Abstract
In the present study, a comprehensive methodology to assess the impact of surface integrity on fatigue behavior is developed. First, an artificial neural network (ANN) model was developed to correlate machining conditions with surface integrity, enabling the design of specific scenarios with varying levels of roughness and residual stress. Accordingly, 42CrMo4+QT steel specimens were machined and subsequently tested in the high-cycle fatigue (HCF) regime at a stress ratio of R = 0.1 until failure. The outcomes of each scenario were critically evaluated, enabling an in-depth discussion on how the coupled effects of residual stresses and surface roughness dictate overall fatigue life. The study of the roughness effect shows that the use of standard roughness parameters, such as Ra, may be inadequate, as differing values may yield similar stress concentration factors. Moreover, experimental data demonstrate that fatigue life is predominantly governed by residual stresses rather than surface roughness. Finally, the feasibility of existing correction strategies is critically discussed, accounting for cyclic residual stress relaxation and microstructural fracture features.