Influence of processing variables on quality of unsaturated polyester/E-glass fiber composites manufactured by double-bag progressive compression method

Abstract

A hybrid process of articulated resin transfer molding and vacuum-assisted resin infusion, called double-bag progressive compression method, has been invented to fabricate composite parts. In double-bag progressive compression method, the secondary bag is divided into several segments. During resin infusion, the double bag is drawn upward by vacuum, and the initial flow resistance offered by the loose preform is low. Once the resin infusion is completed, the vacuum on the segmented bags is progressively released to ambient pressure, and the segmental compression is sequentially performed until unnecessary resin is completely removed. This research is to experimentally investigate the influence of double-bag progressive compression method processing parameters, including vacuum pressure in the cavity, number of segments, initiating time of the next compression, temperature of the heated air, initiating segment of the heated air, initial height of the mold cavity, and excess infused resin, on the mechanical property of the part. The design of experiments adopts Taguchi’s method. Results show that the double-bag progressive compression method significantly reduces total filling time and maximally increases the flexural modulus of the part by 17.81% as compared with the typical vacuum-assisted resin infusion. A preferable parameter condition is proposed by taking both the flexural modulus and the operation complexity into account.

Keywords

Vacuum-assisted resin infusion progressive compression double bag Taguchi’s method liquid composite molding

Introduction

Polymer composites have been broadly applied to automobile and aerospace components as a result of their low weight, high mechanical strength, and good resistance to corrosion. Liquid composite molding (LCM) is an important process for manufacturing high-quality composite components. Both resin transfer molding (RTM) and compression resin transfer molding (CRTM) are familiar LCM processes. In the processes, a fiber preform is preplaced within a set of rigid molds. Thermosetting resin is then injected into the cavity where resin saturates the fabrics. Once the filling process is finished, resin cures and binds the fibers into a solid composite. A main difference between RTM and CRTM is that a partial closed mold is employed in CRTM, and the loose preform is thus present. After the completion of resin injection, the mobile mold compacts the preform and drives the resin through the remaining dry preform.

In the last few decades, many researchers have investigated the mechanical performance of polymer composites. They reported that the presence of voids and poor wetting had severe adverse influences on the mechanical properties in terms of longitudinal tensile strength, transverse tensile strength, and inter-laminar shear strength.^1–4 For the sake of reducing the void formation, the common method is applying vacuum during the filling process. Hayward and Harris⁵ found the significant improvements in the void content and in the mechanical performance for RTM components manufactured with vacuum assistance. In addition, the non-uniform resin flow provides the potential for mechanical entrapment of air as a result of the intra-tow permeability being much less than the inter-tow one.^6–10 Kaynak et al.¹¹ reported that the application of vacuum, an intermediate mold temperature and an increase in the initial resin temperature could attain high mechanical properties of RTM components. Wu and Pan¹² analyzed the unidirectional flow and radial flow in injection/compression LCM and conducted a three-point bending test for the parts. They pointed out that injection/compression LCM could not only reduce the injection pressure but also can improve the part quality. Chang et al.¹³ discussed the influence of mold opening distance, injection pressure, compression pressure, preheated mold temperature, resin temperature, and cure temperature on the ultimate strength of CRTM parts and found that using a low compression pressure and high resin temperature were significant methods for improvements in the mechanical properties of the part. Aurrekoetxea et al.¹⁴ investigated the impact behavior of carbon fiber/epoxy composites made by vacuum-assisted CRTM and showed that a 30.25-J penetration energy threshold was deduced from the energy plot.

