You are on page 1of 10

COMPOSITES SCIENCE AND TECHNOLOGY

Composites Science and Technology 67 (2007) 252261 www.elsevier.com/locate/compscitech

An experimental study of in-plane large shear deformation of woven fabric composite


B. Zhu a, T.X. Yu
a

a,*

, X.M. Tao

Department of Mechanical Engineering, Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong b Institute of Textiles and Clothing, Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong Received 10 June 2006; accepted 2 August 2006 Available online 4 October 2006

Abstract Textile composites used as structural components have attracted more and more attention due to their superior properties and ecient stamping operations. However, formability of textile composite sheets, restricted by failure mechanisms such as wrinkling, remains as a crucial and challenging issue. Although in stamping of real parts, wrinkling depends on many factors, this paper mainly focuses on the in-plane characterization of pure shear deformation and its contribution to wrinkling. A comprehensive experimental study on woven fabric composites is presented. Modied picture frame tests were conducted up to wrinkling. The eect of test conditions was investigated, and the reduction of yarn width was found to be a key for wrinkling. The onset of wrinkling was determined by a densication method and by a laser scanning technique. Moreover, the cross-sectional proles of fabric samples during the test were traced, helping to build up a theoretical model of the composite sheets during their large shear deformation. 2006 Elsevier Ltd. All rights reserved.
Keywords: A. Textile composites; B. Non-linear behavior; C. Buckling; Picture frame test

1. Introduction Compared with traditional metals and laminated composites, textile composites have many advantages due to their high specic stiness, high strength, low weight, nice integral performance, low thermal expansion and good corrosion resistance. Most important is that, textile composites are more exible than metals and possess a high capacity to conform to complicated contours; therefore, they are particularly suitable for manufacturing components with complex shape. Reductions in costs of material forming can be achieved through highly ecient stamping operations [1,2]. Stamping, which deforms a at sheet into a particular shape in a relatively high processing temperature with a pair of punch and die, is a very cheap process with a cycle of only seconds regardless of the size of the parts.
*

Corresponding author. Tel.: +852 2358 8652; fax: +852 2358 1543. E-mail address: metxyu@ust.hk (T.X. Yu).

However, formability of textile composite sheet, restricted by corresponding failures especially wrinkling [3], is a crucial and challenging problem for stamping operations. In-plane characterization experiments have conrmed that wrinkling of the woven fabric will occur when the critical shear angle, named the locking angle, between the warp and weft yarns is reached. These wrinkles have potential to induce numerous processing and strength problems [4]. In the real stamping operations, due to the compression, compaction and friction, wrinkles usually occur well before the locking angle is reached; however, the shear still predominates the deformation mode of the material as well as wrinkling, so a comprehensive study of the in-plane shear characteristic will benet the further understanding of the formability of textile composite sheets under stamping. Most of textile composites, currently used in structural components that have to carry substantial working loads, are woven composite sheets made from glass ber or carbon ber. Because of this, woven fabric reinforced composites have been more widely investigated in recent years.

0266-3538/$ - see front matter 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.compscitech.2006.08.011

B. Zhu et al. / Composites Science and Technology 67 (2007) 252261

253

2. Previous studies The mechanical properties and deformation mechanisms of woven fabric composite have been experimentally investigated and generally understood by a number of researchers through pure shear test approaches [5]. The picture frame test (Fig. 1) is one of the fundamental methods to characterize the in-plane shear behaviour of textile composites [6,7]. A square textile fabric is clamped into a xture with four arms connected by four hinges. Since the arms can rotate freely at the hinges, the angle between two groups of yarns varies, thus resulting in a pure shear deformation of the fabric under an external diagonal tensile load. To normalize test data from dierent picture frame setups, Peng et al. [8] proposed a general formula through an energy approach, which states that at the same shear angle, the consumed external energy in the same area of deformed region is the same and equals to the work done by the external load: P L2 =Lframe fabric p l2 =lframe fabric 1

