Introduction of a “principal stress–weight function” approach to predict the crack nucleation risk under fretting fatigue using FEM modelling

• The fretting fretting crack nucleation prediction is altered by the contact stress gradient effects.
• A non local fatigue approach combining a critical distance - stress gradient function is introduced.
• Very good correlations are observed with plain fretting and fretting fatigue experimental results.
• This approach corrects the stress gradient effects and the FEM surface stress estimation error.
• This approach predicts the fretting crack nucleation risk even using large 100 µm FEM mesh size.

The paper discusses a new proposal to predict the crack nucleation threshold under high cycle fretting fatigue using Finite Element Models. This approach displays two main advantages. First, it takes into account the stress gradient induced by the fretting contact. Secondly, the proposed strategy partly corrects the mesh size dependence of the computed stresses in FEM modelling, so it allows good predictions, and decreases the computation time by increasing the mesh size in the contact area. This original approach is rather simple, combining the study of the maximal stress over a fretting fatigue cycle with a weight function which depends on the stress gradient. The model is tested on quadratic mesh sizes from 5 μm up to 100 μm, and the results correlate well with experiments. It also shows good capability for predicting crack nucleation conditions for various contact configurations, contact sizes but also various plain fretting and fretting fatigue stresses.