EXPERIMENTAL INVESTIGATION OF FIBER SIZING-TEST FLUID INTERACTION FOR IN-PLANE PERMEABILITY MEASUREMENTS OF CONTINUOUS FIBERS.

Transcription

1 EXPERIMENTAL INVESTIGATION OF FIBER SIZING-TEST FLUID INTERACTION FOR IN-PLANE PERMEABILITY MEASUREMENTS OF CONTINUOUS FIBERS A Thesis by Ravi Kiran Dukkipati Bachelor of Engineering, Osmania University, 2004 Submitted to the Department of Mechanical Engineering and the faculty of graduate school of Wichita State University in partial fulfillment of the requirements for the degree of Master of Science December 2007

2 Copyright 2007 by Ravi Kiran Dukkipati All Rights Reserved

3 EXPERIMENTAL INVESTIGATION OF FIBER SIZING-TEST FLUID INTERACTION FOR IN-PLANE PERMEABILITY MEASUREMENTS OF CONTINUOUS FIBERS The following faculty members have examined the final copy of this thesis for form and content, and recommend that it be accepted in partial fulfillment of the requirements for the degree of Master of Science, with a major in Mechanical Engineering. Kurt A. Soschinske, Committee Chair Hamid Lankarani, Committee Member Krishna Krishnan, Committee Member iii

4 ACKNOWLEDGEMENTS I express my sincere gratitude to my advisor and committee chair, Dr. Kurt Soschinske, for his guidance and support, without whom I would not have had such interesting and challenging research. For his valuable help, patience and encouragement, I owe him the deepest gratitude. I also express my appreciation and thanks to my other committee members, Dr. Hamid Lankarani and Dr. Krishna Krishnan. I would also like to thank Mr. Sanjay Sharma of the National Institute for Aviation Research for providing equipment and materials when required, for helping me to understand technical details and overcome difficulties, and for providing suggestions throughout my research, experiments, and documentation. I am thankful to the Department of Mechanical Engineering at Wichita State University for providing space to conduct my research. Finally, and most importantly, I give my appreciation to my family and friends for their encouragement and endless support throughout my life. iv

5 ABSTRACT Many manufacturers are using liquid composite molding (LCM) to manufacture composites because of its simplicity and cost effectiveness. This generic process includes techniques such as vacuum assisted resin transfer molding (VARTM), resin transfer molding (RTM), and Seemann composite reaction infusion molding (SCRIM). VARTM is most commonly used due to its low tooling cost and ease of use. One of the most important factors that affect the manufacturability of composite parts is permeability. Permeability usually affects resin flow within fibers; hence, many researchers have placed importance on the measurement of this factor in various studies. Alternate test fluids such as corn syrup, silicon oil and motor oil are being used to calculate permeability, since they are cheaper, easy to clean, and are not volatile like resins. Permeability is generally affected by fiber sizing-test fluid interaction, fiber volume fraction, and fiber orientation. In the present study, permeability effects due to fiber chemical coating or sizing were investigated using VARTM. Experiments were conducted by inducing high- and low-viscosity corn syrup and silicon oil through uni-directional glass fibers, bi-axial glass fibers, and two types of uni-directional carbon fibers, with and without sizing. Darcy s law was used to calculate permeability. Significant permeability differences found using fibers with and without sizing are reported. Separate saturated flow rate measurements for one of the uni-directional carbon fibers and uni-directional glass fibers, both sized and unsized, were carried out by infusing corn syrup and silicon oil with similar viscosities. It was found that, over time, corn syrup displayed a gradual decrease in flow rate for saturated fiber material for constant vacuum infusion. From previous research studies, the flow rate history in the RTM process showed two stages: initial rapid flow and steady-state flow. However, in the VARTM process, the cumulative mass vs. time v

6 plot was linear. The corresponding flow rate vs. time plot showed a high initial flow rate but was decreased with time when corn syrup was used for both uni-directional glass and carbon fibers, with and without sizing. This effect was not observed in the case of silicon oil. It was concluded that corn syrup should be used with caution as a test fluid for permeability measurements until further investigations can be made. vi

7 TABLE OF CONTENTS Chapter Page 1. INTRODUCTION Simulation of LCM Permeability Sizing Motivation and Objectives LITERATURE REVIEW Permeability Measurement Methods Problems with Permeability Measurements Variations due to Fiber Architecture Validity of Darcy s Law Impact of Sizing EXPERIMENTAL SETUP AND METHODOLOGY Theory Fiber Reinforcements Test Fluids Equipment Experimental Procedure Steps for Running Experiment Data Analysis RESULTS AND DISCUSSION Permeability Measurement Permeability with respect to Fibers Permeability with respect to Test Fluids Permeability with respect to Orientation Saturated Flow Rate Uni-Directional Carbon Fibers Uni-Directional Glass Fibers CONCLUSION Permeability Saturated Flow Rate Future Work...58 vii

8 TABLE OF CONTENTS (continued) Chapter Page REFERENCES...59 APPENDICES...62 A. Fabric Constitution...63 B. Fiber Volume Fraction...64 C. Permeability Calculations for Sized fibers...65 C. Permeability Calculations for Unsized fibers...66 D. Cumulative Mass Flow vs. Time Plots...67 E. Flow Rate vs. Time Plots to Check Time Sensitivity...68 viii

9 LIST OF TABLES Table Page 1. Fabric Characteristics Fibers Porosity, Orientation and number of layers Test Fluid Characteristics Experimental Cases Check List of variables measured and calculated...34 ix

10 LIST OF FIGURES Figure Page 1. Channel flow and radial flow method Different fiber reinforcements Top view and isometric view of experimental injection mold Sealed injection mold Schematic diagram of experimental setup Setup used for conducting flow and permeability experiments Permeability of sized/unsized UD carbon fibers 1 with respect to test fluids Permeability of sized/unsized UD glass fibers with respect to test fluids Permeability of sized/unsized biaxial glass fibers with respect to test fluids Permeability of sized/unsized UD carbon fibers 2 with respect to test fluids Permeability of sized/unsized UD carbon fibers 1 with respect to HVCS and SO Permeability of sized/unsized UD glass fibers with respect to HVCS and SO Permeability of sized/unsized biaxial glass fibers with respect to HVCS and SO Permeability of sized/unsized UD carbon fibers 2 with respect to HVCS and SO Permeability variations in biaxial glass fiber at different orientations using HVCS Flow rate for UD carbon fibers 1 using HVCS at 27 Hg Flow rate for UD carbon fibers 1 using SO at 27 Hg Flow rate for sized UD carbon fibers 1 using HVCS and SO at 27 Hg Flow rate for unsized UD carbon fibers 1 using HVCS and SO at 27 Hg...51 x

11 LIST OF FIGURES (continued) Figure Page 20. Flow rate for UD glass fibers using HVCS at 27 Hg Flow rate for UD glass fibers using SO at 27 Hg Flow rate for sized UD glass fibers using HVCS and SO at 27 Hg Flow rate for unsized UD glass fibers using HVCS and SO at 27 Hg...55 xi

12 NOMENCLATURE v Q A Velocity of the fluid Fluid flow rate in volume per unit time Cross sectional area of material K, K 1 Permeability of specimen in direction of flow µ Viscosity of fluid ε, φ Porosity of material L ρ t Length of material Density of fluid Mold fill time dp/dx Pressure difference with respect to length P S P S V f Re γ SV γ SL γ LV W S W SL W LL Pressure gradient across length Total porous space Pressure Sink function Fiber volume fraction Reynolds number Surface energy between solid and vapor Surface energy between solid and liquid Surface energy between liquid and vapor Work of spreading Work of adhesion Work of cohesion xii

13 CHAPTER 1 INTRODUCTION A composite is a combination of two or more materials of different properties, which when combined result in a material with unique properties. The first modern composite, called fiberglass was developed in the late 1940s and currently makes up approximately 65 percent of all composites produced today [1, 2]. Research in the field of composites has rapidly grown during the past few decades because of its advantage of tailorability over different metals and materials. The most important properties of composites that differentiate them from other materials are high strength, stiffness, damage tolerance, durability, and ease of manufacturing. The material can also be locally stiffened within a product, thus allowing design tailorability. Advanced composites are manufactured by various techniques such as hand-layup, prepreg and resin infusion processes. Of them, Liquid Composite Molding (LCM) is a generic method, wherein resin is induced into a mold containing a bed of fibers using a pressure differential. Resin transfer molding (RTM) process, vacuum-assisted resin transfer molding (VARTM) process, and Seemann composites reaction infusion molding (SCRIMP) process are examples of LCM. In these processes resin, is infused through the preform in the injection mold with the help of externally applied pressure at the inlet, while the outlet is maintained at vacuum. The resin-infused preform is then cured at elevated temperature cycles. After curing, the part is removed (demolded), and the mold is ready for reuse. In another process of manufacturing composites, composite fibers are semi-cured with resin. Hand or machine layup is used to form the prepreg or pre-impregnated layup. The prepreg layup is then vacuum bagged and cured at elevated pressure and temperature cycles. These prepreg composite fibers are expensive when compared to liquid composite molding techniques, since the process requires cold storage and 1

14 tools that can withstand high pressures and temperatures. The main advantage of the prepreg composite is excellent mechanical properties. The choice of different processing techniques is made according to design parameters such as desired mechanical properties, part size, part thickness, fiber volume fraction, fiber viscosity, type of injection mold, curing requirements, and production cost. This current investigation focuses on the LCM process. In a composite, the matrix material bonds with the reinforcement material and forms a new material with unique and enhanced mechanical properties. The strength of the material depends on the bonding between the reinforcement and the matrix. Commercial composites are produced by different polymer matrix materials called resins such as polyester, vinyl ester, epoxy, bismaleimide, etc. The use of different resins with different reinforcement materials is done according to the design and operating condition requirements. According to the design, strength, and stiffness requirements, appropriate fibers and resin systems are chosen. Different resins and fibers are available. Glass fibers, carbon/graphite fibers, and aramid fibers are different types of fiber reinforcements, each of which is available with different tow sizes, sizing, and fabric construction. A tow is a bundle of individual filaments that is stitched or woven to make fabrics. Most of the VARTM processes use 6,000 or fewer tow size (filaments) for carbon fibers. Sizing is the chemical coating on the fiber surface that is crucial for a good fiber-matrix interface. Fiber systems are chosen with compatible resin systems. 1.1 Simulation of LCM Simulation tools have been developed to optimize the composite manufacturing process by predicting the flow of resin through the fibers, which results in maintaining constant quality and reduced production cost. The finite element method (FEM) and boundary element method 2

