Estimation of part-to-powder heat losses as surface convection in laser powder bed fusion

Thermal modeling of additive manufacturing processes such as laser powder bed fusion is able to calculate a thermal history of a build. This simulated thermal history can in turn be used as an input to further simulate temperature related characteristics such as residual stress, distortion, microstructure, lack of fusion porosity, and hot spots. In order to estimate the heat loss to the powder bed during the process, convective heat transfer is widely used as thermal boundary condition in finite element modeling of laser powder fusion processes. However, this convection coefficient is usually selected based on empirical estimation or model tuning. In this work, FEA models of the part and surrounding powder are used as a reference to determine the surface convection BC's for modeling the part only. Seven types of commonly used AM materials with a wide range of thermal conductivities were studied for better testing of the conductivity dependency of the convection coefficient. The convection coefficient values, which predict similar thermal history as the powder model, are found to be a function of thermal conductivity of the deposited material and the cross-sectional thickness of the part feature. A new thickness dependent convection boundary condition is proposed and found to be capable of predicting much closer thermal history to the powder model. These newly developed boundary conditions improve the peak temperature prediction accuracy by 36% while running in 1/4th of the time as the powder model.