Accounting for Geometry and Residual Stresses in Weld Fatigue: A Strain Energy Density Approach to Total Life of Welded T-Specimens

Although many practical examinations of metal fatigue focus on crack initiation or crack growth, in practice many fatigue failures begin by initiating a crack which then grows until the part fractures completely or can no longer support the loads for which it was designed. This two-phase view of fatigue, called “total life”, is intended to aid engineers in designing parts that will meet design requirement while reducing unnecessary weight and cost. The challenge of using the total life to predict component life is magnified when components are welded together. This work, as a part of a larger project, examines the influence of the welding process on fatigue life predictability by predicting the life of welded specimens made from A36 steel as well as specimens machined from the same material using a geometry designed to develop a nearly identical stress distribution when loaded in the same way and comparing the predictions to measurements. Measured weld toe geometry and residual stresses were used and the results are employed in the relevant calculations. The methods used for predicting the strain-life and fatigue crack growth rates of the material are calculated by a strain energy density method [1]. A non-linear method for predicting variable amplitude fatigue crack initiation is also used [2]. The results show promise, in that if the key variables are appropriately accounted for the fatigue life of welds can be predicted using the material properties of the base material if both the fatigue crack initiation and fatigue crack growth phases are considered together as a total fatigue life. This work is a part of a collaborative project of the SAE Fatigue Design and Evaluation committee.