15 (BEM) were used to perform numerical flow simulations in the RTM process by Wang et al [3]. Although finite element analysis (FEA) was found effective for the RTM process design and analysis, it is still in the early stages for the VARTM process. Liquid injection molding simulation (LIMS), developed at the University of Delaware, was validated with respect to onedimensional flow in RTM and VARTM processes [4]. LIMS uses fiber compaction and permeability to estimate the mold fill time and fiber volume fraction in the VARTM process. Process simulation not only helps in predicting the flow but also helps in optimizing the mold design, since it can predict the required external pressure on the resin. There is a need for proper tool design to optimize the LCM process. Mold design is dependent upon the type of process used, for example RTM or VARTM. Part dimensions, including thickness and inlet/outlet port locations (for easy and quick flow of resin through the fibers before it reaches the gel time) are some important tool design considerations. Diameter of the inlet port is also critical for tool design, since a smaller injection port diameter requires higher injection pressure, and vice versa. But higher injection pressures can cause higher flow velocity resulting in fiber wash. RTM processes require closed-injection molds, which are designed such that the two halves of the mold align with each other. The mold is designed to resist the internal pressure without deformation and maintain the pressurized resin without leakage with the aid of a containment seal around the perimeter of the cavity. Mold material is selected in such a way that it withstands the processing and curing temperature cycles. The most important input for accurate simulation during the LCM process is permeability. With permeability as a known parameter, velocity and hence the flow front position can be predicted. The required pressure to maintain flow velocity helps in choosing the appropriate material for the tool. 3

16 1.2 Permeability Permeability (K) is a measure of the ability of a porous material to transmit fluid. It can be measured using Darcy s Law or estimated using empirically derived formulas. Commonly used units for permeability are Darcy (cm 2 ) [1]. Darcy studied the flow of water through a bed of sand [1, 5] and empirically derived the relation know as Darcy s Law. Darcy's Law is a phenomologically derived constitutive equation that describes the flow of a fluid through a porous medium [1]. This method gives one-dimensional measure of fiber permeability. To fully characterize the fiber reinforcement, it is necessary to measure principle permeabilities K 1, K 2, and K 3. Since most of the LCM parts are thin shells, the permeability in thickness direction k 3 is neglected, and in-plane permeabilities K 1 and K 2 are evaluated [6]. For measurement of permeability through porous media, Darcy s Law or Carmen- Kozeny equations are most commonly used. A Carmen-Kozeny equation may not be a good choice for the permeability of fiber reinforcements because it assumes the porous media is isotropic while the permeability of most high-performance fiber reinforcements is anisotropic [7]. Darcy s Law, most commonly used for predicting flow through porous media, is stated as the macroscopic velocity of the flow in a given porous medium that is directly proportional to the pressure gradient and inversely proportional to the dynamic viscosity of the fluid through a proportionality constant K (see equation 1). In an anisotropic medium like a fiber mat, permeability varies in all directions. The three principal permeability values corresponding to the orthonormal directions of the fabric length, width, and thickness are dependent upon many factors which may include fiber thickness and orientation for example. Most common, in-plane permeability measuring techniques are channel flow and radial flow methods. In a channel flow, the fibers are placed in a rectangular cavity, and the resin is injected from one end while a 4

17 vacuum is applied at the other end. In the radial flow method, the resin is injected at the center of the sample such that the flow front is shaped like an ellipse from which the ratio of principal permeabilities can be determined. In both of these techniques, the unsaturated and saturated permeabilities can be measured [5]. Measurement of permeability in the transient flow during mold filling is referred to as unsaturated permeability, and the permeability measured at steadystate flow conditions reached after mold filling is called saturated permeability [5, 8]. This is also used to locate the outlet/inlet ports in a given manufacturing process. Without proper location of these ports, a part can be left without complete resin wetting or contain a number of unsaturated dry spots in the fiber mat. Initially, RTM lacked the consistency in producing aerospace components with low void content and dry spots. Improvements in materials and processes made RTM an effective method for aerospace manufacturing. Implementing RTM in the aerospace industry not only reduced the production time but also reduced the manufacturing cost by more than 60 percent [2]. Low void content is obtained by ensuring that the flow fronts meet at the outlet port location. The location of two or more flow fronts meeting at a point is dependent upon fabric permeability. Therefore, permeability is important in every step of the composite manufacturing process. 1.3 Sizing Research in the field of mechanics of composites focuses on the strength and stiffness of the structures. The strength and stiffness properties of composites are dependent upon the fiber elastic modulus, matrix elastic modulus, and load transfer capabilities between the fiber and matrix [2]. The latter is dependent upon the fiber-matrix bonding. The bonding between fiber and matrix is enhanced by the use of a chemical coating on the surface called sizing. Also, raw fibers are abrasive; sizing protects each fiber from its neighboring fiber. The chemical composition of 5

18 sizing may vary among between individual manufacturers, but its purpose remains the same. Glass fiber sizing is a mixture of several complex chemicals, where each of them contributes to the overall process performance. Composition of the sizing is proprietary, since it distinguishes a manufacturer s products its competitors. Investigation of sizing composition was done by V.M. Karbhari by stripping the fibers with acetone and analyzing the dissolved materials using Fourier transform infrared (FTIR) spectroscopy [9]. Although sizing takes only 0.25 to 6.0 percentage of the total fiber weight, it serves to protect fibers from breakage during the process [2]. The chemical formulation of a sizing interacts with a specified resin system to make a better bond. Hence, for different resin systems, different sizing is required to have good bonding. Silane is coated over glass fibers since it initiates good bonding between glass and epoxy and enhances fiber wetting. Most glass manufacturers offer multi-compatible products, which can be used with multiple resins such as polyester, vinyl ester, and epoxy. Earlier, carbon fibers were sized to be compatible with epoxy resin. But the increase in demand of carbon fibers outside the aerospace industry has compelled manufacturers to produce carbon fibers that are compatible with a broader range of resins and processes [2]. 1.4 Motivation and Objectives As explained previously, to determine the permeability of fibers, tests are usually conducted to observe the flow of resin through the reinforcement. Due to cost and environmental and cleaning problems with traditional resins, various inexpensive test fluids are used in place of resins with similar physical properties. But most importantly, the use of test fluid to measure permeability assumes its validity with all types of fibers. Most of the test fluids used are idealized Newtonian fluids, which are considered with properties that include physical viscosity and density, as well as process conditions such as visibility, availability, and low cost. In studies 6

19 conducted by Baiju and Pillai [10], motor oil was used. Lee and Shin [7], Wang [3], used diphenyl-octyl-pthalate (DOP) oil, while Steenkamer et al. [11], Saraswat et al. [12], Bender et al. [13], Ferland et al. [6], and Correia et al. [4] used a mixture of water and corn syrup. Viscosity can easily be varied in a corn syrup using distilled water. Other researchers [9, 14, 15, 16, 17] used uncatalyzed resin, compared to Shojaei et al. [8] who used silicon oil (SO). These test fluids are accepted in the research community for the measurement of permeability. Currently, there are no specified standard test fluids that can be used to evaluate different reinforcements for permeability measurement. Some relevant investigations have reported an influence of sizing on the flow of resin [9, 14, 16, 17]. Since research on permeability measurement has been continuing for the past few decades, a need to qualify these measurements with new data indicating fiber sizing-resin interaction needs to be incorporated. Various researchers have developed techniques of permeability measurement. Part of their test methodology is based on maintaining constant pressure or constant flow rate in the experiment. Issues regarding permeability investigated by some researchers include [7, 14, 8, 18]: 1. Validity of Darcy s Law 2. Effect of sizing on permeability measurement 3. Effect of sizing on flow rate 4. Effect of sizing with respect to test fluid The principal focus of this current study was to investigate the interaction between fiber sizing and test fluid in the determination of permeability. The overall objectives were to: 1. To study variation of unsaturated permeability for given fiber/sizing configurations with respect to various test fluids. 7

20 2. To compare results to unsaturated permeability measured for the same configurations without the effect of sizing. 3. To study the saturated flow rate variations with respect to test fluid, sizing and fiber material. 8

21 CHAPTER 2 LITERATURE REVIEW Research to standardize the permeability measurement for porous media has been carried out for more than a decade. Various studies have used different techniques, fibers, and test fluids to measure permeability so that flow of resin through the fibers can be predicted. Simulation software packages have been developed to estimate the flow through porous media. These simulation software packages require fiber material dimensions, permeability of the material, and resin viscosity as input parameters. Therefore, accurate permeability measurement is critical in the field of liquid composite molding simulation. The permeability of the fibers depends on various factors such as fiber tow size, fiber orientation, fiber diameter, type of sizing, packing density, and resin system used. 2.1 Permeability Measurement Methods In-plane permeability is usually measured using either the channel flow method or radial flow method, as introduced in the previous chapter and shown in Figure 1. In each case, saturated and unsaturated permeabilities are measured by monitoring the flow of resin through the fiber mat. In channel flow measurement method, a very common approach to measuring the permeability of the fibers [4, 6, 8, 13, 15], unidirectional flow is created through the fibers placed in a long rectangular cavity of known dimensions. Resin is infused through the fiber mat from one end and, air is allowed to escape through the other end. Permeability is measured in two ways: either by monitoring the flow front as the cavity is being filled, or by measuring the pressure drop between two points once the preform is saturated with the fluid. Monitoring of flow front or pressure difference is done in various ways by different researchers [13, 15]. 9

22 Permeability is calculated by substituting the obtained values in Darcy s Law for one dimensional flow [5, 6, 8, 15] as Q K dp v = = (1) A µ dx where v is the average velocity of the fluid in (m/sec), dp/dx is the pressure difference with respect to length (Pa/m), Q is the volumetric flow rate (m 3 /sec), A is the cross-sectional area (m 2 ), K is the permeability measured in (m 2 ), and µ is fluid viscosity (Pa-sec). A noticeable point about Darcy s Law is that it is based on macroscopic flow behavior and does not take porosity of the medium into account. Another approach to determine the permeability of the fibers is with the help of the radial flow method. In this approach, the resin is injected at the center of the fiber sample. The injection is carried out by placing the resin inlet port on top of the fiber pre-form. Just like in the unidirectional flow, there are two approaches to measure the permeability: one by monitoring the flow front position as it fills the cavity and the other by monitoring the pressure drop during saturated flow. In the radial flow experiments, the flow front takes the shape of an ellipse with major and minor axes along the principal directions. The relationship between the principal permeabilities K X and K Y is given by [3] K K X Y 1 2 = length of minor elliptic axis length of major elliptic axis (2) Using equation (2) with Darcy s Law for two-dimensional radial flow gives the two permeabilities corresponding to the principal directions, X and Y. Lee et al. [15] measured in-plane permeability of two different fiber materials by inducing resin in a unidirectional channel flow. The VARTM technique was used, and permeability was calculated by obtaining the flow front locations with respect to time. Flow front 10

