ANALYSIS OF SEEMANN COMPOSITE RESIN INFUSION MOLDING PROCESS

(SCRIMP) Xiudong Sun, Ling Li , Shoujie Li, Jun Ni and L. James Lee Department of Chemical Engineering The Ohio State University, Columbus, OH 43210 Abstract
The objective of this paper is to analyze mold filling and curing in the Seemann Composite Resin Infusion Molding Process (SCRIMP). Mold filling experiments were carried out to explore the influence of various molding conditions on filling pattern and filling time. Computation models were developed to predict the flow pattern and filling time. In the cure study, a kinetic model based on the free radical polymerization mechanism was developed for simulating the reaction kinetics of a vinyl ester resin. A heat transfer model combined with the kinetic model was solved to simulate the cure behavior in SCRIMP. The effects of room temperature and mold materials on the curing process were discussed.

Experimental
Mold Filling
In the first set of SCRIMP mold filling experiments, the fiber preform (QM6408 from Brunswick Technologies, Inc. with permeability of Kx = Ky = 0.6e-10 m2 and Kz = 0.12e-10 m2, and thickness of 0.2 cm per layer,) was placed on a glass platform. A layer of high-permeable medium was placed under the vacuum bag and a peel ply was laid between the permeable medium and the upper surface of the fiber preform. The length of the fiber preform was 64.77 cm and the width was 11.43 cm. Three fluids were used in the experiment: DOP oil (viscosity 43cp at 24.4
C), extra heavy Mobil oil (viscosity 320cp at 26.2
C) and Mobil BB oil (viscosity 530cp at 24.5
C). Since both the top vacuum bag and the bottom mold were transparent, with the aid of two tilted mirrors the resin infusion patterns could be recorded by a CCD camera. In the second set of SCRIMP mold filling experiments, a grooved acrylic plate (3 grooves of 0.3175 cm x 0.3175 cm size, 2.54 cm and 5.08 cm spacing) was used as the 'core', and placed on the glass platform. The inlet channel perpendicularly linked to all of the grooves was connected to the fluid tank. The fiber preform of the same size as that in the first set of experiments was wrapped around the acrylic plate and covered with a large vacuum bag. The outlet channel connected to the vacuum pump was 6.35 cm away from the end of the grooves.

Introduction
SCRIMP [1-2] is a new and promising resin infusion technology that meets increasingly stringent environment regulations and is capable of delivering large size composite parts at low cost. In SCRIMP, the fiber preform is placed on top of the mold and covered with a plastic vacuum bag. The vacuum bag is sealed tightly around the mold periphery. The air in the mold cavity is pulled out and the resin is infused into the dry fiber preform by vacuum. The vacuum pump is kept running until the resin in the fiber preform gels. Generally, there are two types of resin distribution systems to facilitate the resin infusion as shown in Figure 1. One involves a highly permeable medium placed between the fiber preform and the mold surface, while the other involves grooves cut in the core or the mold. Although SCRIMP has already been used in industry for small quantity productions, the process has not been thoroughly analyzed. Since SCRIMP applications are often for very large composite parts like sail boats, refrigerated cargo boxes and bridge decks, the raw materials used are in large quantity and expensive, and the molding process is often time consuming. To make SCRIMP more economically attractive, the process needs to be optimized and the product rejection rate has to be kept very low. This requires a thorough understanding of both mold filling and curing, and the development of efficient computation tools for process design and material selection. Through both experimental and numerical analyses, this paper will address the resin flow and curing in the SCRIMP process.

