INVESTIGATION OF A TWO-STAGE INJECTION PROCESS TO REDUCE THE EFFECTS OF IN-PLANE RESIN FLOW

Transcription

1 AIAA INVESTIGATION OF A TWO-STAGE INJECTION PROCESS TO REDUCE THE EFFECTS OF IN-PLANE RESIN FLOW Erik Larsen, Douglas Cairns, John Mandell, and Daniel Samborsky Montana State University Bozeman, Montana, ABSTRACT This paper is a discussion of a manufacturing technique that reduces dependency on in-plane resin flow, allowing greatly reduced injection times, increased injected resin volume per port (reducing the number of ports required), and allows relatively higher fiber volume content. This method uses a two-stage injection process. The first stage injects resin into a vacuum evacuated pool outside of the plane of the fabric. Pressure is then applied outside of a flexible film, forcing resin into the fabric in the thickness direction. Since resin is not required to flow in the plane of the fabric, relatively high volumetric flow rates are possible during the injection stage. The hydrostatic pressure on the flexible film results in part-to-part and spatially consistent high volume content. This process has also shown to be less sensitive to fabric and manufacturing inconsistencies than traditional twosided mold RTM, and has been used to successfully manufacture parts using difficult-to-rtm materials such as bonded fabrics. Drawbacks of this process include increased mold and process complexities, poor surface finish on one side of the part, and increased complexity of process tuning. INTRODUCTION The introduction of Resin Transfer Molding (RTM) methods for composite production has allowed dramatic improvements to the quality of composite parts used in wind turbines [1]. However, traditional RTM using a two-sided mold has drawbacks. Actions taken to achieve relatively higher fiber volume content tend to drastically increase injection times or increase the number of injection ports required to fill a mold. To achieve high fiber volume content, the fabrics need to be squeezed relatively tightly in the mold during injection. Resin is then forced to flow in the plane of the fabric during injection. This in-plane resin flow introduces several limitations to the standard RTM process. Specifically, a pressure gradient along the fabric is needed to cause resin flow. In the standard RTM process, resin is required to flow in the plane of the fabric via Darcy's law [2]. K ij v = p i µ, j where v i is the velocity of the resin flow front, K ij is the permeability tensor (fabric dependent) µ is the resin viscosity p is pressure i is parallel, j is normal to flow direction. This pressure gradient, along with the entire RTM process and mold design, needs to be carefully designed and controlled to reduce fiber wash, injection times, and the number of injection ports required. Fabric architectures have been introduced that allow resin to flow in channels in the fabric, alleviating some of these limitations [3]. This architecture, however, can cause undesirable inhomogeneities in the composite. Twosided RTM molds require critical interior surface spacing and support for flow characteristics and acceptable product fiber volume content. This investigation focused on a process that injected resin into a fabric in two stages. First, a stage that injected resin into a mold cavity with a larger spacing than the desired final part thickness, then a final pressure stage when a hydrostatic pressure was applied to the resin by way of a bagging film, completing the resin distribution process. The larger spacing injection process is governed by channel flow type resin distribution. The equations for motion for channel flow, neglecting the effects of gravity are [4, 5]. v t i ρ + v j vi, j = p, i + µ v i, jj 1) 2)

