Detectability by X-ray micro computed tomography of fatigue initiating void defects in laser powder-bed additively manufactured Ti-6Al-4V coupons

Abstract

Non-destructive inspection is important for the quality assurance of additively manufactured components. In this study, X-ray micro Computed Tomography (µCT) was assessed for its capabilities and limitations to detect fatigue-initiation porosities in Ti-6Al-4V additively manufactured coupons. Cylindrical coupons were built using laser powder-bed fusion (L-PBF) from three different vendors and four different L-PBF machines that utilized their default process parameters. Using OEM prescribed default process parameters, there was significant variability in the defect distribution between machine to machine and also coupon to coupon that was on the same build plate, primarily due the difference in the internal quality assurance procedures and scan strategies adopted by each vendor. The probability of detection for conventional µCT (cubic voxel size 17.77 µm) was investigated by comparing with what was detected using a Synchrotron Radiation Micro Computed Tomography (SRµCT) at cubic voxel size 5.91 µm. It was found that µCT’s detectability relative to SRµCT was approximately 90% for defect diameters of the circumscribed sphere of 230 µm and above and 100% for diameters of 340 µm and above (12–14 times the voxel size of the µCT scan). The type of defect was a key factor to the detectability of large sized defects (>200 µm in diameter) due to the flatness and elongated shape factors. Fatigue results showed the critical defect √Area at the surface was approximately 100 µm at 500 MPa stress level, where significant scatter in fatigue life was observed. Post-fatigue test fractography identified different types of fatigue initiating defects, of which a ‘flat’ lack-of-fusion type defect was undetectable by either µCT or SRµCT. This type of defect was extremely flat and occupied very little spatial resolution through the build direction which limited the detection by automated threshold-based defect detection algorithms.