Experimental and computational studies of particle scavenge flow in direct laser metal sintering

Powder Bed Fusion (PBF) cross-flow systems are designed to flow gas across the build plane and entrain metallic powder particles that are ejected during the build, due to the thermal and attendant released kinetic energy of the laser melt process. It is important that these particles be removed from the build chamber so that they do not redeposit on the build surface, as this uncontrolled particle deposition can degrade the part quality. Optimal design of these sub-systems involves tailoring a cross-flow jet such that most of the ejected particles are entrained and removed from the build chamber, while the top layer of particles that are freshly spread on the build plate are not entrained. Accordingly, a combined experimental and CFD study has been executed with the goal of developing engineering design guidance for these cross-flow systems. The closed loop small footprint wind tunnel incorporates a 0.305 m x 0.305 m x 0.915 m test section, a variable height build plate upon which powder can be spread, a variable geometry inlet nozzle, and variable flow rate so that a variety of cross-flow configurations can be tested. Helium bubble particle tracking velocimetry (PTV) was used to characterize the single-phase flow at a number of these operating conditions / configurations. In addition, high speed videography was used to study particle liftoff and entrainment at these same conditions. Using these measurements and attendant CFD models, critical particle liftoff Shield numbers were obtained using CFD predictions of friction velocity. Specifically, close agreement between CFD and measurements were obtained, so that predicted Shields numbers, S_h, could be correlated with particle Reynolds number, Re_t. In this paper we present details of the experimental facility and test program, experimental results including uncertainty/error analysis for the PTV measurements, as well as the videography results for an aluminum alloy powder. The results of the CFD modeling are compared to the single phase measurements. Since very good agreement is observed, predicted wall-shear stress values are used to estimate S_h vs. Re_t at flow rates where incipient particle lift-off is observed experimentally.