Composites: Part A 39 (2008) 1455–1466
Contents lists available at ScienceDirect
Composites: Part A journal homepage: www.elsevier.com/locate/compositesa
Non-isothermal stamp forming of continuous tape reinforced all-polypropylene composite sheet N.O. Cabrera a,b, C.T. Reynolds a, B. Alcock a, T. Peijs a,b,* a b
Queen Mary, University of London, Centre for Materials Research, Mile End Road, E1 4NS, UK Eindhoven University of Technology, Eindhoven Polymer Laboratories, P.O. Box 513, 5600 MB Eindhoven, The Netherlands
a r t i c l e
i n f o
Article history: Received 12 December 2005 Received in revised form 14 May 2008 Accepted 16 May 2008
Keywords: A. Thermoplastic resin E. Forming A. Recycling A. Polymer(textile) fiber E. Tape
a b s t r a c t This paper describes the thermoforming behaviour of a self-reinforced composite based on co-extruded polypropylene (PP) tapes. In contrast to traditional continuous woven glass fabric reinforced polypropylene (GF/PP) materials, where the sole mode of deformation is either inter- or intraply shearing, all-PP composites have an additional mode of deformation as here the fibres (or in this case tapes) can still be deformed. The importance of this additional deformation mode is investigated in a range of stamping experiments in combination with 3D strain mapping experiments. Non-isothermal thermoforming experiments revealed that all-PP woven fabric laminates based on flat tapes deform in a different manner to traditional GF/PP. Although the main mode of deformation of both all-PP and GF/PP for the investigated dome parts was intraply shearing, a much lower energy was required to deform the all-PP laminate. Whenever possible, deformation of the tape by drawing should be avoided as it requires higher energy which may lead to higher residual stresses in the final part. However, tape drawing may prove an essential benefit when complex shapes are involved. Ó 2008 Published by Elsevier Ltd.
1. Introduction Solid state thermoforming of isotropic materials such as metallic or plastic sheets can involve several forming operations, e.g., bending, biaxial stretching, plane-strain stretching and drawing [1]. Bending occurs in almost all forming operations. Biaxial stretching is caused by in-plane tensile stresses in excess of the yield stress in perpendicular directions of the sheet. When stamping a geometry such as a dome, the main mode of deformation is balanced biaxial stretching (a dome is defined as hemisphere, or similar section of a sphere). Plane-strain stretching corresponds to elongation in one direction of a sheet and a non-dimensional change perpendicular to that direction. Plane-strain stretching is the weakest deformation phenomenon of many materials. Drawing of a sheet corresponds to tensile deformation in one direction and compression in the perpendicular direction. A typical example is the drawing along the wall of a flat-bottomed cylindrical cup. Thermoplastic composites can also be stamped non-isothermally [2,3]. A typical thermoplastic composite laminate consists of an arrangement of stacked inextensible fibre rich layers alternated with resin rich layers. The preconsolidated laminate comprises a stack of plies (‘‘prepreg”) which can be heated above the
* Corresponding author. Address: Queen Mary, University of London, School of Engineering and Material Science and Centre for Materials Research, Mile End Road, E1 4NS London, UK. Tel.: +44 (0) 20 7882 5281. E-mail address:
[email protected] (T. Peijs). 1359-835X/$ - see front matter Ó 2008 Published by Elsevier Ltd. doi:10.1016/j.compositesa.2008.05.014
softening temperature of the matrix (e.g., melting temperature for a semi-crystalline polymer), then shaped and cooled in a cold mould. Friedrich et al. [4] studied non-isothermal diaphragm forming of continuous glass fibre/PP laminates (PlytronÒ). This work showed that glass fibre reinforced polypropylene (GF/PP) can be melt-phase stamped until 120 °C on cooling because of the large supercooling effect of PP (30–40 °C). As a result, transfer times are not critical. A mould at room temperature can be used and low cycle times of less than 30 s are common. Solid-state stamping of GF/PP is possible for limited deformations if the sheet is heated between the melting temperature and the recrystallisation temperature of the PP matrix [5]. During melt-phase stamping of continuous woven fibre reinforced thermoplastics, the main deformation and flow mechanisms are matrix percolation, squeeze flow, intraply shear and interply shear [3,6,7], as shown in Fig. 1. The laminate deforms by interply shearing (or interply slipping) when it is bent along one of its fibre directions: the quite unstretchable fibre-rich layers will slip relative to each other. Intraply shearing (the ‘‘trellis” effect [3,8]) occurs when tensile forces are generated in fabrics in directions other than the principal directions of the fabric (maximum at 45° of fibre orientation). Not shown in Fig. 1 are fabric wrinkling or buckling mechanisms which can also apply during stamping of fabric reinforced composite laminates. Quite recently, all-polymer or self-reinforced composites have been proposed to replace traditional fibre reinforced PP composites (glass fibre reinforced (GF)/PP or natural fibre reinforced (NF)/PP)
1456
N.O. Cabrera et al. / Composites: Part A 39 (2008) 1455–1466
Fig. 1. Forming mechanisms in textile composites depending on the forming force direction. (a) Interply slip deformation and matrix percolation (or squeeze flow). (b) Intraply shearing.
