1. Introduction
In the modern aerospace domain, the creation of top-tier composite elements stands as a fundamental practice which is frequently accomplished through the utilization of autoclave prepregs. [
1]. The traditional autoclave methodology involves the application of substantial consolidation pressure, typically within the range of 3–12 atm, to prepregs within an autoclave. This controlled pressure disrupts macro- and mesovoids due to trapped air or volatiles within the prepreg laminate. This fragmentation leads to their subdivision into smaller fragments that are then effectively extracted through resin flow. This intricate process yields components that are characterized by a formidable mechanical strength, minimal porosity, and exceptional uniformity [
2,
3]. However, the autoclave route carries inherent expenses linked to its acquisition and maintenance, coupled with the substantial power requirements and potential occupational hazards associated with high temperatures and pressures. These limitations underscore the need to explore alternatives to the autoclave process.
This motivation has driven the exploration of the Out-of-Autoclave (OOA) approach within the aerospace industry. Notable for its independence from the use of an autoclave during the fabrication process, the OOA approach offers many benefits, including cost-effectiveness, a reduced environmental impact, and heightened adaptability to diverse component geometries. This avenue of research has been the subject of extensive investigations by scholars and researchers in recent decades. Among these alternative approaches, the vacuum-bag-only (VBO) process has emerged as a promising avenue. This process offers enhanced control over the volume content and orientation of fibers, making it an attractive, cost-effective alternative to traditional autoclaving. Unlike the autoclave process, the VBO process relies solely on a vacuum pressure of 1 atm to consolidate prepreg laminates. This significant departure from high consolidation pressures yields cost savings and allows for the adoption of economically viable curing methods such as heat ovens, blankets, and molds [
4,
5]. It is important to note that the lack of substantial consolidation pressure in the VBO process precludes the intricate expulsion mechanisms found in autoclave processing. A review of the out-of-autoclave process was reported in [
6].
A pivotal component of the VBO process involves the use of specialized, semi-impregnated prepregs known as out-of-autoclave (OOA) prepregs. Unlike their fully impregnated counterparts, these prepregs feature engineered vacuum channels (EVaCs) integrated into the dry fiber regions. EVaCs enhance the in-plane air permeability of the laminate, thereby facilitating the escape of air, moisture, and volatiles during vacuum application [
5,
7]. Recent advancements have introduced diverse variations of OOA prepregs to enhance process stability and elevate the quality of the resulting components [
4,
8,
9]. Empirical evidence underscores that under optimal processing conditions and when paired with OOA prepregs, the VBO process can yield composite components whose mechanical attributes are on par with those of their autoclave-manufactured counterparts [
5,
7]. The vacuum-bag technique can be used to manufacture primary structures such as decks, hulls, superstructures, and bulkheads and secondary structures such as partition panels and interior joint work [
10].
However, the VBO process is sensitive to various process parameters, including debulking cycles, curing profiles, humidity levels, and air permeability characteristics [
11,
12,
13]. The execution of the VBO process under suboptimal conditions can result in an elevated void content within components, consequently impacting crucial mechanical properties such as transverse tensile properties, bending characteristics, interlaminar shear strength, and elastic attributes; the risk of composite delamination is also accentuated under such circumstances [
14,
15,
16]. A double-bagging process was also proposed to facilitate the degassing process from the collapse of prepreg stacking [
17].
