Liquid Composite Molding (LCM) processes such as Resin Transfer Molding (RTM) are increasingly used to manufacture high performance composites structures in a large number of industrial applications (aerospace, ground transport, etc.) [1
]. This manufacturing method consists of injecting a thermosetting liquid resin in a rigid mold that contains a dry fibrous reinforcement. The saturation level resulting from this injection stage is critical to ensure adequate mechanical performance of the final product [4
]. As a matter of fact, the presence of voids in fiber reinforced composites can reduce significantly the mechanical properties of the material [5
Voids of different origins can be created during composite processing by resin injection. The first cause is mechanical entrapment of residual air during mold filling. The second reason is connected with the chemical reaction of polymerization, during which volatiles are created in some resin systems during cure. Several technological issues contribute also to void creation: leaks in the mold cavity, improper resin degassing, air entrapment in the injection system, use of non-compatible release agents, etc. The analyses presented in this investigation focus on void creation by mechanical entrapment during the injection stage.
Engineering fabrics used as fibrous reinforcements generally possess a dual scale architecture in terms of porosity. Microscopic pores can be found between the filaments inside the fiber tows, while macroscopic pores stand between the tows. The pioneering work of Patel et al. [8
] showed that the progression of the resin through such complex architectures generates uneven flow velocities that can create voids via entrapment of (residual) air bubbles at the flow front. As depicted schematically in Figure 1
, the shape of the front depends on the impregnation velocity, which in turn causes the creation of different types of voids. At low injection velocity, the capillary action induces a leading flow front inside the fiber tows, and macroscopic voids are created in the open channels between the bundles. On the other hand, viscous forces between the bundles are dominant for high impregnation velocity, and the open channels are filled first. This configuration leads to the creation of microscopic voids inside the fiber bundles during the delayed impregnation of the tows. These phenomena have been the subject of several publications as summarized in a review paper by Park and Lee [10
]. Experimental [11
] and numerical studies [14
] showed that the void content created at the flow front varies nearly logarithmically with the velocity and follows a V-shaped curve. As shown in Figure 2
, this behavior indicates the existence of an optimal velocity that minimizes the overall void content in the final part.
Several approaches can be considered to minimize the void content, such as bleeding after mold filling or applying a consolidation pressure. However, these methods come at a cost and are not always adapted to large components or parts of complex geometry. Because of the stringent requirements for high performance composites encountered especially in aerospace applications, the development of practical strategies to produce composites of high impregnation quality represents an important industrial goal [16
]. As proposed by Trochu et al. [4
], control of the injected flow rate to ensure that the front velocity is close to the optimal conditions is an interesting approach to reach that goal. This strategy has been successfully applied in numerical studies [17
], but it has not yet been tested by manufacturing complex parts with the recommended flow rate profile. Furthermore, this approach requires adequate characterization of the fiber/resin system to quantify the optimal injection speed that minimizes the residual void content. To evaluate this important parameter, the void content of test samples manufactured under varying conditions can be measured experimentally [13
]. However, such a procedure is highly time consuming and is only valid for a given resin/reinforcement system. To speed up the characterization stage, an alternative approach based on capillary imbibition was proposed by Lebel et al. [18
The capillary rise method has already been used to characterize the permeability, the architecture, and the capillary pressure at equilibrium in different types of granular porous materials. This approach has also been implemented to study the microscopic and macroscopic properties of fibrous reinforcements used in high performance composites [19
]. In recent work, a new technique based on fluorescent Dye Penetration Inspection
) has been devised to study capillary imbibition in fiber tows and dual scale fabrics [23
]. This allowed gathering automatically and simultaneously the capillary front positions and the uptake fluid mass in time with a high degree of accuracy and repeatability. Moreover, comparison of the void content deduced from image analysis during capillary rise experiments with the porosity measured on rectangular composite panels suggested that capillary characterization can give a satisfactory estimate of the optimal flow front velocity [25
The goal of this investigation is to develop further the capillary characterization method to predict the optimal injection velocity during RTM manufacturing. The paper is divided into two main sections. Following the approach proposed by Lebel et al. [18
], an improved experimental and data processing procedure is first presented, and the new methodology is applied to study the wicking behavior of a glass fabric reinforcement with a reference fluid. The results are then modeled to predict the optimal injection velocity range during mold filling with typical RTM thermosetting resins. In the second part of the investigation, this prediction is tested by manufacturing composite sample parts at the recommended velocity. The specimens possess an irregular geometry to evaluate the potential benefits of adjusting the flow rate during the injection. The influence of the mold configuration (i.e., divergent versus convergent flows) is also evaluated.