for automotive applications. The absence of foreign reinforcements in all-PP composites means that they can be melted down into a polypropylene blend which may be reused to remake all-PP composites, or alternatively, be used for other PP-based applications. However, recycling at the end of the product life for any material is complicated by issues of recovery and separation [9]. Although recyclability is currently considered a very important issue, mechanical performance and product cost remain of key importance for many engineering applications, notably the automotive industry. The issue of recycling is of particular relevance to the automotive industry with the recent End-of-Life Vehicle Directive of the European Union aiming to increase the rate of recycling of all vehicles to at least 95% by average weight by 2015 [10]. The concept of ‘‘self-reinforced” polymer composites has been presented in literature. The ‘‘hot-compaction” process developed at the University of Leeds [11] is a possible route, which has also been investigated for a range of polymers in addition to polypropylene [12–14]. This processing method is highly innovative as it involves welding fibres together rather than impregnating fibres with matrix polymer. It therefore eliminates one of the key problems in thermoplastic composite manufacturing. However, the process suffers from a rather small temperature window for the compaction process of the fibres or fabrics [15]. In order to overcome this problem an alternative route was proposed by Peijs and co-workers [16–23]. The technology reported by Peijs et al. is based on the use of co-extruded polypropylene tapes. The co-extruded tapes consist of a highly oriented isotactic PP homopolymer core and a thin PP copolymer coating or skin with a lower melting temperature than the PP core material. Subsequent solid-state drawing of the co-extruded tapes leads to orientation of the polymer molecules (particularly in the homopolymer core), which results in a large increase in tensile stiffness and strength of the tapes. The thin copolymer skin is present to allow the tapes to be bonded together at a temperature below the melting temperature of the oriented homopolymer core and hence acts as an adhesive layer [22]. This difference in melting temperature between the homopolymer core and the copolymer skin layer provides a larger processing temperature window (of around 30 °C [21]) for the creation of ‘‘all-polypropylene” (all-PP) composites products. The choice of PP copolymer determines the lower limit of this processing window (melting temperature of the copolymer) while the upper limit is the melting temperature of the PP homopolymer. Besides creating a large processing window, one of the other main innovations in terms of composite processing technology of this
novel concept is that reinforcement manufacturing and matrix impregnation are achieved in a single step. This is a very different approach compared to traditional composite systems, as they all rely on impregnation processes. The reinforcement to matrix ratio in these all-PP composites is also extremely high compared to traditional composites resulting in unique properties. In the all-PP composites described in this paper, 90% of the composite volume consists of the highly oriented PP phase, which is the reinforcement phase of the composites, while the remaining 10% is the copolymer skin layer which is the matrix phase of the composites. If based on bidirectional woven fabrics, these all-PP composites possess Young’s moduli of around 5–6 GPa and tensile strengths of 180–200 MPa which makes them an interesting alternative to replace traditional engineering composites such as glass-fibrereinforced PP in a wide range of applications, notably those requiring high impact resistance [20,24]. This paper describes the thermoforming behaviour of this type of self-reinforced polymer composite based on co-extruded PP tapes. All-PP woven fabrics based on co-extruded tapes can be compacted in a hot-press or laminated in a continuous process using an isobaric double-belt press. These laminates can then be stamped into shell products. The same thermoforming equipment as for other thermoplastic composites, can be used, i.e., diaphragm, rubber or matched die forming. Woven laminates are heated above the melting temperature of the PP copolymer skins but below the melting temperature of the homopolymer core of the co-extruded tape. This temperature processing window in which all-PP composites can be processed has been previously shown to be >30 °C [21]. Under these conditions, the oriented PP tape may also be further elongated by solid-state deformation (i.e., below the melting temperature of the matrix phase) on cooling. This means that below the matrix phase melting temperature, all-PP composites are likely to possess an additional mode of deformation compared to GF/PP since the oriented PP reinforcement phase of all-PP composites can be plastically deformed, unlike the glass reinforcement phase in GF/PP. However, it has been reported possible to stamp more complex geometries from other non-extensible fibre composite laminates by stretch breaking fibres during forming [25,26]. The excellent thermal stability of glass fibres in the temperature ranges used for stamping mean that GF/PP composites, while being non-extensible in the fibre direction, are inherently much more temperature stable and benefit from being formable at higher temperatures and so lower matrix viscosities. The forming mechanisms involved in the stamping of all-PP composites are not clear and so tape drawing can be considered an additional deformation mechanism. Such post-drawing of all-PP constituent tapes could theoretically improve mechanical properties and counteract shrinkage effects, which may occur in the tapes at elevated temperatures. The objective of the present work is to prove the feasibility of stamping all-PP composites into a simple hemispherical geometry and investigate the deformation modes which occur in these operations.
2. Materials and methods 2.1. Materials The all-PP composites were manufactured from a three layer co-extruded tape, consisting of an A:B:A (copolymer skin: homopolymer core: copolymer skin) structure. The tape was manufactured at Lankhorst Indutech BV, The Netherlands. The thermal and mechanical tape properties have been characterised previously [17,19,22], and are summarised in Table 1. The tapes were woven into their respective weaves by BW Industrial BV, the Netherlands. In all weaves, the crimp of the fab-
1457
N.O. Cabrera et al. / Composites: Part A 39 (2008) 1455–1466 Table 1 Mechanical properties of PP precursor tapes used for the production of all-PP composites Width (mm) Thickness (lm) Density, q (g cm3) Draw ratio, k Tape composition (A:B:A) Tensile modulus (GPa) Tensile strength (MPa) Strain to failure (%)
2.15 60 0.732 17 5.5:89:5.5 15 450 7
rics along the warp direction is similar to the crimp along the weft direction. Five different weave styles (shown in Fig. 2) have been initially investigated: a plain weave (1/1T), three twill-weaves (2/ 1T, 4/2T2, 3/3T) and a satin weave (S5/1). These fabrics possess a nominal weight of 105 g/m2. To create single ply all-PP composite sheets for mechanical characterisation of thin sheets of different weaving styles, one layer of fabric from each weaving style was pressed in a lab-scale hot-press at 145 °C and 4 MPa [19] and then cooled to 40 °C under the same pressure. The resulting single ply sheet had a thickness of approximately 0.15 mm. Sheets of greater thickness: 0.43 mm (3 plies of all-PP fabric), 1.3 mm (9 plies of all-PP fabric) and 1.7 mm (12 plies of all-PP fabric) were produced for stamping experiments. These thicker laminates were pressed in a continuous, 2 m long, double belt press at 155 °C under a pressure of 2.5 MPa, with a throughput of approximately 1 m/min. This production was performed at the Institut für Verbundwerkstoffe GmbH (Kaiserslautern, Germany). Only the plain weave fabric was used for these thicker laminates, as it will be shown later that it is the most suitable for stamping all-PP plates. Photographs of typical all-PP laminates, together with micrograph cross sections have been presented elsewhere [21]. To compare all-PP composites with conventional glass reinforced materials, some commercial glass fabric reinforced polypropylene plates (GF/PP) have also been stamped. These GF/PP plates were selected to have similar mechanical properties to the all-PP laminates. The GF/PP plates were 1.3 mm thick, commingled (2/2 twill) woven, pre-consolidated sheets with a glass fraction of 40 wt%. The mechanical properties of the all-PP sheets compared with the GF/PP composite plates are reported in Table 2.
Fig. 2. The different weaving styles used to produce the all-polypropylene woven tape fabrics used in this study: plain weave (1/1T), twill weaves (2/1T, 4/2T2, 3/3T) and satin weave (S 5/1).
