Development of MGI Software for Materials Calculations and Its Application on Cladding Materials

Du Shiyu^1,2*

¹ Harbin Engineering University

² Ningbo Institute of Materials Technology and Engineering, Chinese Academy of Sciences

ABSTRACT: With the promotion of Materials Genome Initiative technology, how to promote the efficient application of this technology in the development of nuclear energy materials has become an important topic in the nuclear industry. In accordance with the requirements of National Key R&D Program of MGI, our group designed and developed the MGI simulation software platform MAGMISS, and carried out in-depth research on the nuclear structural materials FeCrAl multi-scale coupling models and the application of artificial intelligence algorithm in material science field. Specific research contents are as follows:

(1) Construction of high-throughput and multi-scale theoretical simulation platform. According to the requirements of major projects, we have developed a multi-scale genome design software platform for advanced nuclear fuel cladding materials. The platform includes three core functions: high-throughput multi-scale theoretical simulation task submission and management, material artificial neural network based gene optimization design, and material gene database management. The platform could help users to realize the submission of high-throughput and multi-scale coupling theoretical computing tasks on supercomputing platform through web browser, and has the functions of computing task fault tolerance and process monitoring, and supports the visualization of calculation results and the optimization design of materials based on artificial intelligence algorithm, which provides software support for the development of material genomic method.

(2) Multi-scale coupling method of alloy materials. Since the single-scale theoretical model is difficult to describe the changes of microstructure and properties of alloys from micro to macro, we have established a series of coupled theoretical models of nano atomic scale, micro defect scale, meso grain size scale and macro scale, and predicted the genetic parameters of materials through the first principles, molecular dynamics, phase field and finite element coupling multi-scale simulation method. In particular, a multi-scale coupling algorithm based on crystal plasticity is designed for FeCrAl and ZrNb alloy cladding materials and applied to predict the influence of material genetic parameters on mechanical properties such as yield strength. The accuracy of the model is verified by comparing with the experimental results.

(3) Application of artificial intelligence algorithm in material development. We have applied artificial intelligence algorithms to the development and optimization of new materials, and gained new knowledge in material preparation and performance prediction. In the development of artificial neural network force field, we found that adding nuclear charge information can effectively improve the resolution and calculation efficiency of force field for different elements and materials. At the same time, our Hume-Rothery criterion model could accurately judge the thermodynamic stability of MAX phase materials, which provides theoretical guidance for the selection and preparation of MAX phase materials.

Keywords: Materials Genome Initiative;High throughput calculation; Multiscale simulation; FeCrAl Alloy; artificial intelligence algorithm.