Research on Forming Technology of Large-Scale Composite Special-Shaped Components based on Vacuum Infusion Process

 

In the production process of composite materials, mold manufacturing and product processing account for a high proportion of the cost. SCRIMP (Seamann Composites Resin Infusion Mnufacturing Process) has the advantages of low cost, environmental friendliness and applicability to large components.

 

In the SCRIMP process, ensuring that the matrix resin fully impregnates the reinforcing material is the key to ensuring product quality, and the layout of the guide channel is the key factor to ensure the progress and quality of the resin impregnation of the fiber reinforcement material.

 

The necessary condition for SCRIMP molded parts to avoid quality problems such as white spots, delamination, and uneven thickness is that the resin fully impregnates the fiber preform. In practical applications, parts gradually develop towards more complex structures and larger sizes, which makes the flow channel design difficult, the resin flow is not easy to control, and the resin has poor impregnation of the fiber.

 

The design of the resin injection flow channel is a key link in the SCRIMP process. Reasonable injection flow channel design can not only shorten the penetration time of the resin in the preform, but also avoid the formation of defects such as dry spots and rich resin during the penetration process.

 

Traditional resin flow channel design is mainly based on engineering experience, supplemented by trial and error. However, when actually manufacturing parts, the complexity of the flow state and flow trajectory of the resin in the complex structure prefabricated body is difficult to accurately grasp only by engineering experiments, and it also requires a lot of manpower and material resources.

 

The operation time of the resin is an important parameter that needs to be considered in the flow channel design. Before the flow channel design, the resin system should be subjected to a gel test to determine the gel time to avoid invalid flow channel design. Permeability is a very critical parameter for surface filling simulation, which is generally obtained through experiments using a permeability test device.

 

The molding of large components with a teardrop-shaped cross-section must ensure that the glue flows quickly on the fabric surface while ensuring that the glue fully penetrates between the fabric layers.

 

Based on Darcy’s law, this paper uses an experimental method to measure the permeability of SCRIMP process glass fiber, establishes a finite element model of a typical component with a teardrop-shaped cross-section through simulation analysis software, and analyzes and compares different flow channel design schemes.

 

 

 

Figure 1. Schematic Diagram of Typical Component Structure

 

Analysis Scheme

 

Different from known large thin shell structures (such as wind turbine blades), wind turbine blades are willow-shaped open structures. When forming, the upper and lower parts are formed separately and then bonded. The typical component is a continuous closed structure, as shown in Figure 1.

 

Post-processing and bonding are not allowed. Currently, there is no literature reporting a molding method for similar structures. When designing the flow channel, consider that the surface layer is a symmetrical structure, and the injection port is symmetrically set for the upper and lower layers. Three schemes are analyzed, as shown in Figure 2.

 

Scheme 1: The injection port is set at the center of the upper and lower layers, and the outlet port is set at the rudder tip corner;

 

Scheme 2: A line injection flow channel is set along the edge (green dotted line), and a line outlet flow channel is set at the guide edge (blue dotted line);

 

Scheme 3: A line injection flow channel is set in a ring along the center line of the axial direction (blue dotted line), and a line outlet flow channel (green dotted line) is set at the guide edge and the edge.

 

 

Figure 2. Infusion scheme design

 

Finite Element Model and Parameter Setting

 

The finite element model was established using 3D modeling software. Since the surface layer is a thin shell structure, the mesh is divided into shell units, with a total of 3445 units, as shown in Figure 3.

 

 

Figure 3. Finite Element Model of a Typical Component

 

During the skin molding process, the effect of gravity on the flow of the glue is considered. The setting of gravity acceleration should be consistent with the actual mold filling process and set to the X+ direction in the software. It is also necessary to set the material parameters and process parameters of the resin flow area, as shown in Table 1.

 

 

Table 1. Material Properties and Process Parameters

 

Calculation Results and Analysis

 

The resin flow process of the upper and lower layers at the center point is shown in Figure 4. Under the action of vacuum pressure, the resin continuously pushes around with the injection point as the center. The diffusion boundary is elliptical. After the front and rear ends merge, it gradually fills the surface layer and surrounds and intersects at the tip. It takes about 360 minutes.

 

The filling process of injection from the guide edge is shown in Figure 5. The front of the glue flow converges at the edge, which takes about 240 minutes.

 

The filling time is shown in Table 2.

Figure 4. Surface Layer Center Injection Molding Process

 

Figure 5. Single-side injection molding process

 

Table 2. Injection Process Simulation Results

 

The typical component is a closed structure. The upper and lower layer injection channels should be set synchronously to ensure the consistency of the flow of the glue front. In Scheme 3, the injection pipe is set in a ring along the rudder axis. The surface layer centerline injection is shown in Figure 6. The center injection channel is used as the starting edge and gradually flows to both sides, intersecting at the guide edge and the trailing edge. It takes about 60 minutes.

