Microstructure-statistics-property relations of anisotropic polydisperse particulate composites using tomography.

In this paper, a systematic method is presented for developing microstructure-statistics-property relations of anisotropic polydisperse particulate composites using microcomputer tomography (micro-CT). Micro-CT is used to obtain a detailed three-dimensional representation of polydisperse microstructures, and an image processing pipeline is developed for identifying particles. In this work, particles are modeled as idealized shapes in order to guide the image processing steps and to provide a description of the discrete micro-CT data set in continuous Euclidean space. n-point probability functions used to describe the morphology of mixtures are calculated directly from real microstructures. The statistical descriptors are employed in the Hashin-Shtrikman variational principle to compute overall anisotropic bounds and self-consistent estimates of the thermal-conductivity tensor. We make no assumptions of statistical isotropy nor ellipsoidal symmetry, and the statistical description is obtained directly from micro-CT data. Various mixtures consisting of polydisperse ellipsoidal and spherical particles are prepared and studied to show how the morphology impacts the overall anisotropic thermal-conductivity tensor.