Most composite materials are fiber-reinforced polymers. This category includes materials obtained from various types of fibers, available in several architectures, and impregnated with different polymers through several processes. Unfortunately, the materials exhibiting the best mechanical properties (unidirectional continuous-fiber-reinforced thermosetting resins) cannot be manufactured by processes conducive to more complex shapes (such as short-fiber-reinforced thermoplastic injection molding). Consequently, different fibrous architectures, matrix materials, and impregnation processes are being investigated to find materials with intermediate properties.

Woven composites have excellent mechanical properties. However, the draping stage is labor-intensive and time-consuming, especially for parts with complex shapes.¹ As an alternative to these woven reinforcements, knitting provides automatically net-shaped preforms² (e.g., 3D complex shapes, cylinders, and holes) whose draping is facilitated by their stretchability.³Development of such reinforcements allows introduction of low-density, lightweight composite materials in industries involving mass production. These reinforcements must be compatible with series-production processes, such as liquid-composite molding and, especially, resin-transfer molding, which are under development for high-throughput production.⁴

Knitted preforms are especially suitable for liquid-composite molding of complex-shaped parts thanks to net-shape knitting and high deformability. Part quality is related to the impregnation process, because porosity formation depends on flow-front velocity.⁵Process modeling and optimization require quantifying the resin flow velocity through the preform, which we have achieved by measuring permeability.⁶ However, the permeability value is affected by preform deformation (see Figure 1) because of draping and changes to the local fibrous architecture induced by knitting 3D shapes.⁷ In addition, preform stretching and local mesostructure (i.e., millimeter scale) modify local mechanical properties.⁸ Addressing these issues requires determining how preform deformation affects liquid flow.

Although this topic has been well studied for woven composites,⁹ less is known about knitted reinforcements, because of their limited use in the manufacturing industry. Our group previously worked on improving the mechanical properties of knitted reinforcements to make them suitable for complex-shaped parts.^10,11 Here we describe subsequent efforts to optimize impregnation and preform stretching while draping. This work¹² focused on the relationship between stretching and liquid flow in glass knitted preforms.

We measured preform permeability by monitoring the flow front. We fed canola oil into the reinforcement through a central injection inlet under atmospheric pressure while creating a vacuum in the knit under a vacuum bagging at the periphery of the sample. We positioned a CCD camera above the setup to monitor the evolution of the flow front. We extracted the size of the elliptical flow front in the course and wale directions from the camera images to provide wale- and coursewise radii. We identified permeability in both principal directions from this data using an analytical method involving the equivalent isotropic system (i.e., an analytical tool that deforms the part). We characterized the in-plane permeability of a glass-fiber 1×1 rib knit for nonextended, and wale- and coursewise extended samples.

Our results show that coursewise permeability is significantly affected by preform stretching, whatever the extension direction. Stretching the preform 15% in the course-wise direction increases permeability by 50%. In contrast, walewise permeability is quite insensitive to preform stretching. Kozeny-Carman fitting of the walewise permeability values measured on each sample shows that the influence of preform stretching reflects modification of fiber-volume fraction (as a result of changes in thickness). Consequently, the anisotropy (i.e., vertical/horizontal permeability) ratio increases almost linearly with induced or applied walewise deformation. Macroscopic observations of stretched and nonstretched knitted architectures show how walewise stretching induces a homothetic mean transformation of the unit cell (i.e., the size changes but not the shape), while coursewise stretching more profoundly reorganizes the rich and poor fiber areas of the unit cell. These observations are consistent with measured permeability changes.

Our work contributes to improving knit-reinforced composite manufacturing by studying the link between preform stretching and liquid flow through the reinforcement. In future, we plan to investigate the effect of preform shearing on permeability and the influence of architectural singularities (modification-induced effects) in knitting complex 3D parts. This information is essential to properly model and control the actual fibrous architecture during knitting and draping, and to optimize both the part and the impregnation process in terms of porosity, mold filling, and predicting local mechanical properties.