Metrological evaluation and classification of porosity in metal additive manufacturing using X-ray computed tomography

• A new modular artefact with real defects is proposed for porosity characterization. • Method proposed for X-ray computed tomography porosity dimensional measurement uncertainty evaluation. • Advanced porosity algorithms are benchmarked to relate pore data with measurement uncertainty. • Over five voxels per pore ensures subvoxel error in porosity dimensional measurements. • Over and under-melting pores are identified with >95 % accuracy. Ensuring the structural integrity of metal additive manufacturing (AM) components is challenging due to inherent porosity, which critically affects mechanical performance. The impact of porosity depends on its morphology, determined by the formation mechanism, making it a core focus for characterization and classification through advanced Non-Destructive Testing methods. This study introduces a novel X-ray computed tomography (XCT) based methodology for the metrological assessment of porosity in laser powder bed fusion manufactured AlSi10Mg components. A new modular artefact, specifically designed to contain real porosity, enables the application of the substitution method for porosity characterization, including the assignment of measurement uncertainty, representing a novel application for evaluating real porosity morphology using XCT. Advanced porosity analysis algorithms, including ’VGDefX’, ’OnlyThreshold’, and ’VGEasyPore’, are benchmarked to determine the one yielding the lowest dimensional measurement uncertainty across varying material thicknesses. The VGDefX algorithm, which incorporates a probability-based and iterative approach, demonstrates superior repeatability, achieving subvoxel systematic errors and measurement uncertainties for dimensions exceeding five times the voxel size. Based on the pore characterization results, a classification methodology using the most discriminating porosity parameters is developed. Leveraging volumetric mean gray value, this method distinguishes over-melting from under-melting porosity in a real component with >95 % accuracy