Zeitschrift Kunststofftechnik Journal of Plastics Technology

Size: px
Start display at page:

Download "Zeitschrift Kunststofftechnik Journal of Plastics Technology"


1 Zeitschrift Kunststofftechnik 4Autor Journal of Plastics Technology Titel (gegebenenfalls gekürzt) archivierte, peer-rezensierte Internetzeitschrift archival, peer-reviewed online Journal of the Scientific Alliance of Plastics Technology (2014) 4 eingereicht/handed in: angenommen/accepted: Reinhold Meier, Dr. Andrew Walbran, Christoph Hahn, Swen Zaremba, Prof. Dr.-Ing. Klaus Drechsler Institute for Carbon Composites, Technische Universität München, Faculty of Mechanical Engineering, Boltzmannstraße 15, D Garching b. München, Germany Methoden zur Bestimmung der Permeabilität von Verstärkungstextilen Die Permeabilität trockener Verstärkungstextilien ist eine wichtige Materialeigenschaft, die zu einem besseren Verständnis des Infiltrationsprozesses in Flüssigimprägnierverfahren beiträgt. Zur Bestimmung der Permeabilität textiler poröser Medien sind verschiedene Methoden bekannt, jede mit ihren Vor- und Nachteilen. In diesem Aufsatz werden Prüfstände zur Bestimmung der gesättigten und ungesättigten Permeabilität vorgestellt, welche den Prinzipien für eindimensionalen (1D) und radialen (2D) Fluss folgen. Neben den Messergebnissen werden der dazu notwendige Zeit- und Materialeinsatz der verschiedenen Methoden zur Bestimmung der Permeabilität in Bauteilebene verglichen. Außerdem wird ein Simulationsansatz zur Bestimmung der Permeabilität vorgestellt, welcher auf der digitalen Abbildung des Materials mit einem Scanner beruht. Methods to Determine the Permeability of Textile Reinforcements Reinforcing textile permeability is an important material property used to better understand the infiltration phase of Liquid Composite Molding processes. A range of methods exists to determine the permeability of textile porous media all with their respective advantages. In this work, facilities to characterize the saturated and unsaturated in-plane permeability using rectilinear (1D) or radial flow (2D) methods are presented. A comparison of the in-plane permeability results obtained using each test method was carried out, together with the required testing time and material usage. Furthermore, a simulation approach to predict the permeability based on scanned images is presented. Carl Hanser Verlag Zeitschrift Kunststofftechnik / Journal of Plastics Technology 10 (2014) 4

2 Methods to Determine the Permeability of Dry Textile Reinforcements R. Meier, A. Walbran, C. Hahn, S. Zaremba, K. Drechsler Reinforcing textile permeability is an important material property used to better understand the infiltration phase of Liquid Composite Molding processes. A range of methods exists to determine the permeability of textile porous media all with their respective advantages. In this work, facilities to characterize the saturated and unsaturated in-plane permeability using rectilinear (1D) or radial flow (2D) methods are presented. A comparison of the in-plane permeability results obtained using each test method was carried out, together with the required testing time and material usage. The results of the two in-plane methods are in good agreement. The time required to fully characterize a material using the 1D method was more than three-times the time required when using the 2D method. The material used for testing with the 1D method was also three times greater than for the 2D method. Furthermore, a simulation approach to predict the permeability based on scanned images is presented. 1. INTRODUCTION AND MOTIVATION In Liquid Composite Molding (LCM) processes, a textile reinforcement structure is placed in a mold cavity and is impregnated with mostly thermoset resins [1]. After complete infiltration of the porous structure and curing of the resin the part is demolded. The quality and the mechanical properties of the part are defined by the quality of the infiltration process and the degree of cure of the matrix material. Porosity or dry spots have therefore to be avoided. In LCM processand tool-design, the permeability of the fibrous reinforcement is a very important parameter to consider. Theoretically, permeability is a geometric property and quantifies the hydraulic conductivity of porous media to fluid flow. One widely accepted model to describe the impregnation step in LCM processes is Darcy s law, where the permeability of the textile reinforcement together with the dynamic viscosity of the liquid matrix material represent the characteristics of the materials for fluid flow [2]. The most straightforward use for permeability measurement is the determination of material data to design the filling process of complex parts. The results can be used to predict process properties, such as fill time and injection pressures. In addition, tool design and part quality can be optimized and flow in areas with complex fiber architecture (such as T-junctions and overlapping areas) can be simulated. In series production processes, Journal of Plastics Technology 10 (2014) 4 91

