Enabling Part-Scale Scanwise process simulation for predicting melt pool variation in LPBF by combining GPU-based Matrix-free FEM and adaptive Remeshing

Abstract

This work proposes to combine matrix-free finite element modeling (FEM), adaptive remeshing, and graphical processing unit (GPU) computing to enable, for the first time, scanwise process simulation of the Laser Powder Bed Fusion (L-PBF) process with temperature-dependent thermophysical properties at the part scale. Compared to the conventional FEM using the global stiffness approach and a uniform mesh running on 10 CPU cores, l-PBF process simulation based on the proposed methodology running on a GPU card with 5,120 Compute Unified Device Architecture (CUDA) cores enables a speedup of over 10,000x. This significant speedup facilitates detailed thermal history and melt pool geometry predictions at high resolution for centimeter-scale parts within days of computation time. Two parts consisting of various geometric features are simulated to reveal the effects of scan strategy and local geometry on melt pool size variation, which correlate well with melt pool and lack-of-fusion porosity measurements obtained via experiment.

Keywords

• Laser Powder Bed Fusion (L-PBF)
• GPU Computing
• Process simulation
• Geometry effects
• Thermal history
• Adaptive remeshing