Another popular LCM process is the vacuum-assisted resin infusion (VARI), also known as the vacuum-assisted resin transfer molding (VARTM), where a flexible vacuum bag replaces one side of the mold in RTM/CRTM. When vacuum is drawn in the cavity, the resin is driven into the compressed preform. Once the preform is totally saturated, the resin inlet is sealed while unnecessary resin is forced to flow out of the cavity through the vacuum vent, called post-filling stage, in order to achieve the uniform part thickness. Some research was available concerning the properties of composites made from polyester resin and glass fiber by VARI.^15–18 Tekalur et al.¹⁵ reported that the E-glass fiber composite had continuous damage progression when subjected to shock blast loading; various failure modes were observed such as permanent deformation, fiber breakage, and delamination. Lee et al.¹⁶ employed Taguchi’s method and an axiomatic design to study the optimizing process for glass fiber–reinforced plastic parts, and the optimal conditions were deduced. Rodriguez et al.¹⁷ compared the characterization of composites based on natural and glass fibers and reported that natural fiber composites, with 30% volume of fiber content, had lower mechanical properties than glass fibers composites. Agarwal et al.¹⁸ investigated the effect of structure on mechanical properties of vinyl ester resins and their glass fiber–reinforced composites and found that with increase in bridge length, mechanical properties decreased significantly; however, with increase in the number of methylene groups in the order 2–8, there was increase in the tensile strength and flexural strength. Shivakumar et al.¹⁹ described the processing of the composites reinforced with carbon and glass fibers and then compared the specific properties of these composites. They reported that the glass composites had higher specific strength but lower specific modulus than marine steel. Cecen and Sarikanat²⁰ investigated the anisotropy of glass fabric and polyester composites and discussed the effects of directions of fiber orientation on the strength of laminates fabricated with chopped strand mat and woven roving fabric. They pointed out that the direction of the fibers played a decisive role in the inter-laminar shear strengths for the case of woven fabric laminates. Kedari et al.²¹ investigated the influences of mold temperature, inlet pressure, and vent pressure on the void formation for polyester/E-glass fiber composites and found that using high mold temperature, high vacuum, and appropriately reduced inlet pressure could produce a VARTM part with high fiber volume fraction and low void content.

In VARI, resin infusion process is relatively long because the maximum infusion pressure is only one atmospheric pressure and the flow resistance offered by compressed preform is high. Some variants of the process have been employed to reduce the filling process. Two common methods are the Seemans’ resin injection molding process, where a distribution medium is placed on the preform as a flow enhancement, and “channel in the core” method in which the channels cut into the core of the sandwich accelerate resin flow.^22,23 Some methods are to create resin distribution channels by vacuum pulling the bag upward such as the fast remotely actuated channels system²⁴, flow flooding chamber method²⁵, vacuum-assisted CRTM,²⁶ progressive compression method,²⁷ and so on. Heider et al.²⁸ developed an infusion methodology where the distribution media were placed on both sides of the preform and the inter-laminar flow media were integrated into the preform. Chang and Li²⁹ placed several air cushions beneath a vacuum bag and then created the resin channels between air cushions when the cavity is evacuated.

This study explores a compression method of resin delivery, called double-bag progressive compression method (DBPCM), which can be regarded as a variation of the articulated RTM.³⁰ This process shares the concept with the fixed rigid mold like RTM, double-bag,²⁴ non-isothermal filling,²⁷ and segmented compression.³⁰ Figure 1 illustrates the resin progression for a case of three-segment compression DBPCM. The secondary bag is divided into two segments and every segment can be controlled individually. During resin infusion, the secondary and primary bags are lifted by vacuum utilized in between the bags and the upper mold. Resin easily flows into the loose fabrics as shown in Figure 1(a). Once the prescribed quantity of resin is infused, the resin inlet is closed. The vacuum on the segmented bags is progressively released to ambient pressure allowing the inflated bags to compact the wetted preform. The sequential inflation of the segmented bag squeezes the surplus resin from the wetted preform and forces the resin through the loose dry preform, as shown in Figure 1(b) and (c). The heated air is inhaled on the primary bag to force resin to completely fill the preform and then remove the unnecessary resin for achieving uniform part thickness, as shown in Figure 1(d). In a practical application, the heated air can be initiated at any segment for adjusting heated resin time and reducing the filling process.

Figure 1.

Schematic diagram of DBPCM: (a) resin infusion, (b) the first segmented compression, (c) the second segmented compression, and (d) the last compression.

