Computer Sciences and Information Technology
Title : | Computational Approach using Multiscale Modeling and Machine Learning for the Accelerated Design of Complex Concentrated Alloys |
Area of research : | Computer Sciences and Information Technology |
Principal Investigator : | Dr. Anuj Dash, Indian Institute Of Technology (IIT) Hyderabad, Telangana |
Timeline Start Year : | 2024 |
Timeline End Year : | 2026 |
Contact info : | anuj.dash92@gmail.com |
Details
Executive Summary : | Recent research focuses on designing Complex Concentrated Alloys (Refractory High Entropy Alloys) to replace current Ni superalloys for higher efficiency engines. The presence of multiple principal elements like Nb, Ti, Mo, Zr, Ta, V, Al, and Cr necessitates robust and efficient methods to screen compositions for high homologous temperature strength, sufficient RT ductility, phase stability, and environmental resistance in the vast unexplored central composition space. A computational framework is proposed to screen possible alloy combinations using CALPHAD with traditional alloy design ideas and supplemented with physics driven first principles or molecular dynamics and phase field models. Machine learning models are then used along with best available experimental data and high throughput experiments to predict both possible properties for specific compositions and the inverse problem of obtaining unknown compositions given a desired set of properties. The initial screening of the composition space will be done with CALPHAD, creating a dataset of alloy chemistries with desired phase distributions and certain parameters such as melting points and modulus. First principle calculations with cluster expansion and Special Quasi-random Structures (SQS) will be used to estimate free energies and alloy phase stabilities and modulus. High throughput experiments, such as constrained diffusion couple techniques combined with nanoindentations, will be used to check the accuracy of the former and supplement data for additional compositions. Statistical methods and multiobjective optimization strategies will be used to rank and identify the most promising candidate alloys with superior properties. |
Organizations involved