2 where ñ is the mass density, v is the velocity, p is the pressure, õ is the resin viscosity i is parallel, j is normal to flow direction. Typically, channel flow provides much higher resin flow front velocities than Darcy's law flow as discussed in Reference [6]. This is the mechanism of flow that was utilized in the two-stage injection process described herein. This process has shown to be able to inject relatively larger volumes of resin from a single port, fill larger areas of fabric, reduce injection times, and successfully inject longer distances from a port. The final product has shown to have consistently high fiber volume percents. THE PROCESS The mold configuration for this process is shown in Figure 1. The bottom mold part in the figure has the finished mold surface. The dry fabric is placed in the bottom part in the desired layup. The flexible film is placed over the fabric. The top part is fastened to the bottom, dividing the mold into two sealed cavities, one with the fabric, and one above the film. The top is rigid to support both vacuum and pressure during the different stages of the process. The breather fabric is used to prevent the bagging film from closing off the vacuum port during the injection stage. Without the vacuum in the top cavity, this process is essentially the SCRIMP [7] process. Before injecting, pressure is reduced in both cavities to near (but above) the vapor pressure of the styrene in the resin. Reducing the pressure in both cavities results in no net pressure on the bagging film during injection. This is different from SCRIMP where atmospheric pressure remains on the film during injection. Resin is injected into the lower cavity of the mold, above the top layer of fabric. The resin pools near the port in the relatively large cavity (Figure 2). After injecting a predetermined amount of resin, the injection port is closed. The pressure in the top cavity is increased (typically to 70 kpa). The bottom cavity is typically left at the reduced pressure until resin first appears at the vacuum port. When the resin exiting at the vacuum port appears relatively less foamy, the injection port is opened to allow excess resin to drain from the vicinity of the injection port. It may be possible to eliminate this injection port draining for smaller parts and/or with proper process tuning, resulting in a near-net resin injection. Since the injection stage typically only takes a few minutes, there is relatively large amount of time for the resin to distribute itself during the second stage before the resin gels. During this time, the resin distributes itself through the fabric. PROCESS VARIABLES The following is a partial list of variables that have been identified to affect the process and/or the results. Fabric/Layup. Since the second stage of this process does depend to some lesser degree on in-plane resin flow, the same flow performance characteristics that affect typical RTM apply to this process as well. The fabric chosen for baseline experiments was CollinsCraft UC1015GV, a unidirectional bonded fabric. This material was chosen because of its identification by previous work [1] as being relatively difficult to RTM because the architecture of the fabric has no in-plane channels to aid in resin flow, causing relatively low in-plane permeability. Also, the binders typically used to hold the material together during handling are styrenesoluble. Therefore, fiber wash tends to occur at lower injection pressures than traditional stitched or woven fabrics. It was used in a [0/45/-45/0]s layup with Owens Corning fabric DB120s used for the +/-45s. Resin Type The resin used for baseline testing was Interplastics Corporation s orthophthalic polyester 63-AX-051. This resin was catalyzed with 1.5 to 2 percent by volume MEKP. Injection Port Design and Flow Rate Since the resin is not injected directly into the fabric, relatively high volumetric flow rates are realized. However, with a relatively small injection port diameter, the higher resin velocity (momentum) was enough to cause fiber wash during initial attempts to reduce injection times. A small tab of bagging film was situated over the fabric immediately adjacent to the injection port to reduce the possibility of velocity induced fiber wash (Figure 3). Further investigations could reveal that extremely short injection times are possible with proper port design. Gel Time Attempts to achieve high injected volume (per port) and longer distances injected from a port resulted in a more traditional RTM flow of resin in the fabric during the second stage. This resulted in a time-critical problem to 2

