Effect of inhomogeneous temperature distribution on the impregnation process of the continuous compression molding technology

Abstract

The demand for continuous fiber-reinforced thermoplastics has steadily increased during the last years due to their specific properties. But considering the applications in large-scale production, the price for the organic sheets is still too high in order to compete with the metal counterparts. A starting point for the reduction of the total costs is the acceleration of the impregnation process by in-plane polymer flow. For this reason, this scientific report provides an insight into the production process using a continuous compression molding machine. In order to analyze the in-plane polymer flow and its driving forces, a method for the evaluation of the pressure distribution is presented in a first step. The examination revealed a significant inhomogeneous pressure distribution for the whole pressing area, which could be affected by different temperature settings. Afterward, the impregnation quality was correlated with the different settings, followed by the illustration of a huge potential for increasing the impregnation speed.

Introduction

The demand for continuous fiber-reinforced thermoplastic composites has steadily increased due to their specific properties, such as high impact and delamination strength, good chemical resistance, and unlimited shelf life. ¹ These materials are used for versatile applications in different industries, for example, as sole or ski reinforcements in sport industry, bumper beams in the automotive industry, or leading edge in the aeronautic industry. Hence, thermoplastic composites offer the advantage of short cycle times for the forming process due to the use of semifinished products, the so-called organic sheets. But having a look at the price, these organic sheets often cannot compete with their metal counterpart. One reason for this disadvantage is the high process costs due to low output of the production processes. In order to improve the economic situation, optimization of the production process involving big cost-saving potential is necessary.

For the production of organic sheets, there are mainly three different types of processes available. These are the static press, semicontinuously working, and continuously working processes. ² Each process has a unique advantageous field of application. A static press is favorable for small numbers of parts. For the cost-efficient large-scale production, either semicontinuous or continuous manufacturing method is a reasonable possibility. But commonly in all the methods the output is restricted by the necessary impregnation time, which is determined by a dramatically high melt viscosity of the thermoplastic matrix material. In order to improve this, some efforts have been made to optimize the used thermoplastic material with regard to viscosity, for example, by the development of reactive thermoplastic materials. ^{3 –5} But in contrast to thermoset production methods, no significant process improvements have been made for thermoplastic composite materials. The general rule in practice is that higher temperature (=lower viscosity) and higher pressure leads to reduced impregnation time.

The basic idea behind this research study is the investigation of an accelerated impregnation process caused by specific process parameters, which can create different shaped flow fronts of liquid polymer inside the tool. This might lead to in-plane and through-the-thickness impregnation simultaneously. Different theoretical analysis have revealed that the permeability of a fiber bundle is up to 200 times higher for a polymer flow parallel to the fibers compared to an orthogonal flow direction. ^6,7 Thus, in-plane polymer flow provides the potential for an accelerated impregnation. In order to use this potential, a fundamental analysis of the impregnation process with respect to the existing process conditions is necessary.

State of the art

Thermoplastic impregnation process

For the thermoplastic impregnation process, it is mostly assumed that there are two separated areas of fiber and matrix material. With regard to examinations carried out in the past by numerous authors, ^{8 –12} the matrix flow velocity $\left(\overline{v}\right)$ can then be described for an ideal reinforcement textile on the basis of the D’Arcy’s law for fluid flow:

$\overline{v}=-\frac{K}{\eta }\times \frac{\partial P}{\partial x},$

where K is the permeability of the fiber bed, η is the viscosity of the polymer matrix, and ∂P/∂x is the pressure drop with regard to the distance x. In contrast to this theoretical approach, there are some differences in reality. The permeability is changing during the whole impregnation process, that is, the permeability is not constant. This effect has been examined, for example, by Michaud and Manson ¹² : After the polymer is molten, the reinforcement fibers are compressed by the application of external pressure P _m (see Figure 1). Subsequently, the matrix material starts to infiltrate the fibers and the compaction of the textile is reduced. After the saturation of the fiber bed is reached, the impregnation is completed, but the fibers are not yet homogeneously distributed across the laminate thickness. Thus, fiber bed relaxation occurs. In her study, Michaud and Manson furthermore revealed a conflict between relaxation and impregnation of the reinforcement fibers with regard to the applied pressure. ¹² She suggests useful settings between both extreme points. Thus, a higher process pressure does not necessarily lead to an accelerated impregnation speed. This effect has also been found by Gibson who suggested useful pressure settings between 2 MPa and 4 MPa for the thermoplastic impregnation process. ¹³

Figure 1.

