Since the early days of additive manufacturing, computed tomography has been the quality assurance tool of choice. Now CT validates its strengths as an informational tool in R&D.
Additive manufacturing: new possibilities, new challenges
By offering enormous flexibility in what is geometrically possible, additive manufacturing (AM) has changed the way engineers think about product design. However, the more complex the designs get, the more challenging it becomes to inspect the components for flaws and irregularities. Since additive manufacturing and computed tomography are both 3D technologies, they have always worked together naturally. As a non-destructive testing technology (NDT), computed tomography virtually eliminates evaluation error and provides an accurate visual understanding of internal spatial structures, densities, and wall thickness.
For AM, CT is more than just a quality assurance tool
While computed tomography has taken its rightful place as a powerful technology for non-destructive testing in quality assurance and quality control, manufacturers have increasingly discovered its capabilities as a process improvement tool. During R&D, computed tomography enables engineers to collect valuable information about the quality of 3D printed products and to feed this information back into the production loop for a continuous improvement of AM processes – reducing cost, risk, and lead time.
Using Dragonfly deep learning
for faster, reliable segmentation
AI and deep learning models offer numerous benefits for segmentation and CT image analysis in additive manufacturing. By accurately and efficiently identifying patterns, objects, and anomalies in large volumes of images, researchers can save time and effort. Additionally, the smart technology helps improve the reproducibility and reliability of image analysis by reducing operator bias – with significant impact on product quality and safety.
Dragonfly helps you unlock the power of deep learning and extract more value from your images. It offers a wide set of tools for any image analysis requirement, including one pre-trained deep learning model to segment pores in AM parts. This model can be adapted and refined with wizard-based training to best fit your specific defect types.
Typical flaws in additive manufacturing
- Nominal to actual deviations
- Residual powder
- Bonding defects
- Gas pores, cracks, inclusions
- Excessive surface roughness
- Micro-structural issues
Which are the applications for industrial CT in AM?
Industrial CT can typically scan around 50,000 particles in one single sequence and determine their sphericity, size distribution, and the existence of open or closed pores in the individual grain. These data can then be plotted and analyzed quantitatively.
Wall thickness and surface roughness analyses
One of the most valuable analyses is wall thickness analysis. Using advanced analysis software, CT inspection systems create images that display different thicknesses and their distribution within the part in color code. High-resolution scans even allow you to make statements about the surface roughness of AM parts, which plays a significant role in their fatigue behavior.
Nominal-actual comparison and GD&T
A CT dataset contains all geometric data of the inside and outside structures of a sample. So, the actual geometry of a part can be easily compared with a CAD file, providing a color-coded model of geometric deviations. It is also possible to conduct geometric dimensioning and tolerancing (GD&T) checks on the real part data.
In addition to nominal-actual comparisons between scan data and CAD file, CT also measures internal and external features in three dimensions. All measurable features with their respective tolerancing can be defined using the CAD model to create a measurement template, which can be applied to the actual dataset by the click of a button.
Porosity and defect analysis
Since computed tomography uses X-rays to inspect the part, delamination, pores, loose powder, cracks, and many other internal defects can be visualized and analyzed. The appropriate software displays pores in color code , while the pore size distribution can be plotted to a histogram or scatter plot.
Mechanical simulation (on “real” data)
For a first assessment of the structural integrity of a given part, finite element (FE) stress simulations can be performed directly on CT scan data. This procedure is particularly suitable for highly complex structures such as foams, lattice structures, or components with microporosity.
CT data allows the calculation of local and global fiber statistics in the form of fiber orientation tensors or histograms, which include porosity analysis results to determine matrix material porosity. In addition, the principal orientation of woven fabrics or lay-up materials can be determined.