High cycle fatigue of welded joints: The TCD experience

Abstract

This paper reports on the use of two methods for the prediction of high cycle fatigue failure in welded joints. The development of these methods – known as the crack modelling method and the theory of critical distances – is described, summarising previously published work on test specimens using steels and aluminium alloys. It is concluded that the methods work well, predicting fatigue strength with errors of less than 20% in the great majority of cases. Certain types of joint were found to give rise to large, conservative errors, but similar errors also occurred when using other, well established, methods of prediction. Some restrictions also arise with the use of the TCD for bodies having very small section thicknesses. Finally, two case studies, described here for the first time, illustrate the issues that arise when making predictions during investigations into the failure of welded joints in real engineering components in service.