About this Abstract |
| Meeting |
MS&T26: Materials Science & Technology
|
| Symposium
|
Additive Manufacturing Modeling, Simulation, and Machine Learning: Microstructure, Mechanics, and Process
|
| Presentation Title |
Accelerating High-Fidelity 3D Melt Pool Simulation with Latent Diffusion |
| Author(s) |
Shohom Bose-Bandyopadhyay, Francis Ogoke |
| On-Site Speaker (Planned) |
Shohom Bose-Bandyopadhyay |
| Abstract Scope |
The stochastic formation of defects during Laser Powder Bed Fusion (LPBF) introduces variation in the mechanical and fatigue properties of printed parts. Multi-physics simulation methods can describe the three-dimensional dynamics of the melt pool , but are computationally expensive at the mesh refinement required for accurate predictions of defect-causing behavior. Diffusion models have been used to cost effectively approximate high fidelity 2D melt pool cross sections through upscaling low fidelity simulations. We extend this framework to 3D melt pool behavior, providing high fidelity melt pool morphologies across process regimes and enabling prediction of defects including keyhole collapse, balling, and lack of fusion. Through iterative denoising in a low dimensional latent space, latent diffusion is used to upscale low-fidelity melt pool dynamics while maintaining multi-physics relationships and reducing computational expense. Rapid, high fidelity estimation of melt pool structure and keyhole structure enables defect detection and mitigation across process regimes. |