Development of real-time melt pool depth estimation and porosity reduction techniques by process monitoring in directed energy deposition (DED) 3D printing

Abstract

Porosity is one of the most critical defects in 3D printing, and the type and degree of the porosity is determined by the melt pool depth. However, existing porosity reduction techniques relied on post-processing methods due to the limitation of melt pool depth estimation, which cannot be directly measured. Therefore, in this dissertation, we developed a real-time melt pool depth estimation technique based on online melt pool monitoring and artificial neural network, as well as the melt pool depth feedback control system to control porosity. First, the melt pool surface model was established using the melt pool width and length measured by a coaxial infrared camera and the melt pool height measured by CCD cameras. The laser-induced heat conduction equation was then used to calculate the internal temperature distribution and melt pool depth. In multi-layer and multi-track printing, the melt pool profile was further measured by a laser line scanner. The monitoring data were used as inputs to the artificial neural network model, and the melt pool depth was estimated online with an accuracy of 26 µm. Afterwards, a feedback system was developed to control the melt pool depth during 3D printing. The laser power was adjusted to maintain the melt pool depth constantly, and the porosity was reduced 81 % compared to the uncontrolled part. The result indicates that the proposed technique improves the part quality in DED 3D printing.