Permeability of non-crimp fabrics : a computational fluid dynamics approach

Abstract: In the present market for manufacturing of high performance structural parts for various applications such as aerospace, transportation, infrastructure, marine, sports and recreation, low weight, high performance composites are preferable. Non-Crimp Fabrics (NCFs) are often used for such applications due to its high mechanical properties and often cost effective production. NCFs usually consist of layers of straight fibre bundles oriented in different directions stitched together. Apart from the fibre structure of the fabric, the properties of the formed composite material are dependent on the resin matrix, which bonds the fibres within the fibre bundles as well as between the fibre bundles together. Hence, in order to maintain the high physical properties of NCFs after the composite has been formed, control of the manufacturing is vital in order to minimize defects in the resulting material. An understanding of the manufacturing also contributes to more effective productions and prediction models are therefore highly desirable. One important stage is the impregnation of the fibres by the liquid resin. In order to control the filling of resin through the fabric, the permeability of the fabric must be known. The permeability is directly dependent of the geometry of the detailed fabric and will therefore generally vary within a fabric due to local geometry variations. The permeability can be determined either by experiments or by mathematical models describing the fabric geometry. Experimental work is often very time-consuming and the prediction models that are often used to predict the permeability are relations originating from granular porous media or with homogenous fibre arrangements. In order to predict the permeability more accurately the complex geometry of the NCF, that is its three dimensional nature as well as the effects from the stitching process, has to be taken into account as well as perturbations of the fabric geometry. In the present work Computational Fluid Dynamics (CFD) is used to solve the flow field through the complex geometries. The thesis consists of three papers. In the first paper (Paper A) the permeability of two unit cells of a biaxial NCF are studied. A profound study of the numerical accuracy is carried out of the CFD simulations together with verification by analytical expressions. The results are furthermore validated to previously produced experimental data. The verification showed good agreement with the analytical expressions whereas the calculated permeability showed a large over-prediction of the permeability compared to the experimental data. The presence of a thread proved to have influence on the unit cell permeability. In the second paper (Paper B) the local permeability of a fabric was studied by CFD simulations of unit cells in order to determine how local geometrical features and variations affect the local permeability within a fabric. It was shown that the width and height of the inter-bundle channels have great influence on the permeability as well as the presence of thread and fibres crossing the inter-bundle channels between stitches. The results indicate that there is a significant distribution of the local permeability within a fabric due to the stitching process and its geometrical perturbations. In the third paper (Paper C) the local permeability distribution, due to the effects from the stitching process and geometry perturbations, in the fabric was included in a statistically based network model for the global permeability of the NCF. The results from the network model, including the thread, fibre crossings, perturbations of the inter-bundle channel dimensions, showed that the fibre crossings, which result from the stitching process, have high influence on the global permeability as well as the mean value of the inter-bundle channel width. The result is validated against experimental permeability data and shows good agreement when the fibre crossings are included in the model.