Performance testing of dimensional X-ray computed tomography systems

Abstract

X-ray Computed Tomography (XCT) has become a common tool for dimensional analysis as it allows for non-destructive internal and external measurements. Manufacturers often specify the accuracy of dedicated metrology XCT systems as a maximum permissible error (MPE) statement determined using workpieces, test lengths and positions that are significantly different between manufacturers. The VDI/VDE 2630-1.3:2011 guideline provides specification and respective test methods to verify performance against these MPE statements and has been applied in this study to evaluate four different commercial XCT systems in a uniform way. For this examination a multi-sphere test object was developed that explicitly complies with these guidelines and maximises the use of the whole measurement volume, which has then been scaled to test both high and low magnifications. The study consisted of two parts: scans according to a protocol to allow for a fair comparison between all the systems and free scans where manufacturers could show the best capabilities of their system. With these particular objects no system complied with its own MPE statement, however sub voxel accuracies were found. The maximum error in terms of voxels ranged between 0.16 voxel to 0.43 voxel for low magnification (voxel size of 100 μm) and 0.49 voxel to 0.82 voxel for high magnification (voxel size of 14 μm) between systems. This indicates the need for greater standardisation and transparency on how accuracy statements are determined, and more directed protocols for testing the performance of a system. In particular due to the extreme range of measurement volumes and voxel sizes/resolutions XCT are capable of, it is demonstrated that it maybe reasonable to consider an MPE dependent on voxel size.