Calibration of ductile fracture models used in the numerical simulation for sheet metal forming process

Abstract

The increasing complexity of sheet metal forming applications and the shortening of the development cycles have introduced new challenges for the forming industry. One of the challenges confronting the manufacturing engineering is the numerical prediction of the onset of ductile fracture. The development of new approaches and reliable computational tools has grown increasing interest in recent years, to enable exploiting the formability of new materials. The main objective of this thesis is to study the predictive ability of phenomenological uncouple ductile fracture models and to improve the understanding about the influence of the plasticity model adopted on the ductile fracture prediction, for different loading conditions. The novelty of the work resides in a detailed analysis of the influence of material and numerical parameters, commonly adopted in uncoupled fracture models, on their predictive performance. The research work has been developed combining extensive databases of experimental results available in literature with detailed finite element models. All numerical simulations were performed with the in-house solver DD3IMP. A comparative study on the predictive ability of several uncoupled ductile fracture models was performed, for the 2024-T351 aluminium alloy and the DP600 steel. The calibration of the selected models is performed, for the 2024-T351, based on different types and number of experimental tests, performed under proportional, plane stress, conditions. This enables to highlight the crucial importance of an accurate experimental evaluation of the fracture strain and of using a set of experimental tests that cover a wide range of stress states. The variables that characterize the stress state also need to be properly determine, including their evolution during the test. The analysis of their evolution is studied for the 6061-T6 aluminium alloy (tubular and sheet specimens) showing that, under proportional loading conditions, the stress state mainly changes due to the onset of strain localization. Therefore, the initial and the average values of the stress triaxiality are considered to calibrate a ductile fracture model for alloys highlighting. Significant differences in the 2D fracture locus are obtained. Moreover, the numerical results highlight the importance of the damage integration, to consider the evolution of the variables that characterize the stress sate. This evolution can affect the ductile fracture parameters calibration process and, consequently, the failure prediction for non-monotonic stress paths. The damage integration is analysed considering different tests, in the low and high range of stress triaxiality, and three uncoupled models, for the 2024-T351 aluminium alloy. The predictive performance of these models is evaluated by comparing the numerical with experimental results available in literature, regarding the location for fracture initiation, the displacement at fracture, the equivalent plastic strain to fracture and the stress state values. Moreover, the results are also compared with the ones obtained with other uncoupled and coupled models, highlighting the effect of the damage accumulation on the onset of ductile fracture, which can help to explain some inconsistencies in the predictive performance. Ductile fracture is controlled by the plastic behaviour of the material, in particular the yield criterion adopted. In this context, three yield criteria, von Mises, Hill’48 and Barlat91, are used to describe the plastic behaviour of the 6K21-T4 aluminium alloy, for tests considering proportional and non-proportional loading conditions. The numerical simulations show substantial differences in the strain paths and in the equivalent plastic strain obtained with the three yield criteria, which are also connect with the stress state evolution. Thus, the yield criterion adopted to analyse the experimental data and perform the calibration of the ductile damage parameters also impacts the performance of uncoupled fracture models.