and transversely incompressible, free to rotate, but not to translate. It is proposed that the minimum possible value for the locking angle occurs when the gap between yarns reaches zero. According to a purely geometrical relation, the locking angle, hc, is given by w 2 hc arcsin w g0 where w is the yarn width, and g0 is the initial gap width between yarns. In order to evaluate the formability of various textile fabric preforms, Yu et al. [16] conducted characterization experiments. The conclusion shows that loading rate and processing time are signicant for composites during stamping, which was also studied by Wu et al. [17]. In the picture frame test, corner cut-o is typically needed to allow rotation of the hinges and prevent immediate wrinkling [18]. In order to eliminate the contribution from the marginal parts, yarns, which are not clamped into the xture in each part between all the crossovers, are removed. However, when shear deformation becomes signicantly large, the fabric composite may still interfere with the xture, as seen in Fig. 2. The two joints of the xture press the middle portion of the test sample, imposing an undesired lateral compressive load to the specimen and accelerating the onset of wrinkling. In this paper, a modied xture was designed, so that the lateral pressure to specimen can be avoided, leading to a more reliable and accurate characterization of material. To investigate the size eect, dierent samples were tested and the relevant inuence is analysed. It has been shown that the pin-joint model does not truly represent the shear behavior of woven fabrics, and lateral compression to yarns, which is relevant to the boundary clamping condition in the picture frame test, aects the onset and evolution of wrinkles. These factors should be considered so as to establish a more precise wrinkling criterion.

where P and p are diagonal tensile forces, respectively; L and l are referred to Fig. 1. It should be noted that Eq. (1) is only valid under the same loading rate. In the case of rate-dependent shear behaviour, a more general result was reported by McGuinness and Bradaigh [9]. Lussier and Chen [10] conducted a series of pure shear characterization experiments on plain woven composites. They analyzed the detailed stages that may occur during the large shear deformation, and made corresponding physical explanations. The inuence of temperature in the range from 20 to 190 C on shear behaviour was also studied. The distribution of temperature on the sample during stamping was measured using a FLIR Systems infrared camera. The tensile and shear properties were likewise studied by other researchers [11,12]. In-plane shear deformation is limited by local wrinkling, when yarns reach the so-called locking angle. This locking angle has been experimentally measured and analytically modelled by many researchers [1315]. Prodromou and Chen [13] used a pin-joint model, namely, yarns are pinned together at points of intersection or joints, while the yarns are assumed to be longitudinally inextensible

Fig. 1. Set-up of picture frame test.

Fig. 2. Compression between xture and fabric in a picture frame test.

254

B. Zhu et al. / Composites Science and Technology 67 (2007) 252261

3. Modied picture frame test 3.1. Composite test sample

sample was still made, and yarns not clamped into the xture were removed. 3.2. Loading xture

Plain-weave textile fabrics comprise continuous E-glass laments as the reinforcement and thermoplastic polypropylene laments as the matrix. The volume fraction of the glass ber is about 0.35. The fabric is made of two groups of yarns initially orthogonal to each other. The glass and polypropylene laments are commingled together in the yarns at room temperature; therefore, the fabric itself is a composite material. The material properties of glass laments and polypropylene are given in Table 1. The average geometric parameters of the specimen are shown in Table 2. Fig. 3 represents its structure and appearance. To minimize the marginal restriction and facilitate the clamping of fabric, moderate corner cut-o on the square
Table 1 Properties of glass bers and polypropylene Property Polypropylene Glass type E Diameter (lm) Density (kg/m3) Tensile modulus (GPa) Tensile strength (MPa) Elongation (%) Coecients of thermal expansion (106/C) Thermal conductivity (W/m/C) Specic heat (J/kg/K) Glass transition temperature (C) 900 11.4 2538 300 110 0.2 20 to 5 814 2540 72.4 3450 1.83.2 5.0 1.3 840 C 2490 68.9 3160 4.8 7.2 S 10 2490 85.5 4590 5.7 5.6

