An automated procedure for geometry creation and finite element mesh generation: Application to explicit grain structure models and machining distortion

Explicit grain structure models within finite element (FE) computational tools are essential for conducting crystal plasticity simulations to elucidate the three-dimensional (3D) topological effects of microstructural evolution on micromechanical fields during plastic deformation. Such full-field framework can predict not only texture evolution but also the development of intra-and inter-granular misorientation, grain shape, and grain boundary character distribution in polycrystalline metals. This paper develops an automated procedure for the geometry creation and FE mesh generation of explicit grain structures to facilitate full-field crystal plasticity modeling of complex shapes other than cuboids. The procedure consists of: (1) generation of a synthetic voxel-based microstructure of cuboidal shapes using the software called digital representation environment for the analysis of microstructure in 3D (DREAM. 3D), (2) cutting of the model into a final shape by Boolean operations in Abaqus software, and (3) generation of volume FE mesh for the shaped polycrystalline aggregate using a custom-built toolset involving Patran software. Explicit grain structure FE models for micropillar compression and for microtube forming are created as examples. To further demonstrate the utility and robustness of the automated procedure for making multiple cuts while attaining the state of stress equilibrium after each cut, machining distortion of an Inconel 718 aircraft engine disk is modeled. The starting material for machining had a distribution of bulk residual stresses resulting from prior thermo-mechanical processing. Specifically, turning and broaching operations are carried out using the procedure. During machining, the disk distorts as material is removed, while bulk residual stress fields evolve to reach a new equilibrium. The disk distortion during the material removal is predicted using the developed automated procedure in Abaqus. (C) 2017 Elsevier B.V. All rights reserved.