The effect of plasticity theory on predicted residual stress fields in numerical weld analyses

Abstract

Constitutive plasticity theory is commonly applied to the numerical analysis of welds in one of three ways: using an isotropic hardening model, a kinematic hardening model, or a mixed isotropic–kinematic hardening model. The choice of model is not entirely dependent on its numerical accuracy, however, as a lack of empirical data will often necessitate the use of a specific approach. The present paper seeks to identify the accuracy of each formalism through direct comparison of the predicted and actual post-weld residual stress field developed in a three-pass 316LN stainless steel slot weldment. From these comparisons, it is clear that while the isotropic hardening model tends to noticeably over-predict and the kinematic hardening model slightly under-predict the residual stress field, the results using a mixed hardening model are quantitatively accurate. The level of inaccuracy in isotropic models is shown to be largely dependent on the extent of thermal cycling experienced by the material. Even though the kinematic hardening model generally provides more accurate results when compared to an isotropic hardening formalism, the latter might be a more appealing choice to engineers requiring a conservative design regarding weld residual stress.