Measurement uncertainty evaluation of an X-ray computed tomography system

X-ray computed tomography (XCT) systems are already well established in the field of non-destructive testing and material characterization, but they are currently not fully accepted in the field of dimensional metrology, since there is still a lack of international standards. Measurement results obtained without specifying the measurement uncertainty are incomplete for metrological applications. For this particular reason, the measurement uncertainty has been determined for a selected specimen from automotive industry, representing a typical measurement task for coordinate metrology by XCT. In addition, the influence of the system stability regarding axis stability, focal spot drift and thermal stability has been analysed. The objective of this work was to estimate a task specific uncertainty for XCT measurements according to an approach described in the guideline VDI/VDE 2630-2.1. The feasibility of the recommended procedure has been evaluated in terms of repeatability and reproducibility conditions, leading to a measurement system analysis regarding accuracy and precision. In addition, the practicability of a fast uncertainty determination has been tested. The system stability is evaluated by a stationary ruby sphere mounted at the non-rotating part on the rotary stage and by a pair of steel wires mounted directly on the tube housing. Moreover the temperature in the measurement room and the temperature rise of the X-ray tube itself has been tracked through the entire measurement time. Results for repeatability conditions show that the uncertainty in measurement is within the sub-voxel region for all measures, with few exceptions. User influence could be minimized by defining a systematic procedure for dimensional measurements with XCT systems leading to traceable measurement results. Furthermore, the correction of systematic errors by calibrated features has been investigated. Reference measurements have been done by a tactile coordinate measurement machine. This correction leads to remarkably better expanded uncertainties and subsequently to improved process suitability, since systematic errors are the main contributor for uncorrected measurement results of XCT.