Non-destructive detection of critical defects in additive manufacturing

Abstract

Abstract Non-destructive examination (NDE) based structural integrity assessment of additively manufactured (AM) metallic parts depends on the reliable detection of volumetric defects and the accurate representation of their detrimental features. Failure to detect critical defects and their features during NDE can result in dangerous non-conservativeness in fatigue design and part qualification, leading to dire engineering consequences. This work highlights an often overlooked property of X-ray computed tomography (XCT) in prevailing NDE practices, i.e., the dependence of the XCT efficacy on the shape of defects within a population, which can manifest in both the probability of their detection and the errors in characterizing their features. By examining an identical material volume of laser powder bed fused AlSi10Mg with XCT at different combinations of voxel size and coupon geometry, it is shown that the performance of XCT deteriorates more significantly for irregular-shaped defects, since their fine features tend to be lost. An approach to recover some important feature information of these irregular defects, such as size, using a distance-based criterion is proposed and is shown to enhance the sizing accuracy of these defects.