A critical comparison of one- and two-fluid approaches for the simulation of laser-induced melt pool formation and vaporisation

Abstract

Physics-based simulations are routinely used to understand and optimize laser-based metal additive manufacturing processes. Across the wide range of model predictive capabilities and computational costs, a lack in understanding of the implications of model choice for simulation accuracy persists. We present the first detailed comparison of results between a leading commercial one-fluid model and an advanced two-fluid model. Only condensed matter is included in the one-fluid model domain, with gas effects being incorporated as phenomenological equations. The two-fluid model directly couples gaseous and condensed phases. Solving both analytical benchmarks and increasingly complex cases of laser heating, melting and evaporation, the two approaches are systematically compared. Finally, both models’ predictions are compared to experimental data for keyhole development. Model differences become visible even at moderate laser intensities, when only a mild vapor depression occurs. The main sources of divergence are the evaporative mass flux and resulting recoil pressure, and surface tension-driven contact angles at the three-phase line. Reducing the evaporative recoil factor in the one-fluid model by 70% gave the best match to experiment in a deep keyhole scenario. However, such one-fluid calibrations must be repeated when process conditions alter, so their applicability to more complex scenarios remains unclear. On the other hand, the greater predictive capabilities of two-fluid models come with a factor of 10 increase in computational cost. This work provides guidance on the implications of model choice for AM simulations and presents groundwork for the development of multi-fidelity models for more complex scenarios.