In this work, an experimental study of the mechanical properties of unsaturated polyester/E-glass fiber composites fabricated by DBPCM is presented. In spite of a large number of investigations on LCM, the processing parameters are much complicated and interrelated in DBPCM. The major factors, including vacuum pressure in the cavity, number of segments, initiating time of the next compression, temperature of the heated air, initiating timing of the heated air, initial height of the cavity, volume of infused resin, resin/fiber properties, and mold geometry, may affect the qualities of the DBPCM parts. The resin/fiber properties and mold factor are excluded so as to correctly evaluate the effect of chosen processing parameters on the component qualities in this study. The flexural modulus of the part measured by a three-point bending test serves as an indicator of the component quality. The design of experiments adopts Taguchi’s method.³¹ An experiment of typical VARI is also performed for comparison purposes.

Experimental study

Materials and equipment

Resin system used the commercial unsaturated polyester resin (Eternal Chemical, 2597PT-6) including 0.5% of the catalyst. Resin and hardener (MEKPO) were mixed in a weight ratio of 100:1 before resin infusion. The resin viscosity in room temperature was 0.6 Pa s. In the work, the manufacture of composite laminate with high–fiber volume fraction and the comprehension of the resin infiltration into small interstices within fiber bundles were attempted. Thus, the preform was made of four plies of the E-glass woven roving with high areal density (Taiwanglass, TGFW-600), which possessed the equal permeability in the main directions, that is, machine direction and cross direction. The mass of the fabric per unit area was 0.6 kg/m².

The mold was composed of a lower mold, two flexible bags, and an upper mold. The bags were made of polyethylene, while the lower mold and upper mold were made of aluminum AL-5083. A rectangular cavity was cut within the lower mold, and the dimensions were 0.28 m × 0.15 m. The resin inlet and one vent were located on both sides of the mold cavity, respectively.

For the sake of segmental compression, the upper mold and the bags were designed as shown in Figure 2. The boundary of the first segment was determined by volume of infused resin. For three-segment compression, the other boundary was set to the middle position of the remaining cavity. In order to enhance circulation of the heated air, two hot air inlets and one air outlet were constructed by drilling holes about 6 mm in diameter on the upper mold for each segment. The air flow was supplied by air compressor, which was adjusted by pressure regulating valve and flow control valve. The compressor pressure was set to 10 kPa higher than the ambient pressure, and the velocity of the air outflow was about 4.24 m/s by measurement of the flow rate.

Figure 2.

Upper mold and double-bag.

Figure 3 shows the experimental equipment schematically. Once the setup was assembled, the double-bag was lifted by vacuum employed in between the bags and the upper mold. The vacuum was then applied in the cavity and the resin was driven into the preform. The approximately unidirectional resin flow was expected by creating a gap of 0.01 m between the three resin inlets and the preform. Once the predetermined amount of the resin was infused, the resin inlets were sealed, and the vacuum on the segmented bags was progressively released to ambient pressure, allowing the inflated bags to compact the wetted preform. The sequential inflation of the segmented bag was to squeeze the surplus resin from the wetted preform and drive the resin through the loose dry preform. When the last compression was initiated, the primary bag could totally compact the preform and the unnecessary resin was then removed from the cavity.

Figure 3.

Experimental setup.

As VARI, the excess infused resin is essential due to the bag flexibility. In the work, the duration of draining the unnecessary resin was estimated by visually inspecting the draining resin volume in transparent tube connected to between the vacuum vent and the vacuum pot. After the completion of the filling process, the composite component was cured in room temperature for 3–4 h.

Mechanical testing

To investigate the mechanical property of the molded part, three-point bending tests were conducted for both VARI and DBPCM samples. The fiber volume fraction of the parts was different resulting by applying various vacuum pressures and imprecise post-filling stage by visual inspection. Hence, the flexural modulus was chosen as an indicator of the part quality to evaluate the effect of various fabrication conditions. Test strips were cut from flat samples and the dimensions were according to ASTM D790. The locations and dimensions of the test strips are shown in Figure 4.

Figure 4.

Locations and dimensions of the test strips.