Fiber relaxation during thermoplastic impregnation process.

Because of the described compaction and relaxation behavior of the reinforcement fibers, efforts have been made in order to describe the fiber volume content V _f as well as the permeability K during the impregnation process. Particularly Gutowski et al. have suggested a mathematical approach, which provides a relation between fiber network stress σ and fiber volume fraction in order to generate more reliable permeability values. ¹⁴ Although this approach is more realistic regarding the permeability values, other differences between the real impregnation process and the model assumptions are still not considered. For example, there are not two separated areas of homogeneous distributed fibers and matrix material in reality. Actually, due to the use of, for example, woven textiles, the impregnation process is characterized by two main phases. The macro-impregnation represents the penetration of the area surrounding the fiber bundles, whereas the micro-impregnation describes the infiltration of the fiber bundles. ¹⁵ Both take place at different permeability levels. ¹⁵ Moreover, there are different types of reinforcement architectures (woven fabrics, non-crimped fabrics, fiber mats, etc.) and different prepreg types in use, which provide a mixture of both materials. This leads to an individual impregnation process for each fiber–matrix combination. Furthermore, for process examinations and optimizations carried out in the past, the pressure is considered to be constant for the complete pressing area. Hence, only polymer flow in through-the-thickness direction has been considered, although the advantageous of in-plane polymer flow has been proven theoretically, ^6,7 and few authors considered this aspect theoretically. ^16,17 For the impregnation process with thermoset material in the resin transfer molding process, Gong ¹⁸ identified a significant influence of the fiber bed distribution on the permeability of the pressure distribution. But no meaningful examinations have been carried out regarding the measurement of the real pressure distribution inside the laminate during the impregnation process with thermoplastic matrix materials. Especially when looking at semicontinuous or continuous production methods, which offer different states of impregnation in the same tool, a homogeneous pressure distribution seems not to be a realistic point of view.

Thus, a main aspect of the process analysis is the measurement of the pressure distribution inside the tool, which is of big importance for in-plane polymer flow.

Continuous compression molding

The investigation focuses on the semicontinuously working continuous compression molding (CCM) process. Due to economic facts, this technology is presently one of the most meaningful possibilities for the production of organic sheets. ^{19 –21} In general, a CCM machine is based on a static press process. Due to the combination with a transport system, a semicontinuously working unit results. Hence, the production process comprises the following four steps: closing the mold, applying the pressure and temperature, opening the mold, and transporting the laminate to a desired distance. This leads to a discontinuous pressure application for each cycle (see Figure 2, detailed view). Within this cycle and along the tool length, the impregnation, consolidation, and solidification of the laminate take place. In order to melt the matrix material and to impregnate the fibers, the first part of the tool is heated. In the second cooled part of the tool, the laminate is cooled down and solidified. Thus, it is exposed to a various temperature profile in longitudinal direction of the tool (see Figure 2, overall view).

Figure 2.

Temperature and pressure application during the CCM process. CCM: continuous compression molding.

Comparable with the double belt process, ^21,22 there are all stages of impregnation and consolidation available in the same tool. At the inlet, the reinforcement fibers are not impregnated, whereas the impregnation is completed at the transition region from the heating to the cooling zone of the tool. After leaving the tool, the laminate is impregnated and consolidated, and the polymer is solidified again. Because the machine is equipped with a continuous tool, these different impregnation stages are leading to a tangential deviation of the tool in process direction (see Figure 3). This effect can be observed in reality, although the tool is maintained in cylindrical bearings at each corner. Due to the high forces applied by the pressing cylinder to the tool, the stiffness of the machine design is not sufficient to prevent deviation.

Figure 3.

Sketch of the tangential deviation of the tool (excessive presentation) due to different impregnation stages.