Curing
The resin used in the mold curing is DERAKANE 411350 vinyl ester resin from Dow Chemical. Cumene hydroperoxide (CHP) in a solution form (Akzo Nobel Chemical) was used as initiator and a solution of 6.0 wt.% cobalt naphthenate (Huls America) was used as promoter. The reaction kinetics and the heat of reaction were determined using a differential scanning calorimeter (DSC, Model 2910, TA Instruments). The reactions were conducted in volatile aluminum sample pans which may stand 2 atmosphere internal pressure after sealing. Iso-thermal reactions were measured at different temperatures with various promoter contents, and two scanning runs were carried out successively from room temperature to 300
C with a heating rate of 5
C/min to determine the residual heat and the baseline. Since DSC can only measure the overall reaction exotherm, a Fourier transform infrared (FTIR) spectroscopy (Nicolet, Magna-IR 550) was applied to

differentiate overlapped multiple reactions of styrene vinyl and vinyl ester vinylene groups. The schematic of the molding set-up is shown in Figure 2. Under the vacuum bag was a 7.62 cm thick foam core with precut grooves. The fiber mats (QM6808) were placed between the composite mold and the foam core. Thermocouples were placed at the surface of each layer of fiber mat to follow the temperature change during curing. Two runs of experiments with different groove design and layers of fiber mats were carried out. In run 1, the groove size was 0.3175 cm x 0.3175 cm, the groove spacing was 2.54 cm and three layers of fiber mats were used. In run 2, the groove size was 0.635 cm x 0.635 cm, the groove spacing was 3.81 cm, and five layers of fiber mats were used.

time. A simple leakage flow model was proposed to solve this problem. The solution procedure of the leakage flow model is as follows: 1). Describe the flow in the grooves as one-dimensional flow through a porous medium, whose equivalent permeability is obtained using a corresponding close-form solution [4]. 2). Divide the fiber area between two neighboring grooves into rectangular blocks. Each block is treated as a sink, receiving the fluid diffused from the groove segments. 3). Calculate the pressure distribution in each groove and the leakage flow from the groove to the neighboring fiber sinks by using an iteration method. Update the flow front positions in the grooves and in the sinks at every time interval. The leakage flow model was first compared with the conventional CV/FEM method using a simple case. A good agreement was found between the two models, and the computation time of the former was 20 times less than that of the latter. The leakage flow model was then used to simulate SCRIMP, and the simulation results were compared with the actual filling pattern taken from the experiment. Figure 4(b) shows the simulated flow patterns obtained from the leakage flow model for SCRIMP with three grooves and 4 layers of fiber mats. The prediction of the flow pattern near the bottom side (i.e. the side with grooves) is very good, while the prediction of the flow pattern near the top side is reasonable.

Results and Discussion

Mold Filling
Figure 3 shows the typical flow patterns obtained from the SCRIMP experiments based on a high-permeable medium. The flow patterns on both the bottom and the top surfaces of the fiber preform are like plug flow, which are in agreement with Darcys law. There is an apparent leadlag distance between the top and bottom flow fronts because of the difference in permeability, which leads to the transverse flow in the thickness direction. It was observed from experiments that except a little decrease at the beginning, this lead-lag distance remained nearly constant during the filling process. Using Control Volume/Finite Element Method (CV/FEM) [3], the continuity equation and Darcys equation governing SCRIMP mold filling could be solved and the filling time could be predicted. Table 1 summarizes the comparison of experimental and simulated mold filling time under various molding conditions. Figure 4(a) shows the typical flow patterns obtained from SCRIMP experiments based on grooves. The filling time for the one-layer fiber preform was very close to that for the two-layer fiber preform, but much shorter than that for thicker fiber preforms. Snapshots of the resin infusion process on both the top and the bottom surfaces showed that the grooves were filled first, then the resin infused from the filled grooves into the fiber preform. For onelayer fiber mat, the flow fronts on the top and the bottom surfaces were almost identical. For four to ten-layer fiber mats, when the bottom surface was almost filled, the top surface was still dry. There were some transverse flows in the thickness direction, and the filling became thickness dependent. For groove type of SCRIMP, the number of grooves can be very large for large parts. It is impractical to use CV/FEM or other conventional numerical methods for process simulation because of unreasonable computation