3 distribute the resin throughout the fabric before it gelled. Mold Gap This is the most unique variable to this process. As shown in Figure 2, the resin pools near the injection port during the injection stage. During early experimentation with a large mold, when the mold gap (distance between the top and bottom mold surfaces, Figure 4.) was larger, more of the resin would remain near the injection port. When the pressure was applied to the bag, resin would have to flow in-plane through more fabric to fill the mold (more like the traditional RTM process governed by Equation 1). However, when this mold gap was reduced, The resin would distribute itself (via channel flow governed by Equation 2) over more of the fabric area during the first stage of injection, reducing the amount of in-plane flow required to properly fill the fabric during the second stage. A more thorough investigation of this variable would be useful in revealing the potential of this process for injecting large volumes and large part surface areas (such as in a wind turbine blade) from a single port. Vacuum During Injection As previously stated, the pressure in both cavities is reduced to near (but above) the vapor pressure of styrene (typically published as 4.5 mm Hg at 20 degrees C). The resin used for these experiments was not degassed, and the pressure during injection was between 75 and 200 mm Hg. Pressure During Stage Two The SCRIMP process relies on reducing the pressure in the fabric which causes a net positive (atmospheric) pressure on the outside of the film. However, since there is a rigid mold top and sealed upper cavity already in place, it is possible and advantageous to apply a positive pressure to the bagging film. A modest pressure would be ideal for keeping mold strength and stiffness requirements at a minimum. Experimentation showed that desirable results were realized with pressures starting at 55 kpa. Pressures as high as 172 kpa were experimented with, but these higher pressures proved to actually reduce performance. Furthermore, mold design considerations for such high pressures would likely be prohibitive. Baseline testing was done with pressures of 70 to 103 kpa. The process and product seem to be relatively insensitive to pressure differences within this range. EXPERIMENTATION Distance From Injection Port The maximum injection distance achieved for the traditional RTM process depends on the fabrics used, specific layup, resin properties, mold wall spacing, injection pressure, and injection time. An experiment was set up to compare injection distance performances of the pressure bag process being investigated to traditional RTM. The layup used was consistent, [0/45/-45/0]s. The resin properties were also consistent throughout this investigation. The zero degree ply fabrics were varied in this experiment. Materials commonly used for wind turbine blades were chosen (Figure 5). A130 is a woven fabric, D155 is stitched, and the UC1015GV is a bonded fabric. An 84 cm x 23 cm flat plate RTM mold was set up with these materials in the prescribed layup. The injection port was located at the center of the mold. The first experiment was conducted with the top plate clamped down directly on top of the fabric. This was to represent a maximum fiber volume percent attainable for these layups with two-sided RTM, the most difficult configuration for resin flow. The second experiment was conducted with the top plate supported at the edges by 2.5 mm shims (creating a controlled mold gap), representing a RTM process designed for slightly lower fiber volume percent but better flow properties. The RTM injection pressure was the same for all of these samples at an average of 138 kpa. The injection time for all plates was 5 minutes. After curing, the maximum injection distances were recorded for all plates. Since flow characteristics are different in the 0 degree direction from the 90 degree direction in the unidirectional fabrics, flow distances were recorded for both of these directions. From this experiment it was seen that the D155 material showed the best permeability in the 0 degree direction, but the worst in the 90 degree direction. The UC1015GV fabric showed similar in-plane resin flow properties in both directions. Investigations of the pressure bag process configured with a 84 cm x 23 cm flat plate mold showed 3 to 5 minute injection times (limited by a small injection port diameter), and no immediately apparent distance limitations. 3

4 To investigate the distance limitations with this process, a 244 cm x 13 cm mold was constructed. The injection port was located at one end of the mold. The same resin, fabrics, layup, etc. were used as in the previous mold. The initial attempts were made with a mold gap of 10.2 mm. It was difficult to get sufficient flow to fill the fabric during the second stage of the process with this gap, so it was reduced by 1.5 mm. This proved to be enough of a reduction of the mold gap to allow the full 244 cm of material to be successfully injected. The resulting plate had a consistent and (relatively) high fiber volume throughout (Figure 6). Results of the distance experiments are tabulated in Tables 1 and 2. Table 1: Maximum Flow Distances (in cm) in The 0 Degree Direction After 5 Minutes at 115 to 140 kpa Injection Pressure. Unidirectional Fabric Clamped Tight 2.5 mm shim A130 (RTM) D155 (RTM) UC1015GV (RTM) UC1015GV (pressure bag process) 244+ Table 2: Maximum Flow Distances (in cm) in The 90 Degree Direction After 5 Minutes at 115 to 140 kpa Injection Pressure. Unidirectional Fabric Clamped Tight 2.5 mm shim A130 (RTM) D155 (RTM) UC1015GV (RTM) Fiber Volume Percent Since the mechanical properties of the matrix materials (resins) are typically poor compared to the fibers, it is usually desirable to maximize the fiber volume percent. However, as stated previously, attempts to increase fiber volume percents in a RTM process tend to degrade flow characteristics, making RTM more difficult. The previous experiment illustrated this. When the hard-sided RTM mold was set up for higher fiber volumes, the maximum flow distances were reduced. The reduced dependency on in-plane resin flow in the process being investigated greatly reduces the effects of this trade-off. From the mold used in the flow distance experiments, samples were produces using both mold configurations; mold halves clamped tight on the fabric and mold halves separated by 2.5mm shims. These samples were measured for fiber volume percent. These fiber volume percents were plotted (against their position along the length of the plate) along with those from a similar mold size with the pressure bag process. The results are shown in Figure 6. The decrease in fiber volume percents near the center of the mold for the RTM samples is likely caused by mold deflection as a result of uneven clamping, and/or deflection caused by fluid pressure on the mold halves during injection. The fiber volume percent distribution along the 244 cm plate was determined and is shown in Figure 7. CONCLUSION The pressure bag injection molding process described in this paper performed well for this investigation. It showed drastic improvements in maximum injection distances for a layup that is relatively difficult to RTM. The flow characteristics in this process governed by channel flow (as compared to Darcy flow for traditional RTM) reduced the limitations caused by in-plane resin flow. This process showed consistent high fiber volumes for all of the samples produced. The positive response to process adjustments suggests that it could be successfully developed for larger, more complex parts required for wind turbine blades. A summary of the impact of varying process parameters is listed in Table 3. Table 3: Qualitative summary of importance of process parameters Relative importance (assuming successful process) Process Parameter For high fiber volume For large injection distance Mold Gap low high Resin Viscosity low high Bag properties medium medium Layup not varied not varied Pressure during injection medium low Pressure during stage 2 low medium Injected volume high medium Injection velocity low low 4