Taguchi’s method

In this work, the authors chose seven important processing parameters that may affect the qualities of the DBPCM parts. All variables and their two levels are listed in Table 1. These factor levels defined the interesting regions of the parameters. Notice that factor A adopts the absolute pressure, that is, zero pressure denotes the absolute vacuum. Number of segments includes the primary bag compression for factor B. Two conditions of the preform states are considered for factor F. One is that the loose preform can fully occupy the mold cavity at low level and the resin flows directly through the preform. The other is that a gap exists between the preform and the bag at high level, and the resin flows above and then is pushed through the preform.

Table 1.

Factors and factor levels selected in the matrix experiment.

Factor	Level
Factor	1	2
Vacuum pressure (kPa)	100	90
Number of segments	2	3
Initiating time of nextcompression (s)	8	16
Temperature of heatedair (°C)	25	50
Initiating segment ofheated air	The firstsegment	The lastsegment
Initial cavity height (mm)	5	7.5
Volume of the infusedresin (mL)	90	95

By calculating the degrees of freedom (DF) of all factors, the L₈(2⁷) standard matrix is applicable and given in Table 2. It consists of eight individual experiments corresponding to eight rows.

Table 2.

L₈(2⁷) matrix.

Expt. no.	Column numbers and factor assignment							Flexuralmodulus(dB)
Expt. no.	A	B	C	D	E	F	G	Flexuralmodulus(dB)
1	1	1	1	1	1	1	1	77.09
2	1	1	1	2	2	2	2	77.38
3	1	2	2	1	1	2	2	77.18
4	1	2	2	2	2	1	1	76.89
5	2	1	2	1	2	1	2	73.56
6	2	1	2	2	1	2	1	76.91
7	2	2	1	1	2	2	1	76.42
8	2	2	1	2	1	1	2	75.20

The objective in the composite project is to maximize the flexural modulus of the sample. For a description of the “larger the better” characteristic, a signal-to-noise (S/N) ratio, also denoted by $η$ , can be calculated by the following equation

η_{i} = - 10 \log_{10} (\frac{1}{n} \sum_{i = 1}^{n} \frac{1}{x_{i}^{2}})

(1)

where n is the number of samples and x_i represents the flexural modulus of the sample in experiment i. The S/N ratio can depict the measured quality characteristics deviating from the desired values. The measured flexural modulus is compiled in decibel (dB) according to equation (1).

Analysis of variance (ANOVA) is a collection of statistical models utilized to analyze the differences among group means and their associated procedures in a sample. Taguchi’s method can also conduct the ANOVA to evaluate the relative contribution of each factor to overall response. It can typically utilize the equation as follows

S S_{j} = 4 \sum_{i = 1}^{2} {(m_{j_{i}} - m)}^{2}

(2)

where m is the average of the eight values of $η$ , and m_ji represents the estimation of the effect of factor j at level i, respectively. SS_j is the sum of squares due to factor j.

The goal of maximizing flexural modulus is equivalent to maximizing $η$ . That is to say, the optimum level for a factor is the level that gives the largest value of $η$ in the experimental region. To predict the value of $η$ under the optimum conditions, denoted by $η_{opt}$ , it is calculated as follows

\begin{matrix} η_{opt} = m + (m_{Aopt} - m) + (m_{Bopt} - m) + (m_{Copt} - m) \\ + (m_{Dopt} - m) + (m_{Eopt} - m) + (m_{Fopt} - m) \\ + (m_{Gopt} - m) \end{matrix}

(3)

Finally, verification experiments are conducted with the predicted optimal variable conditions. By comparing the measured value of $η$ with the prediction, the additive model is valid if both values of $η$ are close to each other. On the contrary, if the experimental value is extremely different from the prediction, a strong interaction exists among the parameters.

Results and discussion

In this work, seven important processing parameters were chosen to investigate the influence on the quality of the DBPCM part. For simplification, the material and mold factor were excluded and all rectangular samples were made from four plies of TGFW-600 mats and unsaturated polyester resin. The fiber volume fraction of the sample was 0.45 ± 0.04.

