
Computational Challenges on Fully Resolving the Compressible Gas Flow around Powder Particles in Powder Bed Fusion Additive Manufacturing
Please login to view abstract download link
Laser Powder Bed Fusion (LPBF) is a promising metal additive manufacturing technique in which thin layers of metal powder are selectively melted to form the cross-section of a final part. The cumulative addition of layers results in the completed component. However, the quality of the final part is significantly influenced by the dynamics of the melt pool, leading to defects such as spattering, lack of fusion, and evaporation-induced porosity. This work focuses on simulating and understanding the complex physical phenomena arising from melt pool and evaporation dynamics, with a particular emphasis on fluid–powder interactions caused by vaporization-driven gas flow. Despite experimental and numerical studies providing insights into these effects, a fully resolved numerical investigation of flow around individual powder particles is still lacking. Such a study is crucial for understanding how key process parameters – such as particle packing density and laser power – affect the resulting fluid-particle dynamics. We present an investigation of the computational challenges associated with resolving the compressible flow field in the gas phase around powder particles in LPBF. Our approach employs a discontinuous Galerkin formulation for the compressible Navier-Stokes equations, combined with a Brinkman penalization approach to enforce interface conditions on the particle surfaces. However, the relevant length and time scales resulting from the typical magnitudes of flow velocity (~10² m/s) and boundary layer thickness (~10⁻⁶ m), introduce severe numerical challenges. This imposes strict time step constraints on explicit time stepping schemes, which are otherwise favorable for computational scaling. To address these challenges, we explore various implicit–explicit and fully implicit time-stepping strategies, comparing their performance and scalability in terms of stability and computational efficiency. Additionally, we give insights into our ongoing work on further enhancing solver performance, ultimately enabling the simulation of gas–powder interaction in larger, practically relevant systems.