Numerous research studies have been conducted to identify optimal process parameters for the VBO technique, with the overarching goal of enhancing the robustness of the process and minimizing porosity within components. One approach involves the integration of heated debulking into the VBO process. While this enhancement augments out-of-plane air permeability, it concurrently introduces the risk of obstructing in-plane air evacuation [
18]. A study conducted by Ridgard et al. [
19] demonstrated this principle by subjecting laminate samples to a 50 °C heated debulk for 4 h before curing. The resulting components exhibited porosity levels comparable to those achieved through a 16 h room-temperature debulk. Hu et al. [
18] delved deeper into the underlying mechanisms, establishing a correlation between out-of-plane permeability, debulk temperature, and prepreg fiber structure. Elevated debulk temperatures were associated with a lower resin viscosity, thereby facilitating the expulsion of inter-laminar voids through the out-of-plane direction. However, this simultaneous infiltration of resin into dry fiber regions obstructed in-plane air evacuation. Their conclusions underscored the applicability of heated debulking to woven fiber laminates with length-to-thickness ratios exceeding 5.5, while unidirectional fiber laminates with lower thicknesses (<1 mm) benefited the most. Maguire and colleagues delved into the significance of prepreg formats and the production process for VBO prepregs [
20]. The manual application of epoxy powder was examined for its potential to result in an uneven distribution of powder, potentially resulting in improved laminate uniformity. The research findings validated the theory that epoxy powder serves as a preventive measure against exothermic reactions in thick composite materials. Edward and his colleagues developed a unidirectional semi-prepreg aimed at enhancing the reliability of VBO processing [
21]. They utilized a fortified epoxy resin for this purpose. The semi-prepreg was tailored to halt the distribution of resin. Consequently, there was an enhancement in through-thickness permeability, enabling more effective gas removal. Laminates created using the semi-prepreg exhibited fewer imperfections compared to those formed using traditional VBO prepregs. The morphology of the resin’s characteristics was noted as a crucial factor influencing the formation of defects.
In a study by Mujahid et al. [
22], a manufacturer-recommended two-stage cure cycle (MRCC) was employed as a benchmark. Recognizing the influence of residual stress and strain effects, two modified cure cycles were introduced. The EMRCC inserted an additional curing stage into the original MRCC progression. At the same time, the DC incorporated an elevated first-stage dwelling temperature. The DC yielded a notable 18.12% reduction in through-thickness voids compared to the MRCC. Furthermore, the EMRCC and DC resulted in tensile strengths 20% and 17.4% higher than the MRCC, respectively.
Hyun et al. [
23] leveraged resin cure kinetics and viscosity modeling, using the “effective flow number” concept to optimize the reference cure cycle. Their modifications, such as the replacement of the 16 h room-temperature debulk with a 2h 60 °C heated debulk (Modified I) and the substitution of the 121 °C dwelling stage with a prolonged 177 °C dwelling stage (Modified II), resulted in the components produced via the Modified II cycle displaying the highest degrees of cure and tensile strength.
It is imperative to highlight that utilizing OOA prepregs within the VBO process entails high costs, extended lead times, and minimum order quantity requirements. Addressing these challenges, Yang et al. [
24] devised a hybrid layup laminate for the VBO process, incorporating fully impregnated prepregs alongside dry fiber fabrics. The inclusion of dry fabrics mimics the action of partially impregnated VBO prepregs, facilitating air and volatile evacuation. Chang et al. [
25] manufactured unidirectional carbon fiber laminates measuring 50 × 50 cm² using the above hybrid laminate which demonstrated tensile strengths akin to their autoclave-manufactured counterparts and minimized thickness deviations. However, it should be noted that the flexural strength of the hybrid laminate was 16% lower due to its heightened porosity, which resulted from protruding dry fabric wefts. This research initiative aims to meticulously analyze the intricate interplay between multiple VBO process parameters and laminate porosity within a prepreg/dry fiber laminate. The goal is to enhance the quality of components for their prospective applications.
2. Methodology
Experiments were carried out in three phases to comprehend the impacts of various process parameters on the hybrid layup laminate parts. In each step, unidirectional carbon fiber panels measuring 15 × 15 cm² were produced using the VBO process, employing different process parameters for quality assessment.
