Modeling complex polycrystalline alloys using a Generative Adversarial Network enabled computational platform.

Area of Science:

• Materials Science
• Computational Materials Science
• Mechanical Engineering

Background:

• Generating statistically equivalent virtual microstructures (SEVMs) for complex polycrystalline materials is challenging.
• Existing methods struggle with multi-modal distributions in morphology and crystallography.

Purpose of the Study:

• To develop an integrated approach for generating SEVMs of complex microstructures in Cold Spray-Formed (CSF) AA7050 and Additively Manufactured (AM) Ti64 alloys.
• To enable robust multiscale micromechanical modeling for predicting material properties.

Main Methods:

• Integration of Generative Adversarial Network (GAN) for multi-modal microstructures with DREAM.3D for grain packing based on Electron Backscatter Diffraction (EBSD) statistics.
• Development of a multiscale model coupling Crystal Plasticity Finite Element Model (CPFEM) for coarse grains and an upscaled constitutive model for ultra-fine grains (UFGs).
• Simulation using sub-volume elements and averaging for overall stress-strain response.

Main Results:

• Successful generation of SEVMs for CSF AA7050 and AM Ti64 alloys with complex microstructural features.
• Demonstration of a robust multiscale model for accurate prediction of material behavior.
• Validation of the approach for image-based micromechanical modeling.

Conclusions:

• The developed method effectively generates SEVMs for challenging alloy microstructures.
• The multiscale modeling approach provides accurate predictions of stress-strain responses.
• This work is crucial for advancing microstructure-property relationship studies in materials science.