Comparison of failure criteria in weld solidification cracking simulations

Finite element simulations of gas tungsten are (GTA) welds in 6061-T6 aluminum are used to examine the application of strain-based, stress-based, and void growth-based failure criteria to weld solidification cracking problems. A strain rate and temperature dependent plasticity model is coupled with a void growth damage model to predict weld solidification cracking. To examine strain-based and stress-based failure criteria, the same constitutive model is used with the damage portion of the model inactivated. Stresses and mechanical strains accumulated during solidification are recorded. Welding experiments are used to provide experimental examples of solidification cracking for comparison with the analyses. Cracking locations were observed in the experiment for different specimen geometries and welding parameters. Stresses and strains at locations in which cracks were observed experimentally are computed and compared to determine if one of these failure criteria appears to consistently predict the correct cracking location. The results of the simulations do not strongly indicate that any of the failure criteria applied here more accurately predict cracking than the others. This conclusion may largely be due to the manner in which these criteria are compared and so this should be modified. Rather than comparing results at individual elements at only a few locations on the specimens it would be more appropriate to incorporate the various failure criteria into the simulations and examine the results on a more global basis. The only criterion that was implemented into the analysis code in this manner was the void growth damage model. It performed reasonably well in predicting where along the weld and where relative to the weld centerline cracking initiates, but the propagation of the cracks relative to the weld centerline is not correct. Possible improvements to the analyses and evaluation procedures are described.