23 positions of the resin were captured by digital cameras placed on the top and bottom of the mold. Bender et al. [13] controlled the flow rate in VARTM process by placing the vacuum sensors on the fibers at a known distance. The vacuum was controlled by pressure controllers connected to a venturi system with feedback from the vacuum sensors. Steenkamer et al. conducted experiments using the radial flow method. In their experiment, resin was injected into the fiber bed at a constant flow rate, and Darcy s Law was used to measure the permeability. A Macintosh SE30 computer with an Analog Connection Quick Log data acquisition program was connected to the injection cylinder to obtain the pressure readings for five times per second. Resin Fiber Resin Fiber θ A B Figure 1: Permeability Measurement: (A) channel flow method; (B) radial flow method [19]. Wang et al. [3] compared the in-plane permeability of fibers using both the radial flow and unidirectional channel flow methods. Experiments were conducted using diphenyl-octylpthalate (DOP) oil as a test fluid with viscosity of 40 to 50 cp on continuous random fiberglass, stitched bidirectional fiberglass, and 8-hardness woven graphite fiber mats. Permeability was calculated in both principal directions. Variations were observed in all the three cases. Numerical 11

24 simulations were conducted with a porosity of 0.56 and flow rate of 425 ml/min, and the results were compared with experimental results of bidirectional fiberglass using the radial flow measurement. A correction factor of 16% was required for simulation results, in order to obtain values closer to the experimental results. Finally, the authors concluded that, to obtain reliable permeability values, both the radial flow and unidirectional flow measurement methods should be conducted for any given fibers. Han et al. [20] developed a new technique for measuring permeability of anisotropic fiber preforms with high fiber content. The radial flow method was used to which determine the principal in-plane orientation and the permeability components in the plane. Permeability measurements were done at saturated state to eliminate the effect of capillary pressure. In the RTM process, capillary pressure exists only in the flow front region due to surface tension between the resin and air. The effect of capillary pressure as the driving force was taken into consideration for predicting the flow using the measured permeability. In this new technique, measurements were carried out at steady state with saturated flow through fibers, and the pressure was measured at four different locations instead of flow front locations. The obtained pressure values were used in determining permeability. Experiments were conducted by injecting silicon oil at a constant pressure on four different fibers: 3K-135-8H carbon fibers, H glass fibers, P-4 chopped glass fibers, and P-4 chopped carbon fibers. Pressure values were recorded from four different positions once the system reached steady state. These values were used in the calculation of fiber permeability. The new technique was validated by measuring permeability in principal direction of H glass fibers using the parallel flow method. A close match between experimental and simulation results was obtained. Using this new 12

25 technique, the permeability of an individual layer in a multilayer preform can be determined, and it can be used also to determine gas permeability in anisotropic porous medium. Lee et al. [15] determined the in-plane permeability of fiberglass chopped mat-300 and plane-woven roving-800 mats by varying the number of layers. Unsaturated polyester resin with viscosity of cps was infused with the help of a vacuum. Darcy s Law was modified to a one-dimensional flow case and was used to measure permeability. The permeability calculation was done by obtaining the linear relation between the square of the length of flow front (L 2 in m 2 ) and the fill time of resin (t in sec) through the cavity. The equation developed to calculate permeability is explained in the Chapter 3, equation 15. In every case, permeability on the top side was larger than the bottom side due to the flow enhancing media, which consisted of the distribution layer D and peel ply P. A decrease in permeability was observed as the number of layers was increased. An additional experiment was conducted to obtain the permeability of the enhancing media K D+P, which was used in obtaining a new variable that represents the change of the permeability in percentage K (%). An increase in the variation of permeability K (%) was observed as the number of layers increased when the distribution layer and peel ply were excluded. Changing the number of preform layers was considered as the change in the porous space provided by the preform for resin flow. A new parameter called the total porous space S was proposed, defined as S = t total φ (3) where t total is the total thickness of the reinforcement, and φ is the porosity of the reinforcement. Once the total porous space S of the designed preform was obtained, K was predicted by interpolating the K-S curve. K was used in the calculation of the new fiber layer order K using the equation 13

26 ( K ) K' = K D+ P K D+ P K = K D+ P 1 (4) This equation was used to predict the permeability of the fibers at the macroscopic level. Out of-plane permeability is the permeability of the fiber reinforcement in the thickness direction, which can be determined by using a unidirectional flow approach or by three dimensional (3D) computer simulation methods [5]. Darcy s Law can be applied to calculate this transverse permeability. However, it is not easy to measure transverse permeability accurately because, for most of the cases, thickness of the part to be fabricated is small, and channel flow and fiber deformation may occur at high pressures. Therefore, mostly 3D computer simulations methods are used where constant flow rate can be applied, and the results are compared with the measured inlet pressure. In-plane permeability results are used as input to obtain transverse permeability with the help of computer simulations [1]. Some studies have used test fluids to measure permeability. To understand the role of test fluids, Steenkamer et al. [11] selected three different test fluids with varying viscosities: diluted corn syrup (190 cp), motor oil (155 cp), and vinyl ester resin (135 cp). These test fluids were used to measure the permeability of two fiberglass reinforcements, where each fabric was tested at two different porosities: a continuous strand mat and a biaxial (±45 ) knitted fabric. Test fluids were injected at a constant flow rate of 16.2 cm 3 /sec using radial flow experiments. Pressure was monitored at each principal direction and opposite to the inlet with the help of pressure transducers. Permeability of both fibers at various porosity levels was calculated using Darcy s Law. In all cases with the continuous strand mat, the measured permeability was higher in weft (fill) than the wrap direction for all three test fluids. For the biaxial (±45 ) knitted fabric, wrap direction permeability was more than weft direction permeability. This was explained on the basis of chain stitches, which hold the +45 and -45 layers together and run in the wrap 14

27 direction. For both mat materials, the highest permeability values were obtained using motor oil and the lowest permeability values using diluted corn syrup. Vinyl ester resin gave values between the two bounds for both the weft and wrap direction, which shows that even motor oil with higher viscosity than vinyl ester resin has more affinity toward glass fibers. Thus, to obtain high-quality composite parts, there should be a balance between macro and micro flow of the resin through the fibers and fiber wet out. Mastbergen et al. [21] created and validated a two dimensional (2D) simulation model for producing large primary structures, such as turbine blades and boat hulls. The simulation and validation were done through every step of modeling: flow through the injection system, channel, and fabric. Diluted corn syrup at constant pressure was used in the experimental validation. This study was conducted using pressure-bag molding, where the distribution system is a channel that covers the whole surface of the fabric. Once the fluid fills the channel, pressure is applied to the vacuum bag to force the fluid into the fabric. Tests were conducted to observe the effect of compaction pressure and fiber architectures. A dramatic effect of compaction pressure on permeability was observed in the obtained results and a maximum variation of 20 percent from the mean value was observed for unidirectional fabric. A Modified Darcy s Law was used, and saturated flow time was calculated. Flow through the hose and channel were represented as Q P µ hose hose = Khose (5) where K hose is the effective permeability of the hose system, µ is the fluid viscosity, and Q hose and P hose are volumetric flow rate and pressure drop, respectively, through the hose systems. The volumetric flow rate through the channel Q chan was defined as 15

28 Q chan = A chan K µ x chan P chan chan (6) where A chan is the cross-sectional area, x chan is the channel length, and K chan is effective permeability of the channel in the channel length directions. These equations were combined to determine the time required to fill the channel, neglecting the fabrics, and were validated experimentally. If the time required to fill the channel was greater than or closer to the time to flow through the fabric, then the process was channel-flow dominated. For the channel flowdominated process, the model and experiments were compared using the time to fill the channel and the position of the flow front on the bottom of the mold. But an error of 10 percent was observed to that of computer-generated results. For the fabric flow domination process, that is, where the flow through the fabric is greater than the time to fill the channel, the analytical equations were much more accurate. The researchers finally concluded that the 2D model developed was suitable for determining the dimensional scaling effects in these types of processes. 2.2 Problems with Permeability Measurements Variations Due to Fiber Architecture Variation in the permeability of fibers with respect to their architecture was observed by some researchers. Shih and Lee [7] measured the effect of fiber architecture on permeability by conducting radial flow tests on four different glass fiber reinforcements, which included a unidirectional and bidirectional stitched mat, a continuous strand mat, and a four hardness woven mat. Porosity was changed by altering the number of layers, and for directional and woven fabrics all layers were laid with the same orientation. DOP oil was infused through fibers at a constant flow rate. High in-plane permeability was observed in the bidirectional fiber mat, while 16

29 the other three fiber mats had similar permeability, even though they had different tow size and structure. Steenkamer et al. [14] compared the permeabilities of five different E-glass fibers and also the role of fiber wetting in permeability measurement with three resin systems. Permeabilities of a continuous strand mat, bi-axial (0 /90 ) knitted fabric, bi-axial knitted fabric plus a layer of chopped strand mat (0 /90 /C), bi-axial (±45 ) knitted fabric, and unidirectional woven fabric were measured at different porosities with vinyl ester resin. Experiments were conducted using the same apparatus used by Steenkamer et al. [11]. To eliminate the effect of channeling and pressure buildup, tests were conducted until the fluid touched the mold wall. The continuous strand mat showed higher in-plane permeabilities than the rest of the mats, which may have been because of the porous nature of the mat relative to the directional fabrics. The second-directional permeability K 2 of bi-axial (0 /90 ) and bi-axial (±45 ) knitted fabrics were similar, but their K 1 differed by a minimum of 146 percent. Changes in permeability due to the fiber material, orientation, porosity, etc., were used to estimate the influence of fiber architecture on processing. The wetting of the fiber surface by fluid was described by three-phase thermodynamic equilibrium between the solid, liquid, and vapor phases calculated by using the Young-Duprè equation [14] γ = γ γ cosθ (7) SV SL + where γ SV, γ SL, and γ LV are surface energies between solid, liquid, and vapor, and θ is the equilibrium contact angle. When θ > 0, liquid is termed non-spreading; when θ = 0, liquid is said to spread over a solid surface and the Young- Duprè equation cannot be used for the case -1 LV 17

