Abstract

In this second part I discuss a coarse-graining technique for amorphous and heterogeneous materials that is currently being developed. By iteratively coarse-graining large microstructural data sets, while maintaining essential statistical information, one eventually reduces the information to a size that is managable computationally. Understanding the relations between the reduced and the initial systems, and the flow of correlations with the upscaling, makes it possible to predict effective macroscopic properties of the original system by investigating its reduced image. This technique can also be used to numerically generate different morphologies of two-phase composites.