First Author, Second Author
Beijing Advanced Innovation Center for Materials Genome Engineering, 100083, China;
Institute of Materials Intelligent Technology, Liaoning Academy of Materials, Shenyang 110004, China.
EXTENDED ABSTRACT: Nickel-based superalloys have excellent properties such as high strength, creep resistance and fatigue resistance at above 600 ·c. The stability, elasticity and lattice mismatch of y'and y" under stress are calculated by using the first principle, and found that the performance ofNi3Al is superior to Ni3Nb and Ni3V, ideal for they" phase. Based on the criterion of high-temperature stable y'+y＇'double precipitation phase, y" dissolution temperature and corrosion resistance, highthroughput thermodynamic calculations were used and the Ni-Al-V-Nb-Cr alloy system is determined. 21 alloys were screened from 7040 alloy composition space, and Ni-5Al-11 V-5Nb-8Cr with the Ni3(V,Nb) y" phase was selected as the candidate alloy.Experiments show that the alloy consists of y + y'+ y" three-phase composition with precipitation phase y'-Ni3Al + y"-Ni3 (V, Nb) double precipitation phase and the y" dissolution temperature 1152 ·c.The Vickers hardness of 571.9HV5 and density of 8.181g/ cm3 are superior to alloy GH4169. The yield strength of the alloy at 850 ° C reach to 803MPa. Considering the high-temperature workability, the machine learning technique is applied to build a flow curve prediction model, and the prediction is realized by usinga small amount of flow curve data, with a prediction accuracy of R2 of 0.986 for the stresses, and the average absolute error of37 Mpa. And the hyperbolic sine Arrhenius-type constitutive model of the candidate alloys is built. The dynamic recrystallization volume fraction of the candidate alloy for different heat defom皿ion conditions was predicted with an average absolute error of 7.2 vol.%. The comparison between the ML model and the FEM model proves that machine learning possesses an advantage in the extrapolation prediction of the flow curves.
Keywords: Superalloys; flow; High-throughput computing; Machine learning;
REFERENCES
[1] Kirklin S, Saal J E, Hegde VI, Wolverton C. Acta Materialia, 2016, 102:125-135.
Yin Haiqing: Professor. He is the deputy director of Beijing Key Laboratory of Material Genome Engineering, member of Scientific Data Expert Group of National Technical Committee for Standardization of Science and Technology Platforms, member of Asian Materials Data Committee and Chinese Liaison, member of Powder Metallurgy Industry Technology Innovation Strategic Alliance, member of CSTM Data Committee for Material Genome Engineering, and member of the editorial board of Data in Brief and Powder Metallurgy Industry. He has involved in more than 30 research projects, including the National Key Research and Development Project, the National Natural Science Foundation project of China, the National Science and Technology Basic Condition Platform Construction Project, the Beijing Natural Science Foundation and the Science and Technology Project, etc. He has won two first-class prizes and three second-class prizes of provincial and ministerial-level scientific and technological achievement awards. He has published more than 100 academic papers and authorized more than 10 patents.