Non-Destructive Inspection of AM parts by CT

Yes, this works quite well within the general limits of the method, of course, such as scan resolution. Pores can be seen as dark regions (low x-ray attenuation) in the datasets.

This generally depends on the sample itself. Smaller samples and samples of less x-ray dense materials can be visualized with higher accuracy than large samples and parts from materials that are high absorbing (such as steel, for instance).

For small samples in the range of 1-2 mm, the voxel size of a CT scan can be as low as 0,5 µm. In terms of metrology, highest accuracy is achieved when looking at samples made from polymer or light metals (such as aluminum).

Glass fibers in polymers are very well visible and their orientations can be measured directly in the CT data. Carbon fibers are much harder to see because of the lack of contrast to the polymer matrix material.

This depends on the particular application. Generally, yes, welds can be visualized and imperfections in the weld might be seen. However, one always will need to perform the CT scan on the full geometry of the part and not just the weld so the CT scan resolution is often limiting.

Regarding the speed of measurements, a CT scan might take anything from just a few minutes to several hours. A very fast scan will have a much poorer quality but for particular cases, this might still be decent to visualize the desired level of detail.

Yes, that is possible but it always depends on the size of the voids and how well they can be detected. For voids that are large enough to comprise several voxels in the dataset, this is no problem.

This depends on the power of the x-ray source. We have sources available at 180, 240, and 450 kV. With 450 kV, the limit for steel is in the range of 50-60 mm;

for plastics the limit probably will be in the range of 500 or more mm of solid material. Linear accelerator systems can have x-ray sources

At least for ASME, we can perform non destructive inspections according to the boiler and pressure vessel code (BPVC). This covers more conventional inspection methods such as ultrasonic inspections or standard radiography and not CT.

Yes this is possible if the glueing layer is large enough to be resolved relative to the size of individual voxels.

Most of the times, inclusions can be seen quite well because they are bright spots in the CT data which can be detected even if they only comprise a few voxels.

In a scan with 50 µm voxel size I would expect to be able to see inclusions in the range of 100 µm and larger; in a scan at 10 µm voxel size, inclusion in the range of 20 µm might be visible etc..

In my experience, CT is not a good tool to measure surface roughness unless the roughness is so strong that it spans several voxels. In most cases, however, the surface roughness will be below the resolution limit of the CT scan.

Yes, this is possible. A CT volume can be freely oriented in virtual space and there are multiple ways to align it to references such as a CAD file.

Yes, this is possible as well. Software tools can work with local gradients to define where in the CT volumetric dataset the part surface is located.

This allows to detect surfaces with sub-voxel accuracy, i.e. the surface definition does not have to follow the grid pattern that is determined by the voxels. How much more accurate the measurement is relative to the voxel sizes depends on the image quality again and can differ between samples.