UNDERSTANDING POWDER SPREADABILITY IN POWDER BED FUSION ADDITIVE MANUFACTURING

Open Access
Author:
Snow, Zackary K
Graduate Program:
Mechanical Engineering
Degree:
Master of Science
Document Type:
Master Thesis
Date of Defense:
July 02, 2018
Committee Members:
  • Richard Martukanitz, Thesis Advisor
  • Sanjay B Joshi, Thesis Advisor
  • Timothy William Simpson, Committee Member
  • Guhaprasanna Manogharan, Committee Member
Keywords:
  • Additive Manufacturing
  • Powder
  • Feedstock
  • Powder Bed Fusion
  • Qualification
Abstract:
Additive manufacturing (AM) represents a class of rapidly developing manufacturing technologies in which material is selectively added layer-by-layer as opposed to traditional, subtractive methods. The layered approach employed in AM decomposes complicated, three-dimensional manufacturing designs into simple, planar geometries, allowing for unprecedented design freedom unencumbered by constraints of traditional manufacturing. While topics such as design for additive manufacturing (DfAM), thermal distortion and residual stress, and process modelling efforts have received recent attention, studies related to powder feedstock requirements for powder bed fusion (PBF) and directed energy deposition (DED) systems remain scarce, and only recently have researchers begun to investigate the influence of powder characteristics on the AM process. Furthermore, standard characterization techniques used in industry, having originated in the powder metallurgy industry, fail to capture powder characteristics relevant to AM. Existing powder quality metrics are related to packing efficiency and flowability, but have found little merit when applied to the dynamics of spreading in PBF. While newer techniques such as powder rheometry and dynamic avalanche testing have shown promise, they are encumbered by an excess of output data, and both techniques fail to relate their results to the ability of a powder to spread in a PBF system. To date, no powder characterization technique is able to predict the spreadability of AM feedstock. In fact, no such spreadability metrics exist. This work endeavors to establish viable powder spreadability metrics through the development of a spreadability testing rig that emulates the recoating conditions present in PBF AM systems. These metrics are then correlated to the results of bulk powder characterization methods, such as the angle of repose and powder rheometry, to correlate powder quality indicators to spreadability performance. A belt-driven gantry mounted to a rigid laser table acts as the main frame of the spreadability tester outfitted with a variety of sensing technologies to study powder performance. As no metrics for spreadability currently exist, seven possible metrics are proposed and evaluated in a 24 full factorial experimental design. These seven metrics are (1) a qualitative visual inspection, (2) the percentage of the build plate covered by spread powder, (3) the rate of powder deposition, (4) the average avalanching angle of the powder, (5) the rate of change of the avalanching angle, and the roughness of the spread powder layer as measured by (6) a portable microscope and (7) a laser profilometer. The influence of the layer height, recoating speed, the recoater blade material, and the powder quality on each of the proposed metrics are evaluated under the construct of the experimental design, and each of the proposed metrics are analyzed using analysis of variance (ANOVA) to test their suitability as a powder spreadability metric. Two samples of gas atomized, Al10SiMg PBF powder, representing differing degrees of quality, were used as the high and low levels of the powder quality input variable. As no powder quality metrics has been shown to be indicative of powder spreadability in PBF, the angle of repose, a simple, inexpensive, and accessible bulk powder characterization method, is used as the powder quality indicator. The powder samples chosen had angle of repose values of 30 and 40° with the lower values being indicative of higher quality powder. Of the seven metrics evaluated, the microscope layer roughness, laser profilometer layer roughness, and deposition rate metrics lacked any statistically significant correlation with the results of the spreading testing and were not considered further. Additionally, the rate of change of the avalanching angle, although found to be statistically dependent on the powder quality, displayed poor model fitness to the measured responses with an R2 value of only 58%, too low to be a viable spreadability metric. Of the four remaining metrics, the visual inspection is purely qualitative and subject to external biases. However, the percentage of the build plate covered during spreading and the average avalanching angle of the powder wave are both quantitative metrics capable of predicting spreading performance as a function of both user-defined processing variables and the quality of the powder feedstock. Additionally, the speed of the recoat showed little correlation with either of these two metrics, while the layer height and recoater blade material both had statistically significant impacts on the spread quality. Additionally, powder spreading simulations using the discrete element method (DEM), were performed to investigate the interaction between the particle size distribution and the layer height as well as the impact of interparticle cohesion. A commercially available DEM software, EDEM 2017, was used to record the average deposited powder size as a function of the layer height. Increasing layer thicknesses were found to increase the average deposited particle diameter at every timestep in the simulation while the introduction interparticle cohesion provided powder avalanching behavior indicative of physical spreading experiments. Preferential deposition of smaller particles at the beginning of a spread was also noted.