Additive manufacturing, or 3D printing, is fast becoming a transformative method in fabrication due to its ability to rapidly construct complex components. Several 3D printing technologies have been developed for printing with metal, most commonly involving laser sintering of fine metal powders layer by layer. The aerospace industry, for which volumes are generally too small for traditional mass production methods, has embraced additive manufacturing, so that it is a multi-billion dollar industry today. In this application it remains necessary to perform some type of non-destructive testing (NDT) i.e., detecting defects and analysing their effect on structural integrity/performance before deployment.
X-ray micro-computed tomography (XMCT) is, in principle, an ideal technology for NDT of printed metal components, since it can measure variations in density in 3D and can achieve the resolution to resolve defects at the tens-of-microns scale where they may affect mechanical performance. However, XMCT images of metal parts often contain severe artefacts because the complex, highly-attenuating geometries cause greatly enhanced X-ray beam-hardening (to the point of photon starvation) and X-ray scattering artefacts compared with most objects. For these reasons, XMCT images of 3D printed components are generally of insufficient quality to support fully automated defect analysis.
In this work, we outline several methods for mitigating these issues developed at ANU CTLab, building on our unique instrumentation and reconstruction algorithms. Firstly, we show reduced image noise by down-weighting photon-starved regions of the detector, incorporating this into a multi-grid iterative image reconstruction algorithm. Secondly, we explore the effectiveness of a collimating grid and a beam-stop array to reduce the impact of X-ray scattering. Finally, we tackle X-ray beam-hardening through a single-material model that maximises consistency between the iterative forward-model (that simulates the experiment). We demonstrate that these approaches have complementary benefits and can yield high-quality outcomes when applied together..