Two-dimensional heat transfer and melt flow in the laser-affected zone at selective laser melting of thin walls

Selective laser melting (SLM) is employed for obtaining thin-wall elements including lightweight lattice structures. The so-called single-track walls can be considered as thermally thin in the scale of the laser-affected zone. Thus, heat and mass transfer in such a zone become essentially two-dimensional. An original setup is developed for high-speed imaging of the laser-affected zone. Melt pool dimensions are measured as function of process parameters in 170 μm-thick plates of Sn60Pb40 alloy. A computational fluid dynamics (CFD) model of two-dimensional conductive heat transfer and thermocapillary-driven convection is developed. The conservation laws for mass, momentum, and energy are numerically solved by a second-order Godunov finite-volume method using an original Riemann solver developed for the applied equation of state. The CFD model is validated by comparison with the experiments. Laser processing is numerically simulated for thin walls of AlSi10Mg and Sn60Pb40 alloys and Fe. Formation of two outward vortices in the melt pool is revealed. Surface-active impurities can make the surface tension-temperature function non-monotonous giving raise additional vortices with the opposite inward flow direction. The influence of melt convection on the melt pool size is not considerable in the conditions of selective laser melting (SLM). The flow velocity in the melt pool is around or less than the laser scanning speed. This means insufficient mixing in the melt pool. The correlation between the error of the Rosenthal model and the latent heat of fusion is carefully studied resulting an analytical model for estimating the melt pool depth. In the studied SLM cases, the accuracy of the developed analytical model is within 10 % relative the CFD model. The obtained experimental and theoretical results indicate that the melt depth is approximately proportional to linear energy density LED in the typical conditions of SLM. This can be useful when optimizing the SLM process.