Curing
Kinetic Characterization To obtain the needed kinetic information, DSC isothermal runs were carried out at various temperatures with three different weight ratios of promoter to initiator, 1/10, 1/5 and 1/3. Figure 5 shows the experimental results of the isothermal reaction rate profiles of the vinyl ester resin with the promoter to initiator weight ratio of 1/5. The average heat of reaction measured in the isothermal mode is 1475.5 J/g. Figure 6 shows the FTIR measurements of the reaction rate profiles of styrene vinyl and vinyl ester vinylene groups. The actual reaction mechanism of styrene-vinyl ester system may be complicated. In this study, we assume that the reaction follows the typical free radical co-polymerization mechanism. Based on the diffusion limitation model [5-6], the reaction rates of styrene vinyl and vinyl ester vinylene groups are developed as follows: Styrene vinyls:

d dt

1 
   

1  1




1



2 3 

2 1 / 2








2 3
(1)

Vinyl ester vinylenes:

d dt

2 
   

1  1

'
'
' 1

3   2 

2 1 / 2





'
4


'
' 2 3
(2)

method. Figure 7 shows the comparison of the simulated temperature distribution and the experimental results. As can be seen, the prediction fits the experimental results reasonably well. In SCRIMP, resin curing occurs at room temperature. Hence, the only heat source is the reaction exotherm of the reactive, room-temperature curable resins. Ideally, the reaction exotherm should be stored in the curing composite so that a higher temperature can be reached during curing, which would in turn accelerate the reaction, produce more heat and drive the curing reaction to completion. However, this ideal situation can not always be achieved because of the heat dissipation into the mold and heat loss to the surroundings. In winter when the room temperature is low, such heat loss may result in a very long curing time. Using cure simulation, we investigated various factors affecting the SCRIMP curing, such as room temperature and mold materials. Figures 8 and 9 show the effect of mold materials on the temperature and conversion profiles inside the composite at different room temperatures respectively. For this vinyl ester resin, the gelation occurs at the conversion of 10%. If we assume that demolding can be proceeded when 70% conversion is achieved, Figure 10 reveals the relationship between the gel time and the demold time at various room temperatures for the composite, aluminum and steel molds. It can be seen that at the same room temperature, the demold time is shorter for the composite mold than for the steel mold, and the difference becomes significant at lower room temperatures. The gel time is affected by room temperature, but is not sensitive to mold materials.

where 1, 2, 3, and 4 are lumped-parameters accounting for kinetic and diffusion effects on propagation and termination. For a redox type of initiation system such as cobalt naphthenate and cumene hydroperoxide, the promoter concentration changes little during reaction. According to Hiatt et al. [7], the decomposition rate of initiator can be expressed as follows:

dIi
 k d I p1/2 I i dt

(3)

The kinetic parameters in Eqs. (1)-(3) are estimated from isothermal FTIR measurements by using a dynamic nonlinear search program. Figures 5 and 6 show the comparison of the model results with the DSC and FTIR measurements. Mold Curing SCRIMP molding experiments were carried out at room temperature. Figure 7 shows the measured temperature profiles inside the composite during curing. It can be seen that the curing reaction occurred from the very beginning of molding, and the temperature increased because of reaction exotherm. Depending on the groove design and thickness of fiber mats, the highest temperature, which was in the resin rich area, i.e. grooves, varied. In the thickness direction, the lowest temperature was observed at the mold surface, which means the reaction would be slower there. In practical SCRIMP curing, which starts after the mold has been filled, only cure reaction and heat transfer need to be considered in the analysis. The basic equation of heat transfer is written as follows:

T 2T 2T C p
k( 2  2 )  G t x y

ACKNOWLEDGEMENT
The authors would like to thank Hardcore DuPont and Dow Chemical for financial support and material donation.

