Exploratory research on the correlation of probability of detection and image quality during the tomographic characterisation of additive manufacturing defects

This contribution reports on exploratory research to streamline the Probability of Detection (PoD) calculation for defect analysis with industrial X-ray Computed Tomography (XCT). Reliable, reproducible and universally applicable PoD information is becoming relevant for all non-destructive testing (NDT) methods, like XCT, where many acquisition parameters can affect the final inspection quality. Especially for complex shaped parts, where XCT is often the only suitable NDT inspection system, the a-priori determination of the inspection limit is paramount to confidently meet the required acceptance levels. In this work, three additively manufactured (AM) metal parts are employed as reference measurands to generate a set of PoD curves, using a binary response, for different XCT acquisition conditions. For this goal, an initial high resolution/quality XCT acquisition was considered as ground truth for measuring the size of the components' internal defects. Subsequently, the image quality of the different CT datasets was evaluated employing a set of quality metrics. The measured strong correlation between the PoD response and the image quality metrics suggests that an estimation of the PoD stemming directly from the quantification of CT image quality is viable and could shorten the XCT PoD assessment.

ISBN:978-1-9989991-0-1

Jaar van publicatie:2021

Toegankelijkheid:Open