Components for aerospace applications are commonly qualified through a defect tolerance fatigue assessment, for which fracture mechanics provides a physics-based relationship between stress state, defect size, and fatigue life. The assessment of components can be coupled with non-destructive evaluations, especially X-ray computed tomography, to determine the size, morphology and location of defects. Recent indications by NASA suggest the adoption of probabilistic methods for the assessment of AM components in presence of non-detectable anomalies and legacy methods when anomalies are detectable. In this work we address these two strategies by applying them to a series of demonstrators manufactured in Scalmalloy by laser powder bed fusion. A probabilistic fatigue assessment is formulated by coupling extreme value statistics for defect populations with an explicit fatigue crack growth model calibrated on specimen data through ProFACE software. In parallel, a detection-based reliability assessment is developed for observed critical defects by combining fatigue model scatter and defect sizing uncertainty, and by correcting the resulting failure probability to account for the likelihood of missed anomalies.
Rusnati et al. (Fri,) studied this question.