Co-authors​ :

 Mouadh BOUBAKER (FRANCE), Willsen WIJAYA , Arthur CANTAREL (FRANCE), Gérald DEBENEST , Simon BICKERTON (NEW ZEALAND) 

A proper characterization of fabrics used in composites manufacturing processes is fundamental in order to accurately model the flow through these fibrous reinforcing materials. While there are several analytical models that have been developed to predict the permeability of a given fibre arrangement, their range of applicability remains limited due to their simplifying assumptions, such as assuming that the fibres are perfect cylinders arranged in periodic arrays [1]. Many researchers have worked on the experimental determination of permeability, but this method is not economically and environmentally efficient. In addition, there is still no standard method for experimental permeability determination, and while several international benchmark exercises were launched in recent years, significant differences were found between the results despite the organizers’ efforts to reduce the sources of scatter [2]. Therefore, permeability prediction through numerical simulation appears as an attractive alternative because of advantages such as the ability to explore a large range of parameters efficiently without the need to reproduce the experiment, and the ability to determine the permeability without using large pieces of equipment and generating large amounts of waste.

In this work, a numerical method is proposed to determine the permeability of a given textile geometry at the mesoscale considering single-scale flow (i.e. impermeable tows without fluid flow inside) or dual-scale flow (i.e. considering flow inside permeable tows) using the numerical implementation of Darcy-Brinkman equation in a finite volume method (FVM) open-source software (OpenFOAM). The method was first tested on simple and ideal geometries (square and hexagonal arrangement of cylinders) and compared to analytical models found in the literature, such as the Gebart and Kozeny-Carman models. Good agreement was found between the numerical and analytical values. The model was then used to determine the permeability of a 3D scanned geometry acquired using an X-ray micro-tomography (μCT) scanner from a four layers stack E-glass plain weave specimen [3], and these results will be compared to the experimental values obtained with an experimental set-up, described in the second section, on the same plain weave textile.

In addition, a radial flow experimental set-up is used to characterize the unsaturated and saturated in-plane permeabilities of different textiles: E-glass plain weave, Quad Axial Stitched Fabric, 0/90 Woven Yarn, 0/90 Biaxial stitched fabric and Unidirectional reinforcement fabrics. In order to identify fabrics in which the dual-scale flow effect is stronger, comparisons are made between the measured saturating and saturated permeabilities, along with visual observations of delayed tow saturation during the oil injection stage of the saturating measurement