Fatigue life prediction of weld joints: Microstructural variation can be omitted while residual stress consideration is essential

Abstract

The work focused on residual fatigue life (RFL) prediction of non-standard dog bone specimens made of the low alloy steel P265GH containing a butt weld joint (Double-Bevel-Groove-Weld). Numerical predictions were based on crack growth rate data measured on the standard compact tension (C(T)) specimens fabricated from the three structural parts of the weld joint, the base metal, the weld metal and the heat-affected zone. X-ray diffraction measurement technique was used to determine residual stress distribution in the weld joints. The results were incorporated into the numerical model. Residual stress profiles in the C(T) and the dog bone specimens exhibited significant differences. Even the lower bound of the crack propagation data obtained from the standard laboratory specimens can be non-conservative. When no residual stress was considered in the numerical model, the computed RFLs of the dog bone specimens in the as-welded state were much different from the experimental results. Consideration of different material data for each weld zone, which is the standard method, resulted in highly non-conservative results. Consideration of the residual stress field in the numerical model was crucial to reach good agreement with the experiments (relative error of about 7-10%). Variation of microstructure did not affect crack growth rates significantly. The base metal data alone were sufficient for all simulations, making the demanding procedure of material data acquisition for each of the weld zones unnecessary. The results helped to bring clarity and understanding to the issue of fatigue of weld joints, contributing to both science and applications.