Fatigue Analysis of a Welded Component Based on Different Methods

Abstract

This study is part of a project regarding the accuracy of life predictions based on finite element analyses and four different fatigue design methods. Different fatigue design codes, such as nominal stress, geometrical stress, notch stress and linear elastic fracture mechanics are compared regarding work effort and analysis accuracy. The paper also describes the process of developing and verifying a finite element (FE) model as well as other practical work such as load data acquisition, static strain measurements and full-scale fatigue tests under spectrum loading. The component that was used was a stay from an articulated hauler. The results from the fatigue tests showed that all cracks started from relatively large root defects and that a large scatter of the weld dimensions can be expected when the welds are manually performed. The comparison between the different fatigue design codes showed a large scatter in the estimated fatigue life. The verification of the FE-model showed some interesting and in some cases unexpected results. The non-linear force-strain behaviour close to the weld toe was not expected and caused a large amount of work with the FE-model. Since also the manufacturing tolerances were shown to influence the measurements, the verification of the FE-model became quite hard to accomplish. The influence of different parameters such as weld geometry, defects, misalignments, pre-tension forces and installation accuracy of the gauges should be studied further.