3 permissible variability from the ideal handling- and processing conditions influencing the permeability can be derived, resulting in reduced scrap, lower process time and energy consumption. For example, the effects of varying binder content and variations in the corresponding activation process (activation temperature and time) on the permeability can be quantified [3] and the influence on the filling process can be predicted. As a standard test method for evaluating permeability currently does not exist, a number of different methods have been developed [4]. Experimental and simulation-based approaches are the most promising, in contrast to analytical approaches such as the Kozeny-Carman equation, to capture the effects in complex parts with respect to accuracy and repeatability. Both approaches are presented in this paper. In terms of the experimental methods, measurement results of the most commonly used test methods are presented and compared with respect to time and material effort. 2 THEORY OF PERMEABILITY MEASUREMENT The permeability of a porous material is an inverse measure of the resistance to fluid flow through that material, thus high permeability values account for low resistance and vice versa. In theory, permeability is dependent only on geometrical quantities of the fibrous reinforcement such as the fiber volume fraction V f, the reinforcement architecture (type of textile and degree of shear), the number of layers and the degree of saturation. The permeability of fiber reinforcement typically varies with direction and is commonly described by Darcy s law for flow through porous media [5]: where is the fluid velocity vector, is the dynamic viscosity of the fluid, is the pressure gradient and is the permeability tensor of the material. While Darcy s law assumes fully saturated flow in the porous media, typical LCM processes involve an unsaturated, wetting flow front. However many authors have used Darcy s law to model this case [6]-[9]. Darcy s law also assumes that the fluid is Newtonian and of constant viscosity, whereas resins used in composites manufacturing are distinctly non-newtonian and the viscosity can change during processing. For the general three-dimensional case, permeability is a symmetric positive definite 2nd order tensor [10]-[11], and can therefore be diagonalized. The major permeability tensor can be graphically described as an ellipsoid. In practical applications, especially for thin-walled structures, it is common to simplify the permeability tensor and devide it into an in-plane tensor, which has the shape of an ellipse that is defined by the two principal permeabilities, K 11 Journal of Plastics Technology 10 (2014) 4 92 (1)

4 and K 22, and one out-of-plane or through-thickness permeability, K 33. It has been shown previously that K 33 is typically one or more orders of magnitude lower than K 11 and K 22 [12]. 3 PRINCIPLES OF PERMEABILITY MEASUREMENT Permeability test methods can be divided into three categories: analytical, experimental and simulation approaches. There are various analytical methods available to determine permeability in a fast manner, such as the Kozeny-Carmann equation [13]. However, all analytical approaches have the drawback that they rely on a certain model assumption which does not represent reality in terms of material type and preform architecture. Nevertheless the Kozeny-Carman equation is frequently cited in papers on textile permeability and is often used by textile engineers as a rule of thumb [14]. Furthermore, for an interpolation of experimental results with varying V f, the Kozeny-Carman equation gives promising results for V f s close to the measured ones [15]. The work of the First Permeability Benchmark Exercise [4] shows that two experimental methods are most frequently used to determine the in-plane permeability, the rectilinear (1D) and the radial (2D) flow method, as shown schematically in Figure 1. These methods both neglect flow in the thickness direction of the preform, as the thickness of composite laminates is typically orders of magnitude lower than the in-plane dimensions. Both methods have in common that a Newtonian fluid is injected into the porous fibrous media whereupon the governing flow direction lies within the sample plane. The flow velocity is determined together with the applied injection pressure and the pressure at the flow front (the borderline between wet and dry textile). The flow front pattern generally is of linear and elliptical shape for the rectilinear and radial method, respectively. In the case of isotropic materials, with respect to the in-plane permeability properties, a circular flow front pattern is observed in the radial method. Journal of Plastics Technology 10 (2014) 4 93