5 The greatest drawback of this process compared to SCRIMP is the necessity of a second mold half. However, this additional mold part does not have the surface finish, stiffness, or tight dimensional tolerance requirements of a second mold half for a traditional RTM mold. Work is ongoing to characterize this process. An investigation into the effects of degassing resin prior to exposure to reduced pressure is scheduled. The effects of varying fabrics and layups will be investigated. Three-dimensional parts will be manufactured with this process. The mechanical properties of final products will be compared to traditional RTM products. REFERENCES [1] Skramstad J.D. (1999), Evaluation of hand lay-up and resin transfer molding in composite wind turbine blade manufacturing, Unpublished master s thesis, Montana State University-Bozeman, Bozeman MT. [2] Darcy, H.P.G., Les Fontaines Bibliques de la Ville de Dijon, Victor Dalmont, France, Paris, [3] N.R.L. Pearce (1999), Improving the resin transfer molding process for fabric-reinforced composites by modification of the fabric architecture, University of Plymouth, Plymouth PL4 8AA, UK. [4] Bird, R.B., Stewart, W.E., and Lightfoot, E.N, Transport Phenomena, John Wiley and Sons, Inc., New York, New York, [5] Welty, J.R., Wicks, C.E., and Wilson, R.E., Fundamentals of Momentum, Heat, and Mass Transfer, Third Edition, John Wiley and Sons Inc., New York, New York, 1984, pp. 128, 711. [6] Douglas S. Cairns, Del R. Humbert, and John F. Mandell, "Resin Transfer Molding of Composite Materials with Oriented Unidirectional Plies", Composites Part A: Applied Science and Manufacturing, Vol 30, [7] SCRIMP is a trademark of TPI Technology, Inc., Warren, RI

6 Flexible Film Top Breather Material Vacuum/Pressure Injection Port/ Open Bottom Fabric Vacuum/Open Figure 1: General mold configuration. Resin Pool During Injection (Channel flow) Figure 2: Mold during injection stage. Tab Port Mold gap Figure 3: Tab inserted to control velocityinduced fiber wash Figure 4: Mold gap definition. 6

7 Figure 5: Typical E-glass materials used in this investigation. 55% 50% 45% Fiber volume 40% 35% 30% 25% 20% 15% A130 RTM clamped A130 RTM w/shim D155 RTM clamped D155 RTM w/shim UC1015GV pressure bag molded UC1015 RTM clamped UC1015 RTM w/shim Relative position along plate Figure 6: Fiber volume percents for samples made during injection distance experiments. Traditional RTM samples (displayed with markers) are from Clamped Tight test (hollow markers) and shimmed test (solid markers). Pressure bag process samples (4 samples) shown without markers. 52% 50% Fiber volume percent 48% 46% 44% 42% Average fiber volume = 46.6%. 40% Distance from injection port, cm Figure 7: Fiber volume percent along a 244 cm plate manufactured with the pressure bag process. 7