Table 2 Tensile properties of all-PP and GF/PP woven laminates used in this study Material
Tensile modulus (GPa)
Tensile strength (MPa)
GF/PP composite All-PP composite
7.5 5.8
163 205
2.2. Methods 2.2.1. Tensile characterisation of individual PP tapes A tensile deformation of the PP tape during forming of all-PP composite laminates may occur at elevated temperatures and varying strain rates, and since the PP tape is thermoplastic, it is important to understand how the tape behaves under tensile loading at various strain rates. Individual PP tapes were tested in tension in a universal tensile testing machine (Instron 5584) at various crosshead displacement speeds to achieve strain rates between 0.2 and 0.002 s1. Individual tapes were tested at three different temperatures (100 °C, 120 °C and 140 °C) at a crosshead displacement rate of 10 mm/min and also tested at four different crosshead speeds (10, 50, 100, 500 mm/min) at a temperature of 140 °C. To achieve an elevated testing temperature, a hot plate was suspended behind, but in direct contact with, the tape being tested. The hot plate was 65 mm long and 50 mm wide, giving a constant heated length of 65 mm. The hot plate is shown in Fig. 3. Although the hot plate was thermostatically controlled, the temperature was recorded before each test with a surface temperature probe. A temperature of 140 °C was chosen because it is above the melting temperature of the PP copolymer and was pre-
Fig. 3. Photograph of the elevated temperature testing apparatus used to perform tensile tests on individual PP tapes and single ply all-PP composite sheets. Shown are the specimen grips and the hot plate used to apply elevated temperature by physical contact.
1458
N.O. Cabrera et al. / Composites: Part A 39 (2008) 1455–1466
viously shown to be a reasonable temperature for all-PP composite production [21,22], and therefore relevant to stamp forming. The use of a hot plate rather than an environmental chamber has to be noted as it implies a difference in the determination of strain rate. When a tensile test is carried out in an oven at a constant crosshead speed, provided the specimen does not yield, the engineering strain rate is assumed constant in the middle of the specimen. With the use of a hot plate, the engineering strain rate increases with time while the true strain rate is constant. As the tape is very thin (65 lm), it cools rapidly as soon as it is no longer in contact with the hot plate, therefore any drawing would only be expected to occur within the tape in contact with the hot plate. 2.2.2. Tensile characterisation of pressed single ply all-PP composite sheets Tensile characterisation of consolidated single ply all-PP sheets was necessary to understand the way in which each layer of the all-PP composite behaves when tensile stresses are applied at elevated temperatures. A similar tensile characterisation was not performed on GF/PP sheets because these were considered to be well understood and because the mechanical behaviour of the reinforcing phase (glass) is unaffected by these relatively small increases in temperature. Tensile testing specimen with dimensions of 47 mm width and 250 mm length were cut from the all-PP single ply sheets (0.15 mm thick). Before testing, a hot plate with a surface temperature of 143 °C was placed in contact for 1 min with the specimen between the two grips (Fig. 3), using the same test setup as was used for single tape tests described previously. The loading axis of the specimens was at 45° to the tape directions. Tests were carried out in a universal tensile testing machine (Instron 5584) equipped with a 1 kN load cell. The crosshead speed was set at 2.5 mm/min and the test was stopped at 20 mm extension. 2.2.3. Non-isothermal stamping A laboratory stamping press has been specially designed for the tests described in this paper, and this is shown in Fig. 4. A hydraulic pump (Hussh ENERPAC GPER-5000) was combined with a precision double acting 80 kN ENERPAC cylinder. With this hydraulic pump, the cylinder extension speed depends on which one of two discrete pressure ranges the pump experiences. The cylinder moves at a higher speed at lower pressures (7.5 m/min at pressures below 6 kN), and a lower speed at higher pressures (1.3 m/min at pressures between 6 kN and 80 kN). A displacement transducer was attached to the press to measure the position of the punch
and a pressure transducer was used to measure the pressure of the hydraulic system, which allowed for the derivation of the force developed by the cylinder. A hemispherical dome geometry was chosen for the matched metal die mould in order to draw a comparison to Friedrich et al. [4]. A photograph of a stamped all-PP composite dome produced by non-isothermal stamping, as presented in this paper, is shown in Fig. 5. The dome had an outside diameter of 60 mm. The 3 layer all-PP composite plates (0.43 mm thick) were stamped by a 58 mm diameter punch while a 56 mm punch was used for the GF/PP composite laminates and the all-PP laminates made from 9 plies (1.25 mm thick) and 12 plies (1.7 mm thick). In all cases, the stamping mould was preheated to 40 °C. Although this is below the temperature conventionally used to stamp GF/PP, it was deemed suitable for stamping all-PP plates, and so to aid comparison between GF/PP and all-PP composites, the same temperature is used for both materials throughout this paper. Disks of 140 mm diameter were cut from consolidated all-PP plates. These circular-shaped laminate disks were placed between two steel rings of 100 mm internal and 140 mm external diameter. A torque of 5 Nm was applied to the 12 screws of the clamping rings. Some plates were partially clamped with 4 screws tightened to a torque of 2 Nm and placed at 45° with the fabric directions, allowing some sliding towards the dome along the warp and the weft directions and maximum intraply shearing. The composite laminate held in the clamping ring was placed in a convection oven with a thermocouple attached to the surface of the laminate. The all-PP specimens were heated to 150 °C while GF/PP laminates were heated to 160 °C. Following heating, the composite laminate (still inside the holding frame) was directly transferred to the press for stamping. The transfer time is approximately 3 s, as the oven is adjacent to the stamping press, and so a negligible drop in temperature is expected between removing from oven and the beginning of the stamping operation. The pole of the dome coincided with the centre of the laminate. The temperature of the laminate, the punch force (from the hydraulic oil pressure) and the displacement of the punch were recorded by a data acquisition system (Spider8, HBM). Similar equipment has been used in the past to reproduce the conditions of industrial non-isothermal thermoforming [27,28]. Some partially clamped laminates were heated identically but not stamped in order to use them for testing of mechanical property loss of the fabric during heating. Tensile test coupons of these heated but unstamped laminates were produced with a gauge length of 70 mm and a width of 15 mm. These were tested
Fig. 4. Photograph of non-isothermal hydraulic stamping press (left) equipped with matched metal die moulds, together with the preheating hot air oven (right).