30 cos θ < 0, where the attraction of solid to the liquid is less than the liquid to itself. Direct measurement of wetting is the work of spreading W S, given as [14] W S = W W = γ γ γ (8) SL LL SV SL LV Where W LL is the work of cohesion of the fluid, and W SL is the work of adhesion. Motor oil showed higher permeability and was attracted toward the glass fiber when compared to the vinyl ester and diluted corn syrup. Both the vinyl ester and diluted corn syrup have a negative work of spread W s, which indicates that there was need of external pressure for the test fluids to wet the fibers. Research in the microscopic flow of resin through the fibers due to capillary pressure showed mixed results as reported by various investigators. Microscopic flow is the resin flow inside the fiber bundles, which is dominated by capillary effects, and macroscopic flow is flow around the fiber bundles, which is governed by the applied mechanical pressure [8, 17]. The authors concluded that, until a complete understanding of the test fluid and the fiber interaction is developed, fibers should be characterized with the actual liquid composite resin for a given application [14]. Since permeability measured with other idealized fluids yield only apparent permeabilities. Baiju and Pillai [10] investigated the effect of fiber architecture along with the effect of fiber-mat compression on the inlet pressure profile. Experiments were conducted using the resin transfer molding process, and data acquisition and monitoring was done by Lab VIEW (National Instruments, Inc.). To study the effect of fiber-mat compression on inlet pressure, biaxial stitched mats with five, six, and seven mats stacking were created. They showed a fiber volume fraction of 0.26, 0.31, and 0.36, respectively, in a mold cavity. Motor oil was injected at a constant flow rate, and the obtained inlet pressure profile was compared with theoretically predicted inlet pressure (P in ) profile using 18

31 P in 2 Q µ = t (9) A K ε sat This equation is an analytical expression for the inlet pressure as predicted by conventional physics for one-dimensional mold flow, where Q is the flow rate, A is the crosssectional area of flow, µ is viscosity of the test fluid, ε is overall porosity of the fibers, t is the process time, and K sat is the saturated permeability obtained after steady-state conditions established by using the formula K sat Q µ L = A P (10) where Q is measured from the flow meter, and P is the pressure gradient calculated from pressure transducers placed at a predetermined distance L. The mold fill time t fill-time was calculated using t fill εalmat time = (11) Q where L mat is the length of the fiber mat, and fill time t fill-time is used as t in equation (9) to obtain theoretical steady-state pressure. The results obtained by there experiments showed that an increase in fiber volume fraction reduced the permeability of preform, which was obviously due to the increase in resistance to the flow. The inlet profile obtained for a stack of five mats had increasing pressure at the beginning but drooped down as the flow front advances thorough the fibers, when compared to the predicted results from equation (9). In the case of six and seven fiber mats stacking, there was a higher initial pressure than the predicted profile. When all the three fiber volume fractions were compared together with the predicted inlet pressure, there was a very large margin at the initial stages but excellent agreement at steady-state pressure. From the 19

32 conducted experiments, it was noted that faster steady-state pressures were obtained with increasing compression. More experiments were conducted by Baiju and Pillai [10] to obtain the effect of fibermat architecture on inlet pressure using four different fiberglass materials: biaxial-stitched, biaxial-woven, triaxial-stitched, and unidirectional-stitched mats. In all cases except for unidirectional mats, inlet pressure profiles were obtained for a stack of five mats, where test fluid was injected at a constant flow rate of 2.83 L/min. Inlet pressure profiles for unidirectional 20 and 22 mats stacked was wavy and fluctuating. The biaxial woven mat did not show any droop in inlet pressure. This may have been due to the absence of continuous uninterrupted macro channels; therefore it cannot be classified as dual-scale porous medium. The investigation showed that a change in fiber volume fraction and mat orientation have a significant effect on flow characteristics Validity of Darcy s Law Shojaei et al. [8] used the unidirectional flow method to find the saturated and unsaturated permeability of fibers in the resin transfer mold process method. Silicon oil with a viscosity of 0.1 Pa.s was infused through a glass woven fabric at constant high pressures of 1, 2, and 3 bars. Darcy s Law was used to calculate the permeability of this unidirectional flow. Since Darcy s Law cannot be used to describe flow correctly for higher flow rates, an acceptable flow rate valid for Darcy s Law was defined using Reynolds number as v* ρ k Re = (12) µ where ρ is the density of the fluid, and v* is the average velocity, which is related to Darcy velocity 20

33 v v * = (13) φ Porosity of the fiber reinforcements is defined byφ. It was shown that for Reynolds number values between 0.1 and 75 for general porous media, Darcy s Law holds. In a partially saturated region where microscopic flow existed, Darcy s law was not valid, since it does not govern the flow. Both the steady-state and transient experiments were used to determine permeability. Results showed higher initial permeability and decreasing behavior of unsaturated permeability until it reached a stabilized value. Saturated permeability was higher than unsaturated permeability in all cases. The permeability difference remained constant, even with the change in injection pressure of the fluid and porosity of the fabric. Pillai [18] reviewed results of various researchers who studied unsaturated flow behind the flow front in the liquid composite molding process using certain directional mats. Unsaturated flow made the current LCM simulation process inapplicable to the directional mats, as it over predicted the inlet pressure and also failed to warn about the presence of unsaturated regions. This was due to the dual scale nature observed in composite porous media, as pores present between the tows in a directional mat are much bigger than the microscopic pores present between the fibers. This over-prediction was not observed in random mats, which created a single-scale porous medium because of pore uniformity. Due to this dual-scale porous medium, resin fills around the fiber tows quickly before it starts penetrating through them. Absorption of resin through tows will continue after the flow front has passed, acting as sinks of liquid in macroscopic flow field. To nonzero this sink function, a mass balance equation was introduced v = S (14) where S is the sink function equal to the volumetric rate of liquid absorption. Using this equation in conjunction with Darcy s Law, the same downward droop of inlet pressure was obtained as 21

34 found in the test results. Four different directional mats were investigated using the inlet pressure profile in the one-dimensional mold-filling process to ascertain if the sink effect was due to the delayed absorption of tows. Among these mats, biaxial-stitched mats displayed a clear drooping profile and triaxial-stitched mats showed a partial droop, whereas woven-biaxial and unidirectional-stitched mats were almost linear. Thus, it is clear that many other factors were involved in manifestation of unsaturated flow in directional fiber mats. Uninterrupted channels throughout the mold length should be created for delayed impregnation effects to take place and for a dual-scale porous medium to be created in the directional fiber mat. Unsaturated flow was also caused as a result of bubble formation behind the flow front. The formation of these bubbles near the flow front was a clear indication of two-phase resin and airflow through the porous medium. The Buckley-Leverette equation was used. This equation, based on the two-phase flow of resin and air in a single-scale porous medium, included the generalized Darcy s Law for averaged momentum balance and a transient saturation-based equation for mass balance. Research to validate this equation continues Impact of Sizing To examine the effect of fiber sizing on overall behavior of the flow, Karbhari and Palmese [9] conducted experiments using S-2 glass rovings with sizings designated as 365, 933, 449, and 463. To gain some knowledge on differences in composition of sizings coated on the fibers, an investigation was made by stripping the fibers with acetone and then analyzing the dissolved solution using Fourier transform infrared spectroscopy (FTIR). It was observed that all four sizings showed the presence of silane-derived silicon. Resin 933 contained condensed silane, whereas the other fiber sizings contained some hydrolyzed material. Resins 365, 449, and 463 with sizing contained a considerable amount of unbound epoxy, which was absent in the 22

35 spectrum of 933. Uncatalyzed resin was used as a test fluid over these fibers with different sizings at room temperature. In the tests, resin was infused through fiber bundles with equal fiber volume fraction in a stainless steel tube. It was observed that the 365 system, which was specifically sized for vinyl ester and polyester, showed the highest steady-state flow rate, and the 933-sized system formulated for high temperature resins showed the highest initial flow rate and longest transient flow period. The order values for steady-state flow rates are as follows: Q365 >> Q463 > Q933 ~ Q449. This clearly shows that sizing-resin compatibility plays an important role in the flow rate of fiber wet out. To see the effect of sizing on microscopic flow, Palmese and Karbhari [17] conducted experiments on 12 K Hercules AS4 sized and unsized carbon fibers. It was assumed that microscopic infusion occurred on a similar time scale compared to macro-infusion. The resin transfer molding process was used to run the experiments in a stainless steel tube. Fibers were axially oriented in the tube, and uncatalyzed Dow Derakane 411-C50 vinyl ester was injected through the fibers at 1.38 kpa pressure. Fiber loads were varied by varying the number of layers from six, eight, and ten. Cumulative flow results, plotted with respect to time, showed two flow regimes. One was a transient region characterized by an initial rapid flow rate, and the other was a steady-state region possessing a significantly lower flow rate. Higher initial flow rates were obtained for lower fiber loadings, and steady-state flow rates increased with decreasing fiber loading with respect to time. Results obtained for unsized fibers were similar to those for sized fibers. Cumulative flow rate graphs for sized and unsized fibers showed the same characteristic shape. The sizing showed a dramatic effect on flow behavior, as the initial flow for unsized system was eight times faster when compared to the sized system. In addition, the steady-state flow for the unsized system was greater than for the sized system. However, the duration of the 23

36 transient period for the unsized system was longer. These results show that the effect of sizing has more impact than the effect of fiber loading on both cumulative weight and flow rate with respect to time. Palmese and Karbhari [17] investigated the transient flow region by conducting experiments in clear tubes. Higher initial flow rates were obtained due to macroscopic channeling. Channeling developed due to the arrangement of the fiber tows, as slight variations of fiber tow placements existed even after proper care. Initial flow rate values were greater for lower-fiber loadings. Once the resin wetted the fiber tows, the fibers spread apart, reducing the macroscopic channeling. The ratio of surface area to volume then increased, and the flow rate through the tube decreased, so that steady state flow was reached when fiber spreading ceased. Capillary forces were determined to be the governing factor in determining the transient period of flow. Capillary forces were affected by distance between fibers, physical and chemical interactions between the fiber surfaces, and the resin system. It was expected that for higher capillary pressures, a shorter duration of transient region could be obtained. Experiments conducted to characterize the effect of fiber loading on micro-flow behavior supported this proposition. From discussions, it was concluded that at a microscopic level, surface energy plays a vital role in flow behavior, and sizing-resin interaction influences capillary pressure and micro flow. 24

