The vacuum infusion process was originally developed as an alternative for open-mould processes like spray-up and hand lay-up. Besides the intended environmental and health benefits, a considerable quality improvement was noticed. Especially the fibre volume content was increased and the void content decreased. This led to the assumption that the vacuum infusion process could be developed into a low-cost alternative for the prepregging process. The prepregging process normally results in laminates with a very high fibre volume content. If the process is carried out carefully, almost void free laminates can be produced. But it is also a rather expensive process. The prepreg material itself is relatively expensive, and it has to be stored in a conditioned environment.
For conventional prepregs, an expensive autoclave cycle is required to cure the laminate where the inside dimensions of the autoclave limit the size of the product. If vacuum infusion would result in laminates with the same mechanical properties as laminates made with the prepregging process, this would result in a considerable cost-reduction. It would then also be possible to produce very large structures in one piece, eliminating the need for subsequent assembly of many smaller parts.

This site describes the vacuum infusion process as low-cost fabrication method of two examples of thick-walled carbon-epoxy structures. Subjects like preforming of the dry reinforcement, resin shrinkage, degassing of resin, infusion strategy, design for infusion, consumables for high temperatures and so on will be described.

The vacuum infusion process is a variation of the resin infusion processes, and thus related to the well-known RTM process. The distinction is that where the RTM process uses an increased pressure at the resin inlet (possibly combined with an decreased pressure at the resin outlet), the vacuum infusion process only uses a reduced pressure at the resin outlet. The major consequence is that RTM required two stiff, rigid moulds, firmly clamped together. The vacuum infusion process requires only one stiff mould and the other mould half can be a flexible film.

The first example is a section of a spar of an A321 wing. As part of a technology program of Airbus UK, the feasibility of the vacuum infusion process was investigated as a potential affordable manufacturing technique for an aircraft composite primary structure. The demonstrator for this program is an inward facing C-shaped wing spar with prime dimensions: 30 cm web height, 8 cm flange width and 10 mm thickness. The spar section features a slight taper, a local reinforced area and integrated rib posts, see figure 1.

The baseline technology for the production of the spar is the autoclaving prepreg technique. Vacuum infusion is considered as an alternative, probably low cost manufacturing technique. Since the process uses vacuum and an one-sided tool only and since an autoclave is not required for curing the part, the manufacturing costs can potentially be very low.
The capabilities of the vacuum infusion process were demonstrated on one meter sections of the C-spar. The requirements of the resulting demonstrator were:

Figure 1 shows the spar with part of the surrounding structure. Not shown are the upper skin and ribs that also have to be attached to the spar. Therefore, the only geometrical critical surfaces of the spar are the outside surfaces of the flanges where the skins are attached to, and one side of the rib-post where the rib has to be attached. The outside surfaces of the flanges can be defined by using a negative mould. The surface of the rib-post requires the use of a positioning tool. The inside surface of the spar and the thickness of the spar laminate are in principal undefined but are governed by the number and type of reinforcement layers and the way the vacuum infusion process is carried out. The lay-up of the spar and type of reinforcement were given by Airbus UK.

The vacuum infusion process was developed for the C-spar by the following approach:

The experiments on flat specimens are simple to perform and give a good indication of injection properties of the materials used. The exact type of reinforcements tested is proprietary information, but all laminates were built up from stitched carbon uni-directional layers (0/90o and +/- 45o) with an average areal weight of 600gr/m2. The laminates were symmetrical combinations of 60% +/- 45o layers and 40% 0/90o layers resulting in a 10 cm thick laminate. Since a high fibre volume content is aimed for, no high permeable fibre layers were used in the laminate. To ensure a proper resin flow, a high permeable layer was placed on top of the laminate, separated by a peel-ply. See figure 2.

The length of the specimens was 30cm, corresponding to the flow-length in the actual spar. Since a Tg of above 150oC was required, the resin systems meeting this requirement all had to be infused at elevated temperatures. Both epoxy and cyanate ester resins were tested. To be able to infuse and cure at elevated temperatures, also the consumables used have to be able to withstand these temperatures. All resin systems have to be infused at 60oC and can be cured at 120oC (with an additional post-cure at even higher temperatures). The following consumables have found to suffice:

The following procedure is followed for all infusion experiments:

The resin is mixed, heated and degassed. In previous research (ref. 3) it has been found that the traditional degassing method does not suffice for high quality vacuum infused laminates. An improved degassing method has been developed and validated (ref. 2) where a piece of Scotchbriteä is placed at the bottom of the container with mixed resin. The container is then placed in vacuum room where the pressure is decreased to 5 mbar for 5 minutes. Pressure in mould cavity is set to 20 mbar. This pressure is higher than the degassing pressure, to avoid outgassing of remaining gaseous components in the resin. But the pressure is still low enough to minimise the amount of air in the dry laminate and to ensure a proper resin flow. By temporarily disconnecting the vacuumpump and monitoring the pressure in the mould cavity, the airtightness of the mould is checked. When no leaks are detected, the vacuumpump is re-connected again. Mould is heated. The whole mould system is placed inside an oven for at least 60 minutes. The elevated temperature and the low pressure will ensure that all moisture is removed from the reinforcement. The resin inlet hose is placed in the container with mixed, heated and degassed resin. Resin inlet is opened. The resin container is also placed inside the oven to maintain the temperature of the resin. When laminate is fully impregnated, the injection pressure is increased to 400mbar. By increasing the injection pressure, the pressure in the resin also increases, thus decreasing the size of any bubbles present in the resin. The reinforcement type permits a pressure of 400mbar without increasing in thickness, which might result in low fibre volume fractions. Inlet is closed. The resin container is removed from the oven to avoid exothermic reactions during cure. Then, the cure cycle is started. Although some of the resin systems can be cured at higher temperatures, all cure cycles have a dwell at 120oC. This will ensure a (partial) curing of the laminate without melting of the resin distribution material. When the resin is cured, the remaining cure cycle follows.

The experiments on the flat specimens showed that it is possible to infuse the reinforcement types. Fibre volume contents varied from 55% to 62%. Void content was typically below 2%, although locally, the void content was slightly higher. The experiments also revealed that the permeability of the reinforcement was very low, both in plane as through the thickness. This makes the process very critical for small, local variations in permeability due to stitching patterns, bridging of fibre layers and/or bagging film and fibre orientations.

The injection experiments showed that it is possible to inject laminates with a flow-length of 30 cm. This means that it should be possible to inject from one flange of the spar to the other flange. This could result in the injection strategy shown in figure 3.

This strategy was tested on a small section of the spar. Many of the resulting laminates had areas with a high void content. These voids could result from either resin shrinkage, outgassing of the resin or unforeseen flow phenomenon resulting from a change from a 2D to 3D set-up. To investigate this, a flow simulation program was used and small-scale experiments were performed. The simulations revealed that the infusion could be critical for the ratio between permeability of the resin distribution material and the permeability of the laminate itself. An unfavourable ratio could result in dry spots along the radius of the spar. To find a better, more robust injection strategy, a resin distribution material with a lower permeability can be applied. Such a material could be Injectex (by Hexcel), a glass or carbon fabric especially intended for infusion purposes. Another option was to apply this material also on the mould side of the laminate. Although this material results in a laminate with a locally lower fibre volume content, it could be favourable to have a fabric on the outside of the laminate to prevent fibre breakout during drilling and cutting.
To validate these findings, experiments were conducted in a transparent mould. The experiments revealed that unwanted runner channels due to bridging of the fibre material in the radius of the mould had a much larger influence on the enclosing of air, than the ratio of permeabilities, see figure 4.

Since bridging of fibres in the radius of the mould is very hard to prevent, a different route is suggested. Instead of infusing from flange to flange with a large risk of enclosing air, a different infusing strategy is investigated. The inlet channels are located between the laminate and the mould, see figure 5. This infusion strategy reduces the flow-length by a factor four, thus reducing the fill-time by a factor four. This could mean that the use of resin distribution material to ensure a proper resin flow is not needed anymore. This would make the preparation much easier and would save both material and labour costs.

The modified injection strategy was first tested in the transparent mould. No air was enclosed during infusion and the resulting laminate showed no defects. Therefore, experiments in the actual spar moulds could be performed.

In the actual spar mould, the injection strategy described above is validated. The spar mould features a local taper, resulting in a three dimensional deformation of the dry reinforcement. This makes it even harder to get the reinforcement to follow the geometry of the mould, resulting in more bridging of the reinforcement in the radius than before. In this experiment, no rib-posts or patches were integrated yet. The resulting spar showed no defects and had a fibre volume content of 60%. See figure 6.

The rib-posts are placed in the mould and positioned with a special positioning tool, see figure 8.

The complete, bagged mould system including resin inlet and outlet hoses and the resulting spar after infusion are shown in figure 9. The average fibre volume content is 61% and no voids can be detected visually.

The experiments described in this paper show the feasibility for vacuum infusion for high quality, carbon epoxy structures with a high fibre volume content. Although all initial requirements were fulfilled, the expensive and time-consuming certification procedures could hamper a swift application of vacuum infusion for aircraft primary structures. Another route to achieve cost-savings in aircraft building is foreseen by applying vacuum infusion for secondary structures first. Many structures like fairings can be produced very cost-effective with vacuum infusion by making the use of expensive prepregs and autoclaves redundant. When sufficient confidence in the process is developed, also the chances for successful application in primary aircraft structures will be increased enormously.