Numerical modelling of nanoparticle diffusion and microstructure formation during selective laser melting process

Abstract

Selective laser melting (SLM) is a popular metal additive manufacturing technique that has a wide range of industrial applications lately. This additive process allows the development of new metal matrix nanocomposites by fusing metallic powders with nanoparticles. However, the molten pool flow generated by a moving laser heat source has complex fluid dynamics which redistribute the nanoparticles. Consequently, the microstructures of the solidified molten pool are affected by the local distribution of nanoparticles, which is reflected in their mechanical properties. Smaller grains can increase the strength and isotropic behavior of the solid layers. Therefore, the current research aims to numerically investigate the relationships among the SLM process parameters, nanoparticle transport, and microstructure evolution to explore the formation of nanocomposites. The current study formulated a three-dimensional computational fluid dynamics (CFD) model of the SLM process in a commercial software package, ANSYS FLUENT. A volumetric laser heat source model melted the aluminum alloy powders and the underlying solid substrate. The difference between the powder and the solid or liquid state of the metal alloy was defined using an effective thermal conductivity model. Lagrangian particle transport calculation was performed to track TiB 2 nanoparticles in the molten pool. This model was coupled with a 2D Cellular Automata (CA) model to simulate the solidified microstructure using MATLAB. Finally, a detailed parametric analysis was conducted to study the effects of varying laser power, scanning speed, and preheating temperature. The numerical results showed that the maximum temperature and Marangoni convection in the molten pool increased at higher laser powers, higher preheating temperatures, and lower scanning speeds. The particle-voided region was significantly large with high Marangoni convection but decreased with weaker Marangoni convection. The simulated microstructure was dominated by large columnar grains when nanoparticles were not considered. The introduction of nanoparticles disrupted the columnar grain growth by promoting small, randomly oriented, equiaxed grains. A decrease of 30%-40% in average grain diameter was measured at the cross-section of the solidified layer when nanoparticles were present. The qualitative comparison of the microstructures showed that the grains were smaller in the uniformly distributed particle region compared to the particle-voided region.