The design against fatigue of metallic components is often a tough task since the highly stressed volume and the loading mode are likely to change with the applied boundary conditions and the component size and geometry.
The material variability is another important issue to deal with. For instance, some widespread manufacturing processes, like casting and additive manufacturing, are known to produce a larger distribution in microstructural features compared to wrought products for instance. This not only affects the fatigue strength but also its distribution. Therefore, a quantitative description of the fatigue performance taking into account explicitely the key microstructural parameters is becoming a major concern with regard to component integrity and reliability.
The present review aims at describing the main results of a few comprehensive studies regarding the High Cycle Fatigue of metallic materials showing a complex microstructure, with a special focus on the dependence of some well-known effects, i.e. the size and multiaxial effects, upon several microstructural features including the porosity content. Cast aluminium-silicon alloys and additively manufactured alloys are more particularly used to illustrate the fundamental role played by the defect features (size, type, shape, distribution, distance to the free surface …) on the fatigue response. In the absence of large pores, it is observed that other microstructural features (particles, local crystallographic texture …) control the crack initiation mechanisms and hence determined the fatigue performance.
Vast fatigue experimental campaigns carried out on alloys with different porosity contents and on specimens of different sizes clearly show that the scale effect and the experimental scatter are strongly defect population dependent. In this context, Computed Tomography technique is used to characterize the defect population and to get a better knowledge of the defect criticality.
It is also emphasized that a numerical probabilistic approach using random sampling can be built to adequately reflect several experimental observations including the size effect and the fatigue strength scatter. The parent defect population features, i.e. pore density and defect size distribution, are found to greatly affect the magnitude in scale effect as experimentally observed.