About this Abstract |
Meeting |
2022 Annual International Solid Freeform Fabrication Symposium (SFF Symp 2022)
|
Symposium
|
2022 Annual International Solid Freeform Fabrication Symposium (SFF Symp 2022)
|
Presentation Title |
Phase-field modeling of microstructure evolution in additive manufacturing with a physics-informed graph network approach |
Author(s) |
Tianju Xue, Zhengtao Gan, Shuheng Liao, Jian Cao |
On-Site Speaker (Planned) |
Tianju Xue |
Abstract Scope |
The phase-field (PF) method is considered as an accurate method for the simulation of microstructure evolution. Recently, there has been an increasing interest in applying the phase-field method to metal additive manufacturing processes for the simulation of power melting, solidification, and grain growth. Due to the high computational cost, most existing simulations are either in 2D, or easily takes weeks for a moderate-size 3D problem. In this work, we propose a reduced-order modeling method for phase-field simulation by employing graph networks from machine learning. Our method solves temperature field, liquid/solid phase field, and grain orientation variables simultaneously in a set of coupled partial differential equations. We compare the reduce-order approach with traditional direct numerical simulation based on the finite difference method and show that the reduce-order approach is at least 10x faster while stilling capturing key outcomes such as melt pool evolution, post-process grain size distribution, etc. |
Proceedings Inclusion? |
Definite: Post-meeting proceedings |