Validation of x-Ray Computed Tomography Detection Limits for Stochastic Flaws in Additively Manufactured Ti-6Al-4 V

X-ray computed tomography (XCT) continues to be a primary means of defining flaw populations in fatigue-critical components fabricated by additive manufacturing (AM), and therefore, defining the detection capability of XCT is necessary. Stochastic flaw populations from four samples from a laser powder bed fusion (L-BPF) build of Ti-6Al-4V fatigue specimens were interrogated with XCT scans at various voxel sizes, followed by automated optical serial sectioning (AOSS) with a Robo-Met.3D system as a higher fidelity technique for comparison. Data sets were registered and processed with an automated defect recognition (ADR) algorithm. Comparison of the detected flaw populations showed a two to three orders of magnitude greater quantity in the AOSS data, with significant improvement in the XCT detection rate with refinement of voxel size. Although refined voxel size XCT scans revealed additional flaws, detection of 90% of the “ground truth” flaws present in the AOSS data was not achieved until flaws reached a size of 7-17 times the voxel size of the XCT scan. The need for additional study of targeted flaw sizes to validate and refine these predictions was identified.