Application of optical coherence tomography for improved in-situ flaw detection in nylon 12 selective laser sintering
MetadataShow full item record
Despite significant advances made since the inception of selective laser sintering (SLS), many of the same problems identified by early researchers including high part porosity, inadequate surface finish, and part strength uncertainty persist today. Because of these challenges, quality validation and improved process control continue to be identified as critical areas of improvement in industry roadmaps. To address these issues, an optical coherence tomography (OCT) sensor is investigated for feasibility of use in in-situ flaw detection in SLS. Benchtop OCT imaging of nylon in solid, liquid, and resolidified phases revealed subsurface imaging through liquid and resolidified nylon material was possible. Subsequent benchtop imaging showed that multiple-scattering was the cause of an imaging artifact which contributed to the limited imaging depth in nylon powder. Additionally, nylon powder was continuously imaged before, during, and after melting and resolidification. The resulting images showed scattering was consistent with the presence of crystalline spherulites, suggesting the spherulites are a strong source of scattering in the nylon 12. An OCT sensor was subsequently mounted on a production-sized research SLS machine. Design and implementation information is detailed including artifact correction and noise subtraction strategies. The OCT sensor is then used to detect various common defects in the SLS process. Imaging single layer individual scanlines revealed deeper melt depth due to overheating from galvo deceleration near the end of the scan lines. Additionally, surface curl was able to be quantified and visualized for a build. Finally, an SLS build was performed at higher powder bed temperatures. OCT images collected from the build were compared with X-ray computed tomography (CT) images, and many of the pores in the OCT images are shown to agree well with those detected in the CT images. One pore in the dataset was much larger than the others in the part. This caused the author to hypothesize that a different mode was responsible for creating these pores which a subsequent build confirmed. A summary of contributions and future work is also listed.