X-ray Computed Tomography has gained popularity as a metrology technique for components with detailed internal features. However achieving micron-scale resolution using X-ray CT is challenging. Synchrotron X-ray sources are highly collimated and brilliant, allowing high resolution tomography in metallic components. The X-ray Fuel Spray research at Argonne National Lab is aimed at utilizing synchrotron X-ray diagnostics for providing insights into automotive fuel injection. We present a case study of micro-CT for automotive fuel injectors, with orifices smaller than 100 micrometers. These orifices are imaged with 1 micrometer voxel size with minimal resolvable features of 2 micrometer. Tomographic analysis on large datasets must preserve resolution while being computationally efficient, which is facilitated by deep learning techniques for segmentation developed in-house. The accuracy of segmentation is evaluated using synthetic data. As results, we show high-quality iso-surface extraction and measurements of orifice features for 7 identical fuel injectors, indicating the extent of manufacturing variability. The capabilities developed in our team have potential applicability in the field of metal additive manufacturing, specifically for millimeter-sized components with micron-scale features.
|