Method for assessing measurement variability of pores in X-ray Computed Tomography scans of additively manufactured parts

Porosity in additively manufactured parts limits the widespread adoption of this technology for fracture-critical components. While X-ray Computed Tomography (X-CT) is commonly regarded as the most reliable nondestructive method for porosity inspection, previous comparative studies have reported large discrepancies in extracted porosity measurements. This work investigates imaging noise as one key source of measurement variability. We developed a method for generating statistically equivalent scans by leveraging Central Limit Theorem behavior in frame-averaged projections. Using this approach, variability in part measurements, pore measurements, and pore detectability were assessed against a substitution reference on the same machine. We compared this variability to that observed in a scan acquired on a second X-CT instrument. Our findings reveal that projection-space noise is more severe than noise present within a single reconstructed scan, producing streaking artifacts that substantially degrade measurement reliability. The results demonstrate that achieving consistent measurement values would require a significant number of frame averages if the observed noise were truly representative of system performance. This method provides a valuable tool for characterizing noise with well-defined statistical distributions and improving angular consistency between projections, ultimately contributing to more reliable porosity assessments in additive manufacturing quality control.