5 a) b) Figure 1: Schematics of the rectilinear (a) and radial (b) filling scheme for evaluation of in-plane permeability values. Flow velocity can be determined by monitoring the flow front evolution (for unsaturated measurements) or the mass flow of the test fluid (for saturated measurements).the saturated permeability is consistent with the assumptions of Darcy s law. The difference between saturated and unsaturated permeability values mainly results from capillary effects which contribute to the driving forces of the flow in addition to the applied pressure difference between the inlet and outlet. However, other phenomena e.g. geometrical rearrangements of the fiber bundles or air entrapments can also lead to differences between unsaturated and saturated permeability. The flow front position can, for both methods, be monitored using a clear mold half, often manufactured from perspex, polycarbonate or glass and reinforced to counter the low bending stiffness of such materials. This is necessary to avoid local bending of the clear mold half which affects the cavity thickness on a local scale, leading to a varying V f and as a consequence to inaccurate measurements. Alternatively, dielectric sensors [16], pressure transducers [17], optical fibers [18], ultrasonic transducers [19], or thermocouple sensors [20] can be used to monitor flow front progression for permeability characterisation. The saturated rectilinear flow method can also be used to determine the through-thickness permeability, K 33. Fluid is forced through the thickness of the textile reinforcement and the mass flow is measured to evaluate the average flow velocity according to Equation 1. Other than the interactions between the applied pressure difference and the resulting preform compaction, the principles are the same as for the 1D in-plane method. These three experimental methods are explained more detailed in the following sections, which discuss the testbenches available at the Institute for Carbon Composites (LCC). Journal of Plastics Technology 10 (2014) 4 94

6 3.1 1D In-Plane Facilities For rectilinear in-plane flow measurements, a rectangular sample is assembled, placed in a mold and Newtonian fluid is injected along one short edge of the preform to ensure in-plane fluid flow. The cavity height is determined by the thickness of a spacer frame which is mounted to the lower mold before sample loading. The minimum frame thickness is 2 mm. At the LCC, two separate 1D in-plane setups are available. These differ in the way the flow front position is detected. In one case the flow front is tracked optically (compare Figure 1) through a transparent mold half made of polycarbonate and in the other case by pressure transducers. The latter is necessary as this mold is manufactured from 35 mm thick aluminium to reduce mold deflection and thus local variations of V f (compare Figure 3). The advantage of the rectilinear flow method over the radial flow method is that unsaturated and saturated flow experiments can be conducted. Saturated measurements fulfil the assumptions of Darcy s law. In this case the flow velocity is calculated based on the measured mass flow through the sample. Pressure pot Thermocouple U T U P Camera Mold Preform Pressure transducer Journal of Plastics Technology 10 (2014) 4 95 D Data acquisition unit U P U T Outlet Figure 2: Schematic of the single-cavity 1D test setup with a clear upper mold and a video camera for optical flow front tracking. The rectilinear flow method is very sensitive to race-tracking of the test fluid along the edges of the sample. Race-tracking is the preferential flow of fluid along the higher permeability regions at the edges of the sample which is further promoted by samples not fitting the mold perfectly [21]-[25]. These effects are overcome by placing silicon strips between the side edges of the fibrous preform and the cavity. The silicon strips are slightly thicker (approximately 10 %) than the spacer frame which is used to adjust the desired thickness of the sample. When the cavity is closed, the silicon deforms and fills the gaps at the edges of the preform, eliminating race-tracking. Furthermore, the silicon helps to avoid fiber squeezing between the frame and the mold halves which could lead to deviations in the V f and leakage of the cavity. To determine the major in-plane permeability tensor with the 1D method, three Scale