37 CHAPTER 3 EXPERIMENTAL SETUP AND METHODOLOGY Experiments in the current study were aimed at determining the interaction between the permeability of fiber reinforcements and test fluids using a vacuum as the driving force. In this investigation, industry-typical fiber reinforcements and test fluids were used. Experiments were conducted using the VARTM process in a unidirectional rectangular injection mold. The channel method of measuring permeability was used [8, 10, 13, 15]. This section is divided into the following parts: theory, fiber reinforcements, test fluids, experimental setup, and data analysis. 3.1 Theory Permeability of reinforcement plays a vital role in the prediction of flow through a porous medium. Many researchers calculate the permeability of fibers using Darcy s Law, the Carman- Kozeny equation, or empirical equations. Selection of the method is done according to their available input values. The most commonly used method to calculate flow through a porous medium is by Darcy s Law, which was given in equation (1) in Chapter 2. For a sufficiently large dx, P can be assumed to remain constant. In such a scenario, the equation can be transformed into equation (15). Letting dp/dx to P/L and then integrating dp/dx in equation (1) with respect to time yields [15] L 2 K P = 2 t φµ (15) Where the length of the flow front is L, P is the pressure difference, t is mold fill time, and φ is porosity of the reinforcement. Pressure difference and viscosity of the fluid were noted while conducting the experiments. An L 2 /t graph was plotted with the help of video-captured data on 25

38 flow-front positions versus time, and its slope was used in equation (15) to obtain the permeability values of the tested fiber mat. Porosity φ was obtained from equation (16) and fiber volume fraction V f from test data found in Appendix B. The fiber volume fraction of these fibers was measured using ASTM D792-91, ASTM D , and ASTM D standards, with testing conducted by Sharma [22]. Experiments were conducted on all fibers with and without sizing by infusing different test fluids. Sizing on the fiber materials was removed by washing them in acetone. Then fiber mats were then cut to the dimensions and laid in the mold cavity. The porosity of the fibers φ was calculated with the help of fiber volume fraction V f, by using equation [15], as φ = 1 V f (16) As mentioned in the introduction, the objective of this study was to investigate the fluidsizing interaction and its impact on permeability measurement. In order to separate the effect of sizing-test fluid interaction, two fluids of similar viscosities were used to measure permeability of various fibers. 3.2 Fiber Reinforcements To calculate the variation of fiber permeability with respect to the sizings and test fluids, four types of fiber materials, each with either different sizing or tow sizes and orientations were considered. Figure 2 shows the different fiber systems and Tables 1 and 2 show the fiber characteristics, sizing, porosity, and orientation and number of layers used in the experiment. Fiber samples used in the experiment included a unidirectional (UD) carbon fiber mat manufactured by Saertex with areal weight of 15 oz/yd 2, unidirectional glass fabric manufactured by Saertex with areal weight of 16 oz/yd 2, biaxial glass fabric by Saertex with areal weight of 12 26

39 oz/ yd 2 and unidirectional carbon fabric by V2 with areal weight of 13.3 oz/yd 2. The physical details of these fiber materials are provided in Table 1 and Appendix A. A B C D Figure 2: Different fiber reinforcements used in the experiments: (A) UD Carbon Fiber 1 (B) UD Glass Fiber (C) UD Carbon Fiber 2 (D) Biaxial Glass Fiber TABLE 1 FABRIC CHARACTERISTICS Fibers Filament Diameter (µm) Sizing Areal Weight (oz/yd 2 ) Tow Size (K) Fiber Manufacturer UD Carbon Fiber 1 7 Polyurethane Saertex UD Glass Fiber 17 Silane Saertex Biaxial Glass Fiber 17 Silane Saertex UD Carbon Fiber 2 7 Epoxy V2 27

40 TABLE 2 FIBER POROSITY [22], ORIENTATION, AND NUMBER OF LAYERS USED Fibers Porosity (φ ) Number of Layers Orientation UD Carbon Fiber Uniaxial UD Glass Fiber Uniaxial Biaxial Glass Fiber ±45 Biaxial UD Carbon Fiber Uniaxial 3.3 Test Fluids Two different test fluids were used to conduct the experiments: diluted corn syrup and silicon oil (Dow Corning 200; fluid 50 cst). These fluids were chosen because they have been commonly used in many investigations [4, 5, 6, 8, 11, 12, 13]. Corn syrup (Karo dark syrup) was diluted with distilled water to the required viscosity was used because of its ease in cleanup and disposal. The National Institute of Standards and Technology (NIST) database on fabric permeability has used diluted corn syrup (Karo dark) as the test fluid to calculate the permeability of fibers [23]. Diluted corn syrup is a Newtonian fluid, since most of the resins are Newtonian before they are cured. Viscosity of the test fluids were measured before and after each experiment by the ball-drop method using a GV-2300 Gilmont Viscosimeter at room temperature. Corn syrup with two different viscosities was used, with the density of fluids calculated using a high precision weight scale (Sartorious Instruments, 0.1 grams accuracy). The viscosities and densities of the different test fluids are shown in the Table 3. Corn syrup was mixed just prior to the experiment, since physical properties may change with respect to temperature and time. The experiments were done in a temperature-controlled room where the average temperature was 24 C and varying not more than 1 C to 2 C. 28

41 TABLE 3 TEST FLUID CHARACTERISTICS Test Fluids Manufacturer/ID Mixture Low-Viscosity Corn Syrup Karo Syrup 3 parts water : 1 part Karo syrup Viscosity (cp) Density (gm/ml) High-Viscosity Corn Syrup Karo Syrup 1 part water : 2 parts Karo syrup Silicon Oil Dow Corning/ Equipment Used To obtain a unidirectional flow, a rectangular mold with cavity dimensions 216 mm by 112 mm by 3 mm was used (see Figure 3). The mold was made of aluminum, which can be used as both an open tool as well as closed injection mold with a transparent top. The injection mold consisted of reservoirs at both ends to gain constant flow throughout the width of the cavity. These reservoirs had slots at the bottom to install compound pressure and vacuum measuring devices. Flow media (as seen in Figure 4) was used at these reservoirs for maintaining constant and unobstructed flow. Since experiments were conducted using the VARTM process, a vacuum bag was used on the upper side of the tool, which was sealed with the help of AT-200Y sealant tapes (Airtech Inc.). The dimensions of the mold permitted a rectilinear and smooth flow front during experimentation. 29

42 Pressure measuring slots A Reservoirs Inlet B Slots to clamp Transparent Figure 3: Experimental injection mold: (A) top and side view of injection mold [19]; (B) 3D view of injection mold. Pressure transducers (Cole-Parmer digital compound gauge , 29 Hg/99.9psi) with a pressure measurement range of ± 30" of Hg and accuracy in the range of 0.1 Hg were installed at the inlet and outlet. Vacuum in the mold was maintained at 29 Hg with the help of a vacuum pump (Alcatel SD2010). A sealed, vacuum capable resin collecting pot (Bel-Art 30

43 Products) was used to collect resin at the exit. To maintain inlet pressure, a pressure pot with an injection pressure range of 0 to 80 psi was used. Video cameras were used to capture the readings from pressure transducers, flow front positions of the fluid through the fibers, and flow rate. In order to measure flow rate in saturated and unsaturated conditions, the volume of the outflow fluid was weighed with a weight scale having an accuracy of 0.1 g (CTS 6,000 by MyWeigh) with respect to time. Different valves, connectors, and tubes were used to seal and maintain the flow through the mold. A digital vacuum regulator (DVR 200 from J-KEM Scientific Inc.) was used to maintain a constant vacuum for saturated permeability calculations. The test injection mold and the experimental setup are shown in Figures 4, 5 and 6. Sealant tape Flow media Vacuum bag Figure 4: Sealed injection mold. 3.5 Experimental Procedure Before starting the experiment, the injection mold and pressure pot were cleaned with soap and water, rinsed, and underwent a final cleaning with acetone. Pressure transducers were 31

44 then fixed at the inlet and outlet of the mold. Fiber mats were cut the required measurements using a sharp box knife to and placed in the mold cavity. A mat width of 112 mm was required to minimize race tracking of flow along the edges, and orientation was maintained by handling and installing the fiber mats into the mold consistently. In order to obtain uniform flow and to avoid race tracking of flow along the mold edges, the side walls of the mold were sealed by adhesive paste (AT-200Y, Adhesive Tech). Strips of meshed flow medium (Figure 4) were placed in both reservoirs to maintain smooth and constant fluid flow. An injection mold vacuum bag was used over the tool to seal the fibers in the cavity. Sealant tape was used to seal the mold/fiber interface. Test fluids were prepared according to the required viscosities of the study. In this study, test fluids with three different viscosities and fibers with different porosities and orientations were used, as shown in Tables 2 and 3. The highviscosity corn syrup was tailored to have viscosity similar to silicon oil. Once the test fluid was ready, its viscosity and density were measured, and the fluid was kept in the pressure pot. One camera was placed to capture the top view (that is for flow front position) and the other camera was positioned to obtain the rate of outflow fluid in the resin collection pot. A vacuum was created in the mold, resin-collecting pot, and connecting tubes by opening the outlet valves connected between the mold and the vacuum pump. After pulling a 29 Hg vacuum, the pump was shut off, and the entire setup was kept steady for 15 to 20 minutes for a vacuum integrity check to identify and reduce vacuum leakage through connectors, the mold, and the resin collecting pot. After the setup was ready, the inlet valve was opened to let the test fluid enter into the fiber mat in the mold. The fluid was made to flow through the fibers for 30 minutes duration, considered the average gel-time. All the recorded data was used for calculating permeability and 32

45 flow rate of the fluid with respect to the fibers. Separate experiments were conducted to measure the permeability of various fiber reinforcements with and without sizing using different test fluids, as shown in Table 4. Refer to Appendices C for calculated permeability values. Sealant Tape Pressure Gauge Line Gate (Resin Inlet) Vacuum Bag Line gate (Resin outlet) Vacuum Gauge Ω Ω Resin flow direction Pressure Pot with Test Fluid Preform Layup One-Sided Tool Vacuum pot to Collect Test Fluid Vacuum Pump Figure 5: Schematic diagram of experimental setup. 33

