Co-authors​ :

 Rajinth SHANTHAR (UNITED KINGDOM), Robert PROSSER (UNITED KINGDOM), Prasad POTLURI , Chamil ABEYKOON (UNITED KINGDOM) 

The permeability of fibre bundles plays an important role during the manufacturing of fibre-reinforced composites. In these processes, textile preforms made of synthetic or natural fibres are impregnated with a liquid resin under a pressure differential. The complete wetting of the fibres by the resin is mandatory for the final part to perform at its design capacity. Given the significant usage of composite materials in engineering applications in recent decades, computational simulations of these processes are employed to reduce the costs associated with experimental prototyping. These simulations typically assume the fibre preform to be a porous medium through which the resin percolates. The governing equations for such flows require prior knowledge of the preform permeability, which greatly influences the resin flow behaviour. The permeability can be calculated analytically, for example using the Gebart permeability model which expresses the axial and transverse permeabilities of periodic fibre distributions as a function of the fibre volume fraction. However, since the fibres in a realistic preform follow a random non-periodic distribution, the conventional Gebart model needs to be modified to account for the irregular nature of the flow channels between fibres to ensure the accuracy of results. This work proposes a novel method to achieve this where the statistical data of the fibre distributions are incorporated into the workflow. The nearest neighbour (NN) distances of the individual fibres will be used when calculating the transverse permeability using the Gebart permeability model. The transverse liquid flow across these distributions was also solved via the OpenFOAM flow simulation package to calculate the permeability numerically. The analytical permeability from the conventional Gebart model was then compared with the numerical permeability, and the difference was cross-checked against the NN distances of the fibre distributions to see whether a correlation exists. Initial simulations indicate the following; the Gebart model tends to overestimate the transverse permeability of random fibre distributions by an order of magnitude. The permeability is strongly dependent on the particular nature of the fibre distribution, which is overlooked in the conventional Gebart model. Even for distributions with similar volume fractions, the transverse permeability decreases with the mean NN distance and vice versa, supporting the previous point. When volume fractions corresponding to the mean NN distances are used instead of the actual volume fractions, the analytical permeability tends towards the numerical permeability. Further work will be carried out to deduce the dependence of the permeability on the NN distances more specifically, which can then be used to improve the accuracy of the conventional Gebart model.