N.O. Cabrera et al. / Composites: Part A 39 (2008) 1455–1466
Fig. 5. Photograph of an all-PP composite dome formed by non-isothermal stamping. The clamping frame is clearly visible around the edge of the stamped all-PP composite dome. The steel rings shown have an internal diameter of 10 cm and an external diameter of 14 cm.
using universal testing equipment (Instron 5584) as described earlier, but were tested at a crosshead speed of 5 mm/min. 2.2.4. 2D and 3D optical strain measurement An optical strain measuring technique [29] was used to determine the strain distribution of the uniaxial tensile tests and of the stamped domes. The Aramis HR strain mapping system developed by GOM GmbH (Germany) is a 3D deformation measurement device using the grating method [30]. A fine random pattern with good contrast is sprayed onto the surface of the test object with spray paint or with an airbrush. One (2D) or two (3D) high-resolution CCD cameras (Type: Vosskühler CCD 1300F) record the deformation of the surface under different loading conditions. The 3D coordinates, the displacement and the plane-strain tensor of the surface can be derived using digital image processing. This technique has also been applied to measure the surface strains on all-PP pipes loaded internally by hydrostatic pressure [31], and to measure the plastic deformation resulting from low velocity impact events [20]. Martin et al. [32,33] using a similar technique (grid strain analysis) to analyse the deformation of stamped domes. They showed that the surface strains describe accurately the deformation behaviour of bi-directional composite sheets. 2D strain mapping was used for the sheets and single tape uniaxial tensile testing. The camera was triggered by the load and displacement analogue signals from the tensile machine so that each strain state is associated to the load and displacement data. For the fabric sheets, the major strain, the minor strain and the shear strain (angle reduction/2) were calculated. The X and Y-directions coincide with the tape orientations at 45° of the loading axis. For the nonisothermal stamping tests, two cameras were used to map the 3D surface deformations of the domes by taking a picture of each disk before and after stamping. The angle between the cameras was 30° and a 50 40 mm calibration window was used.
1459
both the reinforcement and the matrix of all-polypropylene composites are based on the same polymeric material, plastic deformation can occur due to forming stresses at elevated temperatures. Fig. 6 shows the effect of elevated temperature on the tensile properties (specifically the strain to failure) of the tapes. However, the melting temperature of polypropylene must not be approached as this would result in greater molecular mobility, and a reduction in the molecular orientation of the tapes. Hence the tapes would experience a loss of anisotropy and so a decrease in the mechanical properties. For this reason, it is desirable that deformation should occur at lower temperatures and so ideally in the ‘‘solid state” [34]. The co-extruded polypropylene tape has a strain to failure of approximately 100% at 140 °C and a strain rate of 2.5 103 s1. Under these conditions, a stress of 100 MPa (20% of the tensile strength at room temperature) is sufficient to plastically deform the material. The drawability of the tape decreases rapidly from 140 °C to 100 °C, while the strain to failure is lowered by a factor of 3 and the yield stress is multiplied by the same factor. However, the strain rate of 2.5 103 s1 used in these tests is fairly low compared to the forming speeds involved in non-isothermal stamping, which are typically of the order of 101 s1. The effect of strain rate at 140 °C on both the yield stress and the strain to failure is shown in Fig. 7. An increasing strain rate on a logarithmic scale has a similar effect as lowering the temperature. The strain rate range in Fig. 7 compares with the range which is observed during non-isothermal stamping because the deformation is not uniform over the part as this depends on the geometry of the mould and the temperature distribution. The stiffness and yield stress of the tape increase considerably as the temperature decreases (Fig. 6). It can then be assumed that during the test, at any time, only 65 mm of the tape (length of the hot plate) is deformed. A crosshead speed of 10 mm/min implies a true strain rate in the tape of 0.15 min1, equivalent to 2.5 103 s1. For that reason, true strain rates have been used for Fig. 7. 3.2. Tensile characterisation of pressed single ply all-PP composite sheets Single plies of differently woven all-PP fabrics, all with a lay-up of ±45° were pressed to yield single ply sheets, as described earlier. The weave styles considered were: plain weave (1/1T), three twill weave (2/1T, 3/3T and 4/2T2) and one satin (S5/1). Following pressing, some of the weave styles exhibited warpage. Fig. 8 shows the degree of warpage resulting from the pressing process. The
3. Results and discussion 3.1. Influence of strain rate on the tensile behaviour of PP tapes at an Elevated Temperature Textile composites are mostly based on glass or carbon fibre yarns. These fibres are brittle and non-extensible under solid state or melt processing conditions of the polymeric matrix. Because
Fig. 6. Tensile tests at elevated temperature for individual all-PP tape (crosshead speed 10 mm/min), showing that yield stress decreases and strain to failure increases with increasing temperature.
1460
N.O. Cabrera et al. / Composites: Part A 39 (2008) 1455–1466
Fig. 7. Influence of strain rate on both strain to failure and yield stress of all-PP tape tested at 140 °C.
weave styles 1/1T and 3/3T are balanced woven fabrics and do not exhibit any warpage on pressing, while the fabrics with weave styles 2/1T, 4/2T2 and S5/1 exhibited increasing warpage with the increasing degree of imbalance in the weave architecture. The different sheets were tested in tension at 145 °C; the specimens were loaded along the bisector of the warp and the weft fabric directions, i.e., at 45° to either tape direction. Fig. 9 shows that the plain weave and the twill 2/1T possess more homogeneous deformation behaviour than the other weave styles as these were the only specimens that could be deformed to 20 mm extension (20 mm was the limit of the test). All weave styles except the plain weave reach a plateau after approximately 2 mm extension. This difference was confirmed by visual observation. Tapes display out-of-plane buckling or in-plane wrinkling: Fig. 10a shows a picture of the 3/3T specimen after 3mm extension showing gross wrinkling (square symbol in Fig. 9) but Fig. 10b shows that the plain weave after 10.5 mm (circular symbol in Fig. 9) exhibits no twisting or apparent buckling. Fig. 11a shows the shear strain distribution of the deformed plain weave from Fig. 10b, obtained from the Aramis strain mapping system. The maximum shear deformation is observed in the middle of the specimen as Johnson et al. [35] and Hofstee et al. [36] reported previously. The Aramis system allows the export of each data point with the strain tensor (ex, ey and cxy = /) in a coordinate system where X and Y-directions coincide with the undeformed warp and weft directions of the fabric. In addition, the principal strains (e1, e2) can be related to this strain tensor. Here, the e1 direction is parallel to the loading axis. If it is assumed that as the fabric shears, the tape is regarded as non-extensible [6], and so the major strain (principal strain e1) and the minor strain (principal strain e2) can be derived as a simple trigonometric function of the shear strain (/ lies between 0° and 45°), as shown in Fig. 12.
