Data Mining for Mesoscopic Simulation of Electron Beam Selective Melting
In the electron beam selective melting (EBSM) process, the quality of each deposited melt track has an effect on the properties of the manufactured component. However, the formation of the melt track is governed by various physical phenomena and influenced by various process parameters, and the correlation of these parameters is complicated and difficult to establish experimentally. The mesoscopic modeling technique was recently introduced as a means of simulating the electron beam (EB) melting process and revealing the formation mechanisms of specific melt track morphologies. However, the correlation between the process parameters and the melt track features has not yet been quantitatively understood. This paper investigates the morphological features of the melt track from the results of mesoscopic simulation, while introducing key descriptive indexes such as melt track width and height in order to numerically assess the deposition quality. The effects of various processing parameters are also quantitatively investigated, and the correlation between the processing conditions and the melt track features is thereby derived. Finally, a simulation-driven optimization framework consisting of mesoscopic modeling and data mining is proposed, and its potential and limitations are discussed.
Keywords: Electron beam selective melting, Mesoscopic modeling, Data mining