Fatigue initiation in additively manufactured samples/parts often occurs at processed-induced defects such as lack-of-fusion (LoF), keyhole, or other morphological/microstructural defects that have unique characteristics and measurable qualities. Attempts at identifying and minimizing such defects have utilized optimized processing conditions along with in situ and ex situ characterization that includes metallography and/or X-ray computed tomography (XCT). This paper highlights the benefits of using fracture surface analyses to detect and quantify defects that may not be detected by metallography/XCT due to sectioning and resolution limits. In addition to using manual quantification of fatigue initiating LoF and keyhole defects on fracture surfaces, image-based machine learning using convolutional neural networks such as U-Net were also used to automate the process. Statistical analyses were used to identify the extreme cases of defects that initiated and accelerated fatigue and to model the distribution of defect size and shape characteristics to distinguish the type of defect. Initial results show agreement between trained machine learning models and ground truth data in defect segmentation, and the distributions of defect characteristics are distinguishable to particular process-induced defect types.

As part of a NASA University Leadership Initiative (ULI) program, this work supports the continued development and evaluation of a fatigue-based process window for stress-relieved Ti-6Al-4V specimens produced via laser powder bed fusion (PBF-LB/M). Four-point bend and axial fatigue specimens were fabricated by NASA ULI collaborators across a range of scan velocities (800–2000 mm/s) at a constant power of 370 W using an EOS M290 system. All fatigue specimens were low-stress-ground by a commercial vendor and tested at Case Western Reserve University (CWRU) under load-controlled cyclic loading at a stress ratio of R = 0.1. This paper presents a curated dataset linking PBF-LB/M process parameters to fatigue outcomes across 175 specimens. Of these, 136 fractured and this study includes fatigue crack initiation site identification and defect morphology metrics derived from post mortem SEM analysis. Specimens that reached runout (10⁷ cycles) and did not fracture under subsequent fatigue testing are retained in the dataset, with fractographic fields marked as ‘NA’ to indicate non-applicability. The dataset includes specimen metadata, processing parameters, fatigue life data, fatigue initiation site classification (e.g., keyhole, gas-entrapped pore (GeP), lack-of-fusion (LoF), contamination), defect size and shape descriptors, and spatial location relative to the free surface. These data are intended to support defect-based fatigue life prediction, probabilistic modeling, process–structure–property studies, and machine learning frameworks linking process parameters to fatigue performance in PBF-LB/M Ti-6Al-4V.