Substepped and advected subdomain methods for part-scale LPBF modeling

Abstract

This work presents substepping schemes for part-scale Laser Powder Bed Fusion (LPBF) modeling where the computational domain is divided into a fast partition with a small time-step and a slow one using a large time-step. These schemes are of special interest due to the disparity of time scales present in LPBF applications, which renders simulation prohibitively expensive unless simplifications are made. A Robin–Robin substepper is proposed and compared to the scheme proposed by Hodge (2021). Additionally, the introduction of an advected subdomain in the fast partition is proposed. This approach was first presented in Slimani et al. (2024) and allows for larger time-steps, as the problem in the Heat Affected Zone (HAZ) is posed in the moving reference frame of the laser, where the thermal field is significantly steadier than in the fixed frame. These methods are used in combination to obtain a computationally efficient scheme for LPBF thermal simulations. The proposed methods are implemented in FEniCSx. Their strengths and weaknesses are evaluated through 2D benchmarks, comparison with experimental melt pool data, and 3D LPBF simulations.

Michele Chiumenti gratefully acknowledges financial support from the Spanish Ministry of Science, Innovation and Universities through the R+D project ‘‘Desarrollo de un gemelo digital para monitorización 𝑦 el control activo del proceso de fabricación aditiva DED (DIGITAM)’’ (Ref. CPP2023-010440), as well as from the Center for Technological Development and Innovation (CDTI) through the R+D project ‘‘Machine Learning Driven Metal Additive Manufacturing By Direct Energy Deposition For Repairing Operations (MALE4RAM)’’. Mehdi Slimani is grateful to the FEniCSx project and its community for providing an actively developed open-source FEM platform and for their consistently helpful and responsive support.

Peer Reviewed

Postprint (published version)