Efficient DEM calibration of AM powders using a rotating drum through Froude number scaling

Artikel i vetenskaplig tidskrift, 2026

Abstract: Powder qualification in additive manufacturing (AM) is a complex process, requiring extensive know-how to ensure that powder properties are suitable for a specific hardware solution and AM process. This qualification is typically performed through trial-and-error methods, which are costly, time-consuming, and provide little insight into powder behavior. Without a deeper understanding of powder properties, opportunities for optimizing either the powder itself or the recoating mechanism remain unexplored. The Discrete Element Method (DEM) offers a solution by enabling virtual powder characterization and parameter optimization. A method is proposed for calibrating a DEM model of AM powder that incorporates the real particle size distribution by properly scaling the system geometry rather than relying on coarse-graining techniques. A surrogate model-based calibration approach is employed to adjust DEM model parameters simultaneously, capturing their interdependencies and combined effects. A revolution device serves as the experimental reference for the calibration process. To reduce computational costs, a scaled-down simulation setup is introduced while maintaining dimensions significantly larger than the particle size. The calibration method is demonstrated using two widely used AM powders: Inconel 718 and Ti-6Al-4V (Ti64). The geometry down-scaling approach is validated through simulations with the calibrated model across various scaling factors. Additionally, the simulations provide insights into avalanche progression and size segregation in the revolution device. The contribution of this work is a calibration strategy for DEM-based powder modeling in AM that explicitly accounts for the real particle size distribution and is supported by validation in the revolution device. This provides a basis for reliable representation of powders in rotating drum flow, while future work should clarify the extent to which the calibrated state corresponds to the granular dynamical response during recoating. Graphic abstract: (Figure presented.)