Fig. 9. Stress-extension behaviour of the different weaving styles tested in tension at 140 °C. Loading axis is at 45° of the fabric warp and weft directions. Symbols refer to images in Fig. 10.
e1 ¼ 1 þ cos u þ sin u
ð1Þ
e2 ¼ 1 þ cos u sin u
ð2Þ
Major and minor strains can then be expressed as a function of the shear strain. Eqs. (1) and (2) have been plotted in Fig. 11b together with the data points from Fig. 11a. It is interesting to note in the fabric intraply shearing, that the minor strain (e2) exceeds in absolute values the major strain (e1) in the direction of testing. In spite of a non-uniform shear deformation, the theoretical curves are a good fit to the data, which implies that tape drawing can be neglected during pure intraply shearing. In this case, the entire specimen deforms by intraply shearing independently of the deformation level. Table 3 shows the onset of buckling of the different weave styles. The extension in the second column of the table is the value of the crosshead displacement in millimetres when out of plane buckling is observed. The plain weave can deform homogeneously up to a major strain (e1) of 22%, i.e., 3–10 times more than any of the other weave styles considered here. Not only does the plain weave laminate start to buckle later but the phenomenon is also more limited and more localised. The out of plane buckling of the tapes in a fabric is resisted by the increasing frequency of tapes crossing normal to the tape path, since a warp tape passing under a weft tape (or vice versa) is physically constrained. The warp tapes in a fabric can most easily buckle between crossing weft tapes, and in a plain weave, this distance is smallest (i.e., twice the tape width). In looser weaves such as a satin weave, the warp tapes are crossed by weft tapes less often (i.e., six times the tape width). Therefore, the warp tapes in the satin weave can buckle much more easily buckle. Therefore, due to its tightness and stability, the plain weave exhibits the most uniform intraply shearing behaviour, although higher stresses are required to cause the deformation (see Fig. 9). The plain weave fabric is also the easiest and most stable to manufacture since in-plane
Fig. 8. Single layer laminates from different weave styles showing increasing warpage with increasing weave style imbalance. The laminates shown are square with a side length of approximately 25 cm.
1461
N.O. Cabrera et al. / Composites: Part A 39 (2008) 1455–1466
Fig. 10. Tensile deformation of single ply all-PP composite sheets with a vertical tensile loading direction (a) 3/3T after 3 mm extension and (b) 1/1T after 10.5 mm extension. The high contrast random speckle pattern necessary for optical strain mapping is clearly visible on the specimen surfaces.
Fig. 12. Schematic representation of uniform intraply shearing of a fabric, showing that after deformation the length, L, of the fibre tows remains constant.
Table 3 Buckling onset during elevated temperature tensile tests of the different weaving styles of single ply all-PP composite sheets Weave style
Extension (mm)
Shear strain angle (°)
Major strain (%)
Buckling stability
1/1T 2/1T 3/3T 4/2T2 S 5/1
11.3 3.1 1.0 2.1 1.3
14.9 5.6 1.2 2.4 2.5
22.4 8.4 1.9 3.5 3.8
++ + – – –
Buckling stability is categorised in as ++ (greatest buckling stability), (lowest buckling stability). Weave styles are as defined in Fig. 2. Fig. 11. Strain mapping results of plain weave all-PP composite laminate after 10.5 mm tensile extension. (a) Shear strain distribution corresponding to Fig. 10b. (b) major and msinor strain as a function of shear strain, showing the good correlation between experimental data and the ‘‘trellis effect” theory described by Eqs. (1) and (2), and illustrated in Fig. 12.
wrinkling can also be problematic during fabric weaving. These isothermal tests show that the all-PP laminates can be deformed whether the stress is applied at 45° of the fabric directions or along
them. However, the drawability of the tapes decreases quickly with both temperature and strain rate. 3.3. Non-isothermal stamping of all-PP and GF/PP composites 3.3.1. Influence of the closing speed and clamping on the stamping force and energy during forming of all-PP and GF/PP composite domes Isothermal uniaxial tensile tests are easy to perform. However, the information they provide cannot replace the investigation of
1462
N.O. Cabrera et al. / Composites: Part A 39 (2008) 1455–1466
the forming behaviour under non-isothermal conditions, particularly when both the matrix and the reinforcement are inherently viscoelastic in nature. Therefore, dome geometries have been stamped, and an example of one of these is shown in Fig. 5. Although a dome is a simple geometry, it involves balanced biaxial stretching forces, which makes a dome geometry convenient for the study of anisotropic materials. The effect of stamping parameters such as closing speed, clamping and stamping force was first addressed. Since drawing of the all-PP tapes was expected to be the main difference between stamping all-PP composites and GF/ PP composites, the determination of strain distribution along the warp or weft directions was the main objective of the 3D strain mapping study. The relative importance of tape drawing compared to intraply shearing was finally assessed quantitatively. Before looking in detail at the stamping process, it is important to check that the properties of the tapes are not partially lost during the pre-heating step. This may occur due to relaxation processes that can lead to a loss of molecular orientation and hence mechanical properties. Tensile tests on pre-heated specimens have previously confirmed that no degradation in mechanical properties occurs if the specimens are appropriately physically constrained during heating [37]. Since the all-PP composites were still visibly constrained within the clamping ring during heating, it is assumed that negligible degradation in mechanical properties had occurred. However, it was found that the specimens were 20% thicker after heating, due to deconsolidation of the layers, which may have allowed some shrinkage and even some local loss of the mechanical properties of the all-PP laminates. Deconsolidation during preheating is a common phenomenon in moulding of thermoplastic composites [38], even for thermoplastic composites that do not have such heat sensitive reinforcement phases. The punch force is plotted against the punch position in Fig. 13. The mould is closed when the punch reaches 31 mm while the punch position is zero when it touches the laminate. Both all-PP (9 layers) and GF/PP laminates are approximately 1.3 mm thick. The data-sampling rate was constant and each symbol shown in Fig. 13 represents a data point. The time required to close the mould and the energy under the force–displacement curve are indicated on the right of each curve. For both GF/PP and all-PP composites, the energy required is approximately doubled between the partially clamped (4-point clamped) and the fully clamped (12-point clamped) sheets. Under the same clamping conditions, the GF/PP composites required far more energy for the moulding than the all-PP plates. Although the mould was always closed in less than 1 s, the shortest time to close the mould was for the partially (4-point) clamped all-PP composite laminate which required 0.36 s, compared to 0.70 s for the partially (4point) clamped GF/PP laminate. This is because only the all-PP composite laminate (4-point clamped) could be formed entirely at the highest speed of the press. The press switches automatically to the low speed at 6 kN for both GF/PP and all-PP composite (both 12-point clamped) domes. Such two-speed systems are common in industrial stamping presses. There is a discontinuity in these three curves at 6 kN: the force increases faster above 6 kN. This is probably due to the rapid cooling of the laminates leading to a stiffness increase. The force required to stamp the all-PP composite laminate shows a plateau just before complete closure of the mould. The reason for this is not clear, as the forces involved are complex: non-isothermal large deformations are combined with the stress– strain behaviour of viscoelastic PP, a function of strain, strain-rate, draw ratio and temperature. However it might also be due to tearing of the tapes around the clamping screws. The glass fibre bundles in GF/PP composite laminates were observed to be pulled around the screws whereas the PP tapes in all-PP composite laminates were observed to be torn around the screws in the clamping ring.
