1. Introduction
Carbon fiber-reinforced laminated composite materials have drawn significant attention from the aerospace industry in recent decades due to their high rigidity, high strength/weight ratio, and relatively high endurance to environmental factors. High-quality composite materials that can satisfy the stringent requirements of aerospace standard composites are conventionally manufactured using an autoclave process at high pressure and temperature [
1]. However, the utilization of autoclaves possesses numerous disadvantages, such as high capital investment and operation cost, low energy efficiency, long process times, and constraints in the part size [
2]. The motivation for manufacturing larger structural aerospace-grade composite components at lower costs without compromising the quality of the part has led to the development of next-generation materials and manufacturing processes. Accordingly, the out-of-autoclave (OoA) processes have been introduced and attracted widespread acceptance over the last decade due to their abilities to deliver composites without the need for autoclaves [
2]. Subsequently, vacuum-bag-only (VBO) prepregs are explicitly developed for OoA processes, whereby high-performance primary composite structures can be manufactured through an oven-curing process with a required part quality typically achievable with autoclave processes [
3]. The advantages of VBO compared with autoclave processing are the lower capital investment, the elimination of size constraints (larger parts) and the need for expensive nitrogen gas, and enabling higher energy efficiencies [
4]. On the other hand, composite parts manufactured with VBO processing suspiciously include a high amount of voids caused by trapped air bubbles, which degrade the mechanical performance of the parts [
5]. Unlike the autoclave processes, the maximum consolidation pressure applied during the VBO prepreg processing is the atmospheric pressure. Considering that the fiber bed carries a fair amount of this pressure, the remaining consolidation pressure on the resin may not be sufficient to discharge or suppress voids [
6]. It is, therefore, critical to developing a practical methodology to migrate air bubbles, evaporated moisture, or other volatile substances towards the vacuum outlet port before gelation of the resin to produce low-void (<1 vol%) composite materials.
VBO prepregs have dry and relatively permeable air channels (engineered vacuum channels or EVaCs) that allow air removal when the vacuum is applied. The resin is progressively impregnated into these channels during the process so that it is evenly distributed and carries a low amount of voids [
6]. During impregnation, the resin flow dynamics directly govern the void content in the final product, depending on the removal efficacy of the trapped air within the part [
7]. Therefore, a comprehensive understanding of material constitutive behaviors and resin flow dynamics is required to obtain void-free composite structures produced with VBO processes [
3].
The VBO process, in general, can be divided into three separate but interdependent components: porous media flow, heat transfer, and consolidation mechanisms [
2,
8,
9]. The flow in porous media exists due to the resin flow between fibers. Dry fibers are expected to be impregnated with the resin while enabling air discharge. Centea and Hubert [
6] observed resin impregnation in various stages of the VBO process by adapting the micro-CT imaging approach. Comprehensive mathematical models and experimental verifications involving the resin impregnation and bubble migration for VBO prepregs are available in the literature [
7,
10,
11]. In addition to these studies, Gangloff et al. [
12] evaluated void formations and bubble migration based on time, pressure, and temperature, among the process parameters. Heat transfer is included in the VBO process by several studies in the literature based on the modeling of cure kinetics and resin viscosity. The effect of cure kinetics on mechanical performances and void formation, particularly in resin-rich regions, has been investigated by various studies [
8,
9,
13,
14,
15]. Kratz et al. [
9] characterized the two VBO prepreg systems regarding their cure-dependent properties, cure kinetics, viscosity, and glass transition temperature following the standardized methods outlined by Khoun et al. [
8]. The kinetics model’s role in predicting temperature evolution was investigated to clarify the exothermic heat generated during the curing process in thick composite parts. However, in the literature, the temperature values changing with the effect of cure kinetics were not included in the mathematical modeling, and the instantaneous value of the temperature during the process was not accurately characterized. This deficiency leads to inaccuracies in the viscosity, which also depends upon the temperature. There are also studies in the literature on the coupled effects of impregnation and cure behavior. Centea and Hubert [
16] analyzed the resin impregnation with various models, including the cure kinetics and resin viscosity. They investigated the effects of the fiber architecture, temperature profile during the curing, and the initial degree of cure of the resin system through parametric studies. Additionally, they incorporated several characterization studies to develop a model for the fiber architecture. Moreover, the impact of the initial degree of cure on the degree of impregnation of the resin system, which changes with the out-times of the prepregs at room temperature prior to curing, were studied by neglecting other process parameters and its effects on void formation were reported [
17]. Furthermore, heat transfer is driven by two other thermal properties: specific heat capacity and thermal conductivity. Specific heat can be expressed as the amount of heat that the material absorbs 1 °C temperature per 1 g mass, and it is usually a function of temperature. Dynamic Scanning Calorimeter (DSC) is one of the commonly preferred methods used to measure the heat capacity of materials [
18]. In the literature, specific heat capacity was considered a single input to the models, and its evolution during the curing was not thoroughly characterized [
17,
19,
20]. However, Kalogiannakis et al. [
20] investigated the specific heat capacity behavior of carbon/epoxy and glass/epoxy cross-ply laminates with a Modulated Temperature Differential Scanning Calorimetry (MTDSC). They revealed that heat capacity was almost doubled between pre- and post-glass transition stages, and therefore, the heat capacity of the composites is strongly dependent on the temperature. On the other hand, thermal conductivity is another influential parameter for heat transfer since it measures a material’s capability to transfer heat. Carbon fiber-reinforced composite materials with a unidirectional fiber orientation demonstrate different thermal conductivities in the in-plane and through-thickness directions. Hence, the thermal conductivity of the resin and fiber components of the prepreg could be investigated individually by certain thermal conductivity models [
18]. In addition to the effect of geometrical disposition and fiber/resin fraction, thermal conductivity is a temperature-dependent property, which means that the material can demonstrate different thermal conductivity behaviors in different temperature conditions. According to the literature, although thermal conductivity is a temperature-dependent property, it varies slightly in a limited temperature range [
19,
20]. Therefore, it is preferable to conduct the experiments in a wide temperature range to observe the noticeable differences in the thermal conductivity of the constituents.