Different prepreg types (e.g. film-stacking prepreg and powder prepreg) consisting of woven or nonwoven fabrics or random orientated textiles (glass, carbon or aramid, as well as natural fibers) can be used as reinforcement. The selected thermoplastic matrix material mainly depends on the desired application and the desired long-term service temperature. For example, standard polymers such as polypropylene (PP) and polyamide as well as high-performance polymers such as poyletheretherketone can be applied. The laminate thickness is typically in a range between 0.5 mm and 6 mm, with also sandwich structures having a thickness of up to 40 mm can be processed. Finally, the maximum thickness is mainly limited by the resulting process speed.

For the investigation carried out within this research study, the CCM machine of IVW was used (Producer: Xperion Aerospace GmbH, Immenstaad, former ACM GmbH, Germany). The machine offers the possibility for multifarious parameter settings. In particular, the highly flexible heating system enables various temperature profiles along and perpendicular to the process direction, which might create in-plane polymer flow (see Figure 4).

Figure 4.

Modular heating and cooling unit of the CCM machine enables various temperature profiles. CCM: continuous compression molding.

Because the in-plane polymer flow is basically caused by an inhomogeneous pressure distribution, the article will demonstrate a method for the examination of the pressure distribution inside the tool. Furthermore, the influence of different temperature settings on the pressure distribution will be analyzed. Finally, the effect of the different parameter settings on the in-plane polymer flow and the impregnation speed will be shown.

Measurement of pressure distribution

Preparation of measurement equipment

For the examination of the pressure distribution, FlexiForce^® Sensors (model: HT201) of Texscan^® have been used, which offer a specified operating temperature range from 9°C to 204°C. Because the measuring behavior of each sensor is individually influenced by the temperature, calibration and modeling of the sensor behavior are necessary. The calibration was carried out in an autoclave in order to apply homogeneous pressure on the sensors. The sensors have been placed inside a Kapton foil bag and exposed to autoclave atmosphere in order to transfer hydrostatic pressure to force (see Figure 5).

Figure 5.

Capton foil bag with incorporated FlexiForce^® sensors.

During this cycle, sensor values from 0 bar to 20 bar at temperatures of 50, 100, 150, and 200°C have been recorded (see Figure 6). The generated database was used to create a specific polynomial pressure function for each sensor using Wolfram Mathematica^®. A detailed description of the calibration and the modeling is given in the study by Christmann and Mitschang. ²³

Figure 6.

Temperature–pressure profile for the calibration process of the FlexiForce^® sensors and resulting sensor signals.

For the experimental measurement of the pressure distribution during the CCM process, the FlexiForce sensors again have been placed between two Capton films with a layup identical to the one used for the calibration process (see Figure 5). Eight sensors have been arranged with an equal distance in cross direction of a 640 mm wide foil bag. Additionally, eight thermocouples have been placed inside the laminate beside the FlexiForce sensors, in order to receive detailed information about the temperature distribution. Information about the positioning of the sensors in the Capton foil bag is given in Figure 7.

Figure 7.

Capton foil bag containing sensors and thermocouples.

Materials in use

The pressure distribution inside the laminate during the CCM process is mainly caused by two effects: On the one hand, it is influenced by lateral and through-the-thickness flow of the matrix material during the impregnation process. On the other hand, the machine itself can be a reason for inhomogeneities in the pressure distribution. Thus, the experimental measurement has been carried out using two different material configurations. In the first step, silicon films have been used in order to get information about the machine-related effects in the pressure distribution. The silicon films are necessary, because the FlexiForce sensors would be destroyed, if the total load would be applied to the sensors without any distribution media. The silicon film has a thickness of 1.5 mm, a hardness of 40 Shore A, and a temperature stability up to 230°C.

