Permeability modelling and particle deposition mechanisms related to advanced composites manufacturing

Advanced composite materials reinforced by Non-Crimp Fabrics (NCFs) are becoming increasingly popular for high performance, light weight, complex structures, due to their high mechanical properties and relatively cheap production costs. These materials, not only improve the structural properties, but also induce great possibilities for reduced fuel consumptions for aircrafts and automotives, due to their high strength to weight ratio. In recent years, even more advanced composite materials have been developed, in order to meet the increasing demands of optimized composite structures from the manufacturers. These materials can have functional properties such as thermal resistance, electromagnetic shielding, conductivity, sensor or self healing properties integrated into the material through functional filler-particles, which are pre-mixed into the resin before being injected into the preform. In order to use these kind of materials in high-end applications, they need to be of highest possible quality, meaning few defects and homogeneous distribution of the functionality and often also produced at the lowest possible cost. An extensive control of the manufacturing is therefore required, since defects and inhomogeneous particle distributions are usually initiated there. During the impregnation stage, resin flows at low Reynolds number through a porous medium in the form of a fibre reinforcement. In order to control the filling process, the permeability of the reinforcement need to be determined accurately. Initially, the local permeability distribution of biaxial NCFs is investigated. Three types of unit cells are identified, where each one represents a specific geometrical feature originating from the stitching process. The local permeabilities of these unit cells are computed for various dimensions by combining Computational Fluid Dynamics and Darcy’s law, in order to scrutinize the importance of the different features and fabric dimensions on the local permeability. It is, for example, shown in this study, that the widths and heights of the interbundle channels in NCFs and the fibres crossing them between adjacent stitches, have the greatest influence on the local permeability, while the stitching thread itself and the shape of the fibre bundles affect it less. In order to improve the speed of the local permeability computations, without reducing the accuracy, a model as simple as possible is sought after. An investigation of whether or not the fibre bundles need to be included into the permeability model is therefore performed. Modelling the fibre bundles of NCFs is proved to be irrelevant for the local permeability for high fibre volume fractions inside the fibre bundles, fb, while the fibre bundles is shown to be important for low fb:s. A new model, including the effect from the fibre bundles without modelling them directly, is developed for low fb:s, in order to facilitate faster, but still accurate permeability computations on models with reduced fluid domains for the entire span of fb:s. Knowing the influence from the geometry on the local permeability, a global permeability model is developed for biaxial NCFs. Unlike most other developed permeability models for NCFs, this model comprises the complex geometrical features originating from the stitching process as well as the spatial variations of the fabric dimensions. The model is based on a network of interconnecting unit cells, with local permeability values calculated numerically. Validation of the global permeability model shows that inclusion of the features from the stitching process into the permeability model, together with an accurate determination of the average dimensions of the interbundle channels, are fundamental, in order to predict the global permeability of NCFs. The second topic considered in the present thesis is related to the inhomogeneous particle distribution, resulting in poor mechanical and functional properties of liquid composite moulded functional composites. To be able to control the distribution of filler-particles in the composites, knowledge and control about the particle deposition mechanisms occurring during the filling process are mandatory. The mechanisms and their resulting particle depositions are examined by microscopic imaging and from velocity fields measured by Micro Particle Image Velocimetry, on the flow in simplified miniscule geometries. In particular, two main mechanisms are studied, being filtration during fibre bundle impregnation and filtration induced by stationary flow through fibre bundles. These mechanisms, not only result in particle depositions, but also in particle-free regions, which are also observed in the analysis of a macroscopic vacuum infused, real biaxial NCF. Several suggestions of adjustments of the process and material parameters, such as the injection flow rate, fabric architecture and orientation, are furthermore outlined, with the aim of reducing these depositions.Godkänd; 2006; 20061116 (pafi)