Details are presented of a novel carbon/epoxy spray deposition process for producing high performance, net shape charges for low flow compression moulding. The Bentley-Raycell automated carbon composite charge deposition (BRAC3D) process sprays powdered epoxy and chopped carbon bundles onto three-dimensional (3D) tools, offering a fully automated process with no touch labour. It has been demonstrated that fibre volume fractions of up to 54% are achievable for random discontinuous fibre architectures, with low void content (1?6%). This extends the volume fraction range currently offered by liquid moulding/preforming processes and potentially reduces part scrap rate, since the resin flow direction is through thickness rather than in-plane. Results from an experimental programme are presented, which aims to benchmark the BRAC3D material against commercial advanced moulding compounds. Ultimate tensile strength, tensile modulus and Charpy impact values are reported to be 272 MPa, 44?4 GPa and 128 kJ m 22 respectively, for the random fibre architecture at a fibre volume fraction of 54%. This equates to a 99% stiffness retention and a 59% strength retention compared to a continuous fibre, quasi-isotropic counterpart. Observed trends for increasing fibre volume fraction and fibre length have been compared against finite element predictions and an analytical inclusion model. Simulations from a parametric cost model indicate that the BRAC3D process is cost effective for production volumes exceeding 1100 ppa for a structural demonstrator component, compared with prepreg and resin transfer moulding.

A geometrical modelling scheme is presented to produce representative architectures for discontinuous fibre composites, enabling downstream modelling of mechanical properties. The model generates realistic random fibre architectures containing high filament count bundles (>3k) and high (~50%) fibre volume fractions. Fibre bundles are modelled as thin shells using a multi-dimension modelling strategy, in which fibre bundles are distributed and compacted to simulate pressure being applied from a matched mould tool. FE simulations are performed to benchmark the in-plane mechanical properties obtained from the numerical model against experimental data, with a detailed study presented to evaluate the tensile properties at various fibre volume fractions and specimen thicknesses. Tensile modulus predictions are in close agreement (less than 5% error) with experimental data at volume fractions below 45%.Ultimate tensile strength predictions are within 4.2% of the experimental data at volume fractions between 40%-55%. This is a significant improvement over existing 2D modelling approaches, as the current model offers increased levels of fidelity, capturing dominant failure mechanisms and the influence of out-of-plane fibres.

A three-dimensional geometrical model is presented for generating discontinuous random fibre architectures consisting of high filament count bundles. The fibre network model randomly distributes fibre bundles in a three-dimensional volume using a non-contact algorithm, together with Catmull-Rom spline interpolation, to provide a physically representative material. Only the spines of the fibre bundles are modelled, using truss elements to permit high fibre volume fractions of up to 60%, with no restriction on the fibre bundle aspect ratio. ABAQUS/Standard is used to predict the tensile performance for coupons with varying levels of out-of-plane fibre curvature. The effect of fibre curvature was found to be insignificant for tensile modulus, but a 34% reduction in ultimate tensile strength (UTS) was observed when adjusting the maximum permissible out-of-plane angle of fibres from 1 to 35u. A novel method for characterising the degree of out-of-plane fibre curvature within experimental test coupons is also discussed.