Bridging FEM and Artificial Neural Network in gating system design for smart 3D sand casting

Abstract

A relatively new methodology bridging FEM and Artificial Neural Network (ANN) is proposed and validated in this study to optimize the gating system design for smart 3D sand casting. This methodology was applied on the case of 3D sand casting of a simple plate with aluminum alloy (EN AC-44200). Several mold-filling simulations are performed with the commercial FE code ProCast® by using a combination of the studied gating system design parameters, selected from Taguchi orthogonal array. Signal to noise (S/N) ratio and analysis of variance (ANOVA) are then employed to analyze the contributions of the studied design parameters on the molten metal velocity at the ingate. The significant parameters and their corresponding FE simulations are used to train and validate the ANN model. It is found that ANN simulator can rapidly predict the ingate velocity for any combination of the significant gating system design parameters covering the studied design space.