780

940

A modied xture was designed to implement in-plane shear test as shown in Fig. 4. Most features remain the same as before except the two middle joints, deviated from the plane where the fabric is clamped, so that there will be no contact or no pressure imposed on to the fabric sample during its large shear deformation. As mentioned before, the middle joints of the old xture are in the same plane of the fabric; while in the modied one, there is a gap between the joints and the arms which avoids the contact of screws and sample, thus the xture can be sheared up to a collinear state. The improvement in the result of the modied xture will be discussed in Section 4.2. A problem caused is the moment to the xture under non-collinear forces and the increase of friction within the xture. However, these eects may be neglected if the deviation of joints is relatively small compared with the whole xture whilst the internal friction can be reduced by lubrication. Another eective way is to subtract empty xture loading from that with clamped sample, excluding the inuence on the measured materials behaviour due to the xture itself. The tensile machine used was Universal Testing Machine. The specication follows MTS SINTECH 10/D Frame. During the operation, displacement was controlled and the adopted crosshead speed varied from 10 mm/min to 500 mm/min. The load cell was 1 kN, and the tolerance is 0.005%. Force values were instantaneously recorded by TESTWORKS 4. 3.3. Test conditions

Table 2 Geometry of fabric and specimen Property Fabric Specimen size Large Width of yarn (mm) Gap between yarns (mm) Maximum fabric thickness (mm) O-angle Yarn number Side length (mm) 4.34 0.39 0.65 0.38 1.34 0 27 27 140 140 0 17 17 88 88 Small

Dierent loading speeds were adopted as 10, 100, 200 and 500 mm/min. 3.4. Test results Based on the deformed conguration of the xture, shear angle of the arms is calculated by the following formulae:

Fig. 3. Balanced plain woven fabric.

B. Zhu et al. / Composites Science and Technology 67 (2007) 252261

255

Fig. 4. Modied xture of picture frame test: (a) the front view and (b) the side view.

cph 2 p 2Lframe d cos h 2Lframe 2

where Lframe is the side length of the picture frame (Lframe = 180 mm for the frame used in our tests); d is the displacement of the crosshead; h and c are the current angle and shear angle of the arms, respectively. Fig. 5 represents the results of the picture frame tests at room temperature (20 C) under dierent loading speeds. The overall deformation stiness increases with the loading speed. In addition, photos were taken during the test, as shown in Fig. 6. It is obvious that during the large shear deformation, the yarn width decreases under lateral compression, which oers more space for the material to be sheared before wrinkling. To correlate the shear angle with the occurrence of wrinkling, the values of shear angle calculated from Eq. (3) are compared with the measured ones from the large and small fabrics, as shown in Fig. 7. The shear angle of the fabric does not follow Eq. (3) at large deformation. Instead, it gradually levels and will no longer increase after wrinkling. Further shear of the xture only increases the wrinkling. During small deformation, the shear angle of the fabric almost remains the same as that of the xture. The maximum deviation in this stage is about 9.3%, which is caused by the errors in clamping and measurement. When the xtures displacement reaches about 90 mm, the increasing
600 500 400

Fig. 6. Photos taken during picture frame test.

rate of the measured shear angle decreases compared with the calculated one. Therefore, it may be estimated that wrinkling occurs between a xtures displacement of 80 mm and 100 mm. In order to observe the wrinkling more distinctly, a normalized load vs. measured fabric shear angle curve is plotted in Fig. 8 according to Eq. (1). Initially, the shear stiness remains small. Later, it increases rapidly after a
80

10mm/min 200mm/min

100mm/min 500mm/min

Shear angle (deg.)

70 60 50 40 30 20 10 0

calculated measured (large) measured (small)

Load (N)

300 200 100 0 0 -100 10 20 30 40 50 60 70 80 90 100 110

10

20

30

40

50

60

70

80

90

100 110

Displacement (mm)

Displacement (mm)
Fig. 7. Comparison between calculated and measured shear angles.