46 TABLE 4 CHECK LIST OF VARIABLES MEASURED AND CALCULATED in EXPERIMENTAL CASES Fiber Sized/ Unsized Test Fluid µ (cp) Parameters Measured ρ (gm/ml) Cumulative Mass Rate (gm/min) P ("Hg) K 1 (darcy) Calculated Saturated Flow Rate (ml/min) UD Carbon 1 Sized LVCS X X X X X X UD Carbon 1 Sized HVCS X X X X X X UD Carbon 1 Sized SO X X X X X X UD Carbon 1 Unsized LVCS X X X X X X UD Carbon 1 Unsized HVCS X X X X X X UD Carbon 1 Unsized SO X X X X X X UD Glass Sized LVCS X X X X X X UD Glass Sized HVCS X X X X X X UD Glass Sized SO X X X X X X UD Glass Unsized LVCS X X X X X X UD Glass Unsized HVCS X X X X X X UD Glass Unsized SO X X X X X X Biaxial Glass Sized LVCS X X X X X X Biaxial Glass Sized HVCS X X X X X X Biaxial Glass Sized SO X X X X X X Biaxial Glass Unsized LVCS X X X X X X Biaxial Glass Unsized HVCS X X X X X X Biaxial Glass Unsized SO X X X X X X UD Carbon 2 Sized LVCS X X X X X X UD Carbon 2 Sized HVCS X X X X X X UD Carbon 2 Sized SO X X X X X X UD Carbon 2 Unsized LVCS X X X X X X UD Carbon 2 Unsized HVCS X X X X X X UD Carbon 2 Unsized SO X X X X X X 34

47 3.6 Steps for Running Experiment The following steps were used to perform the test procedures. Figure 6 shows the experimental setup Clean the mold apparatus with soap and water, rinse, and finally cleaning with acetone. Fix the pressure transducers at inlet and outlet valves of the injection mold. Cut the fiber mat to required dimensions and mat orientation, and wash them with acetone, if necessary, to remove sizing. Place fiber mat in the cavity of the mold, install flow media at inlet and outlet reservoirs, and seal it using sealant tape and vacuum bagging material. Prepare the test fluid, measuring the viscosity, temperature, and density, and pour it into the pressure pot. Connect the vacuum regulator between the vacuum pump and resin-collecting pot for maintaining a constant vacuum (for calculating saturated permeability of the fibers). Connect the pressure pot, injection mold, resin-collecting pot and vacuum regulator with ¼ ID tubing and plastic valves. Place the resin-collecting pot over a weight scale to note the flow rate readings by weight with respect to time. Make sure all equipment is of the same height. Start vacuum pump. Induce a vacuum in the sealed mold and resin-collecting pot by opening the outlet valve. Let the vacuum pump run until a vacuum of 29 Hg is reached on both the injection mold and resin-collecting pot. For saturated flow rate measurement, maintain a vacuum level of 27 Hg to reduce the bubbling of test fluid. 35

48 Shut off the vacuum pump, conduct the vacuum integrity check for 15 to 20 minutes, identify and repair leaks, and repeat the vacuum check, if needed to maintain 29 Hg. Set up the video cameras for recording the flow front position, driving pressure, and flow rate. Ensure that all electrical connections are proper, start recording with cameras, and start the vacuum pump, trying to maintain a vacuum around 29 Hg. Open the inlet valve for letting the test fluid into the mold with the fiber mat. Conduct each experiment for 30 minutes. Collect the recorded data (video of outflow fluid accumulation with time, video of flow front position in mold with time and pressure difference between inlet and outlet of mold). Use the recorded data to calculate the permeability and flow rate. Following the experiment, properly dispose of the excess test fluid and used fibers, and clean the equipment for another test. Resin pot with Test Fluid Vacuum collecting pot on weight scale Cameras Injection mold with preform Vacuum Pump Video recorder and Monitor Inlet Valve Pressure measuring devices Outlet Valve Figure 6: Setup used for conducting flow and permeability experiments. 36

49 3.7 Data Analysis In this study, experiments were conducted to determine the permeability of fibers and flow rate of the test fluid in an experimental injection mold. The viscosity µ in Pa-sec and density ρ in gm/ml were measured before the start of experiments using a viscometer and weight scale. Pressure difference P was constantly observed, and its average was used in the permeability calculations. Flow front position with respect to time was tracked and plotted to obtain the slope, that is, L 2 /t in m 2 /sec. Porosity φ of the fiber materials was obtained from equation (16). These values were used as input in Darcy s Law for calculating permeability in units of Darcy (1 Darcy = e-9 cm 2 ). Flow rate was measured by weight (mass in grams) with respect to time of the fluid flowing out of the mold under constant vacuum. Density ρ in gm/ml of the test fluid was used for converting the weight of outflow fluid into flow rate (ml/min). The test case checklist is shown in Table 4 while the results for measured and calculated values in the experiments are shown in Appendixes C. 37

50 CHAPTER 4 RESULTS AND DISCUSSION Part of the objective of this current study was to find the interaction between test fluid and fiber sizing. Permeability of fiber reinforcements was calculated by conducting tests on four different fibers with and without sizing using three different test fluids. To see the effects of test fluid on sizing without the effect of viscosity, two different test fluids (corn syrup and silicon oil) were considered with similar viscosities. As detailed in Chapter 3, permeability of the fibers was calculated, and flow rates of the fluids with respect to fibers were also obtained. This section is divided into two parts: permeability measurements and saturated flow rate. 4.1 Permeability Measurement Permeability with Respect to Fibers The permeability values obtained were compared with Palmese and Karbhari [17] and showed similar trends for sized and unsized fibers. All cases except UD carbon fibers 2 showed higher permeability values for unsized fibers then sized fibers. Unidirectional Carbon Fibers 1 Unidirectional carbon fibers with polyurethane sizing (UD carbon fibers 1, see Table 1) were used in this study. Tests on these sized and unsized fibers were conducted using lowviscosity and high-viscosity corn syrup and silicon oil. Figure 7 shows a plot of the permeability variation with respect to the test fluids and fiber material. For fibers with sizing, low-viscosity corn syrup showed higher permeability than with high-viscosity corn syrup and silicon oil, whereas, high viscosity corn syrup and silicon oil with similar viscosities showed only a minor variation in permeability values with respect to each other possibly within experimental uncertainties. For all test fluids, unsized fibers had higher permeability values when compared 38

51 with sized fibers. It was interesting to observe that unsized fibers showed lower permeability values for low-viscosity corn syrup (LVCS) when compared to two other test fluids with higher viscosities. UD Carbon Fibers 1 Permeability Vs Test Fluids Sized Fibers 15.0 Unsized Fibers Permeability K1 (darcy) LVCS = Low Viscosity Corn Syrup HVCS = High Viscosity Corn Syrup SO = Silicon Oil LVCS HVCS SO Test Fluids Figure 7: Permeability of sized and unsized UD carbon fibers 1 with respect to test fluids. Unidirectional Glass Fibers Tests were conducted using the same three different test fluids over unidirectional glass fibers with silane as sizing (see Table 1). Tests were conducted on both sized and unsized fibers. Plotted permeability values are plotted in Figure 8. Similar to unidirectional carbon fibers 1, the unidirectional glass fibers with sizing showed higher permeability for low-viscosity corn syrup compared to the other two fluids. Variations in permeability were observed between sized and 39

52 unsized fibers in all cases, and the permeability of unsized fibers showed lower values for lowviscosity corn syrup then with the other two test fluids with higher viscosities. UD Glass Fibers Permeability Vs Test Fluids Permeability K1 (darcy) Sized Fibers Unsized Fibers LVCS = Low Viscosity Corn Syrup HVCS = High Viscosity Corn Syrup SO = Silicon Oil LVCS HVCS SO Test Fluids Figure 8: Permeability of sized and unsized UD glass fibers with respect to test fluids. Biaxial Glass Fibers Selection of biaxial glass fiber (see Table 1) was done to study not only the interaction between fiber sizing and test fluid, but also the effect of fiber architecture. All three test fluids were used to conduct tests on these fibers, which also used silane as sizing. Permeability values for both sized and unsized fibers were compared and are plotted in Figure 9. Similar to unidirectional carbon fibers 1 and unidirectional glass fibers, biaxial glass fibers with sizing showed higher permeability values with low-viscosity corn syrup than the other two test fluids. The permeability value for unsized fibers using high-viscosity corn syrup was lower than 40

53 permeability obtained for low-viscosity corn syrup. Silicon oil as test fluid over unsized fibers gave permeability five times higher than permeability obtained for sized fibers, whereas other test fluids gave approximately three times higher values. Bi-Axial Glass Fibers Permeability Vs Test Fluids Sized Fibers Unsized Fibers Permeability K1 (darcy) LVCS = Low Viscosity Corn Syrup HVCS = High Viscosity Corn Syrup SO = Silicon Oil LVCS HVCS SO Test Fluids Figure 9: Permeability of sized and unsized Biaxial glass fibers with respect to test fluids. Unidirectional Carbon Fibers 2 These carbon fibers (refer to Table 1) have tow size, which is higher than all other fibers used in this study, and epoxy as sizing. Test fluids were induced into the fibers, where high-viscosity corn syrup and silicon oil both needed a supplementary external pressure in addition to the vacuum to obtain a continuous flow. Tests were also conducted on these fibers after stripping off the sizing. Permeabilities of fibers both with and without sizing were obtained and are plotted in Figure 10. Permeability results obtained for these fibers were different from 41

54 those obtained for all other fibers. Unsized fibers show lower permeability values then sized fibers in this case for all three test fluids. Permeability values obtained by using high-viscosity corn syrup and silicon oil for sized and unsized fibers were closer when compared to other fibers. Permeability for these fibers with sizing increased from low viscosity corn syrup to highviscosity corn syrup while silicon oil and high-viscosity corn syrup permeabilities were similar; this was not the case for all other fibers. UD Carbon Fibers 2 Permeability Vs Test Fluids Sized Fibers Unsized Fibers Permeability K1 (darcy) LVCS = Low Viscosity Corn Syrup HVCS = High Viscosity Corn Syrup SO = Silicon Oil LVCS HVCS SO Test Fluids Figure 10: Permeability of sized and unsized UD carbon fibers 2 with respect to test fluids Permeability with Respect to Test Fluids To find the difference in permeability with respect to test fluids, high-viscosity corn syrup and silicon oil, both with similar viscosities, were considered. Test results for all four different fibers with high-viscosity corn syrup and silicon oil were compared and plotted. 42

55 Unidirectional Carbon Fibers 1 Figure 11 shows the permeability obtained for high-viscosity corn syrup and silicon oil used on sized and unsized unidirectional carbon fibers. Permeability varied with respect to sizing, but the permeability of fibers for both test fluids had almost similar values. UD Carbon Fibers 1 Permeability Vs Test Fluids Sized Fibers 15.0 Unsized Fibers Permeability K1 (darcy) HVCS = High Viscosity Corn Syrup SO = Silicon Oil HVCS SO Test Fluids Figure 11: Permeability of sized and unsized UD carbon fibers 1 with respect to HVCS and SO. Unidirectional Glass Fibers Permeability for unidirectional glass fibers with respect to high-viscosity corn syrup and silicon oil are plotted in Figure 12. As seen in the figure, the variation in permeability for the sized fibers using two test fluids was lower when compared to unsized fibers permeability. The permeability was 3.3 Darcy higher for the unsized UD glass fibers than the sized fiber when 43