For the investigation of impregnation effects caused by the inhomogeneous temperature distribution, it is important to choose a reinforcement textile, which offers symmetric properties in warp and weft direction. Furthermore, a high area weight with reduced impregnation speed should be preferred to make effects more visible. Thus, a glass fiber (GF) textile of Hexcel (HexForce^® Style: 01038) has been chosen, which has a symmetric Twill 2/2 weaving pattern with a balanced fiber distribution of 50%–50% in warp and weft direction and a weight per area of 600 g/m². PP of Borealis (Borealis PP BJ100HP) has been used as the matrix material, because it offers a low-melting temperature of 163°C in order to use the FlexiForce sensors. The PP has been applied on the laminate in the form of films with a thickness of 300 µm for each layer. A symmetric laminate layup with four layers of textile and four layers of polymer film was chosen (see Figure 8), which is theoretically leading to a laminate thickness of 2.14 mm and a fiber volume content of 44%. Again, this setup guarantees a symmetric impregnation of each textile layer in order to minimize the effects of specific laminate setups.

Figure 8.

Laminate setup including the sensor bag.

Parameter settings for the analysis of the pressure distribution

The pressure investigation with the CCM machine has been carried out for three different temperature profiles (see Figure 9). Firstly, a temperature profile with a temperature of 200°C in the heating zone serves as a reference process. Furthermore, two uncommon settings with inhomogeneous temperatures in cross direction of the tool have been selected. One profile offers lower temperatures at the edges of the tool compared to the center (170°C compared to 200°C) and one profile offers higher temperatures at the edges (200°C compared to 170°C). In the further text, the temperature profiles will be named according to the used temperatures (T (170/200°C), T (200/200°C), and T (200/170°C)). Thereby, each of the three profiles has the same temperature settings in the cooling zone of 90°C after the transition region and 40°C at the end of the tool.

Figure 9.

Temperature profiles of the tool used for the measurement of the pressure distribution.

Due to the intermittently working process and the high flexibility of the CCM machine, there are many possible settings to define the rate of production. For the experiments, the cycle time was set to 21 s, which consists of 18 s pressing time and 3 s transportation time. Finally, the resultant production rate is 5 m/h. The pressure force was set to 960 kN, which leads to 15 bar in case of a homogeneous pressure distribution.

Evaluation of pressure measurement

The experimental measurement of the pressure distribution has been carried out using two different configurations. For each measurement, the Capton film bag, which covers the sensors and the thermocouples, has been placed in the center of the laminate. In the first measurement series, the sensor bag has been placed between two silicon films. This allows to get information about the pressure distribution inside the press, which is not influenced by different laminate thicknesses due to the impregnation progress. Afterward, the sensor bag has been put into the laminate as described in Figure 8. Thus, the real pressure distribution inside the laminate-containing machine and impregnation-related effects can be separated afterward. Both configurations have been carried out using the three defined temperature settings described in Figure 9.

In order to illustrate the resulting pressure distribution, the measurements have been transferred into several 2-D plots. Therefore, the data had to be edited in the first step. As explained in the state of the art, the semicontinuous working principle leads to a discontinuous pressure application. Thus, only the constant pressure during the pressing step has been considered for each sensor and for each process cycle (see Figure 2). Finally, a data matrix of 264 metering points (8 sensors and 33 process cycles) has been generated for each parameter setting. Afterward, these data have been transferred into 2-D plots using the software OriginPro of OriginLab Crop.

Effect of inhomogeneous temperature settings on the pressure distribution

Using the silicon films, an inhomogeneous pressure distribution inside the tool was detected, which can be traced back to the machine design (see Figure 10). Having a look at the pressure distribution for homogeneous temperature of 200°C, a low pressure at the lead in area can be seen. This is caused by the tool design of the machine. The surface of the tool at the inlet edge is sloped in order to avoid damage of the textile by applying too high compaction force to the laminate. Furthermore, the pressure is steadily increasing from the inlet to the outlet, and additionally, there is a pressure aggregation in the center of the tool. This is because the machine is equipped with only one pressure cylinder positioned in the center of the tool and shifted in longitudinal direction to the lead out of the press (see Figure 2). Due to these aspects, pressure values much higher than the calibration range of the sensors (approximately 100 bar) are reached in some areas. Regarding the different temperature settings, no significant difference in the pressure distribution have been recorded with silicon films as pressure distribution media (see Figure 10).

Figure 10.

Pressure distribution inside the tool using silicon films instead of a thermoplastic laminate.

