Automated Design and Optimization of Metallic Alloys
Loading...
Authors
Hohl, Jacob
Issue Date
2021
Type
Thesis
Language
Keywords
Alternative Title
Abstract
The design and optimization of metallic alloys poses a significant engineering challenge. The search space of all possible alloys is sufficiently large that it is impossible to fully explore by traditional methods. In order to address this challenge, physics based computational frameworks linked to advanced machine learning algorithms can serve to automate this process with computational efficiency such that the state of the industry may be rapidly advanced. The work herein presents a suite of computational frameworks leveraged to automate the design and optimization process of advanced alloys. An ab initio alloy thermodynamics system, Molecular Dynamics simulations, a Convolutional-Neural Network system, and a coupled Neural Network and Multi-objective Genetic Algorithm. These algorithms are validated over the set of binary nanocrystalline Al-X alloys, and multi-component High Entropy Alloys (HEA).