Biocomposite modeling by tomographic feature extraction and synthetic microstructure reconstruction
Tuukka Verho, Stefania Fortino, Petr Hradil, Tuomas Turpeinen, Kirsi Immonen, Ali Harlin, David Sandquist
Abstract
In comparison to established glass and carbon fiber models, creating a representative volume element to perform finite element analysis for a biocomposite is a complex undertaking. As the fibers appear in a variety of lengths, shapes and orientations, many parameters are needed to describe the microstructure, and a large sample of fibers is needed for a statistically representative RVE. In this study, we present an analysis procedure based on X-ray microtomography to obtain morphological statistics of biofibers in a composite as well as a synthetic microstructure reconstruction and numerical analysis methodology. To obtain statistics on individual fibers from microtomography images, we apply a dual-threshold segmentation approach and fiber backbone tracking. A synthetic model is constructed by using size, orientation and shape statistics from the analysis. Non-overlapping model geometries with fiber volume fractions up to 25% are obtained by a two-stage Monte Carlo packing method. Finite element analyses with periodic boundary conditions are performed to obtain homogenized elastic moduli to be compared with experimental tests. Put together, these steps constitute a complete modeling workflow that also allows virtual design and exploration of the parameter space.