MeltPoolDG: A High-Performance Framework for Mesoscale Powder–Liquid–Gas Modeling in Metal Additive Manufacturing

  • Schreter-Fleischhacker, Magdalena (Technical University of Munich)
  • Much, Nils (Technical University of Munich)
  • Brotz, Julian (Technical University of Munich)
  • Koch, Andreas (Technical University of Munich)
  • Munch, Peter (Technical University of Berlin)
  • Kronbichler, Martin (Ruhr University Bochum)
  • Meier, Christoph (Technical University of Munich)

Please login to view abstract download link

Computational models of metal additive manufacturing processes, such as laser powder bed fusion (LPBF), are essential for advancing physical understanding beyond costly trial-and-error approaches. These processes feature highly dynamic interactions between melt, gas or vapor, and powder particles that strongly affect stability and part quality. Simulating the resulting coupled powder–liquid–gas dynamics at the mesoscale remains challenging due to complex multiphysics coupling, rapidly evolving interfaces with strong discontinuities, and high computational cost. Consequently, open, high-performance computational frameworks addressing these challenges are still lacking. We present MeltPoolDG, an open-source, high-performance computational framework for consistent modeling of powder–liquid–gas dynamics in metal additive manufacturing. Multiphysics coupling is handled via a partitioned solution strategy. Liquid–gas dynamics are modeled using a Eulerian finite-element (FEM) formulation based on continuous and discontinuous Galerkin methods combined with a level-set-based interface-capturing approach. Both diffuse-interface and sharp-interface coupling approaches are supported, the latter relying on cut finite element techniques. Powder–fluid interactions are modeled using an immersed discrete element (DEM)–FEM approach, allowing fully resolved fluid flow around individual powder particles. MeltPoolDG builds on the deal.II finite element library, providing adaptive mesh refinement near interfaces and scalable MPI-based parallelization through domain decomposition. Efficient iterative linear solvers are employed, including preconditioned Krylov subspace methods with matrix-free operator evaluation. Representative LPBF simulations showcase typical features such as melt pool, melt–vapor and vapor–powder dynamics. While full melt–vapor–powder coupling is still under development, MeltPoolDG already provides an advanced code platform for high-fidelity mesoscale modeling in metal additive manufacturing.