The effect of inhomogeneous temperature settings on the resulting pressure for a laminate consisting of reinforcement and matrix materials is more complex. A theoretical contemplation of the interaction between temperature settings and accelerated impregnation shows an irregular correlation (see Figure 11). The viscosity is reduced due to the increased temperature. Theoretically, this should lead to locally accelerated impregnation and faster reduction of the laminate stack height. But the impregnation speed is limited, because the temperature is locally increased and the surrounding areas offer a higher viscosity and still an increased laminate stack height. Thus, a reduction of pressure might be a result at areas with higher temperature due to bridging effects. This probably compensates the acceleration effect.

Figure 11.

Possible effects of a localized inhomogeneous temperature distribution.

The result of the measurements for a GF/PP laminate and the homogeneous temperature profile with 200°C in the heating zone is given in Figure 12. It can be seen that there is an inhomogeneous pressure distribution across the tool. Comparable to the measurement with the silicon films, there is a loss of pressure at the inlet, and the highest values were found in the center of the tool near the lead out. This result is especially interesting, because the laminate thickness is the greatest at the inlet of the tool due to the marginal impregnation. Another important effect became obvious during the investigation. A considerable lack of pressure in the transition region from the heating to the cooling zone was found, which is not affected by the machine design. Hence, the shrinkage of the laminate in the transition region is responsible for this effect. Thermomechanical analysis measurements of the laminate and of pure PP have been carried out for the determination of the shrinkage below melting point (25–150°C). These tests have shown that the shrinkage of the laminate is mainly determined by the thermal expansion of the matrix material (Δh _Laminat = 3%; Δh _PP = 3.2%). For temperatures above the melting point, the volumetric shrinkage of the polymer has been determined experimentally. Finally, for the maximum cooling step of 110°C (200°C–90°C) in the transition region, the shrinkage has been calculated to be approximately 4% (thermal expansion and recrystallization), respectively, and approximately 0.08 mm for the used laminate thickness. Thus, shrinkage of the laminate has been detected as reason for the loss of pressure in the transition region between the impregnation and solidification zone.

Figure 12.

Pressure distribution for homogeneous temperature setting during the impregnation process.

In the next step, the pressure distribution using the two inhomogeneous temperature profiles was analyzed (see Figure 13). In general, the pressure increases from the inlet to the outlet. But there is an influence of the inhomogeneous temperature on the pressure distribution. The pressure is increased in the region with the lower temperature. That means the pressure is more constant in the cross direction of the tool when the edges are tempered to 170°C (center 200°C). In contrast to this distribution, the pressure is more constant in longitudinal direction when the center is tempered to 170°C (edge 200°C). The reason for the resulting pressure distributions have been found in the impregnation progress, which is faster in the hotter region due to the lower viscosity of the polymer. Thus, the necessary impregnation force is reduced, and the interlaminar pressure increases in the colder region.

Figure 13.

Pressure distribution for inhomogeneous temperature setting during the impregnation process.

Recapitulating the findings, there are two significant facts influencing the pressure distribution. On the one hand, the inhomogeneous pressure distribution is caused by the specific machine design. The distribution is characterized by an increasing pressure from the inlet to the outlet of the tool. Furthermore, there is a pressure aggregation in the middle of the tool below the pressing cylinder. On the other hand, the use of inhomogeneous temperature settings also has a significant effect on the pressure distribution. The pressure is increased in those regions, where the temperature is lower. Furthermore, a significant lag of pressure in the transition region from the heating to the cooling zone is evoked by the shrinkage behavior of the laminate.

Due to this inhomogeneous pressure distribution inside the tool, it can be concluded that the in-plane polymer flow during the impregnation process is most likely. Because of the pressure distribution, this effect should be most significant for the inlet area. Furthermore, a different in-plane polymer flow should appear for each temperature profile. The predicted polymer flow is given in Figure 14. Thus, in the next step of the investigation, the impregnation process is analyzed with regard to the laminate quality.

Figure 14.

Predicted polymer flow caused by pressure distribution inside the tool.