Fig. 13. Force–displacement curves for the hemispherical punch of fully (12-point) and partially (4-point) clamped all-PP and GF/PP sheets. Note that the force required to close the mould is always higher for the woven glass reinforced PP sheets. Labels on the right hand side of graph indicate stamping energy (J) and time to mould closure (s) for each material and clamping arrangement.
Fig. 14. Diagram of the effects of fabric deformation of a plain weave fabric, showing the increase in unit cell size as a result of biaxial stretching.
3.3.2. 3D strain mapping of All-PP and GF/PP of stamped domes The surfaces of all of the domes which were partially clamped were observed to be smoother than the surfaces of the domes that were fully clamped. However, no fibre or tape breakage was observed on either surfaces of the domes. In the case of all-PP composites, the tightness of the plain weave on the surface was lost on the outer surface of the dome because of the drawing of the tapes. For a plain weave, a 2.3 mm wide tape on the outside surface of the undeformed laminate is only visible every 2.3 mm. When the fabric is stretched, the visible part of the warp (respectively weft) tape will be partially pulled under the adjacent weft (respectively warp) tape, as shown in Fig. 14. The distance between the tapes increases and so does the unit cell length of the fabric. Hence the plain weave fabric loses its tightness and part of its stability. Similar behaviour was already observed previously for another type of PP–PP composite during isothermal stamping [39]. Because the outer surface of the dome was not deformed as a continuous surface, no strain mapping could be performed on that surface. The following strain mapping results therefore only deal with the inner surface of the dome. Fig. 15 gives the major strain
N.O. Cabrera et al. / Composites: Part A 39 (2008) 1455–1466
1463
Fig. 15. Major strain mapping of the inner (compressive) surface of stamped dome: (a) an all-PP composite dome and (b) a GF/PP composite dome (both domes are partially (4-point) clamped during stamping). The star shaped strain map reveals the non-homogenous nature of the surface strain.
distribution of both (a) all-PP and (b) GF/PP domes from partially clamped plates. The white lines represent the meridians of the dome that coincide with the principal directions of the fabric. The all-PP dome exhibits compressive strains on the pole inner surface. The isostrain lines have a star shape due to higher deformations at ±45° directions compared to the (0°) tape directions. This star shape is particularly clearly defined for the all-PP composite domes because all-PP stamping involves less matrix squeeze flow and because the all-PP tape is more than twice as narrow as the glass bundles (2.3 mm PP tape width compared to 5 mm glass fibre bundle width) giving a finer weave. The major strain was plotted on the dome surface in Fig. 15 and the Aramis system also allows
Fig. 16. Strains X, Y profile along X-axis of the all-PP composite laminate disk. (a) Partially (4-point) clamped. (b) Fully (12-point) clamped.
plotting the strains on the original surface (i.e., the circular diskshaped laminate). This undeformed surface will be used for the rest of the discussion. The origin is the centre of the circular laminate (which coincides with the pole of the dome), and the X and Y-directions correspond to the tape directions of the fabric. 3.3.3. Deformation (drawing) of PP tapes during forming of all-PP composite domes Fig. 16a and b show the X and Y strain profile from the origin along the X-direction (i.e., 0° direction of the fabric) for the compressive (inner) surface of (a) the 9 layer 4-point clamped dome and (b) the 9 layer 12-point clamped dome, respectively. Therefore both X strain and Y strain describe strain at points along the same X-axis, but X strain describes the strain in the direction of the X axis, while Y strain describe the strain normal to the X-axis. Both of these all-PP composites possess a balanced biaxial compressive strain state close to the centre of the laminate where the strain Y remains constant at 2%. This is particularly true for Fig. 16a. The strain X of the 12-point clamped specimen increases up to 10%, so in this case, the tape was drawn. This increase in the strain away from the pole is probably due to a gradient of both temperature and friction as the centre of the laminate is the first material to be in contact with the punch. The wavy shape of the strain X profile in Fig. 16b is not simply due to scatter since it is a periodic pattern whose period is equal to twice the width of the tape (i.e., the plain-weave unit cell length). The high peaks correspond to
Fig. 17. Compressive strain in the vicinity of the dome pole as a function of sheet thickness. The compressive strain of the all-PP dome increases linearly with its thickness as predicted by the beam theory, while the bending stresses of the GF/PP sheet are relieved by interply shearing.
1464
N.O. Cabrera et al. / Composites: Part A 39 (2008) 1455–1466
the tape oriented along X parallel to the forming forces and the low peaks to the tape oriented along the Y-direction. If a plate of 1.25 mm thick (equivalent to an all-PP composite laminate composed of 9 plies of fabric) is bent around a 28 mm radius, a compressive strain of 2.2% is reached on the inner surface. This is close to the strain Y value for Fig. 16a and b. In order to verify this hypothesis, all-PP composites laminates were also stamped with thicknesses of 0.43 mm (equivalent to an all-PP composite composed of 3 plies of fabric) and 1.7 mm (equivalent to an all-PP composite composed of 12 plies of fabric). These were clamped by 4 screws, which were tightened with a torque of 2 Nm. The major strain value in the region of the pole on the inner surface was derived from the strain mapping. Fig. 17 is the compressive major strain of the pole as a function of the sheet thickness. The correlation between the experimental values and the expected bending compressive strain means that the plates deform by bending in the pole region. In the absence of any visible wrinkling or buckling
of the fabric surface, it seems likely that this compressive strain is associated with tape shrinkage on the compressive (inner) face of the dome. This may result in a reduction in stiffness of the all-PP composite local to the compressive face of the hemisphere of the dome, since even such very small tape shrinkages have previously been seen to be associated with a decrease in stiffness of the all-PP tapes [17]. The GF/PP composites do not have significant compressive strains in that region as bending stresses are relieved by matrix squeeze flow and interply slipping due to the matrix rich regions. Although the strain could not be determined on the outer surface of the dome, it can be assumed that the X, Y strain profile along the X-direction of the outer surface is a translation by +4% strain of the profiles from Fig. 16a and b. Indeed the gap between the tapes (Fig. 14) on the outer surface was greater for the 12-point clamping. The tape was then drawn at very high strain rates (14%/ 0.76 s = 18%/s). This strain rate of 0.18 s1 is in the range of the
Fig. 18. Major strain direction distribution for (a) all-PP partially (4-point) clamped, (b) fully (12-point) clamped and (c) GF/PP partially clamped. The main mode of deformation of these domes is intraply shearing corresponding to a major strain direction of 45°.