56 HVCS was used, while the permeability for the unsized UD glass fibers was 24.9 Darcy higher than the sized fibers when SO was used. UD Glass Fibers Permeability Vs Test Fluids Sized Fibers Unsized Fibers Permeability K1 (darcy) HVCS = High Viscosity Corn Syrup SO = Silicon Oil HVCS SO Test Fluids Figure 12: Permeability of sized and unsized UD glass fibers with respect to HVCS and SO. Biaxial Glass Fibers The Biaxial glass fibers (see Table 1) with ± 45 orientation showed permeabilities with highest variation between sized and unsized fibers than any other fibers tested. The results are plotted in Figure 13, which indicate that the permeability for the sized fibers with both test fluids was similar when compared to the unsized fibers. Even when the sized and unsized fibers are compared, unsized fibers showed permeabilities three to five times higher than for high-viscosity corn syrup and silicon oil, respectively. 44

57 Bi-Axial Glass Fibers Permeability Vs Test Fluids Sized Fibers Unsized Fibers Permeability K1 (darcy) HVCS = High Viscosity Corn Syrup SO = Silicon Oil HVCS SO Test Fluids Figure 13: Permeability of sized and unsized biaxial glass fibers with respect to HVCS and SO. Unidirectional Carbon Fibers 2 Unidirectional carbon fibers 2 (see Table 1) with high tow size and epoxy as sizing gave completely different permeability values when compared to the other fiber test results when comparing sized and unsized results as seen in Figure 14. Unlike the other fibers tested, these fibers had higher permeability with sizing when compared to fibers without sizing. However, the permeability for unsized fibers was similar to sized fibers for the two test fluids with similar viscosities (HVCS and SO). 45

58 UD Carbon Fibers 2 Permeability Vs Test Fluids Sized Fibers Unsized Fibers Permeability K1 (darcy) HVCS = High Viscosity Corn Syrup SO = Silicon Oil HVCS SO Test Fluids Figure 14: Permeability of sized and unsized UD carbon fibers 2 with respect to HVCS and SO Permeability with Respect to Orientation To see the effect of fiber orientation versus permeability of the fiber mat, tests were conducted on biaxial glass fiber with and without sizing by changing the mat orientation in the mold. Orientations of ±45 and 0 /90 were considered in permeability measurements using high-viscosity corn syrup as the test fluid, as plotted in Figure 15. Permeability values for sized fibers were similar between ±45 and 0 / 90 orientations. Permeability for 0 /90 orientation unsized biaxial fibers was higher than that of ±45 orientation. Even a variation in permeability with respect to sizing can be observed between the orientations. It was concluded that not only test fluids were interacting with respect to sizing but also potential fiber architecture interaction with sizing was occurring as well. 46

59 Permeability of Bi-Axial Fibers at different Orientations using HVCS ±45º Orientation º& 90º Orientation 81.0 Sized Fibers Unsized Fibers Permeability K1 (darcy) HVCS = High Viscosity Corn Syrup HVCS HVCS HVCS as Test Fluid Figure 15: Permeability variations in Biaxial glass fiber at different orientations using HVCS. 4.2 Saturated Flow Rate To see the effect of a test fluid in flow-rate measurement, test fluids were made to flow through the fibers for a duration of 30 minutes, which is the average gel time for most resins. A constant vacuum of 27 Hg was maintained with the help of a digital vacuum regulator to induce these test fluids through unidirectional carbon fibers 1 and unidirectional glass fibers, each with a different sizing. The UD carbon fibers 1 had a polyurethane sizing, while the UD glass fibers had a silane sizing. A vacuum of 27 Hg was considered because of the water in diluted corn syrup which will start boiling at room temperature above that vacuum level and air bubbles will form. High-viscosity corn syrup and silicon oil with similar viscosities were infused through these fibers. The vacuum was kept constant throughout the experiment, and the flow rate was 47

60 calculated by weighing the outlet flow with respect to time using a weigh scale with an accuracy of 0.1 gm. The flow rates for both sized and unsized fibers were obtained. In a previous research study [9] the flow rate in RTM process showed three different stages: initial rapid flow, transient region, and steady-state flow, but in the VARTM process, only the initial rapid flow and steady-state flow stages were observed. In all cases using test fluids, a drop in flow rate was observed with respect to time Unidirectional Carbon Fibers 1 Flow Rate with Respect to Sizing Saturated flow rates for unidirectional carbon fiber 1 case using high-viscosity corn syrup are plotted in Figure 16 for both sized and unsized fibers. The amount of fluid flow through the unsized fibers was higher than through the sized fibers. For both sized and unsized fibers, there was an initial flow rate increase, but as the time increased, a drop in flow rate was observed. A high drop in flow rate duration of 30 minutes was observed for unsized fibers over the sized fibers using the same test fluid. To see the effect of time sensitivity, graphs were also plotted for time intervals of 15 and 30 seconds, which are plotted with respect to time and compared in Appendix E. Figure 17 shows the plot of flow rate versus time obtained for the same fibers, sized and unsized, using silicon oil as the test fluid. Flow rate variations from 1.5 to 2.0 ml/min were observed with respect to time for unsized fibers. In this case, the amount of fluid flow through the unsized fibers was higher than through the sized fibers for all flow rate averaged time intervals, as seen in Appendix E. The average drop in flow rate was almost similar for both the sized and unsized fibers using silicon oil as the test fluid. 48

61 Flow rate measurements for UD Carbon Fibers 1 using HVCS at 27" Hg Flow rate (ml/min) Sized Fibers Unsized Fibers Poly. (Unsized Fibers) Poly. (Sized Fibers) HVCS = High Viscosity Corn Syrup Time (min) Figure 16: Flow rate for UD carbon fibers 1 using HVCS at 27 Hg. Flow rate measurements for UD Carbon Fibers 1 using SO at 27" Hg Sized Fibers Unsized Fibers Poly. (Unsized Fibers) Poly. (Sized Fibers) 2.0 Flow rate (ml/min) SO = Silicon Oil Time (min) Figure 17: Flow rate for UD carbon fibers 1 using SO at 27 Hg. 49

62 Flow Rate with Respect to Test Fluid Two test fluids were compared with respect to each other for sized and unsized unidirectional carbon fibers 1. The flow rate vs. time results for sized fibers with HVCS and SO are plotted in Figure 18. It can be noted that HVCS and SO had similar viscosities, and flow through the same fibers were different. Silicon oil showed a higher flow rate throughout the experiment than HVCS. This difference may be due to other factors, such as wetting effects between the fluid and fibers. Both fluids showed a drop in flow rate with respect to time. Flow rate measurements for Sized UD Carbon Fibers 1 using HVCS and SO at 27" Hg Flow rate (ml/min) HVCS SO Poly. (SO) Poly. (HVCS) HVCS = High Viscosity Corn Syrup SO = Silicon Oil Time (min) Figure 18: Flow rate for sized UD carbon fibers 1 using HVCS and SO at 27 Hg. Flow rates for unsized unidirectional carbon fibers 1 using HVCS and SO is plotted in Figure 19. In this case the amount of fluid flow through the unsized fibers was higher than that of 50

63 sized fibers, but as in Figure 18 the silicon oil flow rate is higher than that of high-viscosity corn syrup over the entire test period. The fluctuation in the flow rate was higher for both test fluids when compared to the flow through sized fibers. Both fluids showed a drop in flow rate, but high-viscosity corn syrup had a higher flow rate drop than silicon oil. Flow rate measurements for Unsized UD Carbon Fibers 1 using HVCS and SO at 27" Hg HVCS SO Poly. (SO) Poly. (HVCS) Flow rate (ml/min) Time (min) HVCS = High Viscosity Corn Syrup SO = Silicon Oil Figure 19: Flow rate for unsized UD carbon fibers 1 using HVCS and SO at 27 Hg Unidirectional Glass Fibers Flow Rate with Respect to Sizing Saturated flow rate for unidirectional glass fibers was measured using HVCS and SO. Figure 20 shows a plot of the flow rate versus time of sized and unsized unidirectional glass fibers with HVCS as the test fluid. Unsized fibers show high flow rate versus time when compared to sized fibers. In both the cases, there is an initial rapid flow, but a high drop in flow 51

64 rate was observed with respect to time. The absolute flow rate drop for unsized fibers (3 ml/min) was higher than for sized fibers (1.5 ml/min), but percentage drop for both cases were in a similar range. Flow rate measurements for UD Glass Fibers using HVCS at 27" Hg Flow rate (ml/min) Time (min) Sized Fibers Unsized Fibers Poly. (Unsized Fibers) Poly. (Sized Fibers) HVCS = High Viscosity Corn Syrup Figure 20: Flow rate for UD glass fibers using HVCS at 27 Hg. Tests were conducted to see the affect of silicon oil over these unidirectional glass fibers with and without sizing, and the results are plotted in Figure 21. In this case, the flow rate for unsized fibers was higher than that of sized fibers. There was an initial rapid flow, and as time increased, there was a decrease in flow rate. The drop in flow rate for both the sized and unsized fibers was similar, but was less than the drop in flow rate over time using HVCS (Figure 20). 52

65 Flow rate measurements for UD Glass Fibers using SO at 27" Hg Flow Rate (ml/min) Sized Fibers Unsized Fibers Poly. (Unsized Fibers) Poly. (Sized Fibers) SO = Silicon Oil Time (min) Figure 21: Flow rate for UD glass fibers using SO at 27 Hg. Flow Rate with Respect to Test Fluid To see the effects of test fluids with unidirectional glass fibers, flow rate test results were compared with respect to test fluids for both sized and unsized fibers. Figure 22 shows the flow rate vs. time results for sized unidirectional glass fibers using both high-viscosity corn syrup and silicon oil, both with similar viscosities. It was observed that the silicon oil flow rate was higher throughout the flow than high-viscosity corn syrup. The drop in flow rate was higher for HVCS then for SO, but silicon oil showed high fluctuations within an interval. 53

66 Flow rate measurements for Sized UD Glass Fibers using HVCS and SO at 27" Hg Flow rate (ml/min) Time (min) HVCS SO Poly. (SO) Poly. (HVCS) HVCS = High Viscosity Corn Syrup SO = Silicon Oil Figure 22: Flow rate for sized UD glass fibers using HVCS and SO at 27 Hg. Unsized unidirectional glass fibers were also compared with respect to test fluids, and results are plotted in Figure 23. Flow rates vs. time with these test fluids in unsized fibers were close to that in sized fibers, but even in this case, flow rate drop for HVCS was higher than that for SO. 54

