Consideration of Statistical Approaches Within the Accelerated Assessment of Fatigue Properties of Metallic Materials

Abstract

Understanding the fatigue behaviour of metallic materials is highly important when it comes to a reliable assessment of material degradation as a result of dynamic loading. Because the provision of such data is associated with great testing effort leading to increased time and cost requirements in terms of conventional methods, accelerated lifetime prediction methods are becoming increasingly important. However, the reduced number of fatigue specimens and tests complicates statistical validations of the obtained results. In this contribution, combinatorial approaches are used to estimate both lifetime prediction bands and virtually-determined S-N curves with a reduced number of specimens, displaying the material-related scatter due to microstructural inhomogeneities. In addition, a variable energy dissipation factor based on cyclic deformation curves is presented, which enables evaluation of materials that exhibit more pronounced scatter, for instance cast materials. An in situ evaluation of the cyclic deformation curves is provided via integration of non-destructive testing methods into the testing rig. Unalloyed SAE 1045 steel, low-alloyed 20MnMoNi5-5 steel, and the cast material EN-GJS-1050-6 are investigated in this research, as these materials posses gradually increasing complexity regarding their respective microstructures.