enhanced $J_2$ plasticity model. Seven popular fracture models having various numbers of fracture parameters were selected to reconstruct the three-dimensional (3D) fracture loci of Q355 steel and compared in detail. Also, a specially designed flat plate with three holes was tested to failure as an aid to further evaluation of these models at the structural level. The analysis results in this study indicate that, first, the overestimations of the classical $J_2$ plasticity at low stress triaxialities could be corrected by the enhanced one proposed in this study through the introduction of a Lode parameter–related term. Moreover, the enhanced $J_2$ plasticity enables the postnecking hardening relationship and yield criterion to be identified independently, which makes it more convenient in view of the given practical application. Second, the stress triaxiality-only-related fracture models, like the void growth model and Johnson-Cook model, are sufficient for fracture prediction under axisymmetric tension, but in cases of more comprehensive stress states where shear stress cannot be neglected, the fracture resistances would be highly overestimated, leaving a potential threat to human lives. Third, the Wierzbricki-Xue model is more advantageous with a higher fracture prediction accuracy (smaller than 3%) but fewer fracture parameters to be calibrated. Fourth and finally, a flat plate with three holes is a promising specimen in terms of assisting in the evaluation of fracture models because there would be two distinct fracture modes generated simultaneously.