7 experiments must be conducted in order to identify the three unknown quantities the principle permeabilities K 11 and K 22 and the rotation angle (θ) between the major axis of the ellipse and the warp and weft directions of the textile. The 1D facilities respect the recommendations of the 2nd Permeability Benchmark Exercise [26] in which the LCC took part. The goal of this roundrobin study was to determine the accuracy of the unsaturated 1D permeability measurement method. Identical textile material was sent to 13 academic laboratories around the world from which the unsaturated in-plane permeability tensor for a defined V f and number of layers was determined. One outcome of the exercise was that the variability of the principle permeability values including the ellipse-orientation was below ± 20 % when using the least square fit method for analysis. This method is also incorporated in LCC s analysis tools for the 1D setups. Compared to the results of the other laboratories, the LCC values are in the middle of the range. The variability of the LCC values is around ± 15 % for K 11, ± 17.5 % for K 22 and ± 22.5 % for the orientation of the in-plane tensor. Figure 3: Four-cavity setup that allows four parallel measurements of in-plane permeability values. To reduce the time effort for determining in-plane permeability values with the rectilinear flow method a four-cavity setup was developed at the LCC which allows four parallel measurements. The inlets of the four cavities are connected to the same fluid reservoir providing the same injection pressure. Each cell is equipped with its own thermocouples as well as pressure and force transducers to evaluate the mold temperature as well as the unsaturated and saturated permeability of four different preforms at the same time. Measurement of the mold temperature is required to determine the fluid viscosity based on the average temperature of the test. Race tracking can be detected with pressure sensors at the edges of the sample to guarantee validity of results. All experimental process data (the pressure at the inlet and outlet, mold temperature, mass flow and flow front arrival at the pressure sensors) is gathered by a data acquisition system and is automatically analyzed after the test. As a result, a measurement protocol is generated containing the most relevant data together with a plot of the principle in-plane permeability tensor. Journal of Plastics Technology 10 (2014) 4 96

8 3.2 2D In-Plane Facility The radial-flow in-plane permeability evaluation tool is presented in Figure 4. In addition to permeability values, this test facility allows the measurement of the through-thickness compaction behavior of the textile reinforcement which is important to calculate the clamping force in many LCM processes, such as resin transfer molding (RTM), RTM Light and Compression RTM or to adjust a specific V f in vacuum bag processes. It consists of a lower glass platen with an upper aluminum platen. The facility is mounted in a universal testing machine, providing accurate cavity thickness and mold closure control. The V f of the sample can be adjusted continuously and the compaction response can be investigated as a function of the part thickness and mold closure velocity. The glass platen is mounted in a frame above a camera which is used to record the flow front position as discrete images at a set time interval (typical values are between 3 and 10 s), allowing the flow front speed to be calculated. Test fluid is injected via a central hole in the top platen. The injection pressure at the inlet is measured, and laser sensors monitor the cavity thickness during injection. All experimental data is gathered by a data acquisition system and is automatically analyzed after the test. Laser sensor Fluid injection Locking alignment unit Pressure sensor Upper mould platen Sample Lower glass mould platen a) b) Figure 4: 2D in-plane permeability and through thickness compactions measurement facility: a) installed in testing machine, b) schematic D Out-of-Plane Facility The 1D out-of-plane setup allows the determination of the saturated permeability in through-thickness direction of the preform by measuring the Journal of Plastics Technology 10 (2014) 4 97 Frame Camera

9 mass flow through the compacted fibrous textile via a force transducer connected to the outlet. Flow and compaction of the textile material in the thickness direction is enabled by the application of perforated plates. The fluid pressure at the in- and outlet is monitored by pressure transducers. In contrast to the in-plane flow method, the applied pressure difference leads to an additional preform compaction thus increased V f and decreased permeability. As a consequence, the measured permeability value depends on the applied pressure difference. Although the edge permeability has a minor effect on the bulk permeability of the sample [23], an o-ring is placed between the preform and the spacer frame to reduce the influence of race tracking on the measured permeability. The minimum cavity height is 2 mm and is determined by the thickness of the spacer frame. Journal of Plastics Technology 10 (2014) 4 98 Scale U P Preform Perforated plate U P U Data acquisition unit T U T Pressure pot a) b) Figure 5: a) 1D through-thickness permeability measurement facility, b) schematic of facility. 4 MEASUREMENT PROCEDURE The general procedure for any permeability test involves five main steps; sample preparation, sample loading, mold closure, infiltration and finally analysis. This is common for in-plane rectilinear and radial as well as out-ofplane testing, however each method has specific details which will be discussed in the sections Sample preparation The samples must be cut from the material roll. Generally there are several cutting procedures; manual cutting with a roller cutter or a knife, stamping in a press and an automated CNC cutter. For 1D measurements, the quality and accuracy of the cut edge is very important due to the sensitivity to race-tracking of this method. For this reason, the faster methods (stamping and CNC cutting) might not be appropriate for some materials as for example UDs or satin