Table 3 lists the average $η$ for each level of the seven factors by taking the numerical values of $η$ in Table 2. Performing ANOVA technique investigates the relative effect of different factors on the mechanical properties. In Table 3, the variance ratio, denoted by R, represents the ratio of the mean square (MS) due to a factor and the error MS. Namely, a large value of R indicates that the influence of that factor is large as compared with the error variance. Hence, for the flexural modulus of the sample, the processing parameters are listed in order of importance: vacuum pressure (factor A), initial height of the cavity (factor F), volume of infused resin (factor G), temperature of the heated air (factor D), initiating segment of heated air (factor E), initiating time of the next compression (factor C), and number of segments (factor B). Referring to the SS column, factor A makes the largest contribution to the total SS, that is, (5.18/11.99) × 100 = 43.23%. Factors F, G, D, and E make 27.61%, 16.58%, 4.74%, and 4.71% contribution to the total SS, while factor B makes the least contribution to the total SS, only 0.59%. It is surprising that the effect of number of segments (factor B) on the flexural modulus is insignificant. A potential explanation for it is that the boundary of the second segment is not precisely set to the interface between wetted and dry preform for three-segment compression. Actually, the accurate position of the second boundary depends on the resin volume squeezed from the first segmental preform, which relates to the compressibility of the wetted preform. It is quite complicated as discussed later. In addition, although the initiating time of the next compression (factor C) has a low impact on the flexural modulus, but has a significant effect on the filling time.²⁷

Table 3.

ANOVA table.

Factor	Average η by factor level (dB)		DF	SS	MS	R
Factor	1	2	DF	SS	MS	R
A	77.13*	75.52	1	5.18	5.18	73.29
B	76.24	76.42*	1	0.07**	0.07
C	76.53*	76.13	1	0.31	0.31	4.32
D	76.06	76.60*	1	0.57	0.57	8.04
E	76.60*	76.06	1	0.57	0.57	7.99
F	75.69	76.97*	1	3.31	3.31	46.81
G	76.83*	75.83	1	1.99	1.99	28.11
Error			0	0	-
Total			7	11.99
(Error)			(1)	(0.07)	(0.07)

DF: degree of freedom; SS: sum of squares; MS: mean square; R: variance ratio.

Is the optimum level; **Indicates sum of squares added to estimate the pooled error sum of squares indicated by parentheses.

According to the “larger the better” characteristic, the predicted optimum settings are A₁B₂C₁D₂E₁F₂G₁ by analyzing the S/N values in Table 3. Namely, the optimum conditions are a high vacuum pressure, three-segment compression, premature initiation of the next compression, hot air initiated at the first segment, high thickness of the initial cavity, and low excess resin. This is because a high vacuum reduces the entrapped air as discussed in previously published papers,^6–9 and the multi-segment compression is expected to be helpful for removal of the voids.³⁰ The premature initiation of the next compression and hot air initiated at the first segment can increase the heated resin time and resin viscosity is thus reduced. The loose dry preform reduces the flow resistance when a high thickness of the initial cavity is used.¹³ These conditions are helpful for the resin penetration into the interstices of the preform. Excess infused resin degrades the composite performance because resin may be incompletely degassed.

After the optimum parameter conditions were determined, the predicted response under these conditions could be calculated by equation (3). Then, authors conducted verification experiments using the optimum conditions and compared the measured value of η with the prediction. The predicted $η_{opt}$ is found to be 13.69% larger than the experimental one. Both values are somewhat different and thus the factor interactions exist. In fact, several factors are interrelated in the work. For example, the initiating time of the next compression (factor C) relates to the wetted preform compressibility, which is dependent on compaction force (vacuum pressure), resin viscosity, and loose state of the preform. This is to say, factors A, C, D, E, and F are interrelated. Coincidently, the effect of factors A, D, and E on the flexural modulus of the part is small for current levels of the factors, and thus the experimental value is not drastically different from the prediction.

