In automotive engineering, the main load bearing structure (Body-in-White, BIW) has traditionally been manufactured using metals such as steel and light alloys. In order to satisfy upcoming European regulations on CO2 emissions, considerable weight reduction is necessary and requires the extensive use of lightweight materials and technologies. The European project TECABS focuses on the development of technologies enabling the production of a BIW of an A00 VW-Lupo based on carbon reinforced composites on a production strategy of 50 units per day. Gathering 16 partners, the TECABS project aims at reducing the weight of a BIW by 50% (from 270 to 135kg) and the part count by 70% (from 230 to 70). Liquid Composite Moulding (LCM) is a potentially viable process for producing large 3D complex structures with appropriate production rates, using Non-Crimp Fabrics (NCFs) based on carbon fibres. This PhD work investigated this route, by comparing two LCM processes, one based on a thermoset matrix and one based on a reactive thermoplastic matrix. An analysis of the impregnation mechanisms was carried out. Although often neglected by many authors, capillary effects were found to play a role in the infiltration of the fluid into a porous bed. A method was developed to measure the capillary pressure drop during dynamic infiltration experiments. For the very low viscosity monomer (Lactam12 (LC12)), capillary effects showed negative capillary pressure drops (ΔPγ = -14kPa) leading to spontaneous wicking which could not be ignored when considering injection simulations. Epoxy resins, on the contrary, showed positive values of capillary pressure drop (ΔPγ = 14kPa), demonstrating a resistance to flow. When compared to typical injection pressure, the influence of the capillary effects was shown to be negligible for injection simulations. The influences of the fibre chemistry, of the fluid velocity and of the porous bed architecture were investigated. Each was shown to exert an influence and they were all in accordance with theoretical laws. NCFs were selected as being the reinforcement within the TECABS framework. They showed comparable draping behaviour with respect to woven fabrics, but with very large deformations, NCFs provided the best compromise. The energy required for draping NCF was mainly attributed to the friction occurring between fibres inside the fibre tow, whereas shear proved to be very advantageous. Equivalent locking angles for NCFs appeared at values in the range of 55-65°, being above the 40° found with woven fabrics. The influence of the stitching patterns was also investigated since several architectures were characterised. The process window of the thermoset epoxy resin was drawn using the determination of parameters such as gel points and degree of cure. The post-cure stage was also investigated and results showed that no specific heating ramp as suggested by supplier was required for pursuing the cure outside of the mould. As the low toughness of thermoset matrices is a drawback in using advanced composites in automotive industry, a study was carried out to propose a toughening of the matrix. The use of Hyperbranched Polymers (HBPs) proved to increase the toughness of the neat resin by a factor of 50%. This however was not observed with reinforced composites, since no improvement was measured. Adhesion when using HBPs proved to be a major issue, being responsible for the cohesion at the interface. The effect of the stitching was also investigated and underlined that through thickness reinforcement often presented in literature to improve delamination properties cannot be applied to the polyester stitches used in this work. Production of a demonstrator quarter floor-pan was achieved successfully in-house using thermoplastic RTM technologies. An aluminium mould was designed and manufactured in order to produce large 3D complex shapes by injecting into a reinforcing bed a low-viscosity monomer LC12 that polymerises in-situ. Short shot (incompletes) production proved the existence of capillary effects with the presence of large unsaturated regions at the flow front. Thermoplastic and thermoset RTM were finally compared on several grounds. The first one was the impregnation kinetics and part quality. Overall void content for both thermoset and thermoplastic processes were equal to 6.6%. Intra-bundle void content varied since 5% was measured with epoxy and less than 1% with LC12. The second ground was based on a cost analysis where the production of a full floor-pan was considered. Various production strategies were chosen and production curves could be established, allowing the selection of the most economic system in accordance with any production volume. Cost segmentation was carried out underlining the importance of the carbon fibre cost and NCFs manufacturing (up to 77% of total costs for thermoset components). This cost analysis demonstrated the viability of thermoplastic RTM with respect to traditional RTM, as long as the thermoplastic process can be brought to a more mature stage.