CristalX: Facilitating simulations for experimentally obtained grain-based microstructures

Polycrystalline microstructures occur in nature and also arise in industrial processes. Performing finite element computations on such microstructures is relevant in multiple engineering fields. The present paper describes a highly automated, possibly human-assisted, approach to create high-quality finite element meshes for polycrystalline microstructures. The input is an image, showing grains to be identified. First, the individual grains are obtained by performing several image segmentation techniques. The program allows analyzing the relevant properties of a grain, such as its diameter. The segmented image is then turned to a CAD geometry. Different parametrizations of this geometry lead to different grain smoothness. The advantage of our approach is that existing, robust meshing software can be used to create a high-quality mesh on the grain assembly in a completely automatic way. The developed methodology is illustrated on a real-world example.