Before an additively manufactured component can be safely used in a load bearing application, its mechanical performance must be qualified. Traditional qualification approaches, involving the fabrication and testing of many identical components, negate one of the greatest benefits of additive manufacturing, i.e. the ability to quickly and cheaply fabricate one-off components. Thus, qualification methods that rely less on mechanical testing and more on predictive modeling are of value. This is most true for high cycle fatigue performance, where mechanical testing requires significant resources and produces stochastic results.
High cycle fatigue failure is difficult to predict because it can depend nonlinearly on many parameters, e.g. part geometry, residual stresses, surface characteristics, material defect characteristics, grain and dislocation structures, mechanical and environmental loading characteristics and their history. This has motivated a succession of fatigue models with ever increasing mechanistic fidelity, with some now diving down to the atomic scale. This raises the question of: what level of mechanistic detail is required to sufficiently predict the performance of AM Ti-6Al-4V components? In this talk, I will give my perspective on this question, building from a decade of AM Ti-6Al-4V fatigue modeling and experimentation across scales.