Virtual microstructure generation using Voronoi tessellation

Abstract

3D microstructures help us to understand the correlations between microstructure and properties in a quantitative manner. However, obtaining 3D microstructures for computational modelling is not trivial. In the recent advancement of high-end instrumentations, one is able to obtain actual 3D microstructure via 3D computed tomography. As an alternative to those costly and time-consuming techniques, numerical approaches using various mathematical algorithms offer an elegant way to obtain 3D virtual microstructures taking actual statistics of the real microstructural features. Though some limitations exist, as a whole, the virtual microstructures can be considered as representative to those of actual microstructures. Another advantage of those algorithms is that a number of equivalent microstructures with the same statistics can be generated easily. Analysing a number of equivalent microstructures, one is able to establish a microstructure-property correlation, and can identify the optimized microstructure for better properties. In this presentation, I would like to share my experience on Voronoi tessellation based virtual microstructure generation using an existing open source Voronoi code. With several examples of 2D/3D virtual microstructures, I will demonstrate the possibility of Voronoi algorithms to obtain complex microstructures that are of great interest to some of the working groups in our institute (WF). These virtual microstructures are meshable for FE analysis, which means, one can use any FE software to analyse the microstructure-property correlations without additional pre-processing tools. To my experience, the presented approach is adaptable to many complex microstructures, and thus, it opens a way to overcome some bottlenecks regarding microstructure-property modelling of advanced materials.