Knowledge of the permeability tensor in liquid composite molding is important for process optimization. Unfortunately, experimental determination of permeability is difficult and time consuming. Numerical calculation of permeability from a model reinforcement can circumvent experimentation. However, permeability predictions often rely on a model reinforcement that does not accurately mimic the actual microstructure. A rapid, nondestructive technique called optical coherence tomography (OCT) can image the microstructure of a composite in minutes. Actual microstructural information can be then used to improve the accuracy of the model and therefore the predicted permeability. Additionally, the influence on fiber volume fraction and microstructural variability on permeability can be systematically studied.
In this work, binary images were generated from the low contrast OCT data through image de-noising, contrast enhancement and feature recognition. The resulting data were input to a lattice-Boltzmann model for permeability prediction. The influence of the fiber volume fraction, tow surface area, average mean free channel path, and variable microstructure are discussed in terms of their individual and synergistic effects on permeability. The calculated axial and transverse permeabilities from the images show very good agreement with the experimental values.