Another mechanism that needs to be included in the modeling is consolidation. Consolidation is studied in the literature as the resin flow within a fiber architecture. Various mathematical models and numerical analysis methods were developed for VBO prepregs [
2,
21] to understand the effects of the consolidation mechanism. Gangloff et al. [
10] investigated the interactions between engineered air channels and consolidation, and they demonstrated the influence of the consolidation profile on void formation. Centea and Hubert [
22] performed a parametric study for the consolidation profile under different pressures for the VBO process and analyzed the effect of the consolidation mechanism on the microstructure of the final product. They concluded that the effects of other process parameters need to be taken into account.
The design of the VBO process based on the fluid flow, consolidation, and heat transfer should be linked to accurate material parameters of the prepreg and its constituents to derive acceptable process parameters, as schematically shown in
Figure 1. For this reason, there is a vital need for a systematic and inclusive characterization study for VBO prepreg material characterizations. Considering the studies in the frame of VBO process design with various perspectives, to the best of authors’ knowledge, there is no systematic characterization study in the literature focusing on related prepreg material properties in an integrated manner with the physics of the VBO process. This study intends to establish a methodology that systematically characterizes the VBO prepreg properties and develops constitutive behavior to strengthen the accuracy and reliability of the VBO process model. Accordingly, this approach is applied to characterize the properties of a commercial VBO prepreg system and was carried out in three steps. First, for the resin system, the cure-dependent properties are characterized regarding cure kinetics, glass transition temperature, and viscosity by semi-empirical phenomenological models. Second, the fiber architecture is investigated for the resin film, fibrous region, and void-content change and the fiber volume fraction of the prepreg system through sets of X-ray tomography scans of the uncured and cured samples. This study is followed by the numerical permeability characterization of the initial porous media, modeled through laminar flow analysis of the selected domain. Finally, the specific heat capacity and thermal conductivity of the constituents are measured by a novel experimental design. This novel integrated prepreg material characterization recipe maintains the numerical implementation with improved reliability of the process modeling and optimization towards the success of the VBO process design.
4. Conclusions
In this study, a comprehensive methodology that systematically characterizes material properties and the constitutive behavior of a vacuum bag only prepreg system was developed to develop an integrated Vacuum Bag Only process model and to invigorate its accuracy. This approach was established based on three primary process parameters: heat transfer, flow through porous media, and consolidation.
First, the cure-dependent properties of the epoxy resin were characterized to predict the cure kinetics, rheology, and glass transition temperature behavior under Vacuum Bag Only cure conditions. The cure-dependent model parameters were obtained by adopting the least-squares methods to fit the experimental data determined through Dynamic Scanning Calorimetry, Rheometer, and Dynamic Mechanical Analysis, respectively. For this, a diffusion-controlled cure kinetics model accurately predicted the cure kinetics behavior of the epoxy resin and exhibited excellent agreement with the dynamic and isothermal Dynamic Scanning Calorimetry experiments. Subsequently, a phenomenological viscosity model was applied to successfully estimate the resin rheological behavior as a function of temperature and degree of cure. The model demonstrated a reasonable agreement with the rheology experiments at different dynamic conditions. Lastly, the DiBenedetto equation was employed to describe the glass transition temperature evolution of the epoxy resin system with the degree of cure.
Second, first fiber architecture was investigated as the main microstructural evolution is the resin flow into dry regions. The resulting micrographs exhibited that the domain wherein resin flow and air evacuation occur consisted of elliptical dry fiber tow areas, containing randomly packed fibers, surrounded by resin-rich regions. This was followed by void-content and the fiber volume fraction analyses of the prepreg system through sets of x-ray tomography scans of the laminates processed to different curing stages. Dry fiber tow areas were initially significantly increased after the consolidation and air evacuation and stabilized after the resin reached. Furthermore, microscopic transverse permeability of the fabric was calculated through a numerical analysis for a micro-scale domain consisting of random packs of fibers and resin-rich regions. This predicted permeability value then can be inputted as an initial permeability value for the subsequent mathematical modeling and numerical analysis.
Third, specific heat capacity and thermal conductivity of the prepreg system constituents were characterized to better describe the heat transfer of the laminates during the Vacuum Bag Only prepreg processing and to include them into the integrated process model to be established. A series of experiments were performed through Dynamic Scanning Calorimetry and Thermal Constant Analysis to predict the evolution of specific heat capacity and thermal conductivity of the prepreg system with the temperature, respectively. Hence, the evolution of these two fundamental thermal properties and their effects on the heat transfer of the system were able to be characterized and incorporated into the subsequent process modeling.
Conclusively, the devised approach can be accepted as a starting point towards establishing a process design methodology that integrates characterization, modeling, optimization, and verification to produce high-performance composite structures through Out-of-Autoclave techniques with the allowed void content. In future work, the results of this study will be used as an input to develop integrated Vacuum Bag Only prepreg process modeling.