A universal Bayesian inference framework for complicated creep constitutive equations.

Area of Science:

• Materials Science
• Mechanical Engineering
• Statistical Modeling

Background:

• Heat-resistant steels are crucial for enhancing power plant energy efficiency by enabling higher operating temperatures.
• Creep constitutive equations model the strain-time relationship in creep deformation, estimating rupture time.
• Existing Bayesian model selection frameworks for creep often rely on approximations, limiting their applicability.

Purpose of the Study:

• To develop a universal Bayesian model selection framework for evaluating various creep constitutive equations.
• To overcome limitations of approximate-expression methods in existing frameworks.
• To enable robust model selection for creep processes across different steel types, temperatures, and stresses.

Main Methods:

• Developed a Bayesian model selection framework using the replica exchange Monte Carlo method.
• The framework avoids approximate-expression methods, offering broader applicability.
• Applied the framework to evaluate the Kimura creep constitutive equation using Grade 91 steel creep data.

Main Results:

• The proposed framework provided a more reasonable evaluation of the Kimura model compared to existing methods.
• Successfully applied the framework to creep measurement data of Grade 91 steel.
• Identified a potential model candidate for improving the Kimura model through posterior distribution analysis.

Conclusions:

• The developed universal Bayesian framework is effective for evaluating diverse creep constitutive equations.
• This advancement facilitates better material selection and performance prediction for heat-resistant steels in high-temperature applications.
• The framework's ability to suggest model improvements aids in optimizing material design for enhanced power plant efficiency.