Thermodynamic calculation and machine learning aided composition design of new nickel-based superalloys

Abstract

At present, the operating temperature of the fourth-generation of nickel-based powder superalloys has not reach the target of 815 °C, and the traditional superalloy design and optimization methods are costly and inefficient. To overcome these challenges, phase diagram is used to calculate the thermodynamic parameters of the typical first three generations of nickel-based powder superalloys and representative fourth-generation superalloys. From strengthening processes including solid solution strengthening and γ′ phase strengthening, the effects of alloy compositions on alloy characteristics are examined. Based on this analysis, the thermodynamic standards that satisfy the performance requirements of the fourth-generation nickel-based powder superalloys are established. Then, the standards are applied to JMatPro and machine learning techniques as the foundation for determining the composition selection range, and several new nickel-based superalloys are created. The machine learning aided alloy design concept and theoretical results will provide significant guidance for further study of new nickel-based superalloys.

Keywords

• Nickel-based powder superalloys
• Thermodynamic calculation
• Composition design
• Machine learning