Fig. 5. Picture frame test results at 20 C under dierent loading speeds.

256
600 500 400 300 200 100 0 0 5 10 15 20 25 30

B. Zhu et al. / Composites Science and Technology 67 (2007) 252261

Normalized load (N)

large sample small sample

35

40

45

50

55

60

65

Measured shear angle (deg.)


Fig. 8. Normalized picture frame test results at 20 C under 10 mm/min.

certain shear angle, and such a turning point can be regarded as the locking shear angle as well as the onset of wrinkling. Another conclusion can be drawn from Figs. 7 and 8; that is, the test samples of dierent sizes have almost identical normalized load before wrinkling, which veries Eq. (1), and the locking shear angles of two samples are very close to each other. 4. Analysis of large shear deformation 4.1. Three stages during large shear deformation At room temperature, measurement was taken to nd the relationship between the yarn width and shear angle. In Fig. 9, the calculated curve pertains to the rotation of yarns without change of the clamping side length in the xture, i.e. w w0 cos c
0

Fig. 10. Photo taken at c = 32.

Fig. 11. Photo taken at c = 50.

4 For the current initial weave density of fabric, the stages can be distinguished as follows: 1. Stage I (0 < c 0 < 50): Two groups of yarns rotate with gaps between yarns; 2. Stage II (50 < c 0 < 60): Yarns rotate without gaps under larger lateral pressure; 3. Stage III (c 0 % 60): Wrinkling due to the exhaustion of yarns compressibility.

where w0 is the initial yarn width and c 0 is the measured shear angle of the fabric. Fig. 9 shows that in the rst stage when c 0 < 50, the variation in the yarn width may be slightly dierent from the prediction of the simple theory. In this stage, the lateral pressure to the yarns is small and there are still gaps between them (Fig. 10). When the shear angle reaches about 50, the yarn width decreases more quickly, which stands for the vanishing of gaps and a much larger lateral pressure to the yarn (Fig. 11). Further shear results in local wrinkling on the fabric whilst the yarn width keeps as a constant. Therefore, three typical stages can be identied during the large shear deformation of the woven composite sheets.
5

4.2. Onset of wrinkling Based on the physical mechanism illustrated above, the previously calculated and measured shear angle curves only provide an approximate range for the onset of wrinkling. In order to directly observe the wrinkling, the surface prole of the sample during the picture frame test was measured using Reversa Laser Scanner (Fig. 12). At a certain shear angle, the joints of xture were tightened by screws, so that the shear deformation on the fabric could be frozen. After the scanner went through a 2-D horizontal region, the height values on the surface points of the sample were measured and recorded, thus a 3-D view of the sample surface could be displayed. The scanning region was 50 50 mm2; the step length was 0.1 mm; and the accuracy of measured height was 0.02 mm.

Yarn width (mm)

4 3 2 1 0 0 5

calculated
10 15 20

measured
25 30 35 40 45 50 55 60 65

Shear angle (deg.)


Fig. 9. Variation of yarn width during shear deformation.

B. Zhu et al. / Composites Science and Technology 67 (2007) 252261

257

Fig. 12. Laser scanning technique: (a) equipment and (b) scanning.

Fig. 13. Surface prole of sample (a) c 0 = 0; (b) c 0 % 20; (c) c 0 % 40; (d) c 0 % 55; (e) c 0 % 58 and (f) c 0 % 59.

