Numerical/experimental strategies to infer enhanced liquid thermal conductivity and roughness in laser powder-bed fusion processes

Modelling the thermal behaviour of the melt pool produced in Laser Powder-Bed Fusion (L-PBF) processes is not an easy task, as many complex non-linear thermal phenomena are involved. An effective way to make the computational cost of these analyses affordable is to model powder and molten metal as continuous media, wherein all the heat transfer modes occurring in the liquid are simulated as lumped fictitious heat conduction. The augmentation factor used to enhance the thermal conductivity of the liquid is in general calibrated through experimental estimations of the melt pool size. The present work is aimed at devising a robust method for the calibration of such thermal parameters. A specific point of novelty of the present paper is the definition of a method to correlate surface roughness and numerically predicted melting pool size. This strategy is able to predict with good accuracy the roughness of L-PBF fabricated parts and could pave the way for calibration strategies based on roughness measurements. For this purpose, a 3-factor, 3-level Design of Experiment (DoE) has been carried out to investigate melting pool size and roughness by changing the machine process parameters: laser power, hatch distance, time exposure. In this way, the calibration of the thermal properties is made less sensitive to the large uncertainty usually affecting the melt pool size measurements and the range of applicability of the thermal model is explored over a broad spectrum of L-PBF process parameters. Anisotropic and isotropic enhanced thermal conductivity approaches are applied in combination with a laser source modelled either as a 2D or 3D heat source, respectively. The latter approach proved to be more accurate and robust against experimental uncertainties.