10 weaves. The problem in these cases often is the rough surfaces of the underlay in which filaments easily get caught and as a consequence outer rovings are pulled out of the sample. This issue is less of a concern in manual cutting since smooth glass plates can be used as underlay. For 2D measurements edge quality is less important since the flow front does not reach the outer edge of the preform. The injection holes to ensure in-plane flow in the middle of samples for radial flow measurements are always stamped. The sample orientation must also be considered. For radial flow experiments, each layer within a sample must be oriented correctly. In the case of rectilinear experiments it is also necessary to test samples in different orientations to calculate the in-plane permeability tensor. Therefore, consistency between the samples with respect to fiber orientation must also be maintained. Before the fiber stacks are transferred to the mold, each sample is weighed. This data is required to calculate the actual V f and is a measure of the material variability and the cutting quality achieved. Sample loading When the layers are transferred into the mold, distortion of the layers must be avoided. Furthermore, the samples must be positioned in a repeatable manner. Here, spacer blocks between the mold edge and the sample edge are used. Mold closure During mold closure, distortion and displacement of the layers must be avoided, in particular of the upper most layer of each sample. The stack is compacted in the thickness direction without in-plane displacement of single layers until the desired fiber volume fraction is reached. Infiltration Before infiltration the test fluid should have the same temperature as the mold and the samples. This can most easily be achieved by storing the test fluid, the mold and the samples in the same room. The infiltration pressure must be low enough to avoid fiber washing or distortion of the preform due to the fluid stream. Furthermore, mold deflection is lower at lower injection pressures resulting in a more homogeneous V f of the sample. For tests with constant volume flow and tests with constant injection pressure the fluctuation of these values should be reduced to a minimum in order to measure reliable permeability values especially when the squared flow front approach is utilized in the analysis process of the rectilinear flow method [26]. Analysis To achieve high reproducibility, automated data acquisition and analysis processes are suggested, always following the same criteria e.g. when the flow front has reached a certain position in the unsaturated 1D rectilinear measurement. Journal of Plastics Technology 10 (2014) 4 99

11 4.1 1D In-Plane Testing For measurement of the unsaturated and saturated in-plane permeability, samples of 400 mm x 200 mm are placed in the lower mold platen between two silicon stripes to avoid race tracking. Larger samples are less sensitive to possible perturbations such as race-tracking effects or defects in the raw material. However, with increasing sample size the measurement time and material usage also increases. The aspect ratio of the sample geometry is also important in the case of rectilinear measurements to ensure one dimensional flow even for highly anisotropy textile lay-ups [27]. A minimum number of five layers per sample is suggested, depending on the areal weight of the material, to reduce the influence of the outer most layers on the final permeability value. Having direct contact to the rigid mold halves, these layers are compacted differently compared to the other layers and therefore will influence the permeability of thin fiber stacks. Smaller cavity heights than 2 mm lead to extensive deformations of the spacer frame which compresses the o-ring between the lower mold and the frame. Too high deformations lead to race tracking and as a consequence to incorrect results. Hence, depending on the areal weight of the fabric, a minimum number of layers is needed for the 1D inplane method. To determine the permeability of thin preforms or even single layers, the 2D radial flow method is suggested. When sample loading is completed, the upper platen is placed on the fiber stack with the aid of aligning pins and the bolts are tightened to 50 Nm. The unsaturated measurement starts when fluid starts to fill the linear inlet and ends when the flow front reaches the edges of the sample. As soon as no bubbles are flowing out of the outlet and a constant mass flow is measured, the saturated measurement begins and lasts for approximately five minutes. The duration of the unsaturated measurement depends on the permeability of the textile investigated, the applied injection pressure (usually 1 bar) and the fluid viscosity (approximately between mpas for the applied sunflower oil); typical times are around 25 minutes per sample. This procedure must be undertaken with three samples oriented at 0º, 45º and 90º from the warp direction to calculate the in-plane permeability tensor. For statistical purposes a minimum of three repeats of each test is undertaken. The raw data is automatically analyzed using a Matlab-based analysis tool. At the end of the analysis an Excel-based result sheet is created summarizing the important data of the investigated material D In-Plane Testing For the radial permeability and through-thickness compaction test facility, the sample size is 280 mm x 280 mm, with a 250 mm diameter test surface defined by the upper mold platen. Samples can be comprised of any number of layers, however six layers is typical. Consistence in orientation of the single layers of one stack must be ensured. A 15 mm diameter hole is punched in the center of Journal of Plastics Technology 10 (2014) 4 100

