Fatigue crack nucleation modelling for macroscopic defects D. Khan1,∗, D. Leonetti1, J. Maljaars1,2 1 Department of the Built Environment, Eindhoven University of Technology, 5612 AZ Eindhoven, The Netherlands 2 TNO, Molengraaffsingel 8, 2629 JD Delft, The Netherlands ∗ d.i.khan@tue.nl Keywords: Weld defects, Crack Nucleation, Fatigue Engineering steel structures often contain macroscopic fabrication defects, such as weld imperfections. During service, cracks can emanate from such defects under cyclic loading. The crack nucleation from such defects is not only governed by the geometrical characteristics of the defects but also by the underlying microstructure, which results in a distribution of crack nucleation life. Once a defect is detected during the service of a structure, the prediction of the crack nucleation life and the associated uncertainty is important not only to ensure the safe usage of the structure but also to avoid over-conservative corrective measures. With this motivation, the current work aims to develop a methodology for the prediction of crack nucleation from macroscopic defects using a two-scale modelling approach. For this purpose, two models at different scales are used. At the macroscopic scale, a continuum elasto-plastic finite element model (FEM) of the structure with the defect is used to model the geometrical characteristic of the defect and the local stress field. At the mesoscopic scale, a crystal plasticity FEM of the material in the vicinity of the defect is used, wherein the microstructure is captured. The two models are coupled, such that the boundary conditions of the mesoscopic model are obtained from the macroscopic model. To characterize the crack nucleation in the grains of the microstructure, a plastic slip-based fatigue indicator parameter is used. With the modelling set-up, a comparison of defects of different shapes is made in terms of the crack nucleation life and the associated scatter. The prediction of the modelling approach is found to be in qualitative agreement with empirical equations. 99
RkJQdWJsaXNoZXIy MjM0NDE=