Use of X-ray Computed Tomography for dimensional measurement. Lessons Learned and State of the Art in 2025.

X-ray Computed Tomography (XCT) is generating 3D volumes of the parts scanned that can be used for dimensional measurement. Nevertheless, this ability is rather new for NDT specialists, more focused and skilled onto detection of manufacturing flaws than onto metrology. Nevertheless, to be accurate, dimensional measurements need to respect SI metrology rules and that requires specialists. But metrology specialists are not usually skilled into XCT. To complete that picture, the demand of dimensional measurement by XCT have good chances to be dragged by additive manufacturing, with complex geometry parts that will require internal measuring not achievable with current means (CMM, stereo vision or else). This paper is presenting a State of the Art on dimensional measurements by XCT. Different inputs are used: (a) an analysis of VDI/VDE 2630 standards, the only frame currently existing on metrology by XCT; (b) lessons learned, from an interlaboratory comparison campaign onto a demonstration part commonly used in metrology, called CADCube (paper already published by LNE from the work of Cofrend CT Group); (c) lessons learned from practical studies on real industrial parts. The analysis of VDI/VDE 2630 standards revealed several concerns. First, flat panels (DDA) cannot be considered as non-contact dimensional sensors (no dimensional data is directly generated by the DDA). As well, CT systems cannot be considered as sorts of CMM's because some of their characteristics cannot be verified metrologically (X-ray tube focal spot position, detector sensitive surface position). Second, 3D reconstruction is an approximate mathematical process by nature (consequence of J. Radon's research, 1917). Third, surface determination (or segmentation) is another mathematical process which is not properly addressed by the VDI/VDE 2630 standards, although it is directly influencing dimensional measurements. Last but not least, influence of voxel size (regarding trueness and precision targets) is not properly addressed either. All these findings are proving that XCT process is not in line with SI metrology rules. As a consequence, validation of dimensional measurement by XCT requires specific precautions to reach trueness and precision levels acceptable for industrial inspection. On the same manner, lessons learned on real parts revealed other specificities to be taken into account into the development and validation of dimensional measurement by XCT.<br/>By final, this paper is drawing an enlarged State of the Art of current situation about the use of Computed Tomography (XCT) for dimensional measurements, from an industrial user point of view . This paper integrates Lessons Learned from technical items in terms of standards, precautions, performances and limitations, to optimize trueness and precision values as well as skills required in terms of personnel.