 

Figure 6. Surface Layer Centerline Injection Molding Process

 

The product infusion process should not exceed the applicable period of the resin system to avoid the appearance of gel in the resin flow area during the infusion process, which will lead to the phenomenon of impregnation, resulting in white spots on the product surface, lack of glue and other problems. The resin gel test shows that the molding process should be controlled within 50 minutes.

 

As shown in Table 2, Scheme 3 is closest to the time control requirements of the product molding process. Therefore, Scheme 3 is further optimized. On the basis of Scheme 3, in order to further shorten the injection time, a relay injection scheme is designed,

 

Scheme 3-1: 3 line injection channels are set in a circle along the centerline position of the rudder axis direction (yellow and green dotted line positions), and line outlet channels are set at the guide edge and the edge position (blue dotted line position);

 

Scheme 3-2: 5 line injection channels are set in a circle along the centerline position of the rudder axis direction (yellow, blue and green dotted line positions), and line outlet channels are set at the guide edge and the edge position (red dotted line position).

 

That is, multiple annular injection channels are arranged along the rudder axis in the resin flow area, and the injection points are opened/closed in sequence during mold filling to absorb the glue, so as to achieve the purpose of rapid flow of the glue to penetrate the fabric. The specific scheme of the channel arrangement is shown in Figure 7.

 

Figure 7. Runner Arrangement Scheme

 

Relay injection has strict requirements on the opening timing of the injection port. Opening the next injection port too early will cause local gas envelope and form defects. In scheme 3-1, the runner spacing is set to 700mm, and the runners arranged from the central runner to the two sides are the 2nd runner and the 3rd runner.

 

A sensor is set outside the 2nd runner, and the trigger condition is to send an open/close signal when the filling rate reaches 100%. Using sensor 1 (located outside runner 2) for simulation, the opening and closing process of the injection port is shown in Figure 8. There is an obvious resin flow interference area near the runner. After the mold filling is completed, dry spots are easily formed on the surface layer at this location, as shown in Figure 9.

 

Figure 8. Mold Filling Process Flow Chart

 

 

Figure 9. Schematic Diagram of Local Air Encapsulation in the Surface Layer

 

In order to avoid interference, the timing of opening the injection port is optimized. According to the layer thickness and process test results, the position of the glue flow front is used to confirm the opening timing of the injection port. The mold filling simulation is performed using sensor 2 (located outside flow channel 2).

 

The product mold filling process is shown in Figures 10 and 11. The glue flow front remains basically consistent. There is no turbulence at the time node when the injection port is opened sequentially, and the injection time is about 2320s.

 

 

Sequential filling process (single-sided, 3-gate configuration)

 

Figure 11. Filling process for the relay configuration (single-sided, 5-gate)

 

Figure 11. Relay (Single-sided 5 Injection Ports) Mold Filling Process

 

After the glue flows for a certain distance, its penetration speed will be significantly slowed down due to the increase in resistance.

 

In order to open the next injection port in time before the penetration rate slows down, the flow channel spacing is shortened to 400mm in Scheme 3-2, and the corresponding flow channels arranged from the central flow channel to both sides are 2nd level flow channel, 3rd level flow channel, and 4th level flow channel.

 

The mold filling simulation is carried out using sensor 2 in Scheme 3-1. The results show that the mold filling time is further shortened without affecting the wetting effect of the product, as shown in Table 3.

 

Table 3. Relay Injection Simulation Results

 

Through simulation analysis of different flow channel setting schemes, it is found that the setting of the annular flow channel can ensure that the flow path of the resin glue on the surface and inside of the fabric is controllable, and the front is basically consistent, which is conducive to parameter adjustment of different schemes.

 

The multi-threaded continuous injection scheme can effectively reduce the mold filling time. The mold filling time of scheme 3-2 is reduced by more than 50% compared with scheme 3, ensuring that the mold filling is completed within the applicable period of the resin.

 

SCRIMP Process Verification of Typical Component Prototype

 

The flow channel design in this paper is used to make a 3m×4m typical component prototype. The product molding process is shown in Figure 12. The molding and filling time of the prototype is about 28min, which is about 10% different from the simulation result.

 

The prototype is tested for external water pressure, fatigue, torsion resistance, etc. The product has no damage or leakage. The surface quality of the product after molding is shown in Figure 13.

 

Figure 12. Product Molding Process

 

Figure 13. Surface Quality of the Product after Molding

 

Conclusion

 

In this paper, the finite element method is used to simulate the filling process of a typical large teardrop-shaped cross-section component. The filling situation in the glue flow area can be predicted, and the feasibility of the flow channel design can be analyzed, which has guiding significance for the molding of this type of component.

 

The simulation results show that the multi-threaded relay injection scheme can effectively shorten the product filling time and meet the needs of engineering production; the timing of opening the injection port is designed according to the process test results. Early opening can easily lead to product defects.

 

A 1:1 typical component prototype was made using the multi-threaded relay injection scheme. The theoretical calculation of the filling time is about 10% different from the actual time, and the scheme is feasible.