Fig. 13 depicts the scanning results at discrete shear angles in the test in a 3-D view. The onset and aggravation of local wrinkling was observed. Fig. 14 shows the variation of the surface prole from the horizontal direction. It is seen more distinctly that wrinkling occurs when the shear angle of fabric reaches about 58, which agrees with the result

observed from shear angle curves. This scanning method can not only detect wrinkling, but also measure the shape of yarn surface before wrinkling, as well as the yarn width and gaps between them. Those data are valuable for establishing a theoretical model under large shear deformation. An alternative method to determine the exact onset of wrinkling is to nd a densication point from the experimental loaddisplacement curve. Similar to the densication phenomenon of cellular materials under compression, wrinkling of a woven fabric can be regarded as the result of an in-plane densication (i.e., a generalized densication), while both processes involve a sudden increase of the material stiness. Based on this analogy, the onset of wrinkling could be determined from the point of view of the eciency in the external energy consumption by using the following formula [19]: R s  F ds d 0F df s 0 ) sc 5 ds ds where F and s are the diagonal tensile force andRthe corres sponding displacement, respectively; and f s 0 F is the consumed external energy eciency with respect to current force. Determined by Eq. (5), sc is the critical displacement, at which wrinkling occurs. The integration can be done
F ds

Fig. 14. View of surface prole from horizontal direction (a) c 0 % 55; (b) c 0 % 58 and (c) c 0 % 59.

258

B. Zhu et al. / Composites Science and Technology 67 (2007) 252261

numerically based on the experimental data of F vs. s. As a matter of fact, two extremums for the function f(s) would exist for the whole Fs curve. However, the rst extremum is relevant to the gap-vanishing phenomenon, which also results in a sudden increase of the shear stiness; while the second one appears at the locking angle when wrinkling of the sample takes place. To determine the second one, when the displacement reached the rst extremum, the corresponding force and displacement are purposely re-set as zero. This will eliminate the gap-vanishing eect and re-focus on the wrinkling behaviour in the later stage. Thus, based on the concept of in-plane densication, the onset of wrinkling is shown in Fig. 15. Results indicate that wrinkling occurs at a critical displacement of 93 mm for a large test sample under a loading speed of 10 mm/min; at 95 mm for a large sample under 500 mm/min; and at 92 mm for a small sample under 10 mm/min. This again concludes that loading speed and size eect do not signicantly aect the onset of wrinkling. From the critical displacement obtained, it is found that the locking shear angle of the fabric at the onset of wrinkling is about 58. In the old xture test, wrinkling occurs at about 50 of the shear angle. Considering 16% larger of the shear deformation, the modied xture provides a much better result. On the other hand, compared with 36 for the current sample based on Eq. (2), the actual onset of wrinkling is much later in the picture frame test. This discrepancy is attributed to the fact that yarn width reduces during shear deformation, and the onset of wrinkling is greatly delayed as a result of the lateral compressibility of yarn. To establish a more precise critical condition for wrinkling, the reduction in the yarn width, together with the variation of the gap width, will be analysed in Section 4.3. 4.3. Yarn width and gap width It is noted that the variation of yarn width is closely related to the boundary condition in the picture frame test. When the fabric sample is xed on the xture, the clamped part cannot deform, while the central overlapped part, as well as the free yarns, will experience shear deformation. Therefore, there is an angular discontinuity of bers along the boundary of xture, especially when the shear is large.

Another boundary eect is that during the shear deformation, bers in a single yarn will slip among each other; however, such slip is restricted by the friction between the two groups of yarns in the overlapped region, then such a slip delay will bend the relevant bers in the free yarns. These two factors not only cause the phenomenon as seen in Fig. 16, but also aect the yarn width. Fig. 17 sketches how the yarn width is related to the boundary bending eect, where w0 is the initial yarn width that remains constant inside the clamping arms. If the marginal bers kept straight during the test, the marginal free yarn section would possess a shape of parallelogram ABCD. The relative slip distance between bers AD and BC would be D, while the yarn width would be w = w0 sinh. In fact, however, point C can only take a new position at C 0 with a smaller slip distance D 0 , hence the actual yarn width

Fig. 16. Boundary bending eect in picture frame test.

Fig. 15. Determination of onset of wrinkling by densication method.