The DBPCM part under the optimum conditions has the flexural modulus of 7.84 GPa and total filling time of 444 s, including post-filling stage; contrarily, that under the worse settings has the flexural modulus of 4.77 GPa and total filling time of 887 s. The typical VARI part has the flexural modulus of 6.66 GPa and total filling time of 1238 s. Although factor interactions exist, the DBPCM under the optimum conditions increases the flexural modulus of the part by 17.81% and reduces at least half of total filling time as compared with the typical VARI. In addition, the surface finish of DBPCM part is almost identical to that of VARI by human visual observation. However, the operation of the three-segment compression in DBPCM is complicated, and the boundary of the second segment is not easy to be precisely determined. The quality loss function can be used to make the necessary trade-offs. By changing from B₂ to B₁, $η$ is decreased by 76.24–76.42 = –0.18 dB, and this is equivalent to merely 2.14% reduction in flexural modulus. This is because factor B has the least effect on flexural modulus in this study. Thus, a preferable setting can be A₁B₁C₁D₂E₁F₂G₁ in DBPCM by taking both the flexural modulus and the operation complexity into account. It was verified by experiments.

The surface of composite section was observed by scanning electron microscope (SEM). Figure 5 shows the SEM images of the fracture sample in machine direction and cross direction. No distinct voids were found in the images but the debonding between the fibers and matrix was observed in cross direction maybe due to poor interfacial adhesion.

Figure 5.

Images of the fracture sample in (a) machine direction and (b) cross direction.

Conclusion

The purpose of this article was to investigate the influence of seven processing parameters on the mechanical performance of the DBPCM part by applying Taguchi’s method. These factors included vacuum pressure, number of segments, initiating time of the next compression, temperature of the heated air, initiating segment of the heated air, initial height of the mold cavity, and volume of infused resin. Experimental results show that the optimum settings are A₁B₂C₁D₂E₁F₂G₁ for the flexural modulus of the part and the factor interactions exist. Even so, the optimal DBPCM increases the flexural modulus by 17.81% and reduces over half of total filling time as compared with VARI. The preferable DBPCM settings are A₁B₁C₁D₂E₁F₂G₁ by taking both the flexural modulus and the operation complexity into account. Through observation of the surface of the composite section, no distinct voids are found but the debonding between the fibers and matrix are observed.

Although the functionality of the DBPCM is verified, the investment cost of the DBPCM mold is expensive particularly for a complex mold. In the work, the investment cost in DBPCM is roughly three times more than that in VARI.

Footnotes

Handling Editor: Daxu Zhang

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The study was performed by the financial support of the Ministry of Science and Technology of Republic of China under grant [MOST 105-2221-E-244-006].

ORCID iD

Chih-Yuan Chang

References

Judd

NCW

Wright

. Voids and their effects on the mechanical properties of composites. SAMPE J 1978; 14: 10–14.

Yosida

Ogasa

Hayashi

Statistical approach to the relationship between ILSS and void content of CFRP. Compos Sci Technol 1986; 25: 3–18.

Bowles

Frimpong

Void effects on the interlaminar shear strength of unidirectional graphite-fiber-reinforced composites. J Compos Mater 1992; 26: 1487–1509.

Patel

Rohatgi

Lee

LJ.

Influence of processing and material variables on resin-fiber interface in liquid composite molding. Polym Compos 1993; 14: 161–172.

Hayward

Harris

Effect of process variables on the quality of RTM mouldings. SAMPE J 1990; 26: 39–46.

Pillai

Modeling the unsaturated flow in liquid composite molding processes: a review and some thoughts. J Compos Mater 2004; 38: 2097–2118.

Kang

Lee

Hahn

HT.

Formation of microvoids during resin-transfer molding process. Compos Sci Technol 2000; 60: 2427–2434.

Rohatgi

Patel

Lee

LJ.

Experimental investigation of flow-induced microvoids during impregnation of unidirectional stitched fiberglass mat. Polym Compos 1996; 17: 161–170.

Chen

Macosko

Davis

. Wetting of fiber mats for composites manufacturing: II. Air entrapment model. AIChE J 1995; 41: 2274–2281.

10.