12 each sample to enforce two-dimensional in-plane fluid flow. Material variations close to the inlet hole have to be avoided since they severely influence the measured permeability as the pressure gradient is highest there [27]. For this reason, the analysis program considers only the last third of the images recorded when assessing the final permeability. After sample loading, the upper platen is lowered until it just contacts the fiber stack. The compaction test is then started, beginning with a constant-speed compaction to the desired cavity thickness and hence target V f. The permeability test begins 60 s after the cavity thickness is achieved. This is to allow the majority of relaxation of the compaction stresses to occur, giving a more consistent condition for the fluid flow where rearrangement of the fiber bundles should be avoided. Fluid is injected until the flow front reaches the edges of the upper platen. Typical testing time is around 8-10 minutes per sample, depending on the permeability of the sample, the injection pressure (usually between 1 5 bar) and the viscosity of the test fluid (approximately mpas for the applied sunflower oil). For statistical purposes a minimum of three repeats of each test is undertaken. After completion of the testing, the raw data is automatically analyzed using an Matlab tool developed at the University of Auckland [28]. The tool generates the flow front pattern based on a grey scale analysis of the images recorded from the camera during injection (compare chapter 3.2), fits the ellipse and finally calculates the principle permeabilities and the orientation angle D Out-of-Plane Testing The samples for determining the saturated through-thickness permeability have a diameter of 130 mm. Corresponding to the ASTM standard D5493 Standard Test Method for Permittivity of Geotextiles Under Load a minimum diameter of 50 mm is suggested to minimize the influence of hydraulic edge-effects. Circular samples have the advantage that no care must be taken when considering the orientation during cutting. The correct sample orientation can easily be adjusted during layup in the mold as long as no alternating stacking e.g. 0/90 or +-45 is needed. In this case, the whole preform is stacked outside the mold before cutting. The samples are manually placed on the lower perforated plate. The diameter of the o-ring which seals the preform against the spacer frame is slightly (approximately 10%) larger than the desired cavity thickness. This seal reduces the influence of edge effects especially for low permeability preforms. The top perforated plate is placed on top of the fiber stack. The perforation patterns coincide with each other. The mold is closed and the screws are tightened to 11 Nm. Once the desired injection pressure is adjusted in the pressure pot, the saturation process starts which is considered completed as soon as no bubbles can be observed in the outlet hose and a constant volume flow is measured with the force transducer. The volume flow is monitored for approximately two Journal of Plastics Technology 10 (2014) 4 101

13 minutes together with the pressure in the upper and lower fluid chamber of the setup and the mold temperature. After this period, the injection pressure is increased as the applied fluid pressure leads to an additional compaction of the preform thus to a decreasing permeability [29], [30]. This relationship between compaction state and permeability must be taken into account when flow processes in the through-thickness direction are considered. For statistical purposes a minimum number of three repeats of each test is undertaken. 5 SIMULATION APPROACH Another approach to determine the permeability of sheared and compacted preforms is to use simulation techniques. The method used at the LCC is based on fabric images, image processing and textile modeling [32]. The core of the approach is comprised of algorithms for image processing conducted on images obtained using a scanner together with a transparent compaction mold. This technique provides the advantage of fast and repeatable permeability determination and abstains from time- and material-consuming flow experiments. Figure 6 illustrates the data flow of the simulation approach. Figure 6: Data flow for the simulation approach of permeability prediction. Digital images of the scanned fabric, information such as lay-up thickness and desired fiber volume fraction, together with some numerical parameters for the embedded third party tools WiseTex and FlowTex developed at KU Leuven [33], [34] are required input for the simulation approach. The outcome is a material card that can be directly imported to the desired RTM solver of the Journal of Plastics Technology 10 (2014) 4 102