A contribution to the debate on measurement uncertainty when using X-ray computing tomography

Abstract

Currently, XCT metrological capabilities require further enhancement, due to the absence of a full traceability chain and the presence of different scan parameters that can significantly influence the process. In the last 10 years, substantial progress has been made towards an XCT standardised process for dimensional inspection. However, AM surface measurement and porosity characterisation have been shown to be different from dimensional inspection presenting differing problems. The scanned object for dimensional inspection is affected by XCT measurement uncertainties. These are controlled by the scan voxel size (magnification) and scan parameters and to a lesser extent the surface determination; however, porosity characterisation and surface measurement are greatly affected by the surface determination, in addition to the material type, magnification and scan parameters. To address some of these issues artefacts are used for performance verification of the measuring system and the identification of contributing errors; this is considered one of the essential steps in the metrological process, where true traceability is elusive. Measurement uncertainty is the key to evaluation of the quality of the measurement. In order to ensure measurement traceability, every result should be accompanied by an uncertainty statement. However, several diverse sources can cause uncertainty in the XCT measurements, making it very difficult to establish the overall uncertainty. Furthermore, because the XCT system is a multi-purpose device that can be used for internal and external and material measurements, the uncertainty can vary significantly. The current study evaluates the precision of XCT measurement of internal defects in additively manufactured components through the assessment of reproducibility and repeatability of the internal feature characterisation process. Several sources of variability were assessed: two different XCT machines, three different operators and four different surface determination thresholds. The main aim of the repeatability study was to identify the confidence in successive measurements when scanning under the same conditions. The reproducibility study was used to find the relationship between results when conditions are varied. The results of the repeatability study have a deviation from the reference value within 0.8 voxel, and for the reproducibility study has shown the deviation from the reference value is within 1.6 voxels, with some operators’ results’ less than one voxel from the reference value. The use of different scan parameters and specifically different spot sizes result in the result output deviation. All the operators chose an optimum ISO threshold value for surface determination based on a previously published methodology which was more than 50%, the closest results to the reference value were obtained from a 61% surface determination threshold. The custom ISO threshold method for surface determination could detect the cavities between the powder particles.