Improved grain structure prediction in metal additive manufacturing using a Dynamic Kinetic Monte Carlo framework

Abstract

This work describes a Dynamic Kinetic Monte Carlo numerical modeling framework that can predict the microstructure of metals during powder bed fusion (PBF) and directed energy deposition (DED) additive manufacturing (AM) while considering significant variations in thermal history and heat accumulation that occur during the build. Although the conventional Kinetic Monte Carlo (KMC) method is well-established, it does not accommodate variation in the spatial domains of the melt pool (MP) and heat affected zone (HAZ) with time. Thus, the predicted microstructure remains relatively similar even over large AM build domains. While the existing KMC approach may suffice over spatial regions in which the MP and HAZ remain relatively unchanged, this circumstance is largely contrary to what experimentalists have recently found when imaging different regions in PBF and DED AM builds, thus raising issues with scalability and versatility of the method. The Dynamic KMC framework proposed in this work addresses these concerns by implementing discretized, spatially-varying MP and HAZ at every time increment during the grain structure prediction. The new framework operates in two stages; stage one establishes the 3D spatial MP and HAZ dimensions using either thermal finite element (FE) simulation or through experimental 3D thermal imaging; stage two subsequently integrates these time-varying MP and HAZ dimensions into the KMC algorithm at every time increment during the build. Thus, the Dynamic KMC framework captures the effects that rapid thermal cycles and heat accumulation have on grain nucleation and growth. The method is demonstrated through a case study involving a thin-walled Inconel 625 structure made by the selective laser melting (SLM) type of laser-based powder bed fusion (PBF-LB). The numerically predicted microstructures at various regions and scan layers within the build show strong agreement with experimentally observed trends reported in literature. Significant variations in grain morphology predicted by the Dynamic KMC framework can, according to specific thermal histories, provide investigators with new capabilities in assessing mechanical property variations across different regions of AM parts. • A Dynamic Kinetic Monte Carlo (KMC) framework for microstructure prediction in metal additive manufacturing is demonstrated. • Microstructure evolution is captured based on dynamic melt pool and heat affected zone behavior due to 3D heat accumulation. • Microstructure results using the novel framework for a thin-walled Inconel 625 SLM build agree with experimental observation. • More realistic microstructures in laser-based powder bed fusion and directed energy deposition can therefore be predicted.