3D microstructural reconstruction of heterogeneous materials from 2D cross sections: A modified phase-recovery algorithm

摘要

A novel method for the 3D reconstruction of a microstructure from limited statistical information provided by 2D cross sections is developed. In the proposed approach, first full-set statistical information (two-point correlation functions) are extracted from 2D cross sections, and then an approximate 3D microstructure is realized based on them. The proposed method relies mainly on conditional probability theorem to establish explicit functional forms between two-point correlation functions extracted from 2D cross sections and full-set 3D statistics. For 3D realization, a novel phase-recovery algorithm is developed that captures prominent attributions of the microstructure. The salient feature of the proposed realization scheme is the ability to fully reconstruct 3D microstructures from statistical information provided by just one cross section for isotropic microstructures and two perpendicular cross sections for anisotropic ones. A number of illustrative examples are provided to demonstrate the accuracy and the versatility of the proposed scheme. The application of the method for the 3D realization of microstructure using an experimental dataset is demonstrated. Finally, the accuracy of the method in capturing and retaining essential features including volume fractions and characteristic attributions as well as the state of anisotropy and percolation of the phases is discussed. (C) 2015 Elsevier B.V. All rights reserved.