Abstract
Widespread adoption of Additively Manufactured (AM) components requires reliable characterisation of their fatigue properties to meet safety–critical standards. Damage-tolerant models offer an attractive avenue for alleviating intensive process-specific testing while retaining the inherent benefits of the AM manufacturing path. These models integrate principles of fracture mechanics, allow for non-uniform defect populations, and can incorporate process-specific attributes. In this study, we develop a computational framework that integrates population statistics, Monte Carlo defect sampling, and a modified NASGRO damage-tolerant solver to quantify the stochastic relationship between defect population statistics and fatigue strength. The approach provides a transferable modelling architecture for defect-sensitive design in AM. Specifically, we include statistical representations of multiple features (defect size, shape, location, and clustering of volumetric and surface roughness equivalent defects) to estimate fatigue strength distributions, which are shown to facilitate the formation of subtle, unexpected fatigue failures that better reflect the scatter in real fatigue behaviour. Overall, the nuance of defect population statistics as a source of fatigue scatter is underscored, revealing critical aspects for the design and modelling of AM component fatigue life.