|About this Abstract
||2022 TMS Annual Meeting & Exhibition
||Mechanical Behavior and Degradation of Advanced Nuclear Fuel and Structural Materials
||Deconvoluting Properties of Additively Manufactured Alloy 718 Utilizing Coupled Microscopy and Machine Learning
||Stephen Taller, Ty Austin, Kurt Terrani
|On-Site Speaker (Planned)
Ni-based superalloys are a primary candidate alloy class for high temperature applications because of their intrinsic resistance to creep, their adequate resistance to corrosion and their high strength. The poor machinability and extensive work hardening make additive manufacturing (AM) an attractive option to produce geometrically complex components with distinct microstructures from wrought alloys. The microstructure of an AM-produced superalloy 718 was characterized for precipitates, pores, and dislocations using a dynamic segmentation convolutional neural network (DSCNN) with pixel-wise defect-detection algorithms. Solute segregation contributed to multiple unwanted phases, e.g. Laves, in the as-built microstructure. Three heat treatments were performed to anneal unwanted precipitates, metastable precipitates, or solution anneal the microstructure. Uniaxial tensile tests were performed at room temperature and 300-600°C. The combination of heat treatments and rapid characterization of the microstructure using the DSCNN improve the understanding of the relative contribution of each phase to the strength of AM superalloy 718.
||Additive Manufacturing, Machine Learning, Characterization