Fig. 17. Conguration of boundary bending and yarn width.

B. Zhu et al. / Composites Science and Technology 67 (2007) 252261

259

is w 0 and BC must be bent. Consequently, there will also be a little inconsistency between the shear angles of xture and fabric. Since the shear deformation of fabric is delayed in comparison with the situation of straight marginal bers, from an energy point of view, the delayed energy transfers to the bending of the free bers. Therefore, it is possible to investigate the delay by equating the bending energy of the bers with the delaying shear energy. To estimate the bending energy, the following assumptions are made: (i) the bers are regarded as elastic material with constant Youngs modulus, E, and cross-sectional second moment, I [20]; (ii) the shape of bent ber BC is sinusoidal as described by y a l sin px, where 2l is the 2l projection length of ber on x axis; a is a non-dimensional parameter; and OXY is the local coordinate; and (iii) CD and C0 D are of the same length. Hence, the bending energy of a single ber is Z n=2 2 M t dt 6 Eb 2 2EI 0 where n is the total ber length, and M(t) is the bending moment along the ber: Mx EI
ap2 4l

Finally, according to energy conservation, considering F = dE/ds, we have Eds s Eds s Ds Eb a ) F s f s; a dEb a da da ds 12

b a where F(s), f(s,a) are known and dEda dEb EI . By solving l da the dierential Eq. (12), a(s) as well as w 0 (c) can be obtained. It is quite hard to accurately determine the material property of yarn, EI, and the projection length of bent ber, l; however, if it is assumed that the gaps between yarns vanish at c = 50, i.e., s % 80 mm, which can be obtained from the experimental measurement, the value of EI can be calcul lated from Eb s 80 mm=Eb s 80 mm. Therefore, the delayed crosshead displacement in the picture frame test can be determined by the current crosshead displacement, and the result is expressed approximately as a piecewise tted function: & 0 s < 31 Dss % 13 3:08 lns 30 0:37 s P 31

sin px 2l

1 a 4p cos2

2 2

px 2l

3 2

Yarn width (mm)

Therefore, the bending energy can be expressed as a function of a: Z l 2 r M x a2 p2 px 1 dx cos2 Eb a 2 2EI 2l 4 0 2 4 Z l EI a p2 sin2 px 2l 16l 8 5=2 dx 2 2 0 1 a 4p cos2 px 2l Eq. (8) can be rewritten as a non-dimensional relation as: p4 Eb 16 Z
0 1

It indicates that before displacement reaches 30 mm, the delay in shear angle is indistinct, even negligible, which agrees with our observation. From the delayed displacement function obtained above, the dependences of the yarn width and gap width on the shear angle of the fabric are presented in Figs. 18 and 19, respectively. Good agreement is achieved between the predicted and measured results.
5 4 3 2 1 0 0 10 20 30 40 50 60 70

2 2 1 a 4p cos2

a2 sin2

pt
2

dt pt5=2
2

theoretical measured_2

measured_1 measured_3

where non-dimensional parameter Eb Eb l. Eq. (9) may be EI rewritten in a polynomial expression after curve tting: Eb a % 0:086a4 0:212a3 0:178a2 0:014a 10

Shear angle (deg.)


Fig. 18. Prediction of yarn width during picture frame test.

Gap width (mm)

On the other hand, at a certain deformed state, following relations hold: 8 0 > 2a w w > > Dh > arcsin w0 h <2 w0 p p 11 > Dc arccos sDs180 2 arccos s180 2 >2 360 360 > > : Dh Dc 2 2 where s is the crosshead displacement in the situation of straight marginal bers; Dc is the delayed shear angle; and Ds is the delayed crosshead displacement related to Dc. Ds(s,a) can be numerically derived based on the relationship above.

1.2 1 0.8 0.6 0.4 0.2 0 0 10 20 30 40 50 60

theoretical

measured

Shear angle (deg.)


Fig. 19. Prediction of gap width during picture frame test.

