|About this Abstract
||7th World Congress on Integrated Computational Materials Engineering (ICME 2023)
||A Physics-Informed Multimodal Conditional Generative Model for Linking Process and Microstructure in Metal Additive Manufacturing
||Kang-Hyun Lee, Min Gyu Chung, Yeon Su Lee, Gun Jin Yun
|On-Site Speaker (Planned)
In metal additive manufacturing (MAM), the thermal history and the grain growth associated with complex physics lead to the formation of distinguished microstructure compared with conventional manufacturing methods. A quantitative and robust process-structure (P-S) linkage for AM-processed alloys must be established to tailor the highly anisotropic as-built microstructure for obtaining desired mechanical properties. This work proposes a novel approach to model the P-S linkage for MAM with a deep-learned multimodal conditional generative model. To model the relationship between the source domain (temperature field) and the target domain (microstructure) in MAM, the training data obtained from high-fidelity thermo-fluid analysis and cellular automata (CA) based grain growth simulation is employed. The model can generate multiple inverse pole figure (IPF) maps, which can be controlled by latent space manipulation, for a given processing condition that agrees with the numerical simulation results in terms of grain morphologies and texture.