A Comparison of Two- and Three-Dimensional Fracture Assessments in the Presence of Residual Stresses

The influence of various assumptions on the modeling of cleavage fracture in the presence of residual stresses was investigated. Analyses were undertaken for modified single edge notched bend specimens, manufactured from A533B ferritic steel. The influence of residual stress fields, introduced by a method of in-plane compression, was investigated through the use of a modified $J$-integral, designed to retain path independence in the presence of initial stress and strain fields and nonproportional loading. Application of modified $J$ values to predict fracture using probabilistic methods, and their use in a well-known structural integrity assessment code, showed that assumptions about levels of out-of-plane constraint, material hardening behavior, and the method of crack introduction have a significant influence on the conservatism of the resulting failure predictions. It was found that more realistic modeling of crack introduction had a major effect on the accuracy of failure predictions, with the effects of material hardening being of secondary importance.