260

B. Zhu et al. / Composites Science and Technology 67 (2007) 252261


3.5

4.4. Yarn cross-section during large shear deformation Shear stiness and wrinkling of the sample are closely related to the interaction between the two groups of yarns, such as the lateral compression to yarns and rotary friction at the joint regions. Hence, in order to build up a detailed model for large shear and to understand the mechanism of wrinkling, a key issue is to know the variation of yarn cross-section during the test. To observe the cross-section of yarns, rst, before the shear test, one group of yarns in the fabric was dyed black to distinguish from the other group. Then, the whole fabric was solidied, using epoxy curing agent, at dierent shear stages in a picture frame test. Later, the sample was cut perpendicularly to the yarn direction. Finally, the crosssection of yarn was captured by a digital camera. As shown in Fig. 20, the prole of yarn cross-section remains an olive-shape during the whole shear deformation. The width reduces gradually under the lateral compression, as mentioned before, with a little increase in the thickness. The nominal yarn thickness and cross-sectional area measured from the photos are plotted in Fig. 21. When shear deformation is small, a notable increase in the yarn thickness occurs accompanied by a minor decrease in the cross-sectional area, which indicates a relatively small compression; at a large shear deformation, the yarn thickness remains constant whilst the cross-sectional area is much reduced, resulted from a large compression. Through the shear process, not only the gaps between yarns reduce, but the gaps inside a yarn, so-called inter-bre spacing, also shrink. The interesting thing is that in the rst stage, it is the outer gaps that mostly reduce; while in the second stage after the elimination of the outer gaps, the inter-bre gaps begin to shrink greatly. In this sense, the two kinds of gaps dominate the two shear stages, respectively. Moreover, Fig. 20 exhibits the change of the contact region between the upper and lower yarns, which aects the rotary friction at the joints of yarns and hence the shear stiness of the sample. As revealed in Fig. 20e, there is an angle between the lower boundaries of neighbouring yarns. In other words, the yarn width reduction is restricted by wrinkling, i.e., after the width has been reduced to a certain limit, the fabric cannot be further deformed within the original plane, but buckles out of the plane. Analysis of these two dierent deformation modes will lay a basis for a theoretical model of woven composite sheets under large shear deformation up to wrinkling.
Normalized value

yarn thinkness cross-sectional area

3 2.5 2 1.5 1 0 10 20 30

40

50

60

70

Shear angle (deg.)


Fig. 21. Normalized variation of yarn thickness and yarn cross-sectional area.

5. Concluding remarks Wrinkling acts as a predominate factor in the stamping operation of woven textile composites. The present study focuses on the experiment and characterization of in-plane large shear deformation on the composite fabric, and mainly analyse the large shear deformation close to the locking angle as well as its contribution to wrinkling. Experimental set-up, named picture frame test, was again employed but with a new modication, obtaining more accurate and reliable results. Geometrical parameters of the fabric during the large shear deformation, as well as the onset of wrinkling, were determined. Dierences between a simple model used in previous papers and our observations are identied, with the causes being analysed quantitatively. Through a semitheoretical simulation, the variation of the yarn width is modelled, which agrees well with the measured value. Based on the observation of yarn cross-section, the mechanism of wrinkling behaviour was further investigated. As for the further studies on the topic, a more precise theoretical model is now under construction based on the wrinkling mechanism described in the present paper. Meanwhile, the material behaviour under higher temperatures will be investigated. Besides, other factors related to wrinkling in the real stamping operations, such as composite-tool/mould friction, compression along yarns, normal pressure to the material, and so on, are under research to provide a more general understanding of the wrinkling phenomenon. Acknowledgements The work reported in this paper is a part of a Hong Kong RGC project HKUST6012/02E. The nancial

Fig. 20. Cross-section of yarn at dierent shear angles (a) c = 0; (b) c = 20; (c) c = 40; (d) c = 60 and (e) winkling.