REFERENCES
1. 2. 3. 4. 5. 6. 7. W.H. Seemann, U.S. Patent 4,902,215, Feb. 20 (1990). W.H. Seemann, U.S. Patent 5,316,462, May 31 (1994). W. B. Young, K. Rupel, K. Han, L. J. Lee, and M. J. Liou, Polymer Composotes, 12, 30 (1991) J. Ni, Y. Zhao, L. James Lee and S. Nakamura, Polymer Composites, 15, 134(1997). Y. J. Huang and L. J. Lee, AICHE J., 31, 1585 (1985) Y.J. Huang and L. J. Lee, Polym. Eng. Sci., 30, 687 (1990) R. Hiatt, K.C. Irwin, and C. W. Gould, J. Org. Chem. 33, 1431 (1968)

(4)

Where

is density, Cp heat capacity, k thermal

conductivity and G the heat source term related to reaction exotherm. Since heat transfer occurs in the foam core, the grooves, the fiber preform and the mold wall, and these areas have different thermal properties, separate governing equations are needed to describe the heat transfer in each area. These equations coupled with the kinetic model of the vinyl ester resin were solved by a 2-D finite difference program using the alternating implicit

Key words: vinyl ester resins, SCRIMP, mold filling

and curing

vacuum bag high-permeable medium fiber core fiber mold

Groove spacing

(a)

Vacuum bag Groove Foam core Mold

vacuum bag fiber

(b)

core fiber groove mold

(Middle layer)

T2

T1 (Top layer) T3 (Bottom layer)

Figure 1.Resin distribution systems used in SCRIMP. (a) high-permeable medium, (b) grooves cut in the core

Figure 2. Schematic of cure experiment

top bottom (a) (b)

Figure 3. Flow patterns obtained from SCRIMP based on high-permeable medium, with three layers of fiber mat and fluid of viscosity 270 cp. (a) t=67 s, (b) t=463 s

(a)

top bottom t=2.0 s t=39.0 s

(b)

t=1.9 s

t=40.9 s

Figure 4. Comparison of flow patterns obtained from experiment and the leakage flow model for SCRIMP based on grooves, with four layers of fiber mat and fluid of viscosity of 270 cp. (a) experiment, (b) leakage flow model.

Fiber mat layers 2 2 2 3

Oil type Extra heavy DOP Mobil BB Extra heavy

Viscosity (cp) 325 40 425 270

Filling time (sec) Experiment 807 133 1247 960

Filling time (sec) Simulation 863.4 105.9 1124.5 920.2

Table 1. The molding condition and filling time of SCRIMP based on high-permeable medium.

4.E-04

60

Simu latio n 25 C
temp erature (C) 50 40



reaction rate (1/s )

3.E-04

30 C 2.E-04 35 C 45 C 1.E-04

30 20
co m p o sit e



10

st eel alum inum

0.E+00 0 30 60 90 time (min ) 120 150

0 0 30 60 90 120 time (min) 150 180

Figure 5. Reaction rate vs. time at different temperatures(1.5% wt. CHP, Co+/CHP=1/10)
8.E-04
Sim ulat ion

Figure 8. The effect of mold materials on the temperature profile of the composite (fiber thickness =0.2 cm, i.e. 1 layer, mold thickness=0.635 cm)
0.9 0.8 0.7 co nv ers io n 0.6 0.5 0.4 0.3 0.2 0.1 0
comp osit e
 


reactio n rate (1/s )

6.E-04

St , Co+/CHP =1/5 VE, Co +/CHP =1/5

4.E-04

2.E-04

steel aluminum

0.E+00 0 30 60 time (min ) 90 120

30

60

90 120 time (min )

150

180

Figure 6. Reaction rate profiles of vinyl ester resins at 45C measured by FTIR(1.5% wt. CHP, Co+/CHP=1/5)
100 90 80 temp erature (C) time (min) 70 60 50 40 30 20 10 0 0 30 60 time (min)

Figure 9. The effect of mold materials on the conversion profile of the composite (fiber thickness =0.2 cm, i.e. 1 layer, mold thickness=0.635 cm)
200 180 160 140 120 100 80 60 40 20 0

  


comp osite steel aluminum

T 1-s im u la t io n T2 T3 T 1-e xp e rim e n t T2 T3

90

120

10

20 30 temp erature (C)

40

Figure 7. Comparison of experimental and simulated temperature profiles inside the composite

Figure 10. Demold time and gel time as a function of room temperature for composite, steel and aluminum mold