Chang

Hourng

LW.

Study on void formation in resin transfer molding. Polym Eng Sci 1998; 38: 809–818.

11.

Kaynak

Akgul

Isitman

NA.

Effects of RTM mold temperature and vacuum on the mechanical properties of epoxy/glass fiber composite plates. J Compos Mater 2008; 42: 1505–1521.

12.

Pan

YR.

The study of injection/compression liquid composite molding. J Mater Process Technol 2008; 201: 695–700.

13.

Chang

Hourng

Chou

TY.

Effect of process variables on the quality of compression resin transfer molding. J Reinf Plast Comp 2006; 25: 1027–1037.

14.

Aurrekoetxea

Agirregomezkorta

Aretxaga

et al . Impact behavior of carbon fiber/epoxy composite manufactured by vacuum-assisted compression resin transfer molding. J Compos Mater 2011; 46: 43–49.

15.

Tekalur

Shivakumar

Shukla

Mechanical behavior and damage evolution in E-glass vinyl ester and carbon composites subjected to static and blast loads. Composites Part B 2008; 39: 57–65.

16.

Lee

Jeon

Cha

et al . A study on optimizing the mechanical properties of glass fiber-reinforced polypropylene for automotive parts. Polym-Plast Technol Eng 2011; 50: 95–101.

17.

Rodriguez

Petrucci

Puglia

et al . Characterization of composites based on natural and glass fibers obtained by vacuum infusion. J Compos Mater 2005; 39: 265–282.

18.

Agarwal

Singh

Varma

et al . Effect of structure on mechanical properties of vinyl ester resins and their glass fiber-reinforced composites. J Appl Polym Sci 2008; 108: 1942–1948.

19.

Shivakumar

Swaminathan

Sharpe

Carbon/vinyl ester composites for enhanced performance in marine applications. J Reinf Plast Comp 2006; 25: 1101–1116.

20.

Cecen

Sarikanat

Experimental characterization of traditional composites manufactured by vacuum-assisted resin-transfer molding. J Appl Polym Sci 2008; 107: 1822–1830.

21.

Kedari

Farah

Hsiao

KT.

Effects of vacuum pressure, inlet pressure, and mold temperature on the void content, volume fraction of polyester/E-glass fiber composites manufactured with VARTM process. J Compos Mater 2011; 45: 2727–2742.

22.

Sun

Lee

LJ.

Mold filling analysis in vacuum-assisted resin transfer molding. Part I: SCRIMP based on a high-permeable medium. Polym Compos 1998; 19: 807–817.

23.

Dai

Pellaton

Hahn

HT.

A comparative study of vacuum-assisted resin transfer molding (VARTM) for sandwich panels. Polym Compos 2003; 24: 672–685.

24.

Allende

Mohan

Walsh

SM.

Experimental and numerical analysis of flow behavior in the FASTRAC liquid composite manufacturing process. Polym Compos 2004; 25: 384–396.

25.

Alms

Advani

SG.

Simulation and experimental validation of flow flooding chamber method of resin delivery in liquid composite molding. Composites Part A 2007; 38: 2131–2141.

26.

Chang

CY.

Simulation of non-isothermal filling process in vacuum assisted compression resin transfer molding process. Plast Rubber Compos 2015; 44: 162–170.

27.

Chang

CY.

Numerical study on filling process in progressive compression method. Adv Compos Mater. Epub ahead of print 24 November 2017. DOI: 10.1080/09243046.2017.1405598

28.

Heider

Simacek

Dominauskas

et al . Infusion design methodology for thick-section, low-permeability preforms using inter-laminar flow media. Composites Part A 2007; 38: 525–534.

29.

Chang

SY.

Effect of process parameters on the surface quality of air cushion method. J Reinf Plast Compos 2015; 34: 257–266.

30.

Choi

Dharan

CKH

. Mold fill time and void reduction in resin transfer molding achieved by articulated tooling. J Compos Mater 2002; 36: 2267–2285.

31.

Madhav

. Quality engineering using robust design. Upper Saddle River, NJ: Prentice Hall.