Hierarchical importance of physical factors causing gas porosity in powder bed fusion additive manufacturing

Abstract

Gas porosity affects part quality in additive manufacturing and is influenced by multiple physical factors that are not well-understood. Here, we calculated several mechanistic variables representing the physical factors related to porosity and evaluated their hierarchical influence using machine learning. We found that the mechanistic variables representing the dynamics of the gas bubbles inside the pool, time to rise, Marangoni number, time to solidify, and Stokes’ velocity are more important than the variables indicating the pore nucleation and stability such as heat input ratio, internal pressure of gas bubbles, and activity of gas dissolution. Laser power and scanning speed were the two most important process variables. Porosity prediction was better using mechanistic variables than using process variables and alloy properties.