The purpose of this work is to demonstrate the adequacy of a modeling approach of fretting fatigue for a sphere-on-flat geometry entirely based on fracture mechanics. All stages of damage evolution, from initiation to failure, are encompassed within the framework of fracture mechanics, in contrast with the general use of stress based criteria to predict crack initiation. A comparison with the classical approaches show how this methodology allows to circumvent the problem of the length scale for initiation by using the crack analogue methodology of contact of Giannakopoulos, Venkatesh, Lindley and Suresh. On the other hand, another length scale, identified as the region of dominance of the singular adhesive stresses, is introduced by the model, and experimental methods to validate it are suggested. The comparison of simulations with three sets of experiments performed on a titanium alloy, with a good control of the normal, tangential and axial loads, shows that the qualitative trends are captured. Good quantitative agreement is also obtained for some of them, depending on the details of the crack growth law. Experimentally, results on the same material after shot peening are reported and used to evaluate the ability of the approach to cope with residual stresses. A good qualitative explanation of the fretting fatigue resistance of the shot peened material is demonstrated. These experiments also serve to illustrate possible testing methods and observations which could yield more useful information than the "classical" fretting fatigue test to failure, and to emphasize the need for a careful stress analysis to avoid plastification under certain material and experimental conditions.
展开▼