2.1. Prepreg /Fiber Hybrid Layup
A unidirectional carbon fiber prepreg UD150 (37 wt% resin) and dry fabrics provided by Wah Hong Industrial Corp, Taiwan, were used in this research. The laminate arrangement design principle was established by Chang et al. [
25]. The carbon fiber prepreg and dry fabrics were layered alternatively, with the dry fabrics acting as the EVaCs between the prepregs. During the consolidation process using vacuum pressure, it is essential to ensure that the excess resin in the prepregs is enough to fill the dry fabrics. In the design process, the laminate’s saturation index (
Sindex) is a critical factor to consider. The
Sindex reflects the ratio between the resin and the porosity within the laminate structure. Previous research has shown that achieving an oversaturated state (
Sindex > 1) is essential to ensuring the effective filling of laminate dry fiber regions with resin. The expression for the
Sindex is provided below:
where
represents the total volume of resin in the laminate, and
represents the total volume of voids under 1 atm of pressure on the laminate. Based on the required
Sindex and laminate thickness, the number of layers of prepregs and dry fabrics, along with the stacking sequence, can be determined.
2.2. VBO Process of Producing a Hybrid Layup Laminate
The unidirectional carbon fiber laminates used for the quality assessment were fabricated using the VBO process, as illustrated in
Figure 1. Initially, the prepreg was left at room temperature for 30 min to de-ice it, preventing the accumulation of moisture in the laminate and reducing the viscosity of the resin for stacking. The prepreg was then cut into a rectangular shape measuring 15 × 15 cm², while the dry fiber was cut into 15 × 16 cm²; the additional 1 cm of dry fiber would contact the breather layer to create a continuous air channel, enhancing interlaminar air evacuation.
The next step entailed stacking the prepreg and dry fiber in the designated sequence to form a hybrid layup laminate. A plastic blade was used to scrape the laminate surface, preventing the entrapment of air bubbles during the layup. Simultaneously, a release agent was applied to the mold and left to dry completely for 40 min. Subsequently, the laminate was placed on the mold, followed by the peel ply, separator, and breather. The entire mold assembly was then sealed using a vacuum bag and sealant.
Once sealed, the assembly was subjected to vacuum pressure to debulk, removing the air, moisture, and volatiles inside the laminate. Finally, the laminate was heated based on the selected cure cycle, ensuring that the resin thoroughly impregnated the dry fiber area and ultimately reached a fully cured state.
2.3. VBO Process Parameters
For the VBO process parameters, the cure cycle, saturation index, debulking method, and laminate thickness were chosen as the parameters to be studied, given their easily adjustable natures and potential to significantly influence part quality. The standard parameters listed in
Table 1 were derived from previous research and served as a baseline for subsequent parameter variations where P is one layer of prepreg, Pb is two layers of prepreg, and F is the dry fabric.
First, a comparison was conducted between two-stage cure cycles employing different dwelling temperatures, as listed in
Table 2. The elevated dwelling temperature (140 °C) yielded a lower minimum resin viscosity but a shorter gel time. This translates into an improved resin flow yet a limited time for infiltration into the dry fiber area. Conversely, the reduced dwelling temperature (100 °C) had the opposite effect. Throughout this comparison, all other parameters remained constant.
Secondly, various saturation indexes (
Sindex) were compared. Previous research established that the laminate needs to be oversaturated (
Sindex > 1) to fill all voids adequately. Different
Sindex values, including 1.27 (standard), 1.57, 1.69, and 1.92, were selected to further investigate the impact of excessive resin on part quality. As shown in
Table 3, The varying
Sindex values were achieved by adjusting the number of prepregs in the laminate arrangement. Again, all other parameters were kept consistent during this comparison.
Lastly, three debulking methods—a standard debulk, an elongated debulk, and a heated debulk—were compared to assess the effectiveness of longer debulk times or higher debulk temperatures, as listed in
Table 4. Different laminate arrangements, including standard, elevated
Sindex, and double thickness, were also included in the comparison, as shown in
Table 5. All other parameters remained constant throughout this comparison.
2.4. Multi-Stage Curing Cycle
Table 6 shows that the optimized parameters concluded from the previous section were adopted as the standard parameters, forming a baseline for modifications to the cure cycle. The initial step of the cure cycle was increased to 140 C, and the laminate arrangement with saturation index of 1.57 was used. To establish a three-stage cure cycle, a lower temperature stage (100 °C) was introduced before the initial step in the two-stage cure cycle. A four-stage cure cycle was also developed by introducing a new location at 120 °C to ensure a smoother temperature progression.