B. Zhu et al. / Composites Science and Technology 67 (2007) 252261

261

support from Hong Kong Research Grant Council (RGC) is gratefully acknowledged. The authors would also like to acknowledge the Benchmark group led by Prof. J. Cao of Northwestern University, USA, for supplying composite fabric. References
[1] Cogswell FN. Thermoplastic Aromatic Polymer Composites. Butterworth-Heinemann; 1992. [2] Tucker CL. Forming of advanced composites. In: Advanced Composites Manufacturing. New York: Wiley; 1997. [3] Gutowski TG, Dillon G, Chey S, Li H. Laminate wrinkling scaling laws for ideal composites. Compos Manuf 1995;6:12334. [4] Arndt RD. Fabric preforming for structural reaction injection molding. In: Proceedings of conference on advanced composite materials: new developments and applications, 1991. p. 3540. [5] Long AC. Design and manufacture of textile composites. New York: Woodhead Publishing Limited & CRC Press LLC; 2005. [6] Canavan RA, McGuinness GB, Bradaigh COM. Experimental intraply shear testing of glass-fabric reinforced thermoplastic melts. In: Proceedings of the fourth international conference on automated composites, University of Nottingham, 1995. [7] McGuinness GB, Bradaigh CMO. Development of rheological models and picture-frame shear testing of fabric reinforced thermoplastic sheets. J Non-Newton Fluid Mech 1997;73:128. [8] Peng XQ, Xue P, Cao J, Lussier DS, Chen J. Normalization in picture frame tests of composite materials: length or area? In: Proceedings of the fth international ESAFORM conference on material forming, Krakow, Poland, 2002. p. 28790. [9] McGuinness GB, Bradaigh CMO. Characterization of thermoplastic composites melts in rhombus-shear: the picture-frame experiment. Compos Part A-Appl Sci Manuf 1998;29(A):11532.

[10] Lussier D, Chen J. Material characterization of woven fabrics for thermoforming of composites. In: Proceedings of the fteenth technical conference on American Society for Composites, College Station, TX, USA, 2000. p. 30110. [11] Boisse P. Experimental analysis and modeling of biaxial mechanical behavior of woven composite reinforcements. Exp Mech 2001;41(3): 2609. [12] Lussier D, Chen J. Shear frame standardization for stamping of thermoplastic woven fabric composites. In: Proceedings of Thirty-second international SAMPE technical conference, 2000. p. 15060. [13] Prodromou AG, Chen J. On the relationship between shear angle and wrinkling of textile composite preforms. Compos Part A-Appl Sci Manuf 1997;28:491503. [14] Rozant O, Bourban PE, Manson JAE. Drapability of dry textile fabrics for stampable thermoplastic preforms. Compos Part A-Appl Sci Manuf 2000;31:116777. [15] Long A, Robitaille F, Souter B, Rudd C. Permeability prediction for sheared, compacted textiles during liquid composite moulding. In: Proceedings of ICCM-13, the thirteenth international conference on composite materials, 2001. p. 636. [16] Yu JZ, Cai Z, Ko F. Formability of textile preforms for composite applications. Part 1: Characterization experiments. Compos Manuf 1994;5(2):11322. [17] Wu E, Chang LC. Loading rate eect on woven glass laminated plates by penetration force. J Compos Mater 1998;32:70221. [18] Nestor TA. Experimental investigation of the intraply shearing mechanism in thermoplastic composites sheet-forming. M.Eng.Sc. Thesis, Department of Mechanical Engineering, University College, Galway, 1995. [19] Tan PJ, Harrigan JJ, Reid SR. Inertia eects in uniaxial dynamic compression of a closed cell aluminium alloy foam. Mater Sci Technol 2002;18:4808. [20] Hearle JWS, Grosberg P, Backer S. Structural mechanics of bers, yarns, and fabrics, vol. 1. New York: Wiley Inc.; 1969.

You might also like