Multiscale Modeling of Microstructure Evolution

This dissertation develops a comprehensive multiscale framework to predict microstructural evolution and associated mechanical response of materials by employing mechanistic finite element models or data-driven neural networks techniques. First, a novel approach is presented for simulating microstructural evolution during severe plastic deformation (SPD) in multiphase alloys, integrating dislocation density-based models with Crystal Plasticity Finite Element Modeling (CPFEM) to efficiently capture grain refinement across different phases in multiphase material. Second, a data-driven predictive model leveraging Artificial Neural Networks (ANN) is developed to link morphological attributes of microstructure—such as grain and cell structure—with material properties in additively manufactured AISI 316L, enhancing the ability to accurately predict material performance from microstructural details. Finally, dynamic recrystallization (DRX) is modeled through a finite element approach high-temperature deformation with the cell switching strategy of cellular automata, capturing key phenomena such as grain growth and nucleation events within a scalable multiscale approach. Together, these studies advance predictive capabilities for material deformation, promoting more efficient design and manufacturing processes.