Table 7 shows all the multi-stage cure cycles. The aim of incorporating of an additional cure cycle stages was to provide the resin with an extended gel time, allowing more time for the infiltration of resin into the dry fiber area.
2.5. Fiber Layup Orientation
The optimized parameters and the most favorable cure cycles from previous sections were established as the standard parameters, as shown in
Table 8. This phase explored the influence of different fiber orientations between the prepreg and dry fiber fabrics on the quality of the laminate. Laminates featuring prepreg fibers at 0° and dry fibers at 90° were compared with laminates utilizing a sole 0° layup, as shown in
Table 9.
2.6. Quality Assessment Methods
Three assessment methods—microscopy, an average thickness assessment, and a laminate resin loss assessment—were utilized to comprehend the varying levels of part quality. From each produced laminate, four specimens were cut and polished, as shown in
Figure 2 for the left, right, up, and down parts.
Digital microscopy with a magnification of 22.5 was employed to capture images of the specimen cross-sections, as shown in
Figure 2a. These images were then calibrated using OpenCV and processed through imageJ. In the processing, the images were converted to grayscale to identify voids. The porosity in each photograph represents the void area’s percentage of the total image area. The average porosity of a laminate was calculated from the average porosity in the specimen images while excluding outliers. For accuracy, only specimens with cross-sections parallel to the fibers (L and R samples from the laminate) were used for porosity calculations due to the void shape characteristics of hybrid layup laminates, as shown in
Figure 2b,c. Since most voids were located near the dry fiber region, the images were centered in this area.
To determine laminate thickness, specimens were measured at the same 10 points (40 points in total for L, R, UP, and DN samples) to obtain a precise average laminate thickness, as shown in
Figure 3. For laminates using an oversaturated laminate (
Sindex > 1), the estimated thickness
without considering resin loss was the sum of the laminate resin, prepreg fiber, and dry fiber lumped thicknesses. The expression for the estimated laminate thickness
is provided below:
where
is the resin lumped thickness,
is the prepreg fiber lumped thickness, and
is the dry fiber lumped thickness. The estimated fiber volume content
can be calculated using the lumped thicknesses of the laminate fibers and the estimated laminate thickness
, and is expressed as below:
By measuring the average laminate thickness
, we can then calculate the actual fiber volume content, as shown below:
To compare the laminate resin loss under different process conditions, the weights of the laminates before and after curing were measured. The laminate resin loss ratio
was defined as below:
where
is the weight of the laminate before curing, and
is the laminate’s weight after curing.
4. Conclusions
This research substantiates the impact of VBO manufacturing parameters and cure cycles on the quality of unidirectional carbon prepreg/dry fiber hybrid laminates, successfully reducing porosity and yielding high-quality laminates. In the initial stage, elevating dwelling temperatures (140 °C) during the cure cycle decreased resin viscosity, thus enhancing the infiltration of resin into fiber tows and reducing the laminate’s porosity. Sindex values above 1.57 indicate diminished porosity due to increased resin filling in regions with high thickness deviations and fiberless sections. Heated debulking is unnecessary, resulting in premature resin infiltration, blocking EVaCs, and resulting in higher degrees of porosity in thinner laminates (1.6 mm and 2.6 mm). Increasing the number of cure cycle stages extends the duration of the resin’s low-viscosity state, enhancing the impregnation of dry fiber tows and reducing porosity. Aligning the prepreg fibers with the polymer weft on the dry fiber fabric eliminates weft-induced voids, showcasing the feasibility of tailoring prepreg/dry fiber orientations in laminate production.
In conclusion, to produce unidirectional carbon fiber composite laminates, it is recommended to maintain an Sindex of at least 1.57 to ensure sufficient resin content in the laminate. Employing a higher initial dwelling temperature (140 °C) for the cure cycle, followed by a four-stage cure cycle modification, ensures optimal resin infiltration. If achieving minimal porosity is a priority, stacking prepreg and dry fabric with a 90° directional difference is advisable.