The Enabl3D project, funded under the grant number 03LB5000B, ran from 1 October 2020 to 31 July 2023 and aimed to create a new, efficient quality‑assurance method for biotically optimised metal 3‑D‑printed lightweight components used in aerospace, automotive and medical applications. The core idea was to combine three complementary techniques: penetration testing, in‑situ process monitoring and micro‑computed tomography (µCT). Penetration testing measured the actual material properties—tensile strength, yield limit, ductility and anisotropy—directly on the finished part. High‑resolution process‑monitoring data were used to prove process stability and to extrapolate the locally measured properties to the whole component, while µCT was employed to non‑destructively inspect the critical regions identified by the other two methods.

VisiConsult X‑ray Systems & Solutions GmbH, led by Dr. Frank Herold, developed a robot‑based µCT system that could scan local regions of interest (ROIs) with a manipulator robot, enabling complex scan trajectories beyond conventional circular paths. A digital twin of the entire system—including the X‑ray source, the robot, the additive‑manufactured component and the digital detector array (DDA)—was built and simulated with the aRTist software from BAM. The twin produced realistic X‑ray images and allowed optimisation of the projection directions. By generating a dense set of viewpoints on a sphere and then reducing this set to only those that maximised information gain, scan time was shortened while maintaining image quality.

The detection of lack‑of‑fusion (LoF) defects, where unfused powder remains inside the part, was a particular focus. Because unfused powder has a low contrast against the solid material, phase‑contrast‑motivated filters were applied before reconstruction. Several filters were tested, and the Modified Bronnikov Algorithm (MBA) proved effective. Reconstruction was performed with both filtered back‑projection (FDK) and an algebraic method (ART). The ART approach, combined with phase‑contrast filtering, reduced noise and increased defect contrast. After reconstruction, the correct slice was extracted using the previously determined geometric transformation that aligned the printed part with its CAD model. This alignment was essential for correlating defect locations with the process data.

Segmentation quality was evaluated using four metrics: the number of detected defects, the number of defect pixels, the Hausdorff distance (maximum boundary deviation) and the Dice score coefficient (overlap of binary masks). Across all tests, the combination of ART reconstruction and phase‑contrast filtering improved all four metrics, yielding more true defects, fewer false positives and a smaller Hausdorff distance. Small defects of roughly seven pixels were still occasionally missed, indicating that registration accuracy remains a limiting factor.

Fraunhofer IAPT contributed an in‑situ monitoring technique that could verify the integrity of the entire part and predict high‑risk areas with quantified uncertainty. Imprintec developed a tactile probing system for AM samples, allowing local penetration tests to infer general material properties needed for the simulation models that account for critical porosity.

In summary, the consortium produced a fully integrated quality‑assessment workflow that uses in‑situ monitoring to identify critical regions, robot‑assisted µCT to extract defect information, and penetration testing to determine material properties. The resulting data are fed into simulations that translate local measurements into global part performance, thereby enabling significant cost and time savings and contributing to CO₂ reduction in additive manufacturing. The project’s partners—VisiConsult, Fraunhofer IAPT and Imprintec—worked together over the three‑year period, with VisiConsult as project lead, to deliver this comprehensive solution.