N.O. Cabrera et al. / Composites: Part A 39 (2008) 1455–1466
strain rate used in the tape tensile test shown in Fig. 7. The increased distance between the tapes modifies the unit cell length and a consequence of this phenomenon is undoubtedly a change of the intraply shearing behaviour. Appropriate clamping of the disk during stamping allows drawing of the tapes that are oriented along the meridians of the dome. No interply shearing along the same directions was observed as the drawability of the tapes relieves the bending stresses. Although the plates were composed of 9 layers of fabric, they behaved in bending as a single layer. The significance of this tape drawing to the total deformation will now be discussed. 3.3.4. PP tape drawing vs. Intraply shearing during forming of all-PP composites The Aramis software used to map strain identifies both the major strain value as well as the major strain direction. The angle of the major strain direction with the X-axis varies between 0° and 180°. As the strain distribution of the dome is plotted on the undeformed laminate disk and both the X and Y-axes are parallel to one of the tape direction, it is possible to derive the mode of deformation of a particular point from its major strain orientation. If the major strain angle is 45° or 135°, the plate was deformed by intraply shearing, but if the angle is close to 0°, 90° or 180° then the laminate was deformed by drawing along the main fabric directions. For obvious reasons of symmetry, the 90° to 180° angles will be plotted between 0° and 90°. In order to achieve the same num-
Fig. 19. Example of moulding a corrugated part. The high friction forces (arrows) prevent interply slip and lead to fibre breakage if the textile laminate is based on completely non-extensible fibres/tapes.
1465
ber of points in all directions, all of the mesh points inside a circular of 22 mm radius were considered. The bar graph shown in Fig. 18a is a representation of the distribution of the major direction angle for the 9 layers partially (4-point) clamped dome. The cross symbol graph is the average major strain for each interval and more than 3000 data points were used to plot this graph. 55% of these points had a major direction angle between 35° and 55° while only 6% had an angle above 80° or below 10°. This shows that not only does a very small area of the laminate disk deform by drawing, but also the average strain of that area is nearly zero. The same graph is plotted for the 9 ply fully (12-point) clamped dome, and this is shown in Fig. 18b. Fewer points (46%) have a major direction angle between 35° and 55° while slightly more (9%) have an angle between 80° and 90° or between 0° and 10°. The average strain for the fully (12-point) clamped laminate is higher along the tape direction than the partially (4-point) clamped case in Fig. 18a, but the shift of the average major strain of the points deformed by intraply shearing is even higher. So whether the fabric is fully or partially clamped, the main deformation mode of all-PP composites remains intraply shearing of the fabric. Fig. 18c gives the major direction angle of the GF/PP 4-point clamped dome. The distribution is very similar to that in Fig. 18a. Although the glass fibre bundles cannot be drawn, the average major strain is slightly higher than the 4-point clamped all-PP plate. The average major strain of the all-PP dome along the tape direction is lower than the average major strain of the GF/PP dome because the centre of the dome has a bending compressive strain of 2%. It would probably be higher if the strain on the outer surface was used for strain mapping instead. The results from the fully clamped GF/PP dome could not be processed because the matrix squeeze flow destroyed the random stochastic pattern of the surface which is necessary for optical strain mapping. The angle distributions of Fig. 18 follow a Gaussian distribution. In summary, the fully clamped all-PP composite plate deforms similarly to a plate clamped only at 45° with the tapes direction but requires a higher forming energy. This is clearly disadvantageous in terms of investments (requiring bigger presses) and quality of the part (resulting in higher residual stresses). Clamping should therefore be optimised in terms of energy and product quality. However, tape drawing can be avoided here because a dome is a simple geometry, but this mode of deformation may give all-PP an advantage over glass woven fabric thermoplastics composites because complex shapes with inserts are now possible. Fig. 19 illustrates the moulding of a corrugated profile where
Fig. 20. Car geometry formed from all-PP woven tape fabric showing the feasibility of forming complex geometries from all-PP composites (Courtesy of Lankhorst Indutech bv, Netherlands).
1466
N.O. Cabrera et al. / Composites: Part A 39 (2008) 1455–1466
excessive friction forces are generated [6,40]. Fabrics based on non-stretchable fibres may exhibit fibre breakage, whereas this would be prevented if the fibre could be stretched. Although this paper has focussed on non-isothermal stamp forming of simple dome geometries, more difficult geometries have also been explored to determine the limits of other, more complex forming processes. To illustrate one example of what is possible from all-PP composites, Fig. 20 shows a photo of a car geometry produced from the woven all-PP fabric (courtesy of Lankhorst Indutech bv, Netherlands). This shows that it is possible to manufacture complicated geometries with high radii of curvatures from these materials. However, the complicated geometry makes it difficult to determine the deformation mechanisms which may occur in the woven fabric during more complex forming processes, and the complicated geometry also means that it is difficult to measure the mechanical properties retained in the finished product. 