67 Flow rate measurements for Unsized UD Glass Fibers using HVCS and SO at 27" Hg Flow rate (ml/min) Time (min) HVCS SO Poly. (SO) Poly. (HVCS) HVCS = High Viscosity Corn Syrup SO = Silicon Oil Figure 23: Flow rate for unsized UD glass fibers using HVCS and SO at 27 Hg. 55

68 CHAPTER 5 CONCLUSIONS AND FUTURE WORK In this study, the effect of sizing, test fluid, and fiber interactions in in-plane permeability and flow-rate measurements were investigated. The VARTM process was used in conducting the tests, which considered four different continuous fiber systems used in industry and three different test fluids typically used in liquid composite molding research. Tests were conducted on fiber mats with and without sizing. Permeability was calculated using Darcy s Law and flow rate by weighing the outflow fluid with respect to time. Based on the results of permeability and flow rate with respect to fiber type, test fluid and sizing, the following conclusions were drawn. 5.1 Permeability Permeability values obtained for all fiber mats had high variations with respect to the three sizings and three test fluids used in this study. In particular the following was observed. All fibers showed higher permeability for unsized fibers than sized fibers, similar to the results obtained for Palmese and Karbhari [17], except for unidirectional carbon fibers 2, which showed permeability for sized fibers greater than or equal to unsized fibers. This may be due to the change in wetting angle between the test fluid and unsized fibers. An interesting observation was that the permeability of unsized fibers with low-viscosity corn syrup was less than that obtained with high viscosity corn syrup, except for biaxial glass fibers. It would be expected that for the same materials (constant permeability and constant porosity), the volumetric flow velocity should increase when the fluid viscosity is lower. However, this effect possibly may be due to the change in wetting characteristics. The difference in permeability results obtained for biaxial glass fiber may be due to the change in orientation compared to unidirectional fiber. 56

69 In all cases, unidirectional carbon fiber 2 results were completely opposite of the results obtained for the other fibers, i.e., K Sized > K Unsized for UD carbon fibers 2 and K Sized < K Unsized for other fibers. This may be due to the unique wetting characteristics of epoxy. Test fluids with similar viscosity showed different effects on sized and unsized fiber permeability as well as flow rate, with porosity due to other effects such as microscopic flow effects and wetting interactions. 5.2 Saturated Flow Rate Saturated flow rate was measured for two different fibers sized and unsized with HVCS and SO at 27 Hg vacuum. Viscosities for both test fluids were similar, and tests were conducted with the vacuum kept constant. The following conclusions were made. Unsized fiber mats showed higher flow rates than sized fibers mats for both fiber types, as predicted in the permeability results. High-viscosity corn syrup showed a higher drop in flow rate than silicon oil over the test period for both fiber types. This may be due to wetting interactions at microscopic level similar to Palmese s results [17] where microscopic infusion of fluid into the tows expanded the tows and reduced the macroscopic flow channels. Cumulative mass versus time for the RTM process obtained in research by Palmese have showed two stages: initial rapid flow and steady-state flow, but a representative cumulative mass versus time graph obtained in this study showed a linear plot (refer to Appendix D where an example of current test results was plotted and compared to Palmese s results). While using test fluids for permeability and flow rate measurements, fiber sizing, fiber orientation, test fluid type, and test fluid viscosity should be considered, but other factors 57

70 that may involve microscopic effects are involved in the process. This could include the effects identified by Palmese [17] such as, microscopic resin flow which expands the fiber tows and reduces the macroscopic flow channels. The flow rate in his experiments stabilized over time, but in the current study the flow rates for all experiments had not stabilized. Corn syrup as a test fluid should be used with caution for permeability measurements and flow rate measurements. 5.3 Future Work Additional tests should be conducted over a wide range of fibers and test fluids. An injection mold with more flow area could be considered in order to see the microscopic and macroscopic flow effects with respect to fibers and sizing. Wetting angle and capillarity effects should also be determined for test fluids with sized and unsized fiber types. 58

71 REFERENCES 59

72 LIST OF REFERENCES [1] free online encyclopedia. [2] Information from High-Performance Composites, Composites Technology, and Sourcebook from [3] Wang, T. James, Wu, C.H., and Lee L. James, In-plane permeability measurement and analysis in liquid composite molding, Polymer Composites, Vol. 15, No. 4, 1994, pp [4] Correia, N.C., Robitaille, F., Long, A.C., Rudd, C.D., Simacek, P., and Advani, S.G., Use of resin transfer molding simulation to predict flow, saturation, and compaction in the VARTM process, Journal of Fluids Engineering, Transactions of the ASME, Vol. 126, No. 2, 2004, pp [5] Abrate Serge Resin flow in fiber performs, Applied Mechanics Reviews, Vol. 55, No. 6, 2002, pp [6] Ferland Pierre, Guittard Dominique, and Trochu François, Concurrent methods for permeability measurements in resin transfer molding, Polymer Composites, Vol. 17, No. 1, 1996, pp [7] Shih Chih-Hsin and Lee L. James Effect of fiber architecture on permeability in liquid composite molding, Polymer Composites, Vol. 19, No. 5, 1998, pp [8] Shojaei, A., Trochu, F., Ghaffarian, S.R., Karimian, S.M.H., and Lessard, L., An experimental study of saturated and unsaturated permeabilities in resin transfer molding based on unidirectional flow measurements, Journal of Reinforced Plastics and Composites, Vol. 23, No. 14, 2004, pp [9] Karbhari, V.M., and Palmese, G.R., Sizing related kinetic and flow considerations in the resin infusion composites, Journal of Materials Science, Vol. 32, No. 21, 1997, pp [10] Baiju Z. Babu and Pillai Krishna M. Experimental investigation of the effect of fibermat architecture on the unsaturated flow in liquid composite molding, Journal of Composite Materials, Vol. 38, No. 1, 2004, pp [11] Steenkamer, D.A., Wilkins, D.J., and Karbhari, V.M., Influence of test fluid on fabric permeability measurements and implications for processing of liquid moulded composites, Journal of Materials Science,, Vol. 12, No. 13, 1993, pp [12] Saraswat Manoj, Heider Dirk, and Song Young Ultrasonic monitoring of the void formation during VARTM infusion 60

73 [13] Bender Dominik, Schuster Jens, and Heider Dirk Flow rate control during vacuumassisted resin transfer molding (VARTM) processing, Composites Science and Technology, Vol. 66, No. 13, 2006, pp [14] Steenkamer, D.A., McKnight, S.H., Wilkins, D.J., and Karbhari, V.M., Experimental characterization of permeability and fiber wetting for liquid moulding, Journal of Materials Science, Vol. 30, No. 12, 1995, pp [15] Lee, Y.J., Wu, J.H., Hsu, Y., and Chung, C.H., A prediction method on in-plane permeability of mat/roving fibers laminated in vacuum assisted resin transfer molding, Polymer Composites, Vol. 27, No. 6, 2006, pp [16] Karbhari, V.M., and Palmese, G.R., Effect of fiber sizing on processing and performance of S2-glass composites, International SAMPE Technical Conference, Vol. 27, 1995, pp [17] Palmese, G.R., and Karbhari, V.M., Effects of Sizings on Microscopic Flow in Resin Transfer Molding, Polymer Composites, Vol. 16, No. 4, 1995, pp [18] Pillai Krishna M. Modeling the unsaturated flow in liquid composite molding processes: A review and some thoughts, Journal of Composite Materials, Vol. 38, No. 23, 2004, pp [19] Soschinske, Kurt A. Cellular automata simulation of resin flow through a fiber reinforcement, Ph.D. Dissertation, Department of Mechanical Engineering, Wichita State University, Wichita, KS, [20] Ken Han, K., William Lee, C., and Brian P. Rice Permeability measurements of fiber preforms and applications, American Society of Mechanical Engineers, Materials Division (Publication) MD, Vol. 84, Mechanical Behavior of Advanced Materials, 1998, pp [21] Mastbergen Daniel B. and Cairns Douglas S., Process modeling and validation of resin infusion type manufacturing of composite wind turbine blades, Collection of Technical Papers - 44th AIAA Aerospace Sciences Meeting, Vol. 19, pp [22] Sharma, S., A VARTM process for the fabrication of hybrid laminated composite wind turbine blade, Master s Thesis, Department of Aerospace Engineering, Wichita State University, Wichita, KS, [23] National Institute of Standards and Technology data base on selection of test fluids for permeability measurements. [24] Hunston, D., Phelan, F., and Parnas, R., Flow behavior in liquid molding, NASA Conference Publication, Issue 3176, 1992, p

74 APPENDICES 62

75 APPENDIX A FABRIC CONSTITUTION [22] Saertex style number: S32CU oz/yd 2 Unidirectional Carbon Fabric 1 0 carbon fibers (24,000 tows) of nominal areal weight 502 g/m 2 (14.81 oz/yd 2 ) 90 PES fibers of nominal weight 4 g/m 2 (8 tex) +60 E-glass fibers of nominal areal weight 6 g/m 2 (68 tex) -60 E-glass fibers of nominal areal weight 6 g/m 2 (68 tex) The four layers are stitch bonded with nominal 6 g/m 2 PES fibers. Saertex style number: S15EU oz/yd 2 Unidirectional Glass Fabric 0 E-glass fibers(1,200 tows) of nominal areal weight 528 g/m 2 (15.57 oz/yd) 90 E-glass fibers(68 tows) of nominal weight 35 g/m 2 (1.59 oz/yd 2 ) The two layers are stitch bonded with nominal 18 g/m 2 PES fibers. Saertex style number: U32EX oz/yd 2 Double Bias (±45 ) Glass Fabric +45 E-glass fibers of nominal areal weight 200 g/m 2 (5.90 oz/yd 2 ) -45 E-glass fibers of nominal areal weight 200 g/m 2 (5.90 oz/yd 2 ) The two layers are stitch bonded with nominal 6 g/m 2 PES fibers. V2 composite reinforcements 13.3 oz./yd 2 Unidirectional Carbon Fabric 2 (68,000 tows) 63

76 APPENDIX B FIBER VOLUME FRACTION [22] 64

77 APPENDIX C PERMEABILITY CALCULATIONS FOR SIZED FIBERS 65

78 APPENDIX C PERMEABILITY CALCULATIONS FOR UNSIZED FIBERS 66