Nowadays, Liquid Composite Molding techniques are often used to manufacture high quality fiber reinforced composite parts at a relatively low cost. These involve an infiltration process, in which a liquid resin is forced to ingress into a dry fibrous porous medium confined into a mold. An optimal control of the material and process parameters is hence sought since multiple factors can contribute to process uncertainties that may compromise the mechanical performance of the final composite part. Fabrics commonly employed in composite industry are made of dense fiber tows weaved to form textiles that present a dual-scale architecture with micro-pores in between the fibers and meso-voids in between the tows. A dual-scale flow is known to stem from this particular pore distribution from which a void-trapping mechanism arises. Researchers agree with the existence of an optimal injection condition, corresponding to a given flow velocity at which fluid flows with the same speed in micro-spaces in the tows and in meso-spaces in between tows and the amount of voids is minimum. Unfortunately, this optimal condition is specific for a given fabric/resin system, and is mostly found by trial and error. Given the strong complexity of flow in porous media, due to the interplay of several factors such as heterogeneity of the porous network, fluid properties (viscosity, density, and surface tension), wettability, flow displacement speeds, void formation and transport mechanisms as well as the considered length scales, the quantification and analysis of the spatial distribution of fluid flow in porous media is to date a major scientific challenge in composites but also in many other natural and engineering occurrences. The goal of this thesis is thus to gain a better understanding of flow in fibrous media as found in composite manufacturing, where the porous medium is a non-translucent textile formed of an assembly of fiber tows, and where the infiltrating fluid, due do its viscosity, may show dynamic wetting effects. To this end, the study focuses on (i) designing a novel X-ray based dynamic flow visualization method adapted to carbon based textiles, (ii) developing a two-phase flow model to describe the progressive preform saturation as a function of the flow rate conditions, and identify the velocity dependent capillary pressure and relative permeability, and (iii) investigating the effect of the fluid static wetting properties and the textile pore geometry on flow patterns and pore formation. It was found that the in-situ observation of infiltration dynamics, using X-ray scattering or absorption methods, is a suitable tool to elucidate the progressive saturation of carbon fiber preforms, which could be further extended beyond model fluids to the actual resin systems. The resulting saturation curves, coupled to higher resolution static 3D tomographs, can be exploited using a multiphase flow approach to identify the relevant hydraulic parameters, as a function of the flow conditions, paving the way for a systematic analysis of dynamic wetting and void formation in composite processing.