4. Conclusions It has been shown possible to create (albeit simple) geometries by non-isothermal stamp forming of all-PP composite laminates. Although the hemisphere geometry is a simple one, it is a good model to prove the feasibility of stamping all-PP composites. AllPP woven fabric laminates based on flat tapes deform differently compared to GF/PP when forming forces are applied along the reinforcement directions. In spite of the high strain rates experienced in stamping, the PP tape can still be drawn. No interply shearing is observed and the laminate behaves as a single layer under bending stresses. This was shown by the presence of compressive strains close to the pole of the dome, which were dependent on the laminate thickness. Due to limited deformations in that region, the laminate has a balanced biaxial strain state in compression on the inner surface and in tension on the outer surface. The material behaves similarly to an isotropic material (balanced biaxial stretching). However, the material becomes highly anisotropic away from the pole of the dome. Here, intraply shearing of the fabric is the main mode of deformation. 3D strain mapping experiments showed that tape drawing mechanism was negligible compared to intraply shearing. This was also observed if drawing was favoured by an appropriate clamping although this required a higher forming energy. The overall deformation behaviour of the all-PP laminates was therefore similar to woven glass fibres reinforced PP as intraply shearing is still the main deformation mechanism for all-PP laminates. A lower energy was required to deform the all-PP laminate compared to the GF/PP laminate considered here. The intraply shearing behaviour of different weaving styles of all-PP fabrics revealed that the plain weave exhibited the most homogeneous and stable deformation. Unlike the other styles, it does not exhibit any in-plane wrinkling or out-of-plane buckling of the tape. This result is in contrast to textile composites based on woven glass yarns or rovings where a satin weave is usually recommended when stamping complex parts. The difference is due to the wide and flat geometry of the tape. Since all-PP composites are entirely thermoplastic, special attention must be paid to exposing these materials to elevated temperatures, and elevated temperatures are necessary in non-isothermal stamping. A feature of all-PP composites is that the already highly drawn constituent PP tape can be further drawn up to 50– 100% strain at 140 °C depending on the strain rate. Generally, tape drawing should be avoided since it requires high energy, and may
result in residual stresses in the final part. However, some tape drawing may prove essential when complex final geometries are desired. References [1] Taylor B. Formability testing of sheet metals. In: Metals handbook. Forming and forging, vol. 14. American Society for Metals; 1988. p. 877–99. [2] Marissen R, Hornman H, Wenmakers L, Geleen M, Robroek L. Processing of continuous fibre reinforced thermoplastic composites for multi-purpose engineering. In: Proceedings of the Verbundwerk; 1991. [3] Mallon P, O’Brádaigh C. Compliant mold techniques for thermoplastic composites. In: Kelly A, Zweben C, editors. Comprehensive composite materials; 2000. [4] Friedrich K, Hou M, Krebs J. Thermoforming of continuous fibre/thermoplastic composite sheets. In: Composite sheet forming, composite materials series II, Amsterdam; 1997, p. 91–162. [5] Bigg DM, Preston J. Polym Compos 1989;10(4):261–8. [6] Robroek L. Stamping of thermoplastic matrix composites. PhD thesis. Delft University of Technology; 1994. [7] Cogswell F. Int Polym Process 1987;1(4):157–65. [8] Cutolo D, Savadori A. Polym Adv Technol 1994;5(9):545–53. [9] Leterrier Y. Life cycle engineering of composites. In: Comprehensive composite materials. Polymer matrix composites, vol. 2. Amsterdam: Elsevier; 2000. [10] 2000/53/EC. Directive 2000/53/EC of European Parliament of the Council of 18 September 2000 on End-of-Life Vehicles; 2000. [11] Hine PJ, Ward IM, Olley RH, Bassett DC. J Mater Sci 1993;28:316–24. [12] Rasburn J, Hine PJ, Ward IM, Olley RH, Bassett DC, Kabeel MA. J Mater Sci 1995;30:615–22. [13] Yan RJ, Hine PJ, Ward IM, Olley RH, Bassett DC. J Mater Sci 1997;32(18):4821–32. [14] Jordan ND, Olley RH, Bassett DC, Hine PJ, Ward IM. Polymer 2002;43:3397–404. [15] Hine PJ, Bonner M, Brew B, Ward IM. Plast Rub Compos Process Appl 1998;27(4):167–71. [16] Peijs T. Compos Recycl Mater Today 2003:30–5. [17] Alcock B. Single polymer composites based on polypropylene: processing and properties. PhD thesis. UK: Queen Mary, University of London; 2004. [18] Cabrera N. Recyclable all-polypropylene composites: concept, properties and manufacturing. PhD thesis. Netherlands: Technische Universiteit Eindhoven; 2004. [19] Alcock B, Cabrera NO, Barkoula NM, Loos J, Peijs T. Compos Part A: Appl Sci Manufact 2006;37(5):716–26. [20] Alcock B, Cabrera NO, Barkoula NM, Peijs T. Compos Sci Technol 2006;66(11– 12):1724–37. [21] Alcock B, Cabrera NO, Barkoula NM, Spoelstra AB, Loos J, Peijs T. Compos Part A: Appl Sci Manufact 2007;38(1):147–61. [22] Alcock B, Cabrera NO, Barkoula NM, Loos J, Peijs T. J Appl Polym Sci 2007;104(1):118–29. [23] Barkoula NM, Alcock B, Cabrera N, Peijs T. Polym Polym Compos 2008;16(2):101–13. [24] Alcock B, Cabrera NO, Barkoula NM, Wang Z, Peijs T. Compos Part B 2008;39(3):537–47. [25] McDonnell P, McGarvey K, Rochford L, Ó Brádaigh C. Compos Part A: Appl Sci Manufact 2001;32(7):925–32. [26] Wakeman MD, Zingraff L, Blanchard P, Bourban PE, Månson JAE. Compos Sci Technol 2006;66:19–35. [27] Hou M. Compos Part A: Appl Sci Manufact 1997;28(8):695–702. [28] Nowacki J, Fujiwara P, Mitschang P, Neitzel M. Polym Polym Compos 1998;6(4):215–22. [29] Tyson J, Schmidt T, Galanulis K. Advanced photogrammery for robust deformation and strain measurement. In: Proceedings of SEM 2002 annual conference; 2002. [30] Gom,
; 2005. [31] Cabrera NO, Alcock B, Klompen ETJ, Peijs T. Appl Compos Mater 2008;15(1):27–45. [32] Martin TA, Bhattacharyya D, Pipes RB. Compos Manufact 1992;3(3):165–72. [33] Martin TA, Christie GR, Bhattacharyya D. Grid strain analysis and its application in composite sheet forming (Chapter 6). In: Bhattacharyya D, editor. Composite sheet forming. Elsevier Science B.V.; 1997. p. 217–45. [34] Alcock B, Cabrera NO, Barkoula NM, Reynolds CT, Govaert LE, Peijs T. Compos Sci Technol 2007;67(10):2061–70. [35] Johnson A, Costalas G. In: 4th International conference on automated composites, vol. 4; 1995, p. 341–52. [36] Hofstee J, de Boer H, van Keulen F. Compos Sci Technol 2000;60:1041–53. [37] Barkoula NM, Schimanski T, Loos J, Peijs T. Polym Compos 2004;26(1):114–20. [38] Wakeman MD, Zingraff L, Kohler, M, Bourban PE, Månson JAE. In: Tenth European conference on composite materials (ECCM10); 2002. [39] Prosser W, Hine PJ, Ward IM. Plast Rub Compos 2000;29(8):401–10. [40] Breuer U, Neitzel M. Polym Polym Compos 1996;4(2):117–22.