Automated Microstructure-Properties Characterization and Simulation in Brittle Matrix Continuous Fiber Reinforced Composites

摘要

Microstructure (e.g., fibers, fiber coatings, matrix, porosity, secondary phases) canrnhave a significant influence in the fundamental properties of continuous fiberrnreinforced composites. However, tools for capturing and quantifying keyrnmicrostructure-property relationships remain rudimentary, leaving designers to rely onrnoverly simplistic material descriptors such as volume fraction. In this work, a set ofrnautomated tools has been developed to characterize the key microstructure attributesrnhypothesized to have the greatest influence on the response. Materials arerncharacterized via high resolution optical microscopy. Robotic serial sectioning isrnperformed to obtain microstructural material data. Codes are developed that automaternpost-processing of image data, including gradient smoothing/leveling, image stitching,rnthree-dimensional registrations of image stack, image segmentation, feature extractionrnof individual filaments and generation of microstructure statistics. A novel ellipticalrnHough transform based on a complex convolution filter is introduced to extract thernindividual fibers with high accuracy. Through this process, a quantitative descriptionrnof the microstructure is obtained which can then be used to generate digital materialrnvolumes for simulations. These statistical volumes elements that exhibit variability asrnsampled in the actual material can be simulated to determine stochastic variability ofrnthe response. Targeted 3D materials characterization of these microstructure regionsrninforms the stochastic models employed to describe the microstructure attributes ofrninterest and support the development of the physics based behavior models.