Synchrotron imaging derived relationship between process parameters and build quality for directed energy deposition additively manufactured IN718

Abstract

Laser additive manufacturing is transforming several industrial sectors, especially the directed energy deposition process. A key challenge in the widespread uptake of this emerging technology is the formation of undesirable microstructural features such as pores, cracks, and large epitaxial grains. The trial and error approach to establish the relationship between process parameters and material properties is problematic due to the transient nature of the process and the number of parameters involved. In this work, the relationship between process parameters, melt pool geometry and quality of build measures, using directed energy deposition additive manufacturing for IN718, is quantified using neural networks as generalised regressors in a statistically robust manner. The data was acquired using in-situ synchrotron x-ray imaging providing unique and accurate measurements for our analysis. An analysis of the variations across repeated measurements show heteroscedastic error characteristics that are accounted for using a principled nonlinear data transformation method. The results of the analysis show that surface roughness correlates with melt pool geometry while the track height directly correlates with process parameters indicating a potential to directly control efficiency and layer thickness while independently minimising surface roughness.

Keywords

• Laser additive manufacturing
• Directed energy deposition
• Neural networks
• Meltpool geometry
• In-situ x-ray imaging