Impregnation quality for different temperature settings

Parameter settings and sample preparation

In order to evaluate the influence of in-plane polymer flow on the impregnation quality, organic sheets have been processed using inhomogeneous temperature profiles. For the sample preparation, five different temperature settings have been chosen (see Figure 15). The temperature in the middle of the tool has been set constant to 190°C, and the temperatures at the edges of the tool have been varied in the range of 170°C to 210°C with steps of 10 K. The rate of production was again set to 5 m/h with the same cycle time settings, and the pressure force was increased to 1280 kN, which is leading to an average pressure of 20 bar.

Figure 15.

Temperature settings for the analysis of the impregnation quality.

The impregnation quality was analyzed with cross-sectional micrographs that were taken at three different positions in cross-direction on one side of the tool, because lateral symmetry process conditions have been assumed (see Figure 16). Furthermore, at each position, two samples were taken in order to take micrographs in process direction (0° view) and transverse to the process direction (90° view). In order to freeze the state of impregnation at this position in the press, the process was stopped while the mold was opened and the laminate was pulled out of the tool. Thus, the laminate has solidified without external pressure. Due to the use of an air-impermeable release paper, the deconsolidation of the laminate was minimized, because the air could only penetrate the laminate from the side. Nevertheless, the micrographs might show some deconsolidation effects.

Figure 16.

Position of samples taken of the organic sheets.

Effect of in-plane polymer flow on the impregnation process

The evaluation of the impregnation quality has shown significant differences between the 0° view and the 90° view for the samples processed with lower temperatures at the edges. The micrographs of the polished cross sections for the temperature profile T (170/190°C) are given in Figure 17.

Figure 17.

Nonuniform impregnation process for different fiber orientations (temperature settings T (170/190°C).

The pictures clearly show that there are big voids inside the fiber bundles oriented in cross direction (90° view). In contrast, the fiber bundles oriented in process direction (0° view) don’t show much voids. Because this significant difference can be found for every position (L1, L2, and L3) and the 0° and 90° samples have been taken side by side, this effect must be caused by an unequal impregnation process. The analysis furthermore revealed that no differences in the impregnation quality can be found between the individual samples taken in cross direction of the tool, although the laminate was exposed to a much lower temperature at the edges.

Having a look at the samples processed with the temperature settings T (180/190°C), comparable results have been found (see Figure 18). The micrographs taken in 90° view offer much more voids inside the fiber bundles than the pictures taken in 0° view. Again, there is no difference in impregnation quality for the samples taken in cross direction of the tool.

Figure 18.

Nonuniform impregnation process for different fiber orientations (temperature settings T(180/190°C).

In contrast to the results shown before, for homogeneous temperature profiles and for temperature profiles with higher temperature settings at the edges, no difference between the void content in 0°- and 90° view has been found. Figure 19 shows some micrographs for the temperature settings T (190/190°C) and T (210/190°C) exemplarily.

Figure 19.

Impregnation process for different directions and temperature settings.

When correlating the results of the pressure distribution with the impregnation quality, the reason for the difference in void content becomes obvious (see Figure 14). For lower temperatures at the edges, the pressure is more constant in the cross direction, which mainly creates in-plane polymer flow contrary to the process direction and not in cross direction. Because of this, only the impregnation of the fibers oriented in 0° is accelerated. For homogeneous temperature profiles and for temperature profiles with higher temperature settings at the edges, the in-plane polymer flow takes place along and perpendicular to the process direction due to pressure drop in both directions. Thus, the impregnation of the fibers oriented in 0° and 90° are accelerated.

Conclusion

During the investigation, a new technology for the acceleration of the thermoplastic impregnation process by the simultaneous use of in-plane and through-the-thickness polymer flow was analyzed. Firstly, the article has shown a method to analyze the pressure distribution during the production process of organic sheets using the CCM technique. An interaction between the temperature settings in the heating zone and the resulting pressure distribution was found, with the detection of the resulting in-plane polymer flow. In order to determine the effect of the found pressure distribution on the in-plane polymer flow and the impregnation speed, an analysis of the impregnation quality was carried out. This examination revealed that there is a correlation between temperature setting, in-plane polymer flow, and impregnation speed. It was found that the presence of in-plane polymer flow leads to an accelerated impregnation process. Thus, based on these findings, a selective process development and optimization can be carried out in future work. This will lead to an increased process speed in general and thus offers the possibility for reduced process costs.