A boundary smoothing algorithm for image-based modeling and its application to micromechanical analysis of multi-phase materials

A micromechanical modeling framework is developed to analyze mechanical behavior of multi-phase materials. The framework includes novel and efficient numerical algorithms for describing realistic geometry of microstructures in finite element models. One of the difficulties of the image-based modeling is that the pixel images have jagged phase boundaries. A phase function is introduced to define the smooth phase boundaries from the jagged pixel boundaries of the micrographs. The phase function is used to construct finite element meshes conforming to the phase boundaries. The effectiveness of smooth boundary meshes is verified by the analytical solution of a circular inclusion problem. For verification, image-based RVE models are constructed from the scanning electron micrographs of a dual-phase steel. The predicted uniaxial stress–strain behavior of the RVE models compared well with the measurements. Highly localized regions or shear bands are observed in the ferritic phase of the dual-phase steel.