Probabilistic assessment of space components manufactured by PBF-LB

The use of additive manufacturing in aerospace fields is strictly linked to the capability to assess the harmful effect of process anomalies. In this sense, non-destructive evaluations permit the characterization of flaws, which can be employed in fatigue life estimates of parts considering their service lives. X-ray computed tomography proves particularly effective for metal powder bed fusion process, with the ability to recognize gas porosity and lacks-of-fusion from sub-optimal local printing parameters. Nonetheless, the accuracy of the inspection is strongly dependent on the achievable voxel size, which is determined from the part’s bulk. This study proposes a probabilistic fatigue assessment methodology to evaluate the fatigue life of additively manufactured components. This method discusses the reliability of the X-ray computed tomography in terms of probability of flaw detection and of sizing error. Through a suitable fatigue strength model and considering the various uncertainties within the approach, the methodology can estimate the failure probability of a part depending on the applied stress and on the required life. The relationship between stress, defect and life is based on fracture mechanics concepts, since it is built based on the characterization of material properties from experimental tests conducted on standard specimens. The methodology is then applied to benchmark components, manufactured in laser powder bed fused Ti-6Al-4V. The components are brackets for satellite applications sustaining an optical payload. Once manufactured, the parts were inspected through X-ray computed tomography at two voxel sizes, 16 µm and 60 µm. The components were then tested in fatigue on a shaker in a setup that models the original configuration on the satellite; the fatigue load is injected as a sinusoidal excitation at constant frequency and acceleration. The experimental and predicted lives are hence compared, to evaluate the accuracy and bias of the estimates.