Image Analysis Methods For Additive Manufacturing Applications

Sammanfattning: There is an upsurge of research interest on Ni-based superalloys additively manufactured (AM) in aerospace sectors. However, achieving the accuracy and quality of the AM part is a challenging task because it is a process of adding material layer by layer with different process parameters. Hence, defects can be observed, and these defects have a detrimental effect on the mechanical properties of the material. Also, AM materials commonly portray a columnar grain structure which also makes it difficult to determine the average grain size because while using the commonly used intercept method, the grain boundaries do not intercept to the test line appropriately. It is important to measure the defects and grain size before performing mechanical testing on the material. Defect measurement and grain size measurements are usually measured manually which results in longer lead time. This work is addressed towards testing recipes in the automated image analysis software to optimize the lead time with good accuracy. Haynes 282, a γ' strengthened superalloy is used in this work. It was assumed that 1,5mm of material from the surface will be machined away so defects had to be measured in this region of interest. The image analysis tools used to test its potentials are MIPAR and ImageJ. Initially, five images in MIPAR and Image J were tested keeping the manual measurements as a benchmark. From this part, it was concluded that metallography and image quality play an important role in the automated measurement. Also, basic Image J software cannot give the measurements of lack of fusion in terms of caliper diameter (longest measurable diameter). Hence, MIPAR was chosen for the application because it was more promising. In the next part, 15 samples were used with manual measurements from a stitched sample and batch processing with MIPAR. The total caliper diameter results were plotted to compare manual measurements and MIPAR. It was observed that scratches were measured as lack of fusion defects at few instances by MIPAR which were further refined using a post-processing function. The defect density results were plotted and compared as well. Due to the difference in calculation of region of interest, the difference in results was observed.To perform the grain size measurement, Haynes 282 was used in HIP and heat treated condition, achieving equiaxed grains. The etchant should be appropriate to reveal the grains. Hence four different etchants were used in this study hydrogen peroxide+HCl, Kallings (electro etch), Kallings (swab) and diluted oxalic acid. This measurement was performed on the material which was cut along the build direction as well as 90º to the growth direction. Since there is no standard for additively manufactured material yet, the results were tested with hall-petch equation to be convinced of the results obtained. It was observed that MIPAR recipe portrayed good results. The results of manual measurements and MIPAR measurements were plotted and compared. It was observed that Hydrogen peroxide and Kallings (swab) showed the grains evidently but twin boundaries were revealed as well. MIPAR calculated the twin boundaries as grains so it over calculated than manual measurements. Kallings (electro etch) and diluted oxalic acid did not reveal the grains so it was difficult for MIPAR to identify the grains.