Supercomputers Help Fine-Tune Metal Powders for Additive Manufacturing
Additive manufacturing, or AM, offers a manufacturing capability that conventional methods cannot easily match, the ability to build complex parts layer by layer with remarkable control over form and material use. The process can utilize alloyed feedstock powders formulated for a range of environments, up to the extreme temperatures of hydrogen-burning gas turbines. From its origins as a niche tool for printing CAD prototypes and complex mold patterns, AM has increasingly become more capable, versatile, rapid, and affordable. Yet two major challenges remain: improving energy efficiency and enhancing the properties of AM-built parts.
In 2021, Linde Advanced Material Technologies (Linde AMT) collaborated with Ames National Laboratory to push AM frontiers by boosting energy efficiency, tightening precision, and refining the quality of alloyed powdered-metal feedstocks. While several players in this field are working on improving the properties of AM-built parts, the Linde AMT and Ames team developed a multiprong strategy for applying HPC to the gas-atomization (GA) method of producing feedstock powders. In conventional GA, metals or alloys are superheated and shot through a ceramic nozzle to an atomization zone, where the melt stream is shattered by a supersonic blast of inert gas. The rapidly-cooling droplets in the spray solidify as a spherically-shaped powder, which is separated from the process gas as it exits the spray chamber.
GA produces powders of high purity, and its versatility supports custom alloys. But it is an energy intensive process, can produce too broad a range of powder sizes, and GA powders may be plagued with “satellites” that create friction, interfering with smooth flow through a funnel or layer spread in AM machines. While past computational work by others has focused on the satellite problem, Linde AMT and Ames addressed inefficiencies in powder-size control by concentrating on primary and secondary breakup mechanisms during the melt’s violent interaction with supersonic gas. Using 2D and 3D computational fluid dynamics (CFD), the team examined the effects of variables and breakup mechanisms in producing droplets of desired sizes for new and novel gas-die designs. Rapid methods to evaluate the effects of geometric design and operational parameters on jet coherency length were developed by Ames and calculated on National Laboratory of the Rockies supercomputers to reveal the range in which maximal gas-flow energy can be transferred to the melt for optimal powder formation.
One source of GA inefficiency is energy loss as the gas stream hits the outer surface of the melt pour tube, exciting wasteful shockwaves. The team studied how various converging–diverging geometries promote supersonic gas flow as the melt exits gas-die passages. Supersonic flow can maximize the velocity mismatch between gas and melt and increase gas-energy transfer to the liquid-metal stream. With the help of HPC, the team increased the length of gas-jet coherency for new gas-die designs by up to 80%. In expanding the length of this zone, the team achieved maximal breakup-energy transfer and a ~30% reduction in average powder size, greatly improving the yield of powder appropriately sized for AM feedstock.
The Linde AMT–Ames initiative produced a fully verified, multiscale computational approach to improve GA efficiency, leveraging tools built in C/C++ code and message-passing interface (MPI) parallelization to explore dozens of geometric and operational parameters for gas-die designs. Several novel, nonintuitive configurations that were computationally tested near the end of the project showed significant potential for development and exploration.
This project affords U.S. powder-feedstock manufacturers a low-risk testing technology, i.e., a tool for generating desired powder-size ranges, and an expectation of bettering overall GA efficiency by at least 50%. If employed throughout the industry, these advances in the use of precision gas atomization to convert molten alloys into high-quality spherical powder should realize energy savings of up to 50% compared to using conventional atomization gas-dies.
Fore more information about Linde AMT please visit www.linde-amt.com
