Machine learning predictions of superalloy microstructure

2021/09/28 by Taylor, Patrick L., Conduit, Gareth
#FOS: Physical sciences #Materials Science (cond-mat.mtrl-sci)

paper · doi:10.48550/arxiv.2109.13762

Abstract

Gaussian process regression machine learning with a physically-informed kernel is used to model the phase compositions of nickel-base superalloys. The model delivers good predictions for laboratory and commercial superalloys, with R²>0.8 for all but two components of each of the γ and γ' phases, and R²=0.924 (RMSE=0.063) for the γ' fraction. For four benchmark SX-series alloys the methodology predicts the γ' phase composition with RMSE=0.006 and the fraction with RMSE=0.020, superior to the 0.007 and 0.021 respectively from CALPHAD. Furthermore, unlike CALPHAD Gaussian process regression quantifies the uncertainty in predictions, and can be retrained as new data becomes available.

Related

• Machine learning superalloy microchemistry and creep strength from physical descriptors
• Composition based machine learning to predict phases & strength of refractory high entropy alloys
• Physics-informed machine learning for composition-process-property alloy design: shape memory alloy demonstration
• Physics-Based Machine Learning Approach for Modeling the Temperature-Dependent Yield Strength of Superalloys
• Active Learning Sensitivity Analysis of γ^′